University of Groningen

Anastomotic leakageKarliczek, Anne

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2008

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Karliczek, A. (2008). Anastomotic leakage: risk assessment in gastrointestinal surgery. [S.n.].

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 24-04-2021

lfrnastomotic Risk assessment i

ea age Gastrointestinal surgery

Anne Karliczek

Anastomotic Leakage

Risk assessment in gostrointestinol surgery

Anne Karliczek

Cc-111r=-1le U !\kdjschc M Bihliorheck C Groningcn G

Stellingen

behorend bij het proefschrift

'Anastomotic leakage-Risk assessment in gastrointestinal surgery'

1. Prophylactic drainage of colonic anastomoses does not influence anastomotic leakage. (dit proefschrift)

2. Based on clinical intraoperative risk assessment the surgeon can't predict anastomotic leakage. (dit proefschrift)

3. Low tissue saturation in murine colon is a risk factor for anastomotic leakage from the third postoperative day and beyond. (dit proefschrift)

4. Visible light spectroscopy can adequately detect tissue ischemia in murine colon serosa. (dit proefschrift).

s. An abdominal wall hernia using an exteriorized loop of descending colon is a successful model to apply fractionated irradiation in rats (dit proefschrift)

6. Visible light spectroscopy shows stable measurements during oesophageal and colorectal resection in humans and is a feasible instrument to evaluate serosal (adventitial) tissue oxygenation. (dit proefsch rift)

7. Tissue oxygenation measured with visible light spectroscopy does not predict anastomotic leakage. (dit proefschrift)

8. Curiosity kills the rat.

9. Besides feeling gut, a gastrointestinal surgeon needs gut feeling.

lo. The surgeon must have the heart of a lion, the eyes of a hawk, and the hands of a woman. Uohn Halle, 1 529-1 586)

11. Some things separate boys from men. Often the men are far behind. (naar CDNAjaarverslag, Dutch Birding 2000; 22: 251-271)

12. Mountains are not stadiums where I practise my sport, they are cathedrals where I practise my religion (Anatoli Boukreev, 1958-1997)

C\..11t.·:,le u

lvk,.h·:h� M Bib]1�)1� :�k c

Groningcn G

Anne Karliczek

Anastomotic Leakage - Risk assessment in gastrointestinal surgery

ISBN:978-90-367-3612-1

Printed by: Gildeprint, Enschede

© Anne Karliczek, Bergen, 2008

No parts of this book may be reproduced in any form by print, photo print,

microfilm or other means without permission from the publisher

Financial support for the publication of this thesis was kindly provided by

Covidien Nederland BV

RIJKSUNIVERSITEIT GRONINGEN

Anastomotic Leakage

CmG.

Risk assessment in gastrointestinal surgery

Proefschrift

ter verkrijging van het doctoraat in de Medische Wetenschappen

aan de Rijksuniversiteit Groningen op gezag van de

Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op

woensdag 7 januari 2009 om 16.15 uur

door

Anne Karliczek

geboren op 2 maart 1971

te Hannover

Pro motor:

Prof. dr. T. Wiggers

Copromotor:

Dr. G.M. van Dam

Beoordelingscommissie:

Prof. dr. R.P. Bleichrodt

Prof. dr. J,A. Langendijk

Prof. dr. R.J. Ploeg

Paranimfen:

Dr. Ellen Klinkert

Dr. Evelien Lemstra

Contents

Chapter l Introduction, aim and outline of the thesis 9

Chapter 2 Drainage or non-drainage in elective colorectal 19

surgery- a systematic review and meta-analysis

Chapter 3 Surgeons lack predictive accuracy for anastomotic 35

leakage in gastrointestinal surgery

Chapter 4 lntraoperative ischemia of the distal end of colon 49

anastomoses as detected with visible light spectra-

scopy causes reduction of anastomotic strength

Chapter 5 Preoperative irradiation with 5x5 Gy in a murine 67

isolated loop model does not cause anastomotic

weakening after colon resection

Chapter 6 lntraoperative assessment of microperfusion with 85

visible light spectroscopy in oesophageal an cola-

rectal anastomoses - a feasibility study

Chapter 7 lntraoperative assessment of microperfusion with l 01

visible light spectroscopy for prediction of anasto-

motic leakage in colorectal anastomoses

Chapter 8 Anastomotic leakage - Clinical observations on me- 113

chanical forces and review of the literature

Chapter 9 Summary, conclusions and future perspectives l 21

Chapter l 0 Samenvatting, conclusies en toekomstperspectief 135

Dankwoord 149

Curriculum Vitae 155

Introduction, aim and outline of the thesis

Introduction Anastomotic leakage is a major problem in gastrointestinal surgery. The reported incidence of anastomotic leakage varies greatly among studies published on the subject. When analysing these studies 1 -25 (table 1), the frequency of anastomotic leakage in general ranges between 2 and 19%. In all studies, one most significant finding remains constant: the occurrence of anastomotic leakage is associated with considerable morbidity. After anastomotic leakage in 42%5 of the patients re-operation follows, whereas the remaining patients are managed by radiology guided 11

drainage. During re-operation usually a defunctioning stoma is created with or without dismantling of the anastomosis. Hospital and ICU stay is considerably longer after anastomotic leakage and 30-days mortality rises significantly after anastomotic leakages , 1 7, 26, 27. The latter is in most cases due to abdominal sepsis, which is often discovered late during the postoperative course, particularly in low pelvic anastomoses, as symptoms are subtle and non-specific28. On top of aforementioned morbidity, an increased rate of cancer occurrence has been reported following anastomotic leakage26.

As a consequence, it is important to identify patients at risk for development of anastomotic leakage as early as possible, preferably during the initial operation as the consequences of anastomotic leakage are less severe29. Some studies even show a significantly lower incidence of anastomotic leakage when a defunctioning stoma is used 1 6, 1 7. Currently, the intraoperative risk assessment for anastomotic leakage is made by the operating surgeon. If the risk is considered higher than normal, most surgeons tend to construct a defunctioning stoma in order to reduce the potential complications in case of anastomotic leakage.

Many studies have been conducted to identify the most important risk factors for anastomotic leakage and estimate the risk for anastomotic leakage. In Table 1 several studies describing multivariate analysis of risk factors for anastomotic leakage are shown. When analysing the multitude of risk factors described, many of these factors are associated either directly or indirectly with pre-existing cardiovascular disease. This risk factor might be indicative for the role of (micro-) circulatory disturbances and point at ischemia during anastomotic healing at time of the surgical procedure and thereafter. Furthermore, during resection of a diseased colonic segment including its supplying arteries, the surgically induced altered vascular supply clearly compromises the circulation at both ends of the anastomosis. This alteration as such is responsible for the higher rate of leakage in anastomoses located in the oesophagus and rectosigmoid, when compared to small bowel and other large bowel anastomoses3o, 31 _

In current day-to-day surgical practice the intestinal microcirculation is intraoperatively assessed by the surgeon by means of subjective judgement of colour and pulsatile blood flow of the intestinal margins at the dissected end of the

n ::r PJ

(t) ....

PJ :J

0..

:J (t)

� C"' ;-....

"'tJ-, 0 -, Ill < 0 :::s (I) � g-< 11 3 iii"

"CJ iii"� -, (I) 0 0 o_ _., V'I ,.,. V'I "CJ -, ,.,. (I) -· C: n e!.. o. �- 0 iii" iii -, 11'1 ..,, n..,, ,.,. Ill ::::r C: 11'1 0 0. 'P � -< n -· � 0 :::s :;:o :::SIC Qa �-II -'7':'"

iil g, ii1' :::S Ill n 0. :::s ,.,. 0 Ill O 3 11'1 -, ,.,. Ill -· 0 :::s � 3 e!.. 0. 0-< Q; �- �-::!. (DC"' n Ill-< e!..@3 ,.,. IC C: �- !D � - :;:o "S!,_

� (I)

0 � II�

-, (I) (I) 11'1 ,.,. V'I

a a· 11'1 :::s "CJ Ill (I) :::s !l Ill <"-< (1) V'I

!!i y;·

C: 0. -::-

Agnifili 20041

Alves 20022

Benoist 2000---s

Cubertafond 199zl Eriksen 20045

Folkesson 20046

Kasperk 2000' Katory 20078

Konishi 20069

Kruschewski 200?1°

Law 200011

Law 2004 12

Lefebure 200?13 Makela 200314

Matthiessen 200415

Matthiessen 200?1 6 Peeters 200477

Rullier 199a18 Schmidt 200s19

Smith 200320

Sorensen 199921

T occhi 200022

Ve}'T_ie 200723 Vignali 199?25° Yeh 200�

u, c a. '< '< "C C1>

RCT retro RCT retro

(prosp retro retro retro

lprosp lprosp lprosp lprosp retro case-contr

lprosp RCT retro retro

lprosp lprosp retro RCT

(prosp retro

lprosp tftt�I nr nf ri�lr f�l,tnrr-

:::, ci, Cl> ;ii:" Cl>

cc C1>

-ae. 2. 0

� 126/ 18 (14.3) 707/ 43 (6) 15815-11 (�16) 627/ (3.7) 1958/228 (12) 3316/293 (9) 98/18 (18) 707/ 5 (2.1) 391/ 11 (2.8) 276/ 41 (14.9) 196/ (10.2) 622/ (9.4) 132/ 13 (9.8) 88/44 (n.a.) 432/ 53 (12) 234145 (19.2) 924/ 107 (12) 272/ 32 (12) 933/ 68 (9.8) 3700/ 139 (3.81 333/53 (15.9) 112/ 9 (8) 1750/71(4) 1014/ 29 (2.9) 978/ 28 (3)

aU!llnO pue W!E 'UO!l)npOJlU( L JaldE4) � N

<O CD

5-lll> CD CO -, CD �,�

11111

CD 0 "O 3

ll>c:CDC-

9:3<g� ll) � g g !!!.. g-

en�'<"!:? -o CD

g_�.g{g �[ �- �- � � � g; � � g- g- g- �-CD CD :::::S :::::S :::::Sen

Risk factors for anastomotic leaka�

en

ll) s'"

en "O co

:::::s o· r )> :0

0 g��� � [ � en � ; � � � o· ��;;-��co ;:g.i e!, ;:,;- O'" 0 en 3 )> �- CD -, 1-i �r �- � iii. � g. � � � � �

111

3 ll) 5" & ;::;: c5·

:::::s !!?. � Q: � ;-�:::::SO< CD

0 en-CD.-+:::::S !!?.15: 2" g_ � cg:lg-3 C: -· 0 :::::s - C: rnotn_o ,.... ll>CD:::::senenenen

111

111

111 1 I 1

11111

1-0 co 0

��� CD O c: :::::s "O 0 ...... - 0

a.m� 5 % � ii> � :::!". 0 ..c :5· '< g �- �

1 I I I I I I I I I I 1 I I I I I I 1 I I I I I 1 sl2l2l1l1lol1l1lnl1l1l�l1l�l1l1I�I9I1l�l1IRl1l�l1l�l2lnl2l2l1l1

bowel segments. As such, this practice is considered the gold standard to determine viability of bowel ends and the current indirect risk assessment tool for anastomotic leakage. There are, however, no sound data available on sensitivity and specificity, and thus the reliability of this assessment can be questioned. Several clinical studies have been conducted to investigate mucosal perfusion, including Doppler ultrasound32, laser Doppler flowmetry33-36, scanning laser flowmetry37-39, fluorescence videography4o and intramucosal pH measurements41 -43. However, none of these methods is widely accepted due to lack of various liabilities44-46. A relatively new technique, known as visible light spectroscopy (VLS), has been 13 previously described and thoroughly been validated in different tissues47-5 1 . VLS uses the principle of absorbance and scattering of light in biological tissues. The oximeter emits white light from a handheld probe placed on or near the tissue, collecting any light returning to the probe from the tissue. The light travelling through biological tissues is absorbed by multiple molecules and scattered by cellular and anatomical structures. Each type of molecule has a characteristic absorbance spectrum. As oxygenated and deoxygenated haemoglobin have a different absorption spectrum and haemoglobin is considered a critical derivative of perfusion, tissue microperfusion can be assessed using a technique that detects these differences. The abundant haemoglobin in gastrointestinal capillary beds accounts for the majority of absorption in the mucosa! layer. In the range of wavelength used the light penetrates to a depth of approximately 1 mm. A major advantage of visible light is that it penetrates capillaries well, but not larger vessels, resulting in a capillary-weighted measurement in vivo, most closely correlated to tissue oxygen saturation (StO2)47,4B.

Aim and outline of this thesis The main aim of the present thesis was to investigate the impact of

microcirculatory factors contributing to anastomotic leakage and to develop more objective measures to identify patients at risk of developing anastomotic leakage.

Two main hypotheses will be analysed: i) the impact of drainage of fluid collections near the anastomosis, ii) the role of microcirculatory disturbances in anastomotic healing. It is hypothesized that a fluid collection near an anastomosis has a detrimental effect on anastomotic healing and therefore should be drained. Although many studies have been published on this subject, controversy remains. In Chapter 2 the results of a review conducted according to the Cochrane principles to evaluate efficacy of drainage of colorectal anastomoses are presented.

In Chapter 3 the clinical judgement of the surgeon for the intraoperative prediction of anastomotic leakage is evaluated for its accuracy in terms of sensitivity and specificity. As this judgement can be considered the gold standard for intraoperative estimation of anastomotic leakage, this description might form a basis which other new interventions for risk assessment can be compared to.

JlJ :::s 0.

:::s (D

In order to further elucidate the role of ischemia in the occurrence of anastomotic leakage and evaluate a tool for more accurate prediction of the risk of anastomotic leakage, we conducted a series of both clinical and preclinical studies to evaluate a new and well described technique, called visible light spectroscopy (VLS), for intraoperative measurements of ischemia. In Chapter 4 an experimental study in the rat is described, evaluating intraoperative measurements with visible light spectroscopy in deep ischemia of the colon serosa in rats. This study focuses particularly on postoperative anastomotic strength with and without previous

14 ischemia. As it is clear from human studies describing risk factors for anastomotic leakage, preoperative irradiation of the distal part of the anastomosis (i.e. predominantly the rectum) increases the risk of anastomotic leakage. Chapter 5

n describes a novel animal model to study the effects of irradiation with SxS Gray on

� tissue oxygenation and subsequent anastomotic strength. This animal model closely � simulates the human situation with respect of fractionation schemes and localisation

of irradiation. In the literature earlier described animal models show substantial

� 0 0.. C 8. 0 ::::l

e:!.

PJ ::::l 0..

::::l (1)

mortality due to irradiation techniques. Other studies apply clinically irrelevant irradiation schemes and are therefore less relevant to clinical practice.

At the same time we have evaluated the use of intraoperative visible light spectroscopy to predict anastomotic leakage in humans. First a pilot study to evaluate feasibility and stability of the measurements was conducted. The results of this study are described in Chapter 6. Based on the results of the feasibility study the predictive value of visible light spectroscopy for anastomotic leakage was determined in a clinical study, which is described in Chapter 7.

In Chapter 8 we present two clinical observations, that raise questions about the role of physiological intraluminal pressure and longitudinal tension in anastomotic leakage of the human intestine. In this chapter we review the literature published on this subject .

Chapter 9 summarizes the results of the aforementioned studies and discusses future perspectives, followed by Chapter 10, containing a Dutch version of the summary.

Notes on Chapter 4-7: Financial support: Nijbakker Morra Foundation, Leiden; J.C. de Cock Foundation, Groningen. Conflict of Interests statement: D.A. Benaron M.D. is affiliated with Spectros Corporation, Palo Alto, CA (CEO)

References

l . Agnifili A, Schietroma M, Carloni A, et al. The value of omentoplasty in protecting colorectal anastomosis from leakage. A prospective randomized study in 126 patients. Hepatogastroenterology. 2004;5 l :1694-7.

2. Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic leakage after large bowel resection: multivariate analysis of 707 patients. World J Surg 2002; 26: 499-502.

3. Benoist S, Panis Y, Alves A, et al. Impact of obesity on surgical outcomes after colorectal resection. Am J Surg 2000; 179: 275-281.

4. Cubertafond P, Cucchiaro G, Lesourd-Pontonnier F, et al. Early postoperative complications of resection anastomosis in colonic and colorectal surgery. An analysis of 624 cases. Chirurgie. 15 1992;1 l 8: 86-91.

5. Eriksen MT, Wibe A, Norstein J, et al. Norwegian Rectal Cancer Group. Anastomotic leakage following routine mesorectal excision for rectal cancer in a national cohort of patients. Colorectal Dis 2005; 7: 51-57.

6. Folkesson J, Nilsson J, Pahl man L, et al. The circular stapling device as a risk factor for anastomotic leakage. Colorectal Dis 2004; 6: 275-279.

7. Kasperk R, Philipps B, Vahrmeyer M, et al. Risk factors for anastomosis dehiscence after very deep colorectal and coloanal anastomosis. Chirurg 2000; 71: 1365-1369.

8. Katory M, Tang CL, Koh WL, et al. A 6-year review of surgical morbidity and oncological outcome after high resection for colorectal malignancy with and without splenic flexure mobilization. Colorectal Dis 2007; l 0: 165-169

9. Konishi T, Watanabe T, Kishimoto J, et al. Risk factors for anastomotic leakage after surgery for colorectal cancer: results of prospective surveillance. J Am Coll Surg. 2006; 202: 439-44.

l 0. Kruschewski M, Rieger H, Pohlen U, et al. Risk factors for clinical anastomotic leakage and postoperative mortality in elective surgery for rectal cancer. Int J Colorect Dis 2007; 22: 919-2 7.

11. Law WI, Chu KW, Ho JW, et al. Risk factors for anastomotic leakage after low anterior resection with total mesorectal excision. Am J Surg. 2000;179: 92-6.

12. Law WL, Chu KW. Anterior resection for rectal cancer with mesorectal excision: a prospective evaluation of 622 patients. Ann Surg. 2004 Aug;240(2):260-8.

l 3. Lefebure B, Tuech JJ, Bridoux V, et al. Evaluation of selective defunctioning stoma after low anterior resection for rectal cancer. lntJ Colorectal Dis 2008;23: 83-8.

14. Makela JT, Kiviniemi H, Laitinen S. Risk factors for anastomotic leakage after left sided colorectal resection with rectal anastomosis. Dis Colon Rectum 2003; 46: 653-660.

l 5. Matthiessen P, Hall book 0, Andersson M, et al. Risk factors for anastomotic leakage after anterior resection of the rectum. Colorectal Dis 2004; 6:462-9.

16. Matthiessen P, Hallbook 0, Rutegard J, et al. Defunctioning stoma reduces symptomatic anastomotic leakage after low anterior resection of the rectum for cancer, a randomized multicenter trial. Ann Surg 2007; 246: 207-214.

l 7. Peeters KC, Tollenaar RA, Marijnen CA, et al. Risk factors for anastomotic failure after total mesorectal excision of rectal cancer. Br J Surg, 2005; 92: 211-216.

l 8. Rullier E, Laurent C, Garrelon JL, et al. Risk factors for anastomotic leakage after resection of rectal cancer. Br J Surg 1998; 85: 355-8.

19. Schmidt 0, Merkel S, Hohenberger W. Anastomotic leakage after low rectal stapler anastomosis: significance of intraoperative anastomotic testing. Eur J Surg Oneal. 2003; 29: 239-43.

20. Smith JA, King PM, Lane RH, et al. Evidence of the effect of 'specialization' on the management, surgical outcome and survival from colorectal cancer in Wessex. Br J Surg 2003; 90: 583-92.

21 . Sorensen LT, Jorgensen T, Kirkeby LT, et al. Smoking and alcohol abuse are major risk factors for anastomotic leakage in colorectal surgery. Br J Surg 1999; 86: 927-931.

0. C n !:!. 0 ::J

�-3 PJ ::J 0. 0

::J (D

16

() ::::r PJ

(1) ..,

� 0 Cl.. C: � a·

!::!. 3 PJ ::I Cl..

::I (1)

22. Tocchi A, Mazzoni C, Lepre L, et al. Prospective evaluation of omentoplasty in preventing leakage

of colorectal anastomosis. Dis Colon Rectum 2000;43: 951-5.

23. Veyrie N, Ata T, Muscari F, et al. Anastomotic leakage after elective right versus left colectomy for cancer: prevalence and independent risk factors. J Am Coll Surg 2007; 205: 785-93.

24. Vignali A, Fazio VW, Lavery IC, et al. Factors associated with the occurrence of leaks in stapled rectal anastomoses: a review of 1014 patients. J Am Coll Surg 1997; 185: 113-121.

25. Yeh CY, Changchien CR, Wang JY, et al. Pelvic drainage and other risk factors for leakage after

elective anterior resection in rectal cancer patients: a prospective study of 978 patients. Ann Surg

2005; 241: 9-13.

26. Ptak H, Marusch F, Meyer F, et al. Impact of anastomotic leakage on oncological outcome after

rectal cancer resection. Br J Surg 2007; 94: 1548-1554.

2 7. Matthiessen P, Hall book 0, Rutegard J, et al. Population-based study of risk factors for

postoperative death after anterior resection of the rectum. Br J Surg 2006; 93: 498-503.

28. Nesbakken A, Nygaard K, Lunde OC, et al. Anastomotic leak following mesorectal excision for

rectal cancer: true incidence and diagnostic challenges. Colorectal Dis 2005; 7: 576-81.

29. Gastinger I, Marusch F, Steinert R, et al. Protective defunctioning stoma in low anterior resection

for rectal carcinoma. Br J Surg 2005; 92: 1137-42.

30. Dworkin MJ, Allen-Mersh TC. Effect of inferior mesenteric artery ligation on blood flow in the

marginal artery-dependent sigmoid colon. J Am Coll Surg 1996; 183:357-360.

31. Vignali A, Gianotti L, Braga M, et al. Altered microperfusion at the rectal stump is predictive for

rectal anastomotic leak. Dis Colon Rectum 2000; 43: 76-82.

32. Miyazaki T, Kuwana H, Kato H, et al. Predictive value of blood flow in the gastric tube in anastomotic insufficiency after thoracic esophagectomy. World J Surg 2002; 26: 1319-1323.

33. Ambrosetti P, Robert J, Mathey P, et al. Left-sided colon and colorectal anastomoses: doppler

ultrasound as an aid to assess bowel vascularization. Int J Colorectal Dis 1994; 9: 211-214.

34. Redaelli CA, Schilling MK, Carrel TP. lntraoperative assessment of intestinal viability by laser

Doppler flowmetry for surgery of ruptured abdominal aortic aneurysms. World J Surg 1998; 22:

283-289.

35. Pierie JP, De Graaf PW, Poen H, et al. Impaired healing of cervical oesophagogastrostomies can be

predicted by estimation of gastric serosal blood perfusion by laser Doppler flowmetry. Eur J Surg

1994; 160: 599-603.

36. Boyle NH, Pearce A, Hunter D, et al. lntraoperative scanning laser Doppler flowmetry in the

assessment of gastric tube perfusion during esophageal resection. J Am Coll Surg 1999 May; 188:

498-502.

3 7. Krohg-Sorensen K, Line PD, Horn RS, et al. lntraoperative prediction of ischaemic injury of the

bowel: a comparison of laser Doppler flowmetry and tissue oximetry to histological analysis. Eur J

Vase Surg 1992; 6: 518-524.

38. Boyle NH, Manifold D, Jordan MH, et al. lntraoperative assessment of colonic perfusion using

scanning laser Doppler flowmetry during colonic resection. J Am Coll Surg 2000; 191: 504-510.

39. Hajivassiliou CA, Greer K, Fisher A, et al. Non-invasive measurement of colonic blood flow

distribution using laser doppler imaging. Br J Surg 1998; 85: 52-55.

40. Toens C, Krones CJ, Blum U, et al. Validation of IC-VIEW fluorescence videography in a rabbit

model of mesenteric ischaemia and reperfusion. lnt J Colorectal Dis 2006; 21: 332-338.

41 . Millan M, Garcia-Granero E, Flor B, et al. Early prediction of anastomotic leak in colorectal cancer

surgery by intramucosal pH. Dis Colon Rectum 2006; 45: 595-601.

42. Senagore A, Milsom JW, Walshaw RK, et al. Intramural pH: a quantitative measurement for

predicting colorectal anastomotic healing. Dis Colon Rectum 1990; 33: 1 75-179.Millan 2006,

43. Tarui T, Murata A, Watanabe Y, et al. Earlier prediction of anastomotic insufficiency after thoracic

esophagectomy by intramucosal pH. Crit Care Med 1999; 27: 1824-1831.

44. Krohg-Sorensen K, Lunde QC. Perfusion of the human d istal colon and rectum evaluated with endoscopic laser Doppler flowmetry. Methodologic aspects. Scand J Gastroenterol 1 993 ; 28 : l 04-1 08.

4 5 . Brol in RE, Semmlow JL, Sehonanda A, et al. Comparison of five methods of assessment of i ntestinal viabi l ity. Surg Gynaecol Obstet 1 989; 1 68 : 6- 1 2 .

46. line PD, Mowinckel P, lien B, et al. Repeated measurement variation and precision of laser Doppler flowmetry measu rements . Microvasc Res 1 992; 43: 285-293.

4 7. Benaron, D. A. , Parachikov IH , Cheong WF, et al. Des ign of a visible-l ight spectroscopy c l in ical t issue oximeter. J Biomed Opt 2005 ; 1 0: 44005- 1 -44005-9.

48. Benaron DA, Parachikov IH , Friedland S, et al . Continuous, noninvasive and local ised microvascular tissue oximetry u sing visible l ight spectroscopy. Anesthesiology 2004; 1 00: 1 469- 1 475 .

49. Fried land SR , Benaron D. Reflectance spectrophotometry for the assessment of mucosal perfusion in the gastrointestinal tract. Gastrointest Endosc Cl in N Am 2004; 1 4: 5 39-5 5 5 .

50. Fried land S , Benaron D , Parach ikov I , e t al . Measurement of mucosal hemoglobin oxygen saturation in the colon by reflectance spectrophotometry. Gastrointest Endosc 2003; 5 7: 493-497.

5 1 . Ramirez FC, Sukhdeep P, Medl in S, et al . Reflectance spectrophotometry in the gastrointestinal tract: l imitations and new appl ications. Am J Gastroenterol 2002; 97: 2780-2 784.

17

Drai nage or non-d rai nage i n e lective colorectal

anastomos i s

a systemat ic review and meta-analys i s

Anne Karliczek1 , Eliane C Jes us2 , Delios Matos3 , Aldemar A Castro3 , Alvaro

N Atallah3 , Theo Wiggers 1

1 Department of Surgery, Un iversity Med ical Center G ron ingen, Gron ingen, The Netherlands

2 UNIFOA Centro Un ivestario de Volta Redonda, Volta Redonda, RJ , Braz i l

3 Un iversidade Federal de Sao Pau lo, Escola Pau l ista de Medecina, Discip l ina de Gastroenterolog ica

C i ru rg ica, Sao Pau lo, B raz i l

Colorectal Dis 2006 May; 8: 25 9-65 and i n part pub l ished i n Cochrane L ibrary 2004 Issue 4 CD 002 1 00

- -

Abstract

Background : There i s l ittle agreement on prophylactic use of drai ns i n anastomoses i n elective colorectal surgery despite many randomized c l in ical trials. Once anastomotic leakage has occurred it i s general ly agreed that drai ns should be used for therapeutic purposes. However, on prophylactic use no such agreement exists. The aim of the underlyi ng review is to compare safety and effectiveness of routine drai nage and non-drai nage reg imes after elective colorectal surgery. The primary outcome is cl i n ical anastomotic leakage. Methods: A systematic search was undertaken identifying random ized c l in ical trials. Resu lts : Of the 1 1 40 patients enrol led (6 RCTs) , 5 73 were al located for drai nage and 567 for no drai nage. The patients ass igned to the drainage g roup compared with the ones assigned to non-drainage group s howed: a) Mortal ity: 3% ( 1 8 of 573 patients) compared with 4% (2 5 of 567 patients); b) Cl i n ical anastomotic deh iscence: 2% ( 1 1 of 522 patients) compared with 1 % (7 of 5 1 9 patients); c) Rad iolog ical anastomotic deh iscence : 3% ( 1 6 of 5 22 patients) compared with 4% ( 1 9 of 5 1 9 patients); d) Wou nd infect ion : 5% (29 of 5 73 patients) compared with 5% (28 of 567 patients) e) Re i ntervention : 6% (34 of 542 patients) compared with 5% (2 8 of 539 patients); f) Extra-abdominal compl ications : 7% (34 of 5 2 2 patients) compared with 6% (32 of 5 1 9 patients). None of these d ifferences in outcome were s ignifi cant. Conclus ions: We therefore concluded that there is insuffic ient evidence showing that routi ne drainage after colorectal anastomoses prevents anastomotic and other com pl i cations.

Introduction Colorectal anastomosis is a frequently performed procedure in numerous surgical clinics. A serious and therefore feared complication is an abdominal or pelvic sepsis following anastomotic leakage. Mortality of this complication is as high as 1 2% in acute and elective surgery3• In order to treat or prevent anastomotic dehiscence some surgeons routinely use drains. Others use drains only if in doubt regarding the quality of the anastomosis performed. Some surgeons never use drains l .2 . Once anastomotic leakage has occurred it is generally agreed that drains should be used for therapeutic purposes.

The purpose of drainage is to prevent accumulation of fluids in the pelvic or peritoneal cavity and to permit early detection of anastomotic dehiscence by faecal or purulent discharge from the drain. Pelvic drains in extraperitoneal anastomosis are even more commonly used because of higher leakage rates, particularly after radiation therapy 4, s, 6, 7.

Use of drains dates back to Hippocrates, who used linen to keep wounds open after drainage of thoraxempyema. Ambroise Pare was the first to describe drainage of the abdominal cavity, but probably abdominal drainage had been in practice earlier. Abdominal drainage thus has a long historical tradition as well as the question of its usefullness8• A number of studies have questioned the benefit of routine drainage9, 1 0•

It has even been argued that drains might adversely affect anastomotic healing by causing infection in the anastomotic area and in the abdominal wound, based on animal studies1 1 , 1 2• Moreover, drains may cause pain and therefore result in other infections such as pneumonia.

Several reviews, trials and retrospective studies on the issue whether to drain or not to drain colorectal anastomosis have been published. In particular, a number of randomized clinical trials on the same issue have been carried out. However, the results of these trials are contradictory due to lack of quality and/or statistical power and do therefore not answer the clinical question. Urbach found no benefit from drainage of colorectal anastomoses in a review including 4 14 patients submitted to elective and emergency surgery1 3• As stated by Tekkis et al1 4 the urgency of the operation is a independent risk factor for operative mortality. We therefore conducted a new review excluding emergency surgery.

The primary objective of the current review is to determine whether prophylactic drainage after anastomosis in elective colorectal surgery prevents clinical anastomotic leakage. A subgroup-analysis according to level of anastomosis was performed in order rule out a significant effect of drainage of the low anastomosis. Furthermore, drainage and non-drainage regimes were compared in terms of length of hospital stay, radiological anastomotic leakage and infectious complications.

21

22

N

::J 0 ::J

�-::J PJ

l.O l'tl

Methods

Search strategy The studies were identified from computer searches i n CINAHL, EMBASE, LI LACS and MEDLINE, us ing the fo l lowing search terms: anastomos* d rai n* colorect* or pelvi*. The Contro l led C l i n ical Trials Database, the Trials Reg ister of the Cochrane Colorectal Cancer Group and refe rence l i sts were hand searched. Studies were inc l uded i rrespective of language used o r date of publ ication . Selection criteria Al l randomized control led tria ls compari ng any d rai nage with non-drai nage reg i mes after anastomoses i n co lorectal surgery were reviewed . Trials i nc lud ing patients submitted to emergency colorectal su rgery were excl uded. The pr imary outcome measure was defi ned as c l i n ical anastomotic deh iscence. Secondary outcome measures were mortal ity, rad io logical anastomotic dehi scence, wou nd infection , reoperation , length of hospital stay and extra-abdomi nal comp l icat ions.

Data were independently extracted by two reviewers and cross-checked. The same reviewers assessed the methodological qual ity of each trial . Detai ls of the randomization (generation and concealment) , b l i nd ing , and the number of patients lost to fol low-up were recorded. The resu lts of each RCT were summarized on an i ntention-to-treat bas is in 2 x 2 tables for each outcome. External val id ity was defi ned on the bas is of characteri stics of participants, i nterventions and outcomes. The RCTs were stratified based on experimental group, accord ing to c l in ical homogeneity (external val id ity). If the resu lts obtained from these d ifferent statistical methods were very s im i lar, on ly the resu lts from the Risk Difference method (RD) with the corresponding 95% confidence interval is presented in th is review. The fixed effects model was used if there is no s ign ificant statistical heterogene ity (P>0. l ).

The RCTs were stratified based on experimental g roup , type of d rain (pass ive or suction), and site of the anastomos i s (extra-peritoneal , i ntra-peritoneal and not specified) .

The meta-analys is was pe rformed (Review Manager, vers ion 4. 1 for Windows , The Cochrane col laborat ion. Oxford : Update Software ; 2000) on the bas is of various techniques : i n dichotomous outcome measu res the combined logarithm of Peto odds ratio (fixed effect model) , odds rat io, relative r isk, ri sk difference (al l fixed and random effects mode l s) was used. If the resu lts obtai ned from these different methods were very s im i lar, on ly the resu lts from the ri sk d ifference method with the correspond ing 95% confidence interval are presented in th i s review. The fixed effects model was used if there is no s ign ificant statistical heteroge neity (P > 0 . 1 ) .

Statistical hete rogeneity of the resu lts of the trials was assessed on the bas i s of a test of heterogeneity (standard chi-squared test on N degrees of freedom where N equals the number of trials contri buti ng data m inus one) .

Results Nine RCTs that compare prophylactic anastomotic drainage with non-drainage in colorectal surgery were found. Three of 9 RCTs were excluded: one because inadequate allocation concealment 1 5 and two because patients submitted to emergency surgery were included 1 6, 1 7. Six RCTs were analysed with 1140 patients. The characteristics of the included studies are summarized in table 1. Of the 6 RCTs, four studies were published in English (Hoffman 1987 1 8 , Merad 1998 1 9 , Merad l 99920,

Brown 200121 ) and two were published in German (Hagmu l ler 199022, Mennigen 198923).

In 2 studies only patients with cancer (Brown 2001, Hagmuller 1990) were monitored. The other four studies included cancer, inflammatory bowel disease and non-inflammatory disease in different proportions. In none of them children were included. The intervention groups included several types of drainage regimes: one study described patients with corrugated latex drain (Hoffman 1987), one study described easy-flow-drainage (Hagmuller 1990), one silicone drains (Mennigen l 989), one closed-suction intraperitoneal drain (Merad 1999) and in one study both closed suction drains and silicone drains were used (Merad 1998). The last study (Brown 2001) described extraperitoneal closed suction drains.

One study recorded both intra- and extra-peritoneal anastomoses (Merad 1999), one study described extraperitoneal anastomoses (Brown 2001) and four studies intraperitoneal anastomoses only (Hoffman 1987, Mennigen 1989, Hagmuller 1990, Merad 1998). Mortality (6 studies), clinical anastomotic dehiscence (5 studies), radiological anastomotic dehiscence (5 studies), wound infection (6 studies),

Hoffman Mennigen Hagmuller Merad Merad Brown

Year 1 987 1 989 1 990 1 998 1 999 2001

N 60 99 1 1 3 31 7 492 59

corrugated closed-suction closed closed Drain type silicone Easy-Flow and silicone latex (non-suction) suction suction

Cancer indication (%) 63 77 1 00 70 62 1 00

I nfraperitoneal 0 0 0 not described 27 1 00 anastomosis (%)

Multi/ monocentre mono mono mono multi multi mono

Place of study Denmark Germany Germany France France Singapore

Table 1 Summary of trials included in the meta-analysis

23

n ::r $ll "O ,....

N

24

n ::r SlJ

(l) ..., N

�-:::i SlJ

l.O (l)

0

:::i 0 :::i

�-:::i SlJ

l.O (l)

Vl -<

3 §. n ..., rt) S.

reoperation ( 5 studies) and extra-abdominal complications (4 studies) were recorded. Two studies described the sample size calculations used (Merad 1998, Merad

1999). However, the other three studies (Hoffman 1987, Hagmuller 1990, Mennigen 1989) did not describe the presence or method of sample size calculation.

Of the 1140 patients enrolled (6 RCTs), 573 were allocated for drainage and 567 for non-drainage. The comparison of primary and secondary outcome measures in patients assigned to the drainage group with the ones assigned to non-drainage group is summarized in table 2.

The meta-analysis showed no significant difference in the occurrence of clinical anastomotic dehiscence. This was the primary outcome measure, which was analysed in 5 trials (Brown 2001, Hagmuller 1990, Hoffman 1987, Merad 1998, Merad 1999). The same conclusion can be drawn for radiological anastomotic dehiscence, based on the results of 5 studies (Hagmuller 1990, Hoffman 1987, Brown 2001, Merad 1998, Merad 1999). Stratification by level of anastomosis showed no benefit of drainage for extraperitoneal anastomosis. However, these results are based on only 2 trials describing patients with extraperitoneal anastomoses separately (Merad 1999; Brown 2001) and should therefore not be generalised. Data on intraperitoneal anastomoses are consistent with results stratified by type of drain.

The mortality has been analysed in 6 studies (Hoffman 1987; Mennigen 1989 ; Hagmuller 1990; Merad 1998; Brown 2001; Merad 1999) and showed no significant difference between the drainage and non-drainage group (figure 3). The reoperation rate has been analysed in 6 studies (Hoffman 1987, Mennigen 1989; Hagmuller 1990; Me rad 1998; Merad 1999 ; Brown 2001 ), with special attention to reoperation because of anastomotic complications. Again, there was no significant difference between the drainage and the non-drainage group.

outcome measure drainage no drainage risk difference 95% confidence interval

n/N % n/N % %

mortality 1 8/573 3 25/567 4 0,39-1 ,31

clin. anastomotic dehiscence 1 1 /522 2 7/51 9 0,61 -3,95

rad. anastomotic dehiscence 1 6/522 3 1 9/51 9 4 0,42-1 ,61

wound infection 29/573 5 28/567 5 0 0,60-1,76

reintervention 34/542 6 28/539 5 0,73-2,05

extra abdominal complications 34/522 7 32/51 9 6 0,66-1,85

Table 2 Resu lts of primary and secondary outcome measures of the meta-analysis of primary and secondary outcome measures of six randomized clinical trials summarized in table 1 . Results of random effects model are presented as risk difference.

Comparison: 01 All studies stratified by type of drain Outcome: 01 MortaHty (6 trials, 1 140 patients)

Study Drainage or sub-category n/N

01 Passive drainage Hoffmann 1987 0/20 Mennigen 1989 1/51 Hagmulier 1990 1/60

Subtotal (95% Cl) 139 Total events: 2 (Drainage), 5 (No drainage) Test for heterogeneity: Chl2 = 0.81, df = 2 (P = 0.67), 1• = 0% Test for overaU effect: Z = 1 .18 (P = 0.24)

02 Active drainage Merad 1999 8/24 7 Brown 2001 1/31

Subtotal (95% Cl) 278 Total events: 9 (Drainage), 1 1 (No drainage) Test for heterogeneity: Chl2 = 0.01 , df = 1 (P = 0.93), 12

= 0% Test for overall effect: Z = 0.49 (P = 0.62)

03 Active and passive drainage Mered 1998 7 /156

Subtotal (95% Cl) 156 Total events: 7 (Drainage), 9 (No drainage) Test for heterogeneity: not applicable Test for overan effect Z = 0.45 (P = 0.65)

Total (95% Cl) 573 Total events: 18 (Drainage), 25 (No drainage) Test for heterogeneity: Chl2 = 1 .46, df = 5 (P = 0.92), 12 = 0% Test for overaU effect Z = 1 .09 (P = 0.28)

No drainage n/N

2/32 2/48 1/53

133

10/245 1/28

2 73

9/161 161

567

0. 1 0.2

Peto OR 95% CI

•

•

•

0.5 2

Weight %

4 . 72 7 . 08 4 . 76

16 . 57

4 1 . 93 4 . 73

46 . 66

3 6 . 77 3 6 . 77

100 . 00

5 10

Peto OR 95% CI

0 . 15 [0 . 01 , 2 . 4 5 ) 0 . 4 8 [0 . 05 , 4 . 69] 0 . 88 [0 . 05 , 14 . 36) 0 . 4 1 [0 . 09 , 1 . 82 )

0 . 79 [0 . 31 , 2 . 02] 0 . 90 [0 . 05 , 14 . 82 ) 0 . 80 [0 . 3 3 , 1 . 95)

0 . 80 [ 0 . 29 , 2 . 17] 0 . 80 [ 0 . 2 9 , 2 . 17)

0 . 71 [ 0 . 3 9 , 1 . 31 )

Figure 1 Pooled estimates of mortality risk stratified by type of drain . Squares indicate point estimates of treatment effect (odds ratio), with the size of the square

representing the weight attributed to each study. Ninety-five percent confidence intervals are indicated by horizontal bars. The summary odds ratio from

the pooled data with 95% confidence intervals is represented by the diamond. Values to the left of the vertical l ine at 1 favor the drainage group. Point

estimates are significant if their confidence intervals exclude the vertical line at 1 .

Mi:l!/\i:lJ ) !lEW i:llSAs : a 6'eu ,e.1 p- u o u JO a6EU !EJ O 2'. JaldE4J N c,n

Mi:l!/\i:lJ J !lEWi3lSAS :a6eu ,eJ p-uou JO a6EU !EJO

Comparison: 01 All studies stratified by type of drain Outcome: 03 Clinical anastomotic dehiscence (5 trials, 1041 patients)

Study drainage

or sub-category n/N

01 Passive drainage Hoffmann 1987 1/2 8

Hagmuller 1990 1/60

Subtotal (95% Cl) 8 8

Total events: 2 (drainage), 1 (no drainage)

Test for heterogeneity: ChP = 0.50, df = 1 (P = 0.48), 12 = 0% Test for overaU effect Z = 0.62 (P = 0.53)

02 Active drainage Merad 1999 6/247

Brown 2001 2/31

Subtotal (95% Cl) 278

Total events: 8 (drainage), 5 (no drainage) Test for heterogeneity: ChP = 0.40, df = 1 (P = 0.53), 12 = 0% Test for overall effect: Z = 0. 78 (P = 0.44)

03 Active and passive drainage Merad 1 998 1/156

Subtotal (95% Cl) 156

Total events: 1 (drainage), 1 (no drainage) Test for heterogeneity: not applicable

Test for overall effect: Z = 0.02 (P = 0.98)

Total (95% Cl) 522

Total events: 11 (drainage), 7 (no drainage) Test for heterogeneity: ChP = 1 .05, df = 4 (P = 0.90), 12 = 0%

Test for overall effect: Z = 0.92 (P = 0.36)

no drainage n/N

1/32 0/53

85

3 /245 2/28

2 73

1/161 161

519

0.1 0.2

Peto OR 95% CI

■

0.5 2

drainage no drainage

2'. Jalde4J

Weight %

11 . 13 5 . 67

16 . 80

50 . 35 21 . 52 71 . 87

11 . 34 11 . 34

100 . 00

5 10

N a,

Peto OR 95% CI

1 . 15 [ 0 . 07 , 18 . 8 8 ] 6 . 58 [ 0 . 13 , 333 . 89] 2 . 07 [0 . 21 , 20 . 21]

1 . 95 [0 . 52 , 7 . 29] 0 . 90 [0 . 12 , 6 . 74 ] 1 . 55 [0 . 51 , 4 . 66 ]

1 . 03 [0 . 06 , 16 . 58 ] 1 . 03 (0 . 06 , 16 . 58 ]

1 . 55 (0 . 61 , 3 . 95]

Figure 2 Pooled estimates of cl in ical anastomotic dehiscence, stratified by type of drain . Squares indicate point estimates of treatment effect (odds ratio), with the

size of the square representing the weight attributed to each study. N inety-five percent confidence i ntervals are indicated by horizontal bars. The

summary odds ratio from the pooled data with 95% confidence intervals is represented by the diamond. Values to the left of the vertical l ine at l favor

the drainage group. Point estimates are sign ificant if thei r confidence intervals exclude the vertical l ine at l .

Comparison: 02 All studies stratified by level of anastomosis Outcome: 01 Mortality (4 trials, 589 patients)

Study or sub-category

02 Intra-peritoneal anastomosis

Drainage n/N

Hoffmann 1987 0/20

Mennlgen 1989 1/51

Hagmuller 1990 1/60

Merad 1 998 7 /156

Subtotal (95% Cl) 295

Total events: 9 (Drainage), 15 (No drainage) Test for heterogeneity: ChP = 1 .89, df = 3 (P = 0.60), 12

= 0% Test for overall effect: Z = 1 .20 (P = 0.23)

Total (95% Cl) 295

Total events: 9 (Drainage), 15 (No drainage) Test for heterogeneity: ChP = 1 .89, df = 3 (P = 0.60), 12

= 0% Test for overall effect Z = 1 .20 (P = 0.23)

No drainage n/N

3/32 2/48 1/53 9/161

2 94

2 94

....

0.1 0.2

Peto OR Weight Peto OR 95% CI % 95% CI

12 . 53 0 . 14 [0 . 01 , 1 . 44 ) 12 . 75 0 . 4 8 [0 . 05 , 4 . 69)

8 . 57 0 . 8 8 [0 . 05 , 14 . 3 6 ) 66 . 15 0 . 80 [0 . 29 , 2 . 17)

100 . 00 0 . 61 [0 . 2 7 , 1 . 37)

100 . 00 0 . 6 1 [0 . 2 7 , 1 . 37 )

0.5 2 5 10

Drainage Non drainage

Figure 3 Pooled estimates of mortality risk stratified by level of anastomosis. Squares indicate point estimates of treatment effect (odds ratio), with the size of the square representing the weight attributed to each study. Ninety-five percent confidence intervals are ind icated by horizontal bars. The summary

odds ratio from the pooled data with 95% confidence intervals is represented by the d iamond. Values to the left of the vertical l ine at l favor the

drainage group. Point estimates are sign ificant if their confidence intervals exclude the vertical l ine at l .

Ma!/\aJ J 11ewa1sAs :a5eu 1eJ p-uou Jo afreureJa l Ja1de4J N �

!'

Mi:3!/\i:3J J!lEWa:isAs :a5EU!EJp-uou JO a5EU!EJO c Ja:tdE4J

Comparison: 02 All studies stratified by level of anastomosis Outcome: 02 Clinical anastomotic dehiscence (4 trials, 587 patients)

Study Drainage or sub-category n/N

01 Intra-peritoneal anastomosis Hoffmann 1987 1/28

Mennlgen 1989 1/51

Hagmuller 1990 1 / 6 0

Merad 1998 1/156

Subtotal (95% Cl) 2 95

Total events: 4 (Drainage), 4 (No drainage) Test for heterogeneity: Chi2 = 1 .19, df = 3 (P = 0.75), 12 = 0% Test for overall effect Z = 0.00 (P = 1 .00)

Total (95% Cl) 2 95

Total events: 4 (Drainage), 4 (No drainage) Test for heterogeneity: Chi2 = 1 . 19, df= 3 (P = ct.75), 12 = 0% Test for overaH effect: Z = 0.00 (P = 1 .00)

No drainage n/N

1 / 3 0

1/48

0/53

2/161

2 92

2 92

• •

•

0.1 0.2

■

0.5

Peto OR 95% CI

■ ■

2 5 Favours treatment Favours control

Weight %

� 24 . 7 9

� 24 . 98

12 . 5 8

3 7 . 64

1 0 0 . 0 0

1 0 0 . 0 0

10

N co

Peto OR 95% Ci

1 . 0 7 [ 0 . 0 7 , 17 . 6 1]

0 . 94 [0 . 0 6 , 15 . 2 7 ]

6 . 5 8 [0 . 13 , 333 . 8 9 ]

0 . 53 [0 . 05 , 5 . 11]

1 . 0 0 [0 . 2 5 , 4 . 02 ]

1 . 0 0 [ 0 . 2 5 , 4 . 02 ]

Figure 4 Pooled estimates of cl in ical anastomotic deh iscence stratified by level of anastomosis. Squares indicate point estimates of treatment effect (odds ratio),

with the size of the square representing the weight attributed to each study. Ninety-five percent confidence intervals are ind icated by horizontal bars. The

summary odds ratio from the pooled data with 95% confidence intervals is represented by the d iamond. Values to the left of the vertical l ine at l favor the

drainage group. Point estimates are significant if their confidence intervals exclude the vertical l ine at l .

The rate of wound infection is not associated with the insertion or not of a drain in the abdominal cavity. In 6 studies (Hoffman 1987; Mennigen 1989; Hagmuller 1990; Merad 1998; Merad 1999, Brown 2001) no significant difference has been found. Thus drains did not minimize wound infection, although it was expected that drains would protect the wound by carrying intra-abdominal fluid collections through the drain rather than through the operation wound.

Extra-abdominal infection, such as pulmonary and urinary tract infections, have been reported in 5 (Hoffman 1987; Hagmuller 1990; Merad 1998; Merad 1999; Brown 2001) studies. These studies showed no significant differences in infection rates, 29 whether prophylactic drains were placed or not.

Based on the data of this metaanalysis there is insufficient evidence showing that routine drainage after colorectal anastomoses prevents anastomotic and other complications.

Discussion

I t is generally agreed that drainage in circumstantial situations such as improper hemostasis or fecal contamination might be necessary. Once anastomotic breakdown occurred the treatment of choice is usually drainage and construction of a defunctioning stoma. However, prophylactic drainage after colorectal anastomoses is used by many surgeons worldwide in order to prevent anastomotic dehiscence by evacuating fluid collections. Some authors argue that drains in the proximity of the anastomosis might not be useful in order to prevent anastomotic dehiscence or can even be the cause of leakage?, 24, 25, 33, 34, 35 . Considering earlier studies no clear guidelines could be established for routine placement of drains after colorectal anastomoses. This systematic review however shows no benefits from prophylactic drainage of colorectal anastomoses in general.

Prophylactic drainage has been evaluated in several types of operations, such as drainage after pelvic lymphadenectomy and hysterectomy27, pancreatic resection28 ,

hip and knee arthroplasty 29 , lumbar spine surgery30 and surgery for breast cancer3 1 •

None of these studies showed benefit from prophylactic drainage. Several other studies evaluated the use of different types of drains mainly in animal models. 1 2 , 36

Latex drains have been shown to have a detrimental effect on colonic healing and adhesion formation, but no conclusions can be drawn on other drainage materials in vivo. A trial conducted to evaluate irrigation-suction versus suction drains showed no difference in the regimes. 37 No other studies concerning this issue were found.

In this review stratification by type of drain showed no difference in leakage rates in different drainage regimes, but the results must be interpreted with caution due to lack of statistical power and comparable drainage regimes. Drainage of the pelvic anastomosis is more commonly practiced because leakage is a feared complication in these anastomoses. Earlier meta-analyses identified several risk factors for anastomotic leakage. 4, 26 Apart from factors impairing wound healing one of the most important factors causing anastomotic dehiscence is the level of the

anastomos is . The more d istal the colorectal anastomosis , the g reater its tendencies to leak. I ntroduction of total mesorectal exc is ion (TME) and the i ncreas i ng use of preoperative rad iotherapy for rectal cancer causes add itional problems in anastomotic heal i ng. A large prospective study showed a h igher leakage rate after routi ne i rr igation-suction d ra inage in e lective anterior resection. 33 However, i n a recent retrospective study d rainage and the use of a defu nction ing stoma proved to be beneficial on reoperation rates due to anastomotic leakage after TME32 . Colorectal anastomoses in the pelvic space after preoperative SxS Gy might be the main

30 chal l eng i ng problem due to reactive hyperemia. The d iscussion shou ld therefore focus on the benefit of drainage of low anastomoses and management of the pelvic space.

In th is review we cou ld not prove s ign ificant benefit from drai nage of co lorectal anastomoses in general . We conducted stratificat ion to leve l of anastomosis , but the resu lts after this stratificat ion lack statistical power and shou ld therefore be interpreted with caution . Clearly larger random ized c l in ical trials focus ing particu larly on drai nage of the low colorectal anastomoses are needed to answer the question whether drai nage is usefu l or not.

References

1. Berliner SD, Burson LC, Lear PE, et al. Use and abuse of intraperitoneal drains in colon surgery. Arch

Surg 1 964; 89: 686-690

2. Soeters PB, de Zoete JP, Dejong CH, et al. Colorectal surgery and anastomotic leakage. Dig Surg.

2002;19(2):150-5

3. Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic leakage

after large bowel resection: multivariate analysis of 707 patients. World J Surg 2002; 26: 499-502

4. Vignali A, Fazio VW, Lavery IC, et al. Factors associated with the occurrence of leaks in stapled rectal

anastomoses: a review of 1014 patients. J Am Coll Surg 1997; 185: 113-121.

5. Galandiuk S, Fazio VW. Postoperative irrigation-suction drainage after pelvic colonic surgery: a

prospective randomized trial. Dis Colon Rectum 1991; 34: 223-228.

6. Hilsabeck JR. The presacral space as a collector of fluid accumulations following rectal anastomosis.

Dis Colon Rectum 1982; 25: 680-4.

7. Gingold BS, Jagelman DG. Value of pelvic suction - irrigation in reducing morbity of low anterior

ressection of the rectum. A ten-year experience. Surgery 1982; 97: 394-398 8. Robinson JO. Surgical drainage: an historical perspective. Br J Surg 1986; 73: 422-226

9. Sagar PM, Couse N, Kerin N, et al. Randomized trial of drainage of colorectal anastomosis. Br J Surg

1993; 80: 769-771.

10. Galandiuk S, Fazio V W. Postoperative irrigation-suction drainage after pelvic colonic surgery: a

prospective randomized trial. Dis Colon Rectum 1991; 34: 223-228.

11. Berliner SD, Burson LC, Lear PE, et al. Use and abuse of intraperitoneal drains In colon surgery. Arch Surg 1964; 89: 686-690

12. Smith SRG, Connolly JC, Crane PW, et al. The effect of surgical drainage materials on colonic healing.

Br J Surg 1982; 69: 153-155.

13. Urbach DR, Kennedy ED, Cohen MM. Colon and rectal anastomoses do not require routine drainage, a

systematic review and meta-analysis. Ann Surg 1999; 229: 1 74-180.

14. Tekkis PP, Poloniecki JD, Thompson MR, et al. Operative mortality in colorectal cancer: prospective

national study. BMJ 2003; 327: 1196.

15. Johnson CD, Lamont PM, Orr N, et al. Is a drain necessary after colonic anastomosis? J R Soc Med

1989; 82: 661-664.

16. Sagar PM, Couse N, Kerin N, et al. Randomized trial of drainage of colorectal anastomosis. Br J Surg

1993; 80: 769-771.

17. Sagar PM, Hartley MN, Mcfie J, et al. Randomized trial of pelvic drainage after rectal resection. Dis

Colon Rectum 1995; 38: 254-258.

18. Hoffmann J; Shokouh-Amiri MH; Damm P; et al. A prospective, controlled study of prophylactic

drainage after colonic anastomoses. Dis Colon Rectum 1987; 30: 449-52.

19. Merad F, Yahchouchi E, Hay JM, et al. Prophylactic abdominal drainage after elective colonic resection

and suprapromontory anastomosis: a multicenter study controlled by randomization. French

Associations for Surgical Research. Arch Surg 1998; 133: 309-14.

20. Merad F, Hay JM, Fingerhut A, et al. Is prophylactic pelvic drainage useful after elective rectal or anal

anastomosis? A multicenter controlled randomized trial. French Association for Surgical Research.

Surgery 1999; 125: 529-35.

21. Brown SR, Seow-Choen F, Eu KW, et al. A prospective randomised study of drains in infra-peritoneal

rectal anastomoses. Tech Coloproctol 2001; 5: 89-92.

22. Hagmuller E, Lorenz D, Werthmann K, et al. Effects and risks of drainage after elective colon resection.

A prospective, randomized controlled clinical study. Chirurg 1990; 61: 266-271.

23. Mennigen R, Kusche J, Troidl H. Prophylaktische Drainage von Kolonanastomosen. Colo-proctology 1989;11 :76-80.

24. Beahrs, OH. Low anterior resection for cancer of the rectosigmoid and rectum. Surg Clin North

America 1967; 47: 971-975

25. Gallagher P, Cade P, Whale R, et al. Does infection cause anastomotic dehiscence? J R Coll Surg Edinb

1982; 27: 90-92.

31

:::s 0 :::s

�-:::s llJ

lO (I)

32

n :r /lJ

"C ,..,. (1) ..... N

�-:J /lJ

lCl (1) 0

:J 0 :J

�-:J /lJ

lCl (1)

26. Karl RC, Schreiber R, Boulware D, et al. Factors affecting morbidity, mortality and survival in patients

undergoing Ivor Lewis esophagogastrectomy. Ann Surg 2000; 5: 635-643.

27. Patsner B. Closed-suction drainage versus no drainage following radical abdominal hysterectomy with

pelvic lymphadenectomy for stage lb cervical cancer. Gynaecol Oneel 1995; 57: 232-234.

28. Conlon KC, Labow D, Leung D, et al. Prospective randomized clinical trial of the value of

intraperitoneal drainage after pancreatic resection. Ann Surg 200 1 ; 234: 487-494.

29. Parker MJ, Roberts CP, Hay D. Closed suction drainage for hip and knee arthroplasty. J Bone Joint Surg

2004; 86-A: 1146- 1 1 52.

30. Brown MD, Brookfield KFW. A randomized study of closed wound suction drainage for extensive

lumbar spine surgery. Spine; 2004; 29: l 066-l 068.

3 1 . Jain PK, Sowdi R, Anderson DG, et al. Randomized clinical trial investigating the use of drains and

fibrin sealant following surgery for breast cancer. Br J Surg; 2004; 91 :54-60.

32. Peeters KCMJ, Tollenaar RAEM, Marijnen CAM, et al. Risk factors for anastomotic failure after total

mesorectal excision of rectal cancer. Br J Surg 2005; 92: 211 -2 1 6.

33. Yeh CY, Changchien CR, Wang JV, et al. Pelvic drainage and other risk factors for leakage after

elective anterior resection in rectal cancer patients. Ann Surg 2005; 24 1 : 9- 1 3.

34. Scott H, Brown C. Is routine drainage of pelvic anastomoses necessary? Am Surg 1996; 62: 452-458.

35. Wasey N, Baughan J, de Gara CJ. Prophylaxis in elective colorectal surgery: the cost of ignoring the

evidence. Can J Surg 2003: 46: 279-84.

36. Zografos GC, Simeonidis KM, Parasi AS, et al. Adhesion formation: intraperitoneal

catheters in practice. J Invest Surg 2002; 1 5: 37-43.

M M

Chapter 2 Drainage or non-drainage: systematic review

Su rgeons lack pred ictive accu racy for anastomotic leakage in gastrointest i nal surgery

A. Karl iczek1•2, N.J. Harlaar1,3, C.J. Zeebregts 1, T. Wiggers 1 , P.C. Baas2,

G.M. van Dam 1•3

' Department of Surgery, Un iversity Medical Center Gron ingen, G ron ingen, the Netherlands 2Department of Surgery, Mart in i Hospital G roningen, G roningen, the Netherlands

1BioOptical Imaging Center Groni ngen (BICG), Un iversity of Gron ingen, the Netherlands

Submitted

bstract

Background: The dramatic clinical consequences of anastomotic leakage in gastrointestinal surgery can be reduced by a diverting stoma or drainage of the peri-anastomotic area. Currently, the surgeons' clinical judgement is of major importance in decision making, but reliable data of the diagnostic accuracy are lacking. In this prospective clinical study the surgeons' predictive accuracy for anastomotic leakage was evaluated. Methods : In 247 patients undergoing gastrointestinal resection with anastomosis, the risk for anastomotic leakage was determined by the surgeon on the basis of a visual analogue scale. This risk-assessment was compared to the actual occurrence of anastomotic leakage post-operatively. Resu lts: A total of 34 ( 1 3.8%) patients showed anastomotic leakage. The surgeons' median predicted leakage rate was 7.8% (mean 9% SD 6. 1 %, sensitivity 47%, specificity 85%). Stratification of patients by anastomotic height did not improve the diagnostic accuracy of the surgeon in predicting anastomotic leakage. Diagnostic accuracy was not influenced by the surgeons' training level. Conclusion: The surgeons' clinical risk-assessment appeared to have a low predictive value for anastomotic leakage in gastrointestinal surgery. The low a-priori risk of anastomotic leakage of 14% resulted in a low post-test odds (48%) of correct prediction of anastomotic leakage. This warrants the ongoing search for a better diagnostic test of anastomotic leakage to prevent morbidity and mortality.

Introduction Gastrointesti nal anastomotic leakage causes i ncreased early mortality1 and, fol l owing potential ly curative resection of colorectal cancer, leads to h igher recu rrence rates and a poorer oncologic prognosis2 • Cl i n ical ly re levant anastomotic leakage rates range between 3% and 1 9%1 ,3-9 • Anastomotic leakage may remain local ised caus ing perianastomotic i nflammation or abscess formation, o r may progress to genera l ised periton it is . Several authors have demonstrated that a d iverting stoma, placed in the proximal colon or i l eum duri ng the i n it ial operat ion, does not prevent anastomotic leakage but clearly reduces the i ncidence of genera l ised peritonit is and thus reoperations , ICU-stay and mortal ity9- 1 1 • I n pe lvic anastomoses drai nage might reduce the consequences of anastomotic leakage 1 1 , s im i larly to oesophageal anastomosis with i n the thoracic cavity. Given these poss ib i l it ies to reduce the major consequences of anastomotic leakage during the i n it ial operat ion, the operat ing su rgeon might decide to construct a d ivert ing stoma in case of h igh ri sk for anastomotic leakage if determi ned by a rel iable pred ictive test.

So far l ittle is known about the accuracy of the c l i n ical pred ict ion of the ri s k of anastomotic leakage by the operating su rgeon . Three studies have shown that the surgeons ' g lobal assessment is a pred ictor of compl ications in general 1 2 , 1 3 and is even more accurate than ri sk assessment by POSSUM-scores 1 4 • Unt i l now, no stud ies have been publ ished eval uati ng the su rgeons ' c l i n ical j udgement on the ri s k of anastomotic leakage . The aim of the present study is to assess the accuracy of the surgeons ' "gut fee l ing" for the occurrence of anastomotic leakage on the bas i s of a visual analogue scale .

Methods Patients undergoing gastroi ntestinal resection in a large teach ing hospital and a Un ivers ity hospital between August 2006 and August 2007 were stud ied prospective ly. After complet ing the operation the su rgeon was asked to pred ict the ri sk of c l i n ical ly re levant anastomotic leakage on a visual analogue scale (VAS) as shown in figu re l .

Data on pre- and i ntraoperative ri sk factors for anastomotic leakage and postoperative compl ications were prospective ly col lected by review of patient fi les du ring hospital stay and after visit at the outpatient c l in ic (table l ) .

Risk of clinically relevant anastomotic leakage

0% 1 00%

Figure l Visual analogue scale for prediction of risk for anastomotic leakage to be marked

by the surgeon immediately after completing the operation.

37

38

n :::r f:lJ

(D ..... w

patient related

age gender diagnosis tumor stage diabetes pulmonary disease cardiovascular disease smoking weight loss body mass index corticosteroid use ASA level preoperative irradiation weight loss peritonitis bowel preparation

Table l

procedure related

intraoperative adverse events laparoscopic procedure type of anastomosis hand-sutured/ stapled anastomosis distance form anal verge mobilisation of splenic flexure drain near anastomosis defunctioning stoma duration of operation blood loss blood transfusion intraoperative hypotension intraoperative use of vasoactive drugs emergency operation educational level surgeon surgeon

Risk factors for anastomotic leakage analysed in this study.

Anastomotic leakage was cons idered to be present: i) when descri bed at re laparotomy or endoscopy, i i) when postoperative CT scan showed the presence of ai r- or fl u id col lections or an i nfi ltrate su rround ing the anastomosis , and in case of oesophageal anastomoses rad io logical leakage combined with c l i n ical s igns of med iasti nal or p leural i nflammation ( leucocytos i s, p leural effus ion and /or fever). I n oesophageal anastomoses a routine contrast study with a water soluble contrast agent was performed one week afte r surgery; in other anastomoses no routine test ing for leakage was performed. Fol low up on anastomotic leakage was conti nued u nt i l 3 months after the i n it ial operat ion or u nt i l d i scharge .

A total of 68 patients (2 1 .6%) were exc luded from fu rther analys i s (figure 2) . I n 1 3 (4 . 1 %) patients no anastomosis was performed, i n 5 3 ( 1 7. 5%) no r i sk pred iction VAS was fi l led out with i n 24 hours after the operation . These patients d id not d iffer in respect to the number of anastomotic leakages or i n general ri sk factors . Two cases were excluded as the i r records on postoperative compl ications were miss i ng. The remain ing 247 pat ients were al l i ncl uded i n the analys is .

Data were analyzed with Statistical Package for the Social Sciences software (SPSS 1 5-0, SPSS, Chicago, I L, USA). The data are presented as means (s.d.) un less i nd icated otherwise. Differences between categori cal variab les were tested with Pearson's X2 test. Differences between conti nuous variables were tested with Student's two-tai led test (normal d i stri bution) or Mann-Wh itney U test (skewed d i stri bution). The infl uence of pre- and i ntraoperative pat ient-re lated factors on the

occurrence of anastomotic leakage was i nvestigated by means of un ivariate analysis . Variables that had a d i rect i nfl uence on anastomotic leakage after un ivariate analys i s (p<0. 1 ) were entered i nto a mu ltivariate regress ion mode l and analysed i n a stepwise backward manner. P values <0.05 were regarded as statistical ly s i gn ifi cant.

Results A total of 247 pat ients was analysed (patient characteristics are shown in tabl e 2 and 3). Seven (3%) patients d ied during the study period. In fou r of these pat ients anastomotic leakage was present, th ree d ied of other causes. Th i rty-four pat ients ( 1 3 .8%) showed anastomotic leakage. In fou r cases, leakage occurred in gastric and oesophageal anastomoses ( 1 4% of a l l upper G I anastomoses), 9 ( 1 0%) in smal l bowe l , ascend ing and transverse colon anastomoses, 7 ( 1 4%) i n left colon anastom oses, and 1 4 ( 1 8%) i n anastomoses with i n 1 5 cm of the anal verge. I n 24 (77%) pat ients with anastomotic leakage a re-operat ion was performed. In 63% of the pat ients resect ions were performed for mal ignant d isease. Duration of ICU stay ( 1 sd 3 days vs 9 sd 1 3 days , p� 0.05) and hospital adm iss ion ( 1 2 sd 8 days vs 48 sd 46 days , p� 0.05) were s ign ificantly longer after anastomotic leakage. The operat ions were performed by a total of 32 d ifferent surgeons, e ither i n the role as an ass i stant surgeon or supervisor. If an ass istant surgeon was the first surgeon for a procedure

Eligible 315

8 Hartman's resection for diverticulitis 4 no anastomosis duet o low tumor and unfavourable t------.. patient factors

302

249

Inclusion 247

failure of stapling device, rupture of rectal stump

missing data on surgeons' risk estimation (VAS)

missing data on follow up

Figure 2 Flow chart of included patients. Bold numbers indicate patient numbers . Gray boxes ind icate excluded patients.

39

40

V'l C:

0 ::J V,

Table 2

mean age (sd) male nicotine abuse diabetes corticosteroid use cardiovascular disease pulmonary disease ASA level

I I I I l l IV

acute operation

level of anastomosis esophagus stomach small bowel extraperitoneal closure of stoma colon

ascending and transverse descending and sigmoid

rectum mean height in cm (sd)

n (% of total) 65 (1 5) 1 34 (54%) 55 (22%) 30 (1 2%) 1 9 (8%) 64 {26%) 45 ( 18%)

68 (27%) 1 45 (59%) 32 {1 3%) 1 {0.4%) 22 (9%)

1 8 1 0 1 6 1 1

59 90 43 9.3 {4.6)

Patient characteristics of included patients (n=247). The number of patients is shown, between brackets percentage of all patients.

( 1 1 1 /45%), th is was always supervised by a staff su rgeon. Of all 247 operat ions 1 86 (75%) were supervised by a gastroi ntesti nal su rgeon , and i n 97 (39%) operat ions a gastro i ntesti nal su rgeon was the fi rst su rgeon. There were no s ign ificant d iffe rences between su rgeons and assistant surgeons i n respect to i ncidence of anastomotic leakage and visual analogue scale score. The visual analogue scale values recorded by the operat ing su rgeon showed a med ian estimated probab i l ity for anastomotic leakage of 7.8% (mean 9.0% sd 6. 1 %). For anastomoses in the upper gastroi ntestinal tract the median estimated probabi l ity of leakage was 8 . 5% sd 5 . 5%; i n the smal l bowe l , ascend ing and transverse colon 4 .9% sd 4.8%; and i n the descend ing co lon , s ig moid and rectum 1 0.6% sd 6.4%. There was no d ifference i n estimated probabi l ity of leakage between low (i.e. with in � 1 5 cm from the anal verge) anastomoses and anastomoses i n the s igmoid and descend ing colon.

Patients with anastomoses i n the colon , s ig moid and rectum (n= 2 1 7) showed more often anastomotic leakage when a defu nction i ng stoma was constructed (6 / 3 3%), compared to patients without stoma (24 / 1 2%, p=0 .0 1 ). The estimated ri s k

n 247

anastomotic leakage no 21 3 (86%)

yes 34 ( 14%) wound infection or dehiscence

no 210 (85%)

yes 32 (1 3%)

missing 5 (2%) cardiac complications

ischemic 3 ( 1%) arrhythmias 7 (3%) decompensation 1 2 (5%) none 21 8 (88%) missing 7 (3%)

pulmonary complications none 1 99 (81%) pneumonia 1 8 (7%) pleural effusion 1 2 (5%) miscellaneous 8 (3%) missing 10 (4%)

Table 3 Postoperative complications. The number of patients is shown; between brackets percentage

of all patients.

by the surgeon for anastomotic leakage was 1 2 .0% (sd 7.8%) when a stoma was constructed vs 1 0. 7% (sd 6 . 1 %) without a stoma (p=0.43 7). I n pat ients with an anastomosis with in 1 5 cm of the anal verge (n= 78), 8 (29.6%) pat ients showed leakage after mobi l i sation of the splen ic flexure, compared to 6 ( 1 2 .0%) pat ients without mobi l i sation of the splenic flexure (p=0.06). No s ign ificant d ifference i n estimated risk for anastomotic leakage was seen when comparing mobi l i sation of the splen ic flexure vs no mobi l i sation ( 1 1 . 3% sd 7.0% vs 1 0.7% sd 5 . 9%, p=0. 59 1 ).

A mu ltiple regress ion analys is was performed on preoperative and i ntraoperative factors associated with anastomotic leakage (Table 4 and 5) . Durat ion of operation and tumour stage was associated with anastomotic leakage of anastomoses with in 1 5 cm of the anal verge . For anastomoses above 1 5 cm of the anal verge , mobi l i sation of splenic flexure , the use of i ntraoperative vasoactive drugs and preoperative i rrad iation were s ign ificantly associated with anastomotic leakage. The numbe_r of patient-re lated ri sk factors for pat ients without and with anastomotic leakage are depicted i n F igure 3 . Patients with an anastomotic leakage showed a higher rate in the number of ri sk factors compared to non-anastomotic leakage (5 . 7 sd 2.4 vs 4.6 sd 2 . 2 , p = 0.0 1 ) .

41

UJ

42

lJJ 1 2 3 4 5 6 7 8 9 10 11 12

number of risk factors

no leakage

anastomoses > 15 cm n= 149

tumor stage 2.5 (1 .0)

smoking 31 (22%)

weight loss 41 (34%)

body mass index (sd) 25.8 (94.4)

total preoperative risk factors (sd) 4.1 (2.0)

duration of operation in min 1 76 (1 1 5)

blood loss in ml 258 (323)

anastomoses <15 cm n:64

cardiovascular disease 13 (20%)

preoperative irradiation 20 (32%)

cm from anal verge (sd) 9.9 (4.5)

total preoperative risk factors (sd) 2.5 (1 .5)

blood loss in ml 397 (301 )

intraoperative vasoactive drugs 20 (34%)

mobilisation of splenic flexure 1 9 (30%)

defunctioning stoma 1 0 (1 6%)

Figure 3 The number of risk factors present in patients with ( l ight bars, mean 5 .7 sd 2 .4) and without (dark bars, mean 4.6 sd 2.2) anastomotic leakage are depicted (p = 0.01 ) .

leakage p

n:20

3.2 (0.6) 0.02

8 (40%) 0.08

9 (56%) 0.08

23.5 (4.6) 0.04

5.0 (2.4) 0.06

243 (1 36) 0.02

446 (41 2) 0.03

n:14

6 (43%) 0.08

9 (64%) 0.02

6.9 (4.5) 0.03

3.2 (1 . 1 ) 0.07

583 (502) 0.09

8 (67%) 0.04

8 (57%) 0.06

5 (36%) 0.09

Table 4 Factors associated with anastomotic leakage at un ivariate analysis (n=247). Numbers are depicted as mean (sd). Percentages between brackets indicate the percentage of al l patients with or without leakage

Anastomoses > 15 cm (n=169) B 95% CI p

duration of operation 0.1 7 0.04-0.31 0.01

tumor stage 0.07 0.01 -0.1 2 0.04

Anastomoses :S 15 cm (n=78) B 95% CI p

mobilisation of splenic flexure 0.32 0.1 4-0.49 <0.01

intraoperative vasoactive drugs 0.23 0.06-0.40 0.01

preoperative irradiation 0.42 0.02-0.27 0.02

Table 5 Independent predictors of anastomotic leakage analysed by multiple regression analysis (stepwise analysis).

The accuracy of the surgeons' estimation of risk for anastomotic leakage is shown in a receiver-operator characteristics (ROC)-curve (Figure 4). When considering the median (7.8%) as the cut-off point for elevated risk in the surgeons' prediction of anastomotic leakage, sensitivity was 47% and specificity 8 5% for an accurate risk assessment of anastomotic leakage.

1 ,0

0,8

f0,6

� C Ql en

0,4

0,2

, , , ,

, , ,

, , ,

, , ,

, , ,

, , ,

o,o--+----..------r---�----.----r-�

0,0 0,2 0,4 0,6

1 - Specificity 0,8 1 ,0

Figure 4 Receiver operator characteristics curve of probability estimation of anastomotic leakage by the surgeon for all anastomoses.

43

w

44

Discussion In this study we evaluated the accuracy of the surgeons' judgement in

prediction of anastomotic leakage in upper- as well as lower gastroinstestinal anastomoses. The clinical judgement of the operating surgeon in our study appeared to have a sensitivity of 47% and specificity of 85% for all anastomoses. When taking into account the relatively low exposure to anastomotic leakage of individual surgeons (i.e. a mean of 1 3.8% in this study) it might not be surprising that anastomotic leakage is difficult to predict (i.e. having a low a-priori risk), thus resulting in a low post-test odds (48%) of correctly predicting anastomotic leakage. Nevertheless, the surgeon seems to predict reasonably well non-anastomotic leakage (specificity 85%). The results of our study clearly identify the lack of a reliable intraoperative predictive test for anastomotic leakage by the operating surgeon, whereas several studies identify a large number of risk factors for anastomotic leakage. In general, the global, clinical judgement seems to localize a subset of patients at risk for developing complications in general, whereas many patients with no risk factors at all may develop anastomotic leakage 1 2 •

The risk factors for anastomotic leakage identified by univariate and multivariate analysis in our study corroborate with those found in other studies 1 -8• Similar to these studies, the number of risk factors present in an individual patient appeared to be an important predictor of anastomotic leakage in all anastomoses. This finding was previously described in two large studies 1 , 1 5 • Makela et al. described leakage rates of 76% in patients with three risk factors, 87% with four risk factors and 100% in patients with five risk factors 1 5 • In Alves' study leakage rates increased from 38% when two risk factors were present to 50% in the presence of three risk factors 1 • We found a similar increase in risk of anastomotic leakage with increasing risk factors, although the risk did not increase linearly. Furthermore, data on the weight of individual risk factors and the total amount of risk factors considered relevant are not readily available in the other two studies addressing this issue.

When considering individual risk factors for anastomotic leakage, it should be emphasized that construction of a defunctioning stoma and mobilisation of the splenic flexure might be considered a risk factor, as well as a measure to prevent anastomotic leakage. When conducting separate analysis for leakage rate and risk estimation, we found significantly higher leakage rates after mobilisation of the splenic flexure and construction of a stoma, showing that anastomotic leakage is not prevented by these measures. Moreover, the surgeons did not report a higher estimated risk for leakage in these patients on the basis of the visual analogue scale. This might point towards an underestimation of the leakage rate by the surgeon after mobilisation of the splenic flexure and construction of a defunctioning stoma. Another explanation might be that these procedures are mainly carried out in anastomoses within 1 5 cm of the anal verge and therefore present an epiphenomenon instead of a true causative relationship.

Markus et al. evaluated the clinical judgement of the operating surgeon (denomina�ed as 'gut feeling') as a predictor for general postoperative mortality and morbidity in large surgical interventions, comparing clinical judgement and prediction by POSSUM scores to actual outcome 1 4. In this study the surgeons' general clinical judgement was more accurate than POSSUM scores in upper gastrointestinal and hepatobiliary procedures, but not in colorectal procedures. Pettigrew et al. compared the predictive value of the judgement of the operating surgeon for development of postoperative complications based on a global assessment of the patient, with the assessment of an independent physician 45 performing a complete physical examination and with several anthropometric indices 1 3. In a population of 218 patients undergoing major gastrointestinal surgery they reported that global risk assessment by the operating surgeon is less accurate (sensitivity 32%, specificity 83%) than careful assessment of in particular cardiorespiratory disease and nutritional status of the patient. In a third study using a similar design 1 2, but evaluating the surgeons' judgement before and after surgery as well, they concluded the surgeons' postoperative judgement to be superior to the surgeons' preoperative judgement or assessment of either preoperative patient factors or systematic clinical factors. Based on this striking finding they concluded that surgical performance during the operation is an important predictor of postoperative complications. Considering this latter study 1 2 it might be assumed that the surgeon is able to reliably predict the risk of anastomotic leakage, but as mentioned by Markus et al. 1 4 they found the surgeons' risk estimation after completion of the operation not to be a good predictor of postoperative complications, particularly in colorectal surgery. One should be aware that none of the aforementioned studies addresses anastomotic leakage as primary endpoint in particular.

In conclusion, this prospective study has shown that global clinical risk assessment of anastomotic leakage by the operating surgeon has low predictive value for anastomotic leakage, clearly supporting the need for the development of a more reliable predictive test. Such a test should be carried out intraoperatively, real­time, easy to acquire, to interpret and to act on. Moreover, the test should have a high sensitivity for anastomotic leakage because of the detrimental effect of missing a leak. When taking the possible pathophysiological mechanism of anastomotic leakage into account1 6, probably a diagnostic test which measures microperfusion before and after creation of an anastomosis will be most suitable. At out center we are developing optical imaging techniques such as visible light spectroscopy1 1, 1 8

and multispectral fluorescence imaging techniques1 9 to evaluate realtime microperfusion in gastrointestinal surgery. Future (pre)clinical studies will elucidate the value of these techniques.

w

l./") C:

0 ::I II'

References

l . Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic leakage after large bowel resection: multivariate analysis of 707 patients. World J Surg 2002; 26: 499-5 02.

2. Ptok H , Marusch F, Meyer F, et al. Impact of anastomotic leakage on oncological outcome after rectal cancer resection. Br J Surg 2007; 94: 1548- 15 54.

3. Eriksen MT, Wibe A , Norstein J, et al. ; Norwegian Rectal Cancer Group. Anastomotic leakage following routine m esorectal excision for rectal cancer in a national cohort of patients. Colorectal Dis 200 5 ; 7: 5 1- 5 7.

46 4. Karl RC, Schreiber R, Boulware D, et al. Factors affecting morbidity, mortality and survival in

w

patients undergoing Ivor Lewis esophagogastrectomy. Ann Surg 2000; 5: 635-643. 5. Yiklund P, Lindblad M, Lu M, et al. Risk factors for complications after esophageal cancer

resection: a prospective population-based study. Ann Surg 2006; 243: 204-21 1. 6. Yignali A, Fazio VW, Lavery IC, et al. Factors associated with the occu rrence of leaks in stapled

rectal anastomoses: a review of 10 14 patients. J Am Coll Surg l 997;185 :113- 121. 7. Matthiessen P, Hallbook 0, Andersson M, et al. Risk factors for anastomotic leakage after

anterior resection of the rectum. Colorectal Dis, 2004; 6 : 462-469. 8. Katory M, Tang CL, Koh WL, Fook-Chong SMC, Loi TT, Ooi BS, Ho KS, Eu KW. A 6-year review

of surgical morbidity and oncological outcome after high anterior resection for colorectal malignancy with and without splenic mobilisation. Colorectal Dis, 2008; l 0: 165-9.

9. Matthiessen P, Hallbook O, Rutegard J, Simert G, Sjodahl R.,_ Defunctioning stoma reduces symptomatic anastomotic leakage after low anterior resection of the rectum for cancer: a randomized m u lticenter trial . Ann Surg , 2007; 246: 207-214.

l 0. Gastinger I, Marusch F, Steinert R, Wolff S, Koeckerling F, Lippert H; Working Group 'Colon/ Rectum Carcinoma'. Protective defunctioning stoma in low anterior resection for rectal carcinoma. Br J Surg , 2005 ; 92: 1 137- 1 142.

1 1 . Peeters KC, Tollenaar RA, Marijnen CA, Klein Kranenbarg E, Steup WH , Wiggers T, Rutten HJ, van de Velde CJ; Dutch Colorectal Cancer Group. Risk factors for anastomotic failure after total mesorectal excision of rectal cancer. Br J Surg , 200 5 ; 92: 211-216.

l 2. Pettigrew RA, Burns HJG, Carter DC. Evaluating surgical risk : the importance of technical factors in determining outcome. Br J Surg , 1987; 74 : 791-794.

l 3. Pettigrew RA, Hill GL. Indicators of surgical risk and clinical j udgement. Br J Surg, 1986; 73: 47-51.

14. Markus PM, Marte l l J, Leister I , Horstmann 0, Brinker J, Becker H. Predicting postoperative morbidity by clinical assessment. Br J Surg , 200 5 ; 92: 101-106.

l 5. Makela JT, Kiviniemi H , Laitinen S. Risk factors for anastomotic leakage after left sided colorectal resection with rectal anastomosis. Dis Colon Rectum, 2003; 46: 653-660.

16. Stokes KY, Granger DN. The microcirculation: a motor for the systemic inflammatory response and large vessel disease induced by hypercholesterolaemia? J Physiol 200 5 ; 562: 647-653.

l 7. Benaron DA, Parachikov IH, Friedland S , et al. Continuous, noninvasive and localised microvascular tissue oximetry using visible light spectroscopy. Anesthesiology 2004; l 00: 1469-1475.

l 8. Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosa! perfusion in the gastrointestinal tract. Gastrointest Endosc Clin N Am 2004; 14: 5 39- 55 5.

19. Weissleder R, Pittet MJ. Imaging in the era of molecular oncology. Nature 2008; 452: 580-589.

t Chapter 3 Surgeons lack predictive accuracy

l ntraoperative ischemia of the d i stal end of co lon

anastomoses as detected with vi s i ble l ig ht

spectroscopy causes reduction of anastomotic stre ngth

A. Karl iczekl , D.A. Benaron2 , C.J . Zeebregts 1 , T. Wiggers 1 , G.M. van Dam 1 ,3

1 Department of Surgery, Un iversity Medical Center Gron ingen, Gron ingen, The Netherlands 2 Stanford University School of Medicine, Palo Alto, CA, USA

3 BioOptical Imaging Center Groningen (BICG), University of Groningen, The Netherlands

J Surg Res 2008, E-pub May 8

Background: To explore new methods for i ntraoperative eval uation of t issue oxygenation we evaluated the use of v is ib le l ight spectroscopy (VLS) as a pred ictor of anastomotic strength in an experimental model with ischemic murine colon anastomoses. Methods: Male rats (n =32) were divided i nto two groups ( ischemia and non­ischemia). I n the ischem ia-group the arteries of the distal colon were l igated u nt i l t issue oxygen satu ration (StO2) dropped below 5 5%. A segment of the proximal part of the colon was resected unt i l a we l l perfused area was reached and an anastomosis was performed. I n the non-ischemia g roup, resection of a segment of descend ing colon and a co lon anastomosis was performed. The animals were sacrificed on the third or seventh postoperative day. The anastomosis was tested for bursting pressu re and breaki ng strength. Results: After l igation of the re levant mesenteric arteries StO2 of the d istal part of the colon decreased (54 .6% sd 6 .4% vs 7 1 .2% sd 7.4%, p ::s 0 .05). On the th i rd and seventh postoperative day StO2 had normal i sed . Ad hesion score in the ischemia group was h igher compared to the non-ischemia group ( 1 .6 vs 0 .4 , p::s 0.05) . There were no d ifferences i n bursti ng pressure between both g roups. Breaking strength was l ower in the ischemia group on the th ird postoperative day ( 1 69.8 sd 44.2 vs 2 1 2.6 sd 4 1 . 2 , p ::s 0 .05). Conclus ion : lschemia can i ntraoperatively accu rately be detected by VLS in a murine model . Partial ly ischemic anastomoses showed more adhesions and d im in ished breaking strength in the early phase of heal i ng , whereas bursting pressure was not affected . Low StO2 of a d istal colon anastomosis appeared to be a ri sk factor for anastomotic deh iscence at day 3 and beyond.

I ntroduction

I n gastrointestinal surgery leakage of anastomosis is a chal leng ing problem because of its strong re lation with raised mortal ity and morbid ity1 ,2 ,3. Several general ris k factors for anastomotic leakage have been identified i n c l i n ical studies. Many of these factors suggest local hypoperfus ion to be the main cause of local ische mia and thereby of anastomotic dehiscence. Furthermore, earl ier experimental stud ies have al ready demonstrated that profound i schem ia 4,5 and in particular reperfus ion 6-8 c learly have detrimental effects on anastomotic heal i ng. I n hu man co lorectal anastomosis compromised vascu lar supply of the d istal part of the anastomos is causes major concern. Particu larly i n pelvic anastomosis , where b lood supply to the distal stump of the rectum is compromised by, for i n stance, a total mesenteri c excis ion, ischemia may play an important role i n the development of anastom otic dehiscence9 • Th is fi nd ing , i n wh ich on ly one part of the anastomosis i s ischemic rather than general ised ischem ia of the colon inc lud ing both s ides of the anastomosis , is probably more representative of the hu man intraoperative situat ion . To our knowledge th is phenomenon has not been studied before in an an i mal model.

I n cu rrent dai ly practice i ntraoperative i ntesti nal m icroci rcu lation is assessed by the su rgeon by means of subjective judgement of color and pu l sat i le blood flow of the i ntesti nal marg i ns at the d issected ends of the bowe l segments. As such, th i s practice i s cons idered the go ld standard to determ ine viab i l ity of both bowel ends . There are, however, no sound data avai lable on sens itivity and specific ity, nor whether th is j udgement is re l iable and accu rate. Several studies have been conducted i n order to i nvestigate m ucosa! perfus ion , inc lud ing Dopp ler u ltrasound1 0 , laser Doppler flowmetry1 1 , 1 2 , scann ing laser flowmetry1 3 , 1 4 , fl uorescence videography 1 s and i ntramucosal pH 1 6, 1 7. However, none of these methods i s widely accepted due to methodological or practical drawbacks 1 s-20.

Recently an innovative ischem ia detection device , known as vis i b le l ight spectroscopy (VLS) ox imeter, has been deve loped 2 1 . Th is system uses the princ i p le of absorbance and scatteri ng of l ight i n t issues. I n contrast to near- infrared spectroscopy (NIRS) and pu lse oximetry in VLS oximetry shal low-penetrati ng vis i b le l ight i s used to measure haemog lobi n oxygen satu ration i n th i n , smal l t issue vol umes (denominated in St02). As oxygenated and deoxygenated haemog lob in have a d ifferent absorption spectrum and haemog lobin i s cons idered a critical derivative of perfus ion , t issue m icroperfus ion can be assessed by us ing the VLS oximetry tech n ique that detects these d ifferences. The abundant haemog lob in i n gastroi ntesti nal capi l lary beds accounts for the majority of absorption i n the mucosa! layer. A major advantage of vis i ble l ight is that it penetrates capi l laries we l l , bu t not larger vesse ls , resu lt ing i n a cap i l lary-weighted measurement in vivo, most closely correlated to ti ssue oxygen saturat ion, St02 . In previous stud ies 22-25 vis i b le l ight spectroscopy has been val idated for the detection of t issue ischemia i n human gastric and co lonic m ucosa and might therefore be a usefu l i nstrument for

51

prediction of anastomotic perfus ion and anastomotic strength . We therefore designed th is study to i nvestigate the value of v is ib le l i ght spectroscopy as a predictor of anastomotic heal ing of partial ly i schemic colon anastomosis. I n order to obtain a c l i n ical ly relevant model with regard to the s ituation in low colorectal anastomos is , where it is hypothes ized that the distal stu mp is mi ldly ischem ic with adequate perfus ion of the proxi mal part of the anastomos i s , we induced m i ld i schemia i n the d istal part of the anastomosis.

s2 Material and methods

V, n ::r ro 3 s:u :::l ro ><

-0 ro

3 ro :::l

s:u :::l s:u V, 8 3 0 V, ro V,

Overview of study design

For detection of a 2 0% d ifference i n t issue satu ration (StO2) between ischemic and non-ischemic anastomos is , a total sample of 34 animals was needed (oc=0.05 , �=0.9 5). Rats were randomly d ivided i nto two exper imental g roups (ischemia and non-ischemia). In the ischemia group (n = l 5) an anastomosis was performed after i nduction of ischemia in the d istal part of the colon. In the non-ischemia g roup (n = l 7 ) an anastomosis was performed obvious ly without provoking ischemia. The rats of both groups were term inated after laparotomy on the th i rd or seventh postoperative day. In both g roups the t issue saturat ion (StO2) of the colon serosa was measu red at each laparotomy. After termination , c l i n ical signs of anastomotic leakage were noted , the anastomosis was resected and tested for bursti ng pressure and breaking strength.

Animal subjects

Male Wistar rats (n = 3 2 , mean weight 344 g ; Harlan BV, Horst, The Netherlands) were al lowed to adapt to laboratory conditions for one week before experimental use. The animals were housed two per cage u nder specified pathogen-free conditions and had free access to water and standard rodent food (AB d iets , Woerden, The Netherlands). After laparotomy the an imals were housed in separate cages unt i l passage of stool . The study protocol was approved by the Animal Eth ics Committee of the Un ivers ity of Gron i ngen .

Technical detai ls of measurements and operative procedures Al l rats were anesthetised with a mixture of i sofl u rane 2 . 5% and oxygen and received 0.03 mg/ kg buprenorfi ne s .c. after induction . The abdomen was shaved in the m id l i ne and d is infected. Duri ng the operation , body temperatu re was maintai ned at 38 ° C us i ng a heati ng pad. The operation was performed under semi-steri le condit ions. After m id l i ne laparotomy the descend i ng co lon was local i sed and a vis ible l ight spectroscopy record ing was performed at the anticipated s ite of the anastomos is. To measure StO2 , the probe was held a few m i l l imetres above gauze cleaned serosal su rface u nti l val ues stabi l i zed within a 5% marg i n.

In the ischemia-group, the prox imal marg inal artery (C l ) , the d istal marg inal artery (C2) and the s igmoid artery (C3) were identified and l igated unti l StO2 val ues dropped below 5 5%. Next 1 - 1 . 5 cm of colon was resected, assu ring the d istal part of the colon to remain i schemic, the proximal part however to be wel l oxygenated and thus a wel l perfused area was reached. In the non-ischemia g roup the descend ing colon was transsected with care not to damage the vascu lature and a 1 -1 . 5 cm segment was resected. In both the non- ischemia and the ischemia g roup continu ity was restored by construction of an end-to-end anastomosis us ing 8- 1 0 inverting, interrupted sutures (Prolene™ , 7-0, Eth i con, Amersfoort, The Netherlands). 53 The anastomosis was p laced 2 cm prox imal to the peritoneal reflection. After construction of the anastomosis , again StO2 of the colon proximal and d istal to the anastomosis was measured. StO2 of caecum serosa served as reference measurement. In a recent human study conducted at our center, th is reference value was found to be representative of normal val ues in the rectum (data not shown). The abdomen was closed us ing a continuous resorbable sutu re (Polysorb™ 4-0, Tyco Healthcare, Gosport, UK) for the m usculofascial layer and the skin was closed with an intracutaneous runn ing resorbable s uture (Monocryl™ 5 -0 , Eth icon , Amersfoort, The Netherlands). To compensate for fl u id loss l 0 m l 0 .9% NaCl was administered subcutaneously. Al l operative procedures were performed by one investigator (AK) .

The first 24 hours after operation 0.03 mg/kg buprenorfine s.c. was administered twice dai ly. An imals were weighed dai ly and observed for activity l eve l and signs of i l lness. At sacrifice on day 3 or 7 an imals were fi rst anesthetized with a mixture of isoflu rane 2 . 5% and oxygen and buprenorfine 0.03 mg / kg s.c. , placed on a heating pad maintain ing body tem perature at 38 ° C and relaparotomy was performed. StO2 of the colon proximal and d istal of the anastomosis and the caecum was measured, taking care not to damage any adhesions to the anastomosis . The abdominal cavity was explored for s igns of i leus and deh iscence of the anastomosis. The an imals were then terminated by intracard iac pentobarbital injection.

For VLS measurements a system cons isting of a vis ib le l ight spectroscopy (VLS) oximeter (T-State , model 303 lschemia Detection System ; Spectros Corp. , Portola Val ley, CA, USA) was used. Detai ls and descri ptions of this system have been descri bed previous ly 2 1 -23 . In short, vis i ble l ight spectroscopy uses the princip le of absorbance and scattering of l ight in b io log ical t issues. Each type of molecu le has a characteristic absorbance spectrum. As oxygenated and deoxygenated haemoglob in have a d ifferent absorption spectrum and haemoglobin is cons idered a critical derivative of perfus ion , t issue m icroperfus ion can be assessed us ing a techn ique that detects these d ifferences. In contrast to laser Doppler or pu lse oximetry, in vis ib le l ight spectroscopy no pu lsati le flow is needed.

lnterobserver variabi l ity In order to eval uate measurement stabi l ity, in 5 consecutive an imals of the non­ischemia g roup interobserver variabi l ity of StO2 measurements during laparotomy

V> r, ::, (I)

3 pj. :::s

(I) X

"Cl (I) ..... 3 (I) :::s r1'

llJ :::s llJ V>

8 3 0 V'I (1) V>

prior to termination was determ i ned . Two diffe rent observers performed 3 measurements in each animal at the coecum, proximal and d istal to the anastomos is . Both observers were b l i nded to the resu lts of the other observer.

Anastomotic strength Anastomotic strength was eval uated by measuri ng res istance to i ntral um inal pressure (bursting pressu re) and to long itudi nal forces (breaki ng strength) by us ing wel l descri bed methods26-29 . Anastomotic strength i n rat co lon i s weakest around

54 the th i rd postoperative day due to remode l ing of col lagen fi bers , which resu lts i n decreas ing bursting pressure and breaking strength 26 ,30. After 5 to 6 days postoperatively, bu rsting pressure i ncreases and is considered to be maximal .

VI n :r­m

3 PJ � ro ><

"'O

�-3 ro �

Jl.l � Jl.l VI

8 3 0 VI (0 VI

Breaki ng strength, however, is regai ned much s lower i n the course of several weeks26,30,3 1 . I n this study we tested anastomoses on the th i rd and seventh postoperative day for bursti ng pressure and breaking strength, as i n the c l i n ical s ituation anastomotic leakage most frequently occurs i n the fi rst postoperative week.

After termination of the ani mal s on the th ird or seventh postoperative day the desce nd ing colon with the anastomosis was resected. Adhesions and ad hes ive structures to the anastomoses were carefu l ly left in place and resected together with the colon segment. Severity of adhesions was class ified as none, m i ld , moderate and severe as previous ly descri bed by van der Ham et al.4 • l ntral uminal faeces were removed .

Subsequently, al l co lon segments were tied off at the d istal end . A cann u la was inserted i n the prox imal end and the colon was tied around it. The can nu la was connected to an i nfusion pump and a manometer and i nfused at a rate of l m l /m in with methylene-blue d issolved i n sal i ne . The ri s ing pressure was recorded g raph ica l ly and the bursti ng pressu re (in m i l l imetres of mercu ry, (mmHg)) was determi ned and defined as the maximal pressure the anastomosis or co lon segment res isted. The site of leakage was recorded. Thereafter, the seg ment was placed in a tens iometer system (Mecmesin AFG l 0, Horsham, UK). A constantly i ncreas ing force was appl ied and the l i near breaking strength was recorded . Th is procedure has been previous ly descri bed and val idated. 32

Statistics Data were analyzed with the Stati stical Package for the Social Sciences software (SPSS l 5-0, SPSS, Ch icago, I L, USA) . Statistical d ifferences between g roups were tested by Student T-test (normal d i stri bution), Mann Whitney U- test (skewed d i stri but ion) or Chi-squared test when appropriate . Stat istical s ign ificance was accepted at and be low the 5% leve l.

* 1 1 0 *

*

1 05 ·a;

0

cu

:s � 0

�

1 00

95

90

0

Results General outcome

I

2 ':I �, 4 5 6 7

days after laparotomy

Figure l Change of body weight after laparotomy. Weight loss is expressed as percentage of preoperative body weight. Animals i n the anastomosis group are represented by the dark l i ne, the ischemia group by the grey l ine. The error bars represent standard deviation. Statistical sign ificant differences (p=,:; 0.05 , t-test) are indicated by an asterisk (*).

All rats showed d im in i shed activity on the fi rst postoperative day. In a l l rats a transient period of weight loss was observed , as shown i n Figure l . Rats in the ischem ia g roup showed s ign ificant more we ight loss and a s lower regain of weight compared to the non-ischemia group.

I n the ischemia group 3 an imals , schedu led to be term inated on day 7 , were euthan ized on day 3 due to lack of faecal product ion after construction of the anastomosis ind icati ng the presence of a pers istent bowel obstruction . On laparotomy severe i leus was present but no clear evidence of anastomotic leakage was seen in any of these th ree animals . In the non-ischemia g roup two an imals d ied duri ng recovery i mmediate ly after laparotomy. Necropsy was performed , but no clear cause of death was identified. These two an imals were excluded from fu rther analys i s .

Tissue oxygenation Vis ible l ight spectroscopy measurements were easy to perform and took 5 to l 0

seconds per measurement. Mean St02 after laparotomy and before i ntervent ion to the colon or its feed ing arteries was measured and was s ign ificantly lower compared to the reference St02 measured at the leve l of the caecum (67.0% sd 7.2% vs 71 .2% sd 6 .6%, p � 0 .05) .

I n the i schemia g roup, after success ive l i gation of the prox i mal marg i nal artery (C l ) , the d istal marg i nal artery (C2) and the s igmoidal artery (C3) vis i b le l i ght

55

V'I ("\ ;;;; (I)

3 iii" :::s

(D X

"O (I) ...., 3· (I) :::s

tlJ :::s tlJ V'I

8 3 0 V'I (I) V'I

56

l/l (") :::, (I) 3 pj" ::l (I) X

"O (I) -, 3 ::l

PJ ::l PJ l/l 8 3 0 V, (I) l/l

StOz 90

70 66.8

40

RO co C1

Non- lschemia

C2 C3

Figure 2 Mean StO2 and data d istribution are shown for T-Stat measurements during induction of i schemia by l igation of mesenteric vessels. RO: Reference StO2 at coecum, CO: StO2 after laparotomy, Cl : StO2 after ligation of proximal marginal artery, C2: StO2 after ligation of distal marginal artery, C3: StO2 after ligation of sigmoid artery.

Denotes a significant difference (p:s;0.05, t-test). Boxes represent first and third quartile with the median as a central l ine. The whiskers represent 5th and 951h percenti les.

lschemia

Figure 3 Mean StO2 (standard deviation) in non-ischemic colon and after induction of ischemia by ligation of mesenterial arteries. R is the reference-measurement performed on the coecum, P depicts StO2 of the proximal part of the anastomosis and D of the d istal part. * Denotes significant difference between groups ((p:s; 0.05, t-test).

100

80

60

40

Figure 4

0

RO RJ Rf PO PJ P'7

Non-isdlemia

DO DJ Df RO RJ Rf PO PJ P'7

lsehemia

DO DJ Df

Results for non-ischemic and ischemic anastomosis are shown. RO indicates measurements on the serosa of the caecum on day 0 , considered Reference measurements. R3 shows reference measurements of animals kil led 3 days postoperative, R7 on day 7. PO, P3 and P7 show St02 proximal of the anastomosis on day O and at termination on day 3 or 7. DO, D3, and D7 show St02 distal of the anastomosis on day O and at termination on day 3 or 7. * Denotes s ign ificant difference (ps 0.05, t-test). Boxes represent first and third quartile with the median as a central l ine. The whiskers represent 5th and 95th percentiles. 0 depicts an outl ier (value of 53 .8%).

spectroscopy measurements were performed. After ligation of the three vessels, mean St02 had dropped significantly (71 .2% sd 7.4% vs 54.6% sd 6.4%, p � 0.05) as shown in Figure 2. In the non-ischemia group St02 was measured as described previously. No changes in St02 after construction of the anastomosis occurred.

The results of St02 measurements after resection of a colon segment and construction of an anastomosis in both the ischemia and the non-ischemia group are shown in Figure 3. When compared to the non-ischemia group St02 in the colon distal to the anastomosis was significantly lower in the ischemia group (67.2% sd 7.4% vs 5 3.2% sd 14.8%). St02 of the caecum and the proximal colon remained unchanged in both groups.

Animals were terminated according to protocol on the third or seventh postoperative day. Before termination laparotomy was performed and St02 was measured proximal and distal to the anastomosis and at the caecum. The results of these measurements are shown in Figure 4. In the ischemia group St02 in the colon distal to the anastomosis was significantly lower at the fi rst operation, but normalized at termination on day 3 or day 7. Reference St02 on the third postoperative day seemed higher compared to the initial laparotomy, but this change was not sign ificant. Compared to the recordings during the first laparotomy no significant changes in mean St02 in any part of the colon were seen in the non­ischemia group. St02 of the caecum (reference value) was stable in all measurements. Overall interobserver variability was 1.2% of St02 values when

57

n ::r (D 3 pj' :l (D >< -a �-3 (D :l ,... e:. PJ ::l PJ V\ ,... 0 3 0 V, (D V,

58

n :r PJ

(D ...,

VI l"'I :r (D

3 pj' :l

(D ><

-0 (D ..., 3 (D :l

PJ :l PJ VI

0 3 0 VI (0 VI

Table l

site observer N mean St02 (%) sd

reference A 5 69.4 6.8

B 5 65.0 (94%) 3.9 n.s. proximal A 5 70.4 7.6

B 5 68.2 (97%) 7.6 n.s distal A 5 63.0 4.8

B 5 66.0 (1 05%) 2.1 n.s.

overall A 1 5 67.6 6.9

B 1 5 66.4 (98%) 4.9 n.s.

lnterobserver variab i l ity of St02 measurements performed by 2 different observers (A and B) at different measurement sites (reference measurement on the caecum, colon proximal and distal of the anastomosis) and overall is shown . Number of measurements (N), mean St02, between brackets the difference compared to observer A, and standard deviation are ind icated. n.s . is no statistically s ign ificant difference.

St02 > 61%

St02 < 61 %

total

no complications

6

2

8

complications

7

1 5

22

Table 2 Number of an imals with anastomotic complications (bursting pressure and/or breaking strength below mean and /or earl ier termination because of i leus) occurring after h igh (> mean- sd: 6 1%) and low ( < mean - sd) St02 during construction of the anastomosis .

expressed in abso l ute n umbers (Table l ) , resu lt ing i n 2% d ifference between two observers (not s ign ificant, p=0. 587) . The highest d ifference was fou nd among reference measu rements (4 .4% StO2, p=0.2 51 ) .

As shown i n Table 2 i n e ight an imals no anastomotic comp l ications occurred, which vis ible l ight spectroscopy predicted correctly by StO2> 6 1 % (= mean-sd in the d istal colon at construct ion of the anastomos is) in 6 cases. Twenty-two animals showed signs of anastomotic compl ications (burst ing pressure and/or breaki ng strength be low mean and /or earl ier termination because of i leus) and was predicted by VLS correctly in l 5 cases. Therefore , sens itivity of vi s i b le l i ght spectroscopy for detection of compl ications due to ischemia in th is model is 68% and specificity 75% with an overal l accuracy of 73%.

Anastomotic strength

At termination e leven ani mals showed s igns of i leus (7 in the non-ischemia g roup, 4 i n the ischemia g roup), but the d ifference between g roups was not s ign ificant.

250 600

200 500

§ .S.150 � 0

i4oo C

� C

r100 � 300

50

6

r--;--, ai

0 200

0 1 00

3 7 3 7

Figure 5 Bursting pressure and breaking strength of the resected anastomosis after termination on day 3 postoperative (3) and day 7 (7). lschemic anastomosis are shown in l ight-coloured bars, non­ischemic anastomosis in dark. Significant (p s 0.05, Mann-Whitney-U test) d ifferences between groups are indicated (*). Boxes represent first and third quartile with the median as a central l ine. The whiskers represent 5th and 95th percentiles. O depicts an outlier (value of 1 37 mm Hg).

Adhesion formation was s ign ificantly more severe in the ischemia group (adhesion score 1 .6 vs 0.4, p:$ 0.05).

Bu rsting pressure s ign ificantly i ncreased in both the ischem ia and the non­ischem ia g roup from the th ird to seventh postoperative day (39.8 sd 52. 1 mmHg / 20.3 sd 9.4 to 1 88.4 sd 2 9.2 mmHg / 1 73. 1 sd 2 1 . l , p:$ 0.05), as shown in figure 5 . Between the i schemia and non-ischemia g roup there was no s ign ifi cant d ifference at both t ime poi nts measured. The bursti ng s ite was most frequently fou nd at the level of the anastomosis ( 1 8 cases, 60%), i n 1 0 (3 3%) cases the burst was proximal to the anastomosis and 2 (7%) d i stal to the anastomosis. Again , no s ign ificant d ifferences between the ischemia and the non-ischemia g roup cou ld be detected.

Breaking strength i ncreased s ign ificantly in both the ischem ia and the non­ischemia g roup from the th ird to seventh postoperative day ( 1 62 .3 sd 47. 3 / 2 1 2 .6 sd 4 1 .2 to 327 . 1 sd 1 1 7.0/ 294.4 sd 43 .0 , p:$ 0.0 5). On day 3, however, anastomoses i n the ischem ia g roup showed s ign ificantly (p:$ 0 .05) lower break ing strength as shown in Figu re 5 . The most frequent s ite of d isruption was at the level of the anastomosis (24 cases, 80%), wh i l st in 4 ( 1 3%) cases the d i stal colon and i n 2(7%) cases t he proximal colon d isrupted (Table 3). With regard to the s ite of d isruption no s ign ificant d ifferences between the two groups were found. No s ign ificant corre lations between St02 and bursting pressure or breaking strength were fou nd.

59

V'I n ::, (D

3 PJ

:::l

(D ><

-0 (D ::::!. 3 (D :::l

g_ PJ :::l PJ V'I

8 3 0 V'I (D V'I

60

l/l '"' ::::r (D

3 pj' :::i

(D X

"O (D ..., (D

PJ :::i PJ V,

8 3 0 l/l (D l/l

% disrupture at anastomosis

day bursting pressure breaking strength ischemia 3 1 00 1 00

7 29 72

non-ischemia 3 1 00 1 00

7 1 3 2 5

Table 3 Percentage of anastomoses d isrupting at the anastomosis at the th i rd and seventh postoperative day.

Discuss ion

I n the present study we demonstrated that v is i ble l ight spectroscopy us ing the T­Stat l schemia Detection System can detect St02 variat ions i n the colon serosa that corre l ate wel l with the provoked ischemia through l igation of relevant supplying arteries. To thi s context a s ign ificant decrease i n the mean t issue satu ration up to 54.6% was demonstrated in the serosa of the colon after consecutive l igation of the feed i ng arteries. These fi nd ings are comparable to previous ly described values for ischemic colon mucosa22 ,2 3.

We observed increas ing St02 of the d istal co lon i n the first days after creation of an ischemic anastomosis , ind icat ing rapid convalescence of blood flow through the previous ly i schemic part of the colon. Us ing v is ib le l ight spectroscopy we cou ld not demonstrate a continu ing i ncrease of the m icroperfu s ion and thus e levated St02

leve ls between the thi rd and seventh postoperative day, as suggested by a previous study3 3 describ ing s ign ifi cant rise i n angiogenes is as measured by bisbenzim ide fluorescence stain i ng duri ng this postoperative period. The difference with our resu lt s m ight be exp lai ned by the fact that th is study was performed in non­ischemic anastomoses on ly. No c lear concl us ions can be drawn , however, as no data descri b ing both angiogenes is and t issue oxygenation in ischemic colon anastomoses are reported so far.

The fi nd ing of an i ncreased St02 level i n the caecu m (i .e. upstream of the ischemic anastomosis) m ight be due to an i ncreased b lood perfus ion of the co lonic microcircu lat ion remote from the s ite of ischem ia, as recently suggested by Colak et al 34. In our experiments there is no c lear evidence of an i schemic precondition ing phenomenon, because the anastomosis was created without prior clamping of the mesenteric arteries and ischemic precondition i ng is a consequence by defin it ion. However, th is phenomenon might be an i nteresti ng field for further research in colon anastomosis.

Most authors descri be decreas ing burst ing pressure in the first days after construct ion of profound ischemic anastomoses4-8• As our resu lts do not show th is d ifference in bursti ng pressure , it can be concl uded that ischemia of on ly one part in colon anastomoses m ight be less detrimental to anastomotic strength than ischemia

of the whole anastomosis. Th is effect might be contri buted to a higher degree of m icroperfus ion from the wel l-oxygenated proximal part of the colon as ind icated by a restored St02 in the d istal part of the anastomosis as early as day 3. However, no defin itive conc lus ion based on our data can be d rawn as variabi l ity of bursti ng pressure measurements was relatively high and the sample s ize probably too smal l to provide suffic ient power. The sample s ize of our study was calcu lated based on an St02 difference of 20% between ischemic and non- ischemic anastomosis as measured by VLS.

Although no s ign ificant d ifferences in bursti ng pressure were detected , 61 breaki ng strength was s ign ificantly lower i n partial ly ischemic anastomoses on the th i rd but not on the seventh postoperative day. As breaking strength is cons idered to be an i nd icator of the abi l ity of the bowel wal l to retain sutures in the early phases of anastomotic heal i ng26 , the lower break ing strength fou nd in our study might ind icate weaken ing of the bowel wal l by i schem ia. This i s supported by a high rate of d i s ruption at the anastomosis s ite i n our data. These fi nd ings in our experimental model support the observation that i n the c l in ical context most anastomotic leakages occur in the fi rst postoperative week.

Vis i ble l ight spectroscopy appeared to be an easy and qu ick method to evaluate t issue oxygenation of colon serosa. We demonstrated accu rate and reproducib le measurements of t issue perfus ion with low standard deviat ion i n colon anastomoses, as shown previous ly i n hu man co lon m ucosa22-2s. When compari ng the standard deviat ions of 1 1 % i n non-ischemic and 2 9% i n i schemic colon fou nd in our data with standard deviat ions reported by other authors describ ing methods for measurement of t issue oxygenation in animal models35-38 , al l studies show high standard deviat ions i n ischemic t issue, rang ing from 4% in hydrogen clearance35 to 43% us ing Clark-type e lectrodes38 • In non-ischemic t issue standard deviat ions tend to be lower rang ing from 1 . 5% with laser Doppler39 to 1 6% in a study us ing Clark-type e lectrodes4o.

Overa l l , accu racy of vis i ble l i ght spectroscopy in prediction of comp l ications of ischemia in our study was 70% (sens itivity 68%, specificity 75%). When comparing these resu lts to other non-i nvas ive methods for measuring St02, on ly one study with laser Doppler measurements reported also sens itivity and specificity1 7. Th is study compared laser Doppler flowmetry to i ntramural pH measurements report ing a sens it ivity of 66% vs 1 00% and specifi city of 29% vs 1 0% i n colon anastomosis. Three stud ies report test detai l s for pu lse ox imetry in transmural necros is , describ ing sens itivit ies of 95%1 2 , 2 3%4 1 and 1 00%39 and specificit ies of 79%, 96% and 86%, respective ly. In case of mi ld ischemia sens itivity and specificity were 89% and 5 8% respectively, i n one of these stud ies39. Al l stud ies u sed histolog ical analys is as gold standard. As most studies descri be the use of pu lse oximetry requ i ring pu l sati le flow for measurements i n deep t issue necros is , it i s not surpris ing to fi nd high values for sensitivity and specificity. I n the c l in i cal s ituation however, i t i s not the deep ischemic state that is d ifficu lt to detect, but the m i ld to moderate d im in ished tissue satu ration that causes concern for anastomotic i ntegrity. Looking into non-

V, n ::::r (I)

3

jij.

Ill :::s Ill V, rt

3 0 l/1 (I) V,

62

n ::r PJ "S ro ...,

V, n ::r ro 3 �-::; ro X "O ro ..., 3 ro ::; g_ PJ ::; PJ V,

8 3 0 V, ro V,

invas ive test methods, on ly author addressed the issue of test accuracy in these m i ld to moderate ischemic cond itions in anastomosis and descri bed a test accuracy of 70% i n laser Dopp ler measurement 1 7 .

I n conclus ion, we demonstrated that i schemia can i ntraoperative ly accu rate ly be detected by VLS i n muri ne colon . Partial ly ischemic anastomoses show more adhesions and d im in ished breaking strength i n the early phase of heal ing , but burst ing pressure i s not affected. Low St02 of a d istal colon anastomosis i s a ri sk factor for anastomotic deh iscence at day 3 and beyond.

References

1 . Eriksen MT, Wibe A, Norstein J, et al. Anastomotic leakage following routine mesorectal excision for

rectal cancer in a national cohort of patients. Colo rectal Dis 2005; 7: 51-57.

2 . Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic leakage

after large bowel resection: multivariate analysis of 707 patients. World J Surg 2002; 26: 499-502.

3. Vignali A, Fazio VW, Lavery IC, et al. Factors associated with the occurrence of leaks in stapled rectal

anastomoses: a review of 1014 patients. J Am Coll Surg 1997; 185: 113-121.

4. van der Ham AC, Kort WJ, Weijma IM, et al. Healing of ischemic colonic anastomosis: fibrin sealant

does not improve wound healing. Dis Colon Rectum 1992; 35: 884-891.

5. Garcia JG, Garcia Criado FJ, et al. Healing of colonic ischemic anastomoses in the rat: role of

superoxide radicals. Dis Colon Rectum 1998; 41: 892-895. 6. Posma LEA, Bleichrodt RP, van Goer H, et al. Transient profound mesenteric ischemia strongly

affects the strength of intestinal anastomoses in the rat. Dis Colon Rectum 2007; 50: 1-1 0.

7. Kologlu M, Yorganci K, Renda N, et al. Effect of local and remote ischemia-reperfusion injury on

healing of colonic anastomoses. Surgery 2000; 128: 99-104.

8. Kuzu MA, Tanik A, Kale IT, et al. Effect of ischemia/reperfusion as a systemic phenomenon an

anastomotic healing in the left colon. World J Surg 2000; 24: 990-994.

9. Vignali A, Gianotti L, Braga M, et al. Altered microperfusion at the rectal stump is predictive for rectal anastomotic leak. Dis Colon Rectum 2000; 43: 76-82.

10. Ambrosetti P, Robert J, Mathey P, et al. Left-sided colon and colorectal anastomoses: doppler

ultrasound as an aid to assess bowel vascularization. Int J Colorectal Dis 1994; 9: 211-214.

11 . Redaelli CA, Schilling MK, Carrel TP. lntraoperative assessment of intestinal viability by laser Doppler

flowmetry for surgery of ruptured abdominal aortic aneurysms. World J Surg 1998; 22: 283-289. 1 2 . Krohg-Sorensen K, Line PD, et al. lntraoperative prediction of ischaemic injury of the bowel: a

comparison of laser Doppler flowmetry and tissue oximetry to histological analysis. Eur J Vase Surg

1992; 6: 518-524.

13. Boyle NH, Manifold D, Jordan MH, et al. lntraoperative assessment of colonic perfusion using

scanning laser Doppler flowmetry during colonic resection. J Am Coll Surg 2000; 191: 504-510.

1 4. Hajivassiliou CA, G reer K, Fisher A, et al. Non-invasive measurement of colonic blood flow

distribution using laser doppler imaging. Br J Surg 1998; 85: 52-55.

1 5. Toens C, Krones CJ, Blum U, et al. Validation of IC-VIEW fluorescence videography in a rabbit model

of mesenteric ischaemia and reperfusion. lnt J Colorectal Dis 2006; 21: 332-338.

1 6. Millan M, Garcia-Granero E, Flor B, et al. Early prediction of anastomotic leak in colorectal cancer

surgery by intramucosal pH. Dis Colon Rectum 2006; 45: 595-601.

17. Senagore A, Milsom JW, Walshaw RK, et al. Intramural pH: a quantitative measurement for predicting

colorectal anastomotic healing. Dis Colon Rectum 1990; 33: 175-179.

18. Krohg-Sorensen K, Lunde QC. Perfusion of the human distal colon and rectum evaluated with

endoscopic laser Doppler flowmetry. Methodologic aspects. Scand J Gastroenterol 1993; 28: 104-

108.

l 9. Brolin RE, Semmlow JL, Sehonanda A, et al. Comparison of five methods of assessment of intestinal

viability. Surg Gynaecol Obstet 1989; 168: 6-12. 20. Line PD, Mowinckel P, Lien B, Kvernebo K. Repeated measurement variation and precision of laser

Doppler flowmetry measurements. Microvasc Res 1992; 43: 285-293. 21. Benaron, D. A., Parachikov IH, Cheong WF, et al. Design of a visible-light spectroscopy clinical tissue

oximeter. J Biomed Opt 2005; 10: 44005-1-44005-9.

22 . Benaron DA, Parachikov IH, Friedland S, et al. Continuous, noninvasive and localised microvascular

tissue oximetry using visible light spectroscopy. Anesthesiology 2004; 100: 1469-1475.

2 3. Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosal perfusion in

the gastrointestinal tract. Gastrointest Endosc Clin N Am 2004; 14: 539-555.

24 . Friedland S, Benaron D, Parachikov I, et al. Measurement of mucosa! hemoglobin oxygen saturation

in the colon by reflectance spectrophotometry. Gastrointest Endosc 2003; 57: 493-497.

63

Vl !"'I ::r (1)

3 SlJ

:::s (1) X

"O (I) -,

3 :::s

SlJ :::s A) Vl

8 3 0 Vl (I) Vl

64

2 5 . Ramirez FC, Sukhdeep P, Medlin S,et al. Reflectance spectrophotometry in the gastrointestinal tract:

limitations and new applications. Am J Gastroenterol 2002; 97: 2780-2784.

2 6. Hendriks T, Mastboom WJB. Healing of experimental intestinal anastomoses, parameters for repair.

Dis Colon Rectum 1990; 33: 891-901.

27. Brasken P. Healing of experimental colon anastomosis. Eur J Surg 1991; 566 (supp): 1-51.

2 8. Koruda MJ, Rolandelli RH. Experimental studies on the healing of colonic anastomoses. J Surg Res

1990; 48: 540-515.

2 9. Mansson P, Zhang XW, Jeppsson B, et al. Anastomotic healing in the rat colon: comparison between

a radiological method, breaking strength and bu rsting pressure. Int J Colorectal Dis 2002; 17: 420-

425.

3 0. Jiborn H, Ahonen J, Zederfeldt B. Healing of experimental colonic anastomoses; I Bursting strength

of the colon after left colon resection and anastomosis. Am J Surg 1978; 136: 587-594.

3 1 . Jiborn H, Ahonen J, Zederfeldt B. Healing of experimental colonic anastomoses; II Breaking strength

of the colon after left colon resection and anastomosis. Am J Surg 1978; 136: 595-599.

3 2. de Waard JW, Wobbes T, de Man BM, et al. Post-operative levamisole may compromise early healing

of experimental intestinal anastomoses. Br J Cancer 1995; 72: 456-460.

3 3 . Seifert WF, Verhofstad AAJ, Wobbes T, et al. Quantitation of angiogenesis in healing anastomoses of

the rat colon. Exp Mal Pathol l 997; 64: 31-40.

34. Colak T, Tu rkmenoglu 0, Dag A, et al. The effect of remote ischemic preconditioning on healing of

colonic anastomoses. J Surg Res 2007; 143 : 200-205.

3 5 . Shikata JI, Shida T. Effects of tension on local blood flow in experimental intestinal anastomoses. J

Surg Res 1986; 40: l 05-111.

3 6. Kashiwaga H. The lower limit of tissue blood flow for safe colonic anastomosis: an experimental

study using laser Doppler velocimetry. Surg Today 1993; 23: 430-438.

3 7. Attard JAP, Raval M, Martin GR, et al. The effects of systemic hypoxia on colon anastomotic healing:

an animal model. Dis Colon Rect 2005; 48 : 1460-1470.

3 8. Kimberger 0, Fleischmann E, Brandt S, et al. Supplemental oxygen, but not supplemental crystalloid

fluid, increases tissue oxygen tension in healthy and anastomotic colon in pigs. Anesth Analg 2007;

l 05: 773-779.

39. Erikoglu M, Kaynak A, Beyath EA, et al. lntraoperative determination of intestinal viability: a

comparison with transserosal pulse oximetry and histopathological examination. J Surg Res 2005;

128: 66-69.

40. Shandall A, Lowndes R, Young HL. Colonic anastomotic healing and oxygen tension. Br J Surg 1985;

72 : 606-609.

41 . Ando M, Ito M, Nihei Z, et al. Assessment of intestinal viability using a non-contact laser tissue

blood flowmeter. Am J Surg 2000; 180: 1 76-180.

:a Chapter 4 lschemia in experimental anastomoses

Preoperative i rrad iat ion with SxS Gy i n a m u ri ne

iso lated co lon loop model does not cause anastomot ic

weake n i ng after co lon resect ion

A. Karl i czek 1 , C.J . Zeebregts 1 , D.A. Benaron2 , R .P. Coppes 3 , T . Wiggers 1 ,

G .M . van Dam l ,4 .

1 Department of Surgery, Un iversity Medical Center G roningen, Groningen, The Netherlands 2 Stanford Univers ity School of Medic ine, Palo Alto, CA, USA

3 Department of Cel l Biology, section Rad iation and Stress Cel l Biology Un iversity Med ical Center G roningen, Gron ingen, The Netherlands

4 BioOptical I maging Center Groningen (BICG), Groningen, The Netherlands

Int J Colorectal Dis . 2008; 23: 1 1 1 5 - 1 1 24

... -. - .-.. ..,

. Abstract .

Background: There are confl ict ing stud ies on the i nfl uence of fractionated preoperative 5 days of 5 G ray i rrad iation on t issue oxygenation and subsequent colon ic anastomotic strength. To e lucidate the effect of preoperative i rrad iation on anastomotic strength an iso lated colon loop model was deve loped. Methods : Male Wistar rats (n = 1 64) were random ly d ivided i nto 3 g roups . One group remai ned untreated (control) . I n the other two g roups a loop of descending colon was exteriorized to create a hern ia of the abdominal wal l . After fou r weeks recovery th i s loop was local ly i rrad iated with SxS Gy of y-rays or sham i rrad iated . One week after (sham-) i rrad iation an anastomosis was pe rformed i n a l l g roups. Tissue oxygenation (StO2) was dete rm i ned with vis ib le l i ght spectroscopy (VLS) . The an imals were sacrificed 3 or 7 days after the operation and the anastomos is was tested for bu rst ing pressure and breaking strength . Resu lts: I rrad iated rats showed s ign ificantly more weight loss (90% sd 4 . 3 of in it ial body weight vs 96% sd 2 .8 , p� 0.05) and enteritis ( 1 8% vs 5%, p=0 .0 1 3) compared to sham and contro l an imals . Tissue oxygenation (StO2) was not i nfl uenced by i rrad iation and was not pred ictive for anastomotic strength. The control g roup showed s ign ificantly lower bu rsti ng pressure and breaki ng strength compared to (sham-) i rrad iated animals . Conclusion: We deve loped a new isolated loop model for i nterm ittent i rrad iat ion of the colon. Preoperative i rrad iation of the d istal part of a colon anastomosis was successfu l ly admin i stered with acceptab le s ide effects and did not cause reduced t issue oxygenation , nor c l i n i cal s igns of anastomotic weaken ing , nor objective reduction in bursting pressu re and breaking strength.

Introduction Leakage of anastomosis after co lorectal resection is a serious problem in gastro intestinal surgery1 . After the i ntroduction of preoperative i rrad iat ion , confl icting resu lts have been reported on the i ncidence o f anastomotic leakage2-5 . I n several animal6- 1 1 and cl i n ical4,5 , 1 2 studies decreased anastomotic strength or h igher i ncidence of anastomotic leakage after i rrad iation was demonstrated . Neverthe less , in most of the animal stud ies the i rrad iat ion protoco ls were not comparable o r even completely d ifferent from the c l i n ical s ituation . In studies s imu lat ing c l i n i cal protocols with respect to fractionation of preoperative rad iation doses 1 1 , 1 3 , 1 4 and irrad iation of one or both l imbs of the anastomosis 1 5 , the i rrad iat ion techn ique appeared to be a major chal lenge. The murine colon i s d ifficu lt to i rrad iate without considerable damage to surround ing t issue. I n these studies the colon was irradiated in s itu by plac ing rad iopaque markers and dosimeters du ring a laparotomy1 3, 1 5 or by i rrad iation of the entire pelvis which resu lted i n h igh mortal ity 1 1 , 1 4 . The detrimental effects on normal t issue in the acute phase after i rrad iat ion are thought to resu lt main ly from loss of i ntestinal crypts and an acute i nflammatory reaction, which m ight resu lt in reduced blood flow 1 6 . This reduction in blood flow may play an important role in anastomotic weaken ing as shown by a study investigat ing m icrovascular blood flow afte r i rrad iat ion measu red by laser Doppler i n re lationsh ip to anastomotic strength of the colon6 . A clear decrease i n blood flow four weeks after i rradiation with 5 x 8 . 5 Gy in combination with d im i n i shed anastomotic strength was found . As in cap i l lary beds such as colon mucosa pu lsati le blood flow requ i red for laser Dopp ler ve locimetry is absent, we used a techn ique for measurement of m icroperfus ion and thus t issue oxygenation based on vis ib le l i ght spectroscopy. The system uses the pr incip le of absorbance and scatteri ng of l ight i n t issues, not requ i ring pu lsati le b lood flow or t issue contact. I n previous stud ies 1 7-2 1

vis ible l ight spectroscopy has been val idated to be usefu l for the detection of t issue i schemia i n human gastric and colon ic mucosa.

The aim of this study was two-fo ld . Fi rst, to develop a nove l murine model for local ized fractionated i rrad iat ion of the descend ing colon , based on a techn ique previous ly described for fractionated sma l l bowel irrad iation 22 ,23 . Thus it is i ntended to s imu late a c l in ical ly re levant context in colorectal cancer more closely with respect to fractionation of irrad iation , i rrad iation damage of su rround ing t issue and general ised i rrad iation effects . Second, to eval uate t issue oxygenation and anastomotic strength after preoperative i rrad iation .

Material and methods

Overview of study design An overview of the design of the study is shown in figu re 1 . After inc lus ion an imals were randomly ass igned per cage to receive an abdominal wal l hern ia of the s igmoid

69

n :::r Jl)

ro ..... V1

)> :::s Jl) VI

8 3 0 VI ro VI

:::s Jl)

3 C ..... :::s ro 3 0 Cl. ro

� V1 X V1

C") -<

70

V,

)> :::, Pl Vl

8 3 0 Vl (D Vl

:::,

Pl

3 C ..., :::, (D

3 0 Q.

V, X V,

C) -<

co lon (i rrad iat ion and sham group) or to serve as control group . Four weeks after the hern ia operation the survivi ng animals were d ivided i nto an i rrad iat ion g roup and a sham group. At each operative procedure tissue oxygenation of the co lon segments forming the futu re anastomos i s was eval uated with vis i b le l i ght spectroscopy (VLS). In the week fol lowi ng i rrad iat ion of the hern iat ing exteriorized colon segment with 5x5 Gy or sham i rrad iat ion an anastomosis with in this colon segment was constructed . The animal s were terminated on the th i rd or seventh postoperative day. S igns of anastomotic leakage , s uch as abscess or deh iscence, and adhes ion formation were noted. Anastomotic strength was quantified by measurements of bursti ng pressure and break ing strength. A power analys is based on the resu lts of a p i l ot study revealed that a sample s ize of 23 ani mal s per g roup was needed to detect a d ifference of l 5 mm Hg in bu rsting pressu re between the i rrad iation and the sham group on the th i rd postoperative day (0<= 0.05 , 13= 0 .95 ) . Twentys ix add it ional an i mals were i ncl uded to compensate for antici pated mortal ity of 30% after the hernia operation (based on unpub l i shed data from a p i lot study conducted at our center) . The study protocol was approved by the Animal Eth ics Com mittee of the Un ivers ity of Gron i ngen .

Animals Male Wistar rats (n= 1 64 , mean weight 344 g; Harlan BV, Horst, The Netherlands) were a l lowed to adapt to laboratory cond itions for one week before experimental

--

rn -- M

>-II CQ

C rad i ation 0 I C :.:; "

RI 0 CQ I. -0 cu u +-0. - - - - - - - - - - "tJ 0

,a

rn ·s M

·c E >-CQ

I. 01 +-cu 'in sham In " 'in

0 CQ

� E +-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - .s -In M

,a >-C CQ

+-

con trol 'I "

-- +-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -week 1 2 3 4 5 6 7 8

Figure 1 Overview of the study design . The study groups are shown on the left, the i nterventions for each group are shown on a t ime scale in weeks. Arrows ind icate irradiation, t indicates termi nation.

use. The animals were housed five per cage u nder specified pathogen-free condit ions and had free access to water and standard rodent food (AB d i ets , Woerden, The Netherlands). After each laparotomy the fi rst 24 hours 0 .03 mg /kg buprenorph ine s .c. was adm in istered twice dai ly. To compensate for flu id loss l O m l 0 .9% NaCl was adm in istered subcutaneous ly. The animals were housed i n separate cages unt i l passage of stoo l . Based on the find ings in a p i lot study (data not shown) the animals were euthanized on the fourth postoperative day if no passage of stool was observed . After an operation or (sham) i rrad iation the animals were weig hed daily and inspected for signs of i l l ness . 71

Hernia operation Rats i n the irrad iation and sham groups were anesthetised with a m ixtu re of isoflu rane 2 . 5% and oxygen and received 0 .03 mg/kg buprenorph ine s .c . after induction of anaesthes ia. The abdomen was shaved in the m id l i ne and d i s i nfected . During the operation body temperatu re was maintai ned at 3 8 ° C us i ng a heati ng pad . The operation was performed under sem i-ster i le conditions .

The steps carried out to create an abdominal wal l hern ia contain i ng descend ing colon approximately l . 5 cm from the peritoneal reflection are shown and descri bed in Figu re 2. The skin was closed with an i ntracutaneous resorbable runn i ng sutu re (Monocryl™ 5-0, Ethicon , Amersfoort, The Netherlands).

Vis i ble l ight spectroscopy measurements of StO2 were performed at the hern iating s igmoid colon and at the coecum before and after construction of the hern ia. A system cons i st ing of a vis ible l ight spectroscopy (VLS) oximeter (T-Stat"' , model 303 lschemia Detection System ; Spectros Corp. , Portola Val ley, CA, USA) was used. Detai ls and descri pt ions of this system have been descri bed previously 1 8, 1 9 • I n short, the oximeter em its wh ite l ight from a probe (5 mm d iameter) placed on or near the t issue, col lecti ng any l ight retu rn ing to the probe from the t issue. VLS u ses the principle of absorbance and scatteri ng of l ight i n biolog ical t issues. Each type of molecu l e has a characteristic absorbance spectrum. As oxygenated and deoxygenated haemoglob in have a different absorption spectru m and haemog lob in is cons idered a crit ical derivative of perfus ion , t i ssue m icroperfus ion can be assessed us ing a techn ique that detects these d ifferences (Figure 3). To measu re StO2 the probe was held a few mi l l imetres above the gauze-cleaned serosal su rface unt i l val ues stabi l ized with in a 5% marg i n .

Technique of irradiation After four weeks the i rrad iation group was anesthetised with 40 mg /kg ketam ine and 0 . 1 mg/kg medetomid ine subcutaneous and placed i n lateral position on a 2 cm lead sh ie ld with a 2 x l . 5 cm open ing for the hern ia (Figu re 4) . On 5 consecutive days 5 Gray gamma (Cs- 1 3 7) i rrad iat ion (CIS bio i nternational , IBL 303) with a Focus Subject Distance of 2 5 cm was adm in istered with exposure to scatter rad iat ion of < 5%. The sham group was treated s im i larly to the i rrad iation g roup, but no i rrad iat ion was appl ied. I n case of d iarrhoea the animals were rehydrated with 1 6 m l

V1

)> ::s SlJ V, 0 3 0 V, (l) V,

::s SlJ 3 C: ::::!. ::s ct> 3 0 0.

Q V1 >< V1

C) -< .., .., SlJ 0.

�: 0 ::s

72

n ::; PJ "O ro -, u,

)> ::, PJ V,

8 3 0 V, (1) V,

::, PJ 3 C -, ::, (1)

3 0 CL � 0 -, u, X u, C') -< -, � CL

�-0 ::, Figure 2

G

Creati on of an abdom inal wal l hern ia . F i rst a s u bcutaneous pocket was created (A) . After laparotomy a musculofascial flap from the rectus abdomin is muscle was mobi l ised by sharp d issection (B). The mesentery of the descending colon was opened and the descending colon and its vasculature was l ifted from the abdominal cavity (C). Subsequently, the musculofascial flap was pu l led through the open ing in the mesentery (D, E) and sutured to the contralateral fascia of the rectus abdomin is muscle (F). The skin was closed covering the herniating colon loop. A rat with completed abdominal wal l hernia (encircled in red) is shown (G).

Figure 3 Visible l ight spectroscopy system: handheld probe used in this study (T-Stat• 303 lschemia Detection System; Spectres Corp. , Portola Valley, CA, USA).

0.9% NaCl subcutaneous in 2 doses dai ly unt i l we ight gain and cessation of diarrhoea. Otherwise no s ide effects were noticed .

Anastomosis operation

Three to seven days after i rrad iation al l an imals were anaesthetised and prepared for surgery as descri bed previous ly (see the sect ion on ' hern ia operation ') . The anastomosis operation was performed us ing an operation m icroscope. The skin was opened and the herniat ing co lon loop was carefu l ly freed of adhes ions from fascia and omentum. St02 of the herniated bowe l was measured as soon as serosa became v i s i b l e . N ext, t he abdomen was o p e n e d and a l cm s e g m en t of t h e descend i ng colon was resected ensuring that on ly the d istal part o f the anastomosis consisted of irrad iated s igmoid . Bowel conti nu ity was restored by an end-to-end anastomosis us i ng l 0-1 2 i nverting interru pted non-absorbable monofi lament sutures (Dafi lon ® 8-0, Braun, Tutt l ingen, Germany) . Afte r construction of the anastomosis St02 val ues of the colon proximal and d istal to the anastomosis were measured. The abdomen was closed us ing a conti nuous resorbable suture (Polysorb™ 4-0, Tyco Healthcare , Gosport, UK) for the muscu lofascial layer and the skin was closed with an i ntracutaneous runn ing resorbable sutu re (Monocryl™ 5-0, Ethicon, Amersfoort, The Netherlands). I n the contro l group the same procedure was fo l lowed .

73

l> ::::, llJ V, 8 3 0 Vl (D Vl

::::, llJ 3 C .... :::::5 (I)

3 0 a.. (I)

u, X u, C') -<

74

V1

)> ::i PJ V,

0 3 0 V, (D V,

::i

PJ

3 C: ..... ::i /t)

3 0 0. /t)

V1 X V1

C) -<

hernia

Figure 4 Rat positioned on lead sh ie ld ing for irradiation. To achieve isolated irradiation and to min imize systemic effects on the abdomen, the hern ia contain ing the descend ing colon (arrow) is placed above an opening of 2 x 1 .5 cm in the lead shie ld ing.

Termination Either 3 or 7 days after anastomosis construction the an imals were anesthetised.

After relaparotomy St02 of the colon proxi mal and d i stal to the anastomosis was meas u red. The abdominal cavity was explored for s igns of i leus and deh iscence of the anastomosis. The an imal was subsequently terminated by intracard iac pentobarb ital injection.

Anastomotic strength Anastomotic strength was eval uated on the th i rd and seventh postoperative day by measu ring resistance to intral um inal pressure (bu rsting pressure) and to longitud inal forces (breaking strength) by us ing we l l described methods24-28.

After termination the anastomosis was resected together with adhesions and adhes ive structu res. The severity of adhesions was noted as none, m i l d , moderate or severe as descri bed by Van der Ham et al 29. lntral um inal faeces were removed.

Subsequently, a l l colon segments were t ied off at the d istal end. A cannula was inserted in the prox imal end and the colon was tied around it. The cannu la was connected to an infus ion pump and a manometer and infused at a rate of 1 m l / m in with methylene b lue d isso lved in sal ine. The ri s ing pressu re was recorded g raph ical ly and the bursting pressure (in m mHg) was determined and defined as the maximal pressure the anastomosis or colon segment res isted. The s ite of leakage was recorded.

Thereafter, the segment was placed in a tens iometer system (Mecmesin AFG 1 0 , Horsham, UK). A constantly increas ing force was appl ied and the l inear breaking strength was recorded. Th i s procedu re has been previous ly descri bed and val idated 28.

Statistical analysis Data were analyzed with Statistical Package for the Social Sciences software (SPSS l 5-0, SPSS, Chicago, I L, USA). Statistical d ifferences between g roups were calcu lated us ing one-way analys is of variance (ANOVA), Student t-test and Chi-squared test when appropriate. Corre lations between variables were analysed by Pearso n or Spearman corre lation or mu lt ip le logist ic regress ion analys is when appropriate. Power analys is was carried out by us ing Sample Power 2.0 (SPSS, Ch icago, IL, USA). Statistical s ign ificance was accepted at the 5% leve l .

Results

General outcome An overview of the mortal ity of the hernia operat ion, i rrad iation procedure , and anastomos is operation, respectively, i s shown i n figu re 5 .

Control 44

Anastomosis 44

Inclusion 1 64

Sham 39

Anastomosis 37

Hernia 1 20

5x5 Gy 46

Anastomosis 44

26 obstruction 9 evisceration

2 radiation enteritis 2 anesthesia

1 anesthesia

Figure S Flow chart of i nc luded animals . Contro l : control-group . Hernia: fi rst operation in wh ich an abdominal wall hernia was constructed. t and grey boxes indicate mortality and causes. Sham: sham group. 5x5 Gy: irradiation group, irradiated with SxS Gy. Anastomosis: second operation during which an anastomosis was constructed. Day 3 /7: postoperative day when termination took place. Bold numbers represent numbers included.

75

n :::r llJ

l'tl .... u,

)> ::l llJ Vl 8 3 0 Vl l'tl Vl � llJ

C .... ::l l'tl 3 0 0. (0

Q u, X u,

C) -<

76

V1

)> :J PJ V, 8 3 0 V, (D V, :J PJ 3

:J (D 3 0 a. (D

Q V1 X V1

C) -<

105 * *

*

100

90

BS_._-�-�-�-�-�-�-�-�__,

1# 2# 3# 4# 5# 6 7 8 days after irradiation

Hernia operation

Figure 6

Percentage of body weight change during and after 5x5 Gy irradiation of the isolated colon segment. * indicate significant difference (p:::;; 0.05, t-test) between i rradiation and sham group. Light grey ind icates radiation group, black l ine ind icates sham group. Error bars show 95% confidence interval.

Al l an imals suffered trans ient we ight loss after the hernia operation (mean weight loss 5% sd 5 .4%) . The mean period u nt i l passage of stool was 0 .8 (sd 1 . 3) days . Twentys ix (2 2%) an imals showed pers istent bowel obstruction at the fourth postoperative day and were terminated accord ing to protoco l . On te rm ination, obstruction of the colon at the s ite of the hernia was found i n al l animals and in one an imal an abscess due to a pe rforation at the level of the fascia was seen . In n i ne (7%) animals eviscerat ion after wound dehiscence due to g nawing had occurred and these animals were euthan ized the fi rst day after hernia operat ion. The overa l l mortal ity of the hern ia operation was 2 9% (3 5 ani mals) .

Irradiation During irrad iation ani mals in both the i rrad iation and the sham group lost we ight; this was s ignificantly more pronounced in the irrad iat ion g roup, as shown in figure 6 (90% sd 4 . 3 of in it ial body weight vs 96% sd 2 .8) . In 1 9 (22%) of the ani mals enteritis was observed, 1 5 of them after i rrad iation (sham vs i rrad iation, p=0.0 1 3) as shown in tab le 1 , ind icat ing s uccessfu l i rrad iation of the bowe l . Diarrhoea typical ly started on the last day of the irrad iation scheme and reso lved after 1 -3 days . On the last i rrad iation day three animals were considered too s ick for anaesthesia and transport . Therefore in three cases the last irrad iat ion took place two days later than schedu led. These th ree an imals survived and were analysed in the irrad iation group and separate analyses showed no difference.

no enteritis enteritis p

radiation 31 (36%} 1 5 (1 8%) J 0.01 3

sham 35 (41 %) 4 (5%)

total 66 (78%} 1 9 (22%)

Table 1 Enteritis after irradiation. Between brackets percentages of the total number of animals are indicated.

Du ring i rradiation fou r animals died. Two animals in the sham group died du ring anaesthesia. At abduction no clear cause of death could be identified. Two animals of the i rradiation group showed severe enteritis with persistent weight loss and progressive abdominal distension and were euthanized. On post-mortem examination severe paralytic ileus was seen.

Anastomosis operation In most animals of the i rradiation group mild to moderate oedema of the herniating colon loop was present at laparotomy. During the postoperative period all animals lost weight, which was most pronounced in the sham group. At anastomosis operation the control group (3 84 g sd 29) was heavier compared to the radiation group (3 54 g sd 3 1 , p� 0 .05) and comparable to the sham group (3 77 g sd 2 5 , p= 0.83). During or after the anastomosis operation one animal in the sham group died during anaesthesia. At termination no clear cause was identified, but the animal had suffered considerable weight loss due to enterit is.

The mean period until passage of stool after the operation was 0.4 (sd 0.9) days. None of the animals was killed earlier than scheduled due to ileus. At termination 3 animals (2%, 2 sham and 1 control) showed signs of obstruction at the anastomosis without clear evidence for anastomotic leakage.

At termination adhesions were absent in 4 (3%) animals, mild in 36 (29%), moderate in 44 (3 5%), and severe in 40 (32%) animals. There were no significant differences between i rradiation, sham and control group, nor between early (day 3) and late (day 7) termination.

Colon oxygenation The changes in St02 as measu red by VLS are shown in figu re 7. There were no significant changes in St02 in the colon proximal or distal to the anastomosis. At univariate regression analysis there was no significant correlation between St02 and anastomotic bursting pressu re, nor breaking strength. There were no correlations between St02 and clinical outcome parameters (i .e. adhesions, ileus, enteritis, mortality).

77

)> :::s � VI

0 3 0 VI (I) VI

:::s � 3 C .... :::s (I)

3 0 0..

[ \.n X \.n

C) -<

78

)> :J Ill V\

8 3 0 V\ (I) V\

:J

Ill

:J (I)

3 0 0.. (I)

u, X u,

C) -<

80

9 70 I.I)

hem,a anast

Proximal

tenn hernia anast tenn

Distal

Figu re 7 StOz at proximal and distal part of the anastomos is . The irradiation group is shown in wh ite bars, the sham group in l ight grey and the control group in dark grey. "Hern ia" designates StOz measured during the hernia operation, "anast" the anastomosis operation and "term" prior to termination. Boxes represent first and third quartiles with the med ian as a central l ine . The whiskers represent 51h and 95th percentiles.

Anastomotic strength Anastomotic strength expressed as bursti ng pressu re and breaking strength i ncreased s ign ificantly between the th i rd and seventh postoperative day i n a l l groups, as shown i n figure 8.

On the th i rd and seventh postoperative day bursti ng pressu res were s ign ificantly higher i n both the i rrad iat ion (1 59 mmHg sd 3 2 / 1 96 mmHg sd 39) and sham ( 1 67 mmHg sd 2 5 /209 mmHg sd 40) g roup when compared to the control g roup ( 1 05 mm Hg sd 56/ l 75 mm Hg sd 22). In breaki ng strength measurements a s im i lar patte rn was seen on day 3 with s ign ificantly h igher res istance to long itudi nal forces i n both the i rrad iat ion (244 g sd 37) and sham (2 3 5 g sd 5 7) group compared to the control group ( 1 74 g sd 62). On day 7 the d ifference in breaking strength was on ly s ign ificant between the sham and the control g roup (i rrad iat ion 3 2 3 g sd 65 , sham 364 g sd l 03, control 303 g sd 82).

There was no s ign ificant d ifference i n bursti ng pressu re or break ing strength between the i rrad iat ion and the sham group on the th i rd or seventh postoperative day. In table 2 the percentage of anastomoses burst ing with in the anastomotic l i ne are shown.

Discussion In th is study we descri be a new an imal model with an isolated colon loop to i nvestigate the effects of fractionated preoperative rad iotherapy on colon anastomos is closely resembl ing the c l i n ical s ituat ion.

300 600

ci

$ � f 200

� 1 00 .0

500

�

§400

$ � :Joo

.o 200

1 00

rad sham contr rad sham contr

Figure 8 Burst ing pressu re and breaki ng strength 3 (grey bars) and 7 (wh i te bars) days after colon anastomosis. rad represents irradiation group, sham sham-irradiated group, contr control group. Boxes represent first and th ird quartile with the med ian central. The whiskers represent 5th and 95th

percenti les. * Marks s ign ificant difference (p� 0.05) .

% disrupture at anastomosis

day 3 day 7

irradiation

sham

control :: J] . 82

Table 2 Percentage of rupture with in the anastomotic l ine in bursting pressure measurements. * p= 0.008, # p=0.057 between sham and irradiation vs . control group.

In most earl ier conducted animal stud ies i ntraoperative s ing le-dose i rrad iat ion of both l imbs of the anastomosis6,9, i o,3 o was appl ied. Th is resu lted in weaken ing of the anastomosis in some of the stud ies6 ,9, 1 0 , whi le others d id not detect th is effect 1 3 ,3 0 • I n these stud ies , a l l evaluati ng anastomotic strength after irrad iat ion , h igh i rrad iation doses without fractionation were appl ied, but these were not representative for the c l i n ical s ituation 6-1 0 • I n the c l i n ical s ituation , usual ly CT­planned i rrad iation with SxS Gy is appl ied on the rectum with m in imal i nvolvement of the adjacent t issue. After colorectal resection the remain ing , previous ly i rrad iated rectal stump forms the d i stal l imb of the anastomosis , whi l st the proximal part cons ists of non-i rrad iated descend ing colon , thus creat ing a partial ly i rrad iated anastomosis. Several an imal stud ies were conducted evaluating c l i n ical ly re levant fractionated preoperative i rrad iat ion schemes. Two studies describe confl icti ng resu lts us i ng fractionated i rrad iat ion of the colon i n s itu after previous implantation

79

)> :::s PJ VI

8 3 0 VI (t) VI

:::s

PJ

3 C: .... :::s (D

3 0 0..

� V1 X V1

C) -<

of rad iopaq ue markers and dosimeters 1 3 , 1 5 • First, De Meerleer et al . found weakened anastomoses after cumulative doses h igher than 80 Gy 1 s, wh ich are never used c l in ical ly. I ndeed , Cee len et al . describes no d im i n i s hed anastomotic strength after total doses up to 80 Gy 1 3 • Moreover, De Meerleer et a1 . 1 s stud ied i rrad iat ion of one l imb of the anastomosis vs . i rrad iat ion of both l im bs and found d im in i shed bursti ng pressu re after i rrad iat ion of both l imbs of the anastomosis , whereas no effect was seen after i rrad iat ion of on ly one l imb. Kuzu et a l . appl ied SxS Gy i rrad iation on the pelvis in rats and fou nd d im in i shed bu rsti ng pressu re and hydroxypro l i ne content of

80 the anastomos is , but th is study d id not resemble the c l i n i cal s ituation as both l i mbs of the anastomosis were i rrad iated9• Seifert et al. 6 describe s im i lar resu lts althoug h in the early phases of heal ing al l anastomoses were weaker after i rrad iat ion, whereas

Vl

l> ::::s ,:l.J Vl

8 3 0 Vl (1) Vl

3

::::s (0

3 0 0.. (0

Vl X Vl

C) -< .., .., ,:l.J 0..

at seven days postoperatively only anastomoses after i rrad iation of both l imbs showed d im in i shed bursti ng pressure . In th i s study, however, an i rrelevant i rrad iation scheme as s i ngle-dose i ntraoperative i rrad iat ion with 2 5 Gy was appl ied.

Apart from studies describ ing c l i n ical ly i rre levant i rrad iat ion schemes , previous ly appl ied methods for preoperative fractionated i rrad iat ion of colon anastomoses have a h igh mortal ity. Kuzu et al . reports a 2 7% mortal ity after pelvic i rrad iation 1 1 and De Meerleer et al . found a 26% mortal ity due to imp lantation of dos imete rs 1 5 . Apart form a lower mortal ity (2 2%) the hern ia techn ique presented i n th is paper is the on ly techn ique in which a n iso lated colon loop can b e i rrad iated with re lat ively low mortal ity of the i rrad iat ion itself, making it more su itable for eval uation of h igher i rrad iat ion doses in the cu rrent and futu re stud ies i n order to el uc idate the effect of pro longed i rrad iat ion as g iven i n the hu man s ituation for local ly advanced mal ignancies. The model cou ld also be su itable to evaluate rad ioprotective compounds for protection of the large bowel , such as 4-hyd roxy-2 , 2 ,6 ,6-tetramethylpi perid i nyloxy (Tempol) , a n itrox ide rad ioprotector3 1 .

This study confi rms the resu lts from several other stud ies evaluati ng anastomotic strength after i rradiation with SxS Gy, being that no s igns of anastomotic weaken ing by low dose preoperative i rrad iation of the co lon cou ld be detected . Surpri s i ngly, in our study both the i rrad iated and sham i rrad iated co lon showed even h igher burst ing pressure and breaki ng strength when compared to control an imals. As th is i ncreased strength after i rrad iation has not been descri bed previous ly, th is fi nd ing m ight be explai ned by the observed adhesions caused by the hernia operation , but based on our data no clear concl us ions can be drawn so far. There were no d ifferences between the i rrad iat ion and the sham group with respect to anastomotic strength. Contrary, s ign ificant effects of i rrad iat ion were observed with regard to weight loss and enterit is in the rad iat ion g roup com pared to the sham group , as might be expected .

I n conc lus ion , we successfu l ly deve loped a new an imal model su itab le for fract ionated i rrad iat ion of an isolated colon loop by su rg ical ly plac ing a colon loop i n a n abdomi nal wal l hernia. Partial ly i rrad iated anastomoses i n the s igmoid co lon , i rrad iated with SxS Gy, d id not show d im i n i shed anastomotic strength. The model for i nterm ittent i rrad iation of a short, isolated colon segment is a usefu l mode l to

study the effect of irradiation dosage and fractionation in future studies, either for surgical studies or for evaluation of compounds to diminish irradiation effects of the colon.

81

V,

)> ::J PJ Vl ,..,. 0 3 0 Vl (1) Vl

::J

PJ

3 C ::J. ::J (D

3 0 0. � Q V, X V,

C) -<

82

)> :::l JlJ V> ,-+

3 0 V> I'll V>

:::l

JlJ

3 C � :::l I'll

3 0 0. I'll

� U7 >< U7

Cl -<

References

l . Ptak H, Marusch F, Meyer F, et al . Impact of anastomotic leakage on oncological outcome after rectal cancer resection. Br J Surg 2007; 94: 1 548- 1 5 54.

2. Marijnen C, Kapiteijn E, Van de Velde CJM, et al. Acute side effects and compl ications after short term preoperative rad iotherapy combined with total mesorectal excision in primary rectal cancer: report of a mu lticenter randomized trial . J Clin Oneal 2002 ; 20: 8 1 7-82 5 .

3 . Carlsen E , Schlichting E , Gu ldvog I , e t al. Effect o f the introduction of total mesorectal excision for the treatment of rectal cancer. Br J Surg 1 998; 85 : 526-529.

4 . Eriksen MT, Wibe A, Norste in J , et al . Anastomotic leakage fol lowing routine mesorectal excision for rectal cancer in a national cohort of patients. Colorectal Dis 2005 ; 7: 5 1 -57 .

5 . Bu ie WD, Maclean AR , Attard JA , e t a l . Neoadjuvant chemorad iation increases the ris k of pelvic sepsis after radical excision of rectal cancer. Dis Colon Rectum 2005 ; 48: 1 868- 1 874.

6 . Seifert WF, Wobbes T, Hoogenhout J , et al . l ntraoperative irrad iation delays anastomotic repair in rat colon. Am J Surg 1 995 ; 1 70: 2 56-265 .

7 . Mi lsom JW, Senagore A, Walshaw RK , et a l . Preoperative rad iation therapy produces an early and persistent reduction in colorectal anastomotic blood flow. J Surg Res 1 992 ; 5 3 : 464-469.

8. Winsey K, Simon �. Levenson SM, Seifter E, Demetriou AA ( 1 987) Effect of supplemental vitamin A on colon anastomotic heal ing in rats g iven preoperative i rradiation . Am J Surg l 5 3 : l 53- 1 56 .

9 . Morgenstern L, Sanders G , Wah lstrom E, et al . Effect of preoperative i rrad iation on heal ing of low colorectal anastomoses. Am J Surg 1 984; 1 47:246-249.

l 0 . Weiber S, Jiborn H, Zederfe ldt B. Preoperative irrad iation and co lonic heal ing . Eur J Surg 1 994; 1 60: 47-5 1 .

1 1 . Kuzu MA, Koksoy C, Akyol FH, et al. Effects of preoperative fractionated i rrad iation on left colonic anastomoses in the rat. Dis Coon Rect 1 998; 4 1 : 3 70-3 76.

1 2 . Matth iessen P, Hal l book 0, Andersson M, et al. Risk factors for anastomotic leakage after anterior resection of the rectum . Colorectal Dis 2004; 6 : 462-469.

l 3. Ceelen W, El Malt M, Cardon A, et al . I nfl uence of preoperative h igh-dose radiotherapy on postoperative outcome and co lonic anastomotic heal ing : experimental study in the rat. Dis Colon Rect 2001 ; 44: 7 1 7-72 1 .

1 4. Bed i rl i A, Kerem M, Karahaciog l u E, et al . Effects of two conventional preoperative radiation schedules on anastomotic heal ing in the rat co lon. Eur Surg Res 2007; 39 : 1 4 1 - 1 47 .

l 5 . De Meerleer G , Pattyn P , Fortan L , e t a l . H igh-dose preoperative rad iotherapy does not alter the strength of un i laterally irradiated colon anastomoses in the rat. I nt J Rad iation Oncology Biol Phys 1 999; 44: 1 63-1 70.

1 6 . O'Brien PC. Radiation injury of the rectum. Radiother Oneal. 200 1 ; 60 : 1 - 1 4. 1 7 . Benaron, D. A. , Parach i kov IH , Cheong WF, et a l . Design of a vis i ble-l ight spectroscopy c l in ical

tissue oximeter. J Biomed Opt 2005 ; l 0: 44005- 1 -44005-9. l 8 . Benaron DA, Parach i kov IH , Fried land S, et al . Conti nuous, noninvasive and local ised m icrovascu lar

tissue oximetry using vis ib le l ight spectroscopy. Anesthesiology 2004; l 00: 1 469- 1 475 . 1 9 . Friedland SR , Benaron D. Reflectance spectrophotometry for the assessment of mucosal perfusion

in the gastrointestinal tract. Gastrointest Endosc Cl in N Am 2004; 1 4 : 5 3 9-5 5 5 . 20 . Friedland S , Benaron D, Parach i kov I , e t a l . Measurement of mucosal hemoglobin oxygen saturation

in the colon by reflectance spectrophotometry. Gastroi ntest Endosc 2003 ; 57 : 493-497. 2 1 . Ramirez FC, Sukhdeep P, Med l i n S, et al . Reflectance spectrophotometry in the gastrointestinal

tract: l im itations and new appl ications. Am J Gastroenterol 2002 ; 97 : 2 780-2 784. 2 2 . Peck JW, Gibbs FA. Assay of premorbid murine jejunal fi brosis based on mechanical changes after

X-radiation and hyperthermia. Radiat Res 1 987; 1 1 2 : 5 2 5-543. 23 . Peck JW, Gibbs FA. Mechanical assay of consequential and primary late rad iation effects i n murine

small intestine : alpha/ beta analysis . Radiat Res 1 994; 1 38 : 2 72-28 1 . 24 . Hendriks T, Mastboom WJB. Heal ing of experimental intestinal anastomoses, parameters for repair.

Dis Colon Rectum 1 994; 3 3 : 89 1 -901 .

2 5 . Brasken P. Heal ing of experimental colon anastomos is . Eur J Surg 1 991 ; 566 (supp) : 1 - 5 1 . 26 . Koruda MJ, Rolande l l i RH. Experimental studies on the heal ing of colonic anastomoses. J Surg Res

1 990; 48: 540-5 1 5 . 2 7. Mansson P, Zhang XW, Jeppsson B, et al. Anastomotic heal i ng in the rat colon : comparison between

a radiolog ical method, breaking strength and bursting pressure. Int J Colorectal Dis 2002 ; 1 7 : 420-42 5 .

2 8. de Waard JW, Wobbes T , de Man BM , e t a l . Post-operative levamisole may compromise early heal ing of experimental intestinal anastomoses. Br J Cancer 1 995 ; 72 : 456-460.

29. van der Ham AC, Kort WJ, Weijma IM, et a l . Heal ing of ischemic coloni c anastomosis: fibrin sealant does not improve wound healing. Dis Colon Rectum 1 992; 3 5 : 884-89 1 .

30 . B iert J , Wobbes T, Hendriks T, et al . Effect of i rradiation on heal ing of newly made colonic 83

anastomoses in the rat. lnt J Rad iation Oncology Biol Phys 1 993 ; 2 7: 1 1 07- 1 1 1 2 . 3 1 . Vitolo JM, Cotrim AP, Sowers AL, et al . The stable n itroxide tempol faci l itates sal ivary g land

protection during head and neck irrad iation in a mouse model . Cl in Cancer Res 2004; 1 0: 1 807-1 8 1 2 . n

::,

(tl ..., U1

)> :J PJ Vl

8 3 0 Vl (tl Vl

:J

PJ

3 C: ..., :J (tl

3 0 0.. � Q U1 X U1

C') -<

l ntraoperative assessment of m icroperfus ion with

vi s i b le l ight spectroscopy in oesophageal and colorectal

anastomoses - a feas i b i l ity study

A. Karl i czek 1 ,2 , D .A. Benaron3 , P.C. Baas2 , C .J . Zeebregts l , A. van der Stoe l2 ,

T. Wiggers 1 , J . Th . M . Pl u kkerl , G .M . van Dam 1 ,4

' Department of Surgery, Un iversity Medical Center Gron ingen, Gron ingen, The Netherlands 2Department of Surgery, Marti n i Hospital G ron ingen, Gron ingen, The Netherlands

3 Stanford University School of Medicine, Palo Alto, CA, USA 4Bio0ptical I magi ng Center Groningen (BICG), University of Groningen, The Netherlands

Eur Surg Res 2008 ; 4 1 : 303-3 1 1

Abstract

Background� We evaluated the technical feas ib i l ity and stabi l ity of measurements us ing vi s ible l i ght spectroscopy (VLS) to measure m icrovascu lar oxygen satu ration (StO2) i n gastroi ntesti nal anastomoses . Methods: In consecutive oesophageal (n= 1 4) or colorectal (n= 30) resections, during which an uncompl icated anastomosis was performed, measurements of serosal StO2 were performed duri ng the procedure. Resu lts: I n oesophageal resect ions, median (± standard error) StO2 was stable before and after anastomosis in the proximal oesophagus (before: 66.0% ± 4.6%, after: 68.3% ± 6 .0%) and the gastric condu it (before : 70 .6% ± 8.6%, after: 69.8% ± 8.0%). Mean colorectal StO2 before and after anastomosis i ncreased in the proximal part (7 1 . 3% ± 8.4% to 76.6% ± 8.2%, (p< 0 .005)) . Mean StO2 in the d istal part remained stable (72 .4% ± 6.6% to 74.8% ± 6. 7%). Conclusions: VLS is a feas ible and fast method for i ntraoperative assessment of microperfus ion of the serosa in oesophageal and co lorectal anastomos is . Futu re cl i n ical stud ies wi l l defi ne its role in the pred ict ion of anastomotic leakage .

I ntroduction I n gastrointesti nal surgery leakage o f anastomoses i s a chal leng ing problem because of its strong re lation with raised mortal i ty and morbid ity 1 ,2 . The h ighest i ncidence of anastomotic leakage is fou nd at the most prox imal and d istal parts of the d igestive tract, i . e . the s ites of oesophageal and colorectal anastomoses. I ncreased stra in and l im ited vascular su pply at these anastomotic s ites are the two main reasons of leakage, especial ly i n the absence of a serosal layer3 .4 . Apart from local risk factors , several systemic ri sk factors also contri bute to the occurrence of anastomotic fai l u re , of wh ich smoking, card iovascu lar d isease, gender, age and malnutrition are the most important2 ,5-8 • Most of these general factors suggest vasoconstrictive effects to be the cause of local ischemia and are thereby an important cause of anastomotic dehiscence.

Apart from these co-ex isti ng morbid ities the surg ical ly induced altered vascu lar supply during resection of the d iseased bowel segment, inc l ud ing its su pplying arteries , comprom ises the microci rcu lation at both ends of the anastomosis and is as such respons ib le for the h igher rate of leakage compared to smal l and other large bowel anastomoses9, 1 0 . I n oesophageal resection the perfus ion of the gastric condu it is obvious ly compromised as it is on ly based on the r ight gastroepiploic artery due to resection of the majority of feed ing vesse ls . In add it ion , the gastric condu it is transposed from its anatomical abdominal position i nto the thoracic cavity and cervical reg ion. Thus the t issue oxygenation of the prox imal t ip of the gastric condu it depends on i ntramural vascu larisation on ly1 1 - 1 4 . I n colorectal anastomos is the vascular supply of the rectal stu mp is comprom ised by resection of the proximal feedi ng s igmoidal vessels , the superior haemorrhoidal artery and the marg i nal artery.

I n cu rrent dai ly practice i ntraoperative i ntesti nal microci rcu lation is assessed by the su rgeon by means of subjective judgement of co lou r and pu l sating b lood flow of the i ntestinal marg ins at the d issected end of the bowel segments. As such, this practice i s cons idered the gold standard to determ ine viab i l ity of bowel ends. There are, however, no sound data avai lable on sens it ivity and specific ity and thus re l iabi l ity of this assessment can be q uestioned. Several c l in ical stud ies are conducted in order to i nvestigate mucosal perfus ion , inc lud ing Doppler u ltrasound 1 5 , laser Doppler flowmetryl l - 1 3 , 1 6 , scann ing laser flowmetry1 4, 1 ?, 1 s, fl uorescence videography 1 9 and intramucosal pH measurements20-22 . However, none of these methods is widely accepted due to various l iabi l it ies (Table l ) .

Recently, a new techn ique has been deve loped for detection of i schem ia, known as vis ible l ight spectroscopy (VLS)26. VLS uses the pri nc ip le of absorbance and scatteri ng of l i ght i n biological t issues. VLS measur ing t issue oxygenation has been previous ly descri bed and thorough ly been val idated us ing the T-Stat-system with various probes and in different t issues27, 2s . In humans T-Stat was demonstrated to be an usefu l and val idated instrument for assessment of mucosal perfus ion i n the pathophys iology of several d i sorders , such as gastric u lcer d isease, i nflammatory

87

bowel d isease, chron ic hepatit is and portal hypertension and general ized hypoxia29-3 1 •

VLS has been used endoscopical ly to eva luate mucosal perfus ion in critical care and operative settings during AM-repair3 2 , for assessment of effects of vasoactive med ication33 and in patients undergoing card iopul monary bypass34 . However, no data on the variabi l i ty of intraoperative VLS measu rements are avai lable. The practical appl icab i l ity and prior val idation in comprom ised mucosal m icrocircu lation stud ies in both an i mals and humans lead us to evaluate the techn ical feas ib i l ity of VLS for intraoperative measurement of serosal m icroperfus ion of

88 h igh-r isk gastro intestinal anastomoses. Because oesophageal and colorectal anastomoses have the h ighest inc idence of anastomot ic deh i scence, obvious ly

� (1) ..... O')

"Tl (1) � S!.

�-0

� 0 (1)

:;:;" (D

<

Technique

Laser Doppler flowmetry11-15, 23-25

Scanning laser Doppler flowmetry 1&-18

Near-infrared spectrometry35

Visible light spectroscopy 26-31, 37

(e.g. T-Stat 303, EMPHO II)

Fluorescence videography 19

lntramucosal pH20-22

Assets

• easy to perform during operation

• evaluation of larger surfaces

• evaluation of deep tissue samples

• low standard deviation of measurements

• no pulsatile blood flow necessary

• evaluation of small tissue samples

• no tissue contact required • no pulsatile blood flow

necessary

• no tissue contact required • evaluation of larger

surfaces

• high accuracy in prediction of anastomotic leakage

Liabilities

• repeated measurements necessary; therefore time consuming

• requires tissue contact, might disturb local blood flow

• pulsatile blood flow required

. requires pulsatile flow

. relatively complicated system

• relatively long duration of measurement (~2 min.)

. large tissue samples due to penetration depth of light

• point measurements

• requires injection of fluorescent dyes

• position-dependent measurements

• necessity to leave catheters • no real-time measurements

Table 1 Characteristics of techniques described for i ntraoperative evaluation of tissue oxygenation. Assets and l iabilities described by the authors are l i sted.

th is group of patients wi l l benefit the most of an accurate assessment of serosal microperfus ion. Therefore we selected these patients as our study group. The main issues to be solved were the s ite of measurement in re lation to success ive devascu larisation of t issue during surg ical resection , the types of probes to be used , and the reproducib i l ity of the acqu ired data in serosal or adventitial m icroperfus ion . The aim of th i s study was to develop a practical and usefu l protocol for the use of VLS in h igh-risk gastro-intestinal anastomoses in preparation of a larger study evaluat ing the predictive val ue of St02 for anastomotic leakage. 89

Material and methods

Design of the study The study cohort inc luded a series of patients undergoing e ither oesophageal or colorectal resection without anastomotic leakage in two d ifferent teaching hospitals (Un iversity Medical Center Gron ingen (UMCG) and Martini Hospital Gron ingen (MHG)). A total of e ight oesophageal and four colorectal resections were recorded in the UMCG, wh i le s ix oesophageal and 26 colorectal resections derived from MHG. Tissue oxygen saturation measurements were carried out in all of these pat ients with the use of a vis ib le l i ght spectroscopy (VLS) oximeter. The study protocol was approved by the med ical eth ical committees of the Un ivers ity Medical Center Groningen and the Martin i Hospital Groningen.

Study subjects In total , 44 patients were included in the study, of whom 1 4 underwent oesophageal resect ion and 30 underwent colorectal resection. Patient characteristics, includ ing ind ication for surgery, are shown in Table 2 . Al l but one oesophageal anastomoses were performed after resection of cancer. In one patient, continu ity was restored after emergency resect ion due to perforation. The oesophageal operations were carried out through a thoracolaparotomy with right-s ided resection and reconstruction with a gastric tube in 1 3 cases. The remain ing patient had a transh iatal resection. Colorectal resection was carried out for removal of cancer in 20 cases, for d iverticu l it is or d iverticu los is in five cases, and for restoration of continu ity after Hartmann 's procedure for d iverticu lar abscesses in four cases. Al l anastomoses were located with in 20 cm of the anal verge.

Exc lus ion criteria were major surg ical procedures within 30 days prior to the procedure. In order to e l ucidate reproduc ib i l ity and stabi l i ty of VLS-measurements, and cons idering the smal l number of patients , subgroup analys is of patients who deve loped anastomotic leakage was not considered usefu l and these patients were therefore excl uded from the analys is.

90

,, ro PJ V,

�-0 ...., -, PJ 0

'"Cl ro

<" < r l/)

number of patients male sex age (years) weight (kg) Body Mass Index ASA classification

I I

I l l

type of suture running stapled interrupted

level of anastomosis (cm) >10 cm** 5-1 0 cm** <5 cm**

disease cancer diverticular disease continuity other

operating time (minutes) neoadjuvant treatment

none 5x5 Gy

oesophageal resection

1 4 8 (57%) 64 (47-75) 87 (42-1 1 3) 27 (1 8-35)

2 (1 4%) 10 (72%) 2 ( 14%)

14 (1 00%) 1 1 (1 1 -1 2)* n.a. n.a. n.a.

13 (93%)

1 (7%)

392 (1 34-579)

1 4 (1 00%) n.a.

colorectal resection

30 21 (70%) 66 (35-85) 79 (50-103) 27 (21 -32)

11 (37%) 15 (50%) 4 (1 3%)

11 (41%) 19 (59%)

11 (4-12)** 21 (70%) 6 (20%) 3 (1 0%)

20 (67%) 5 (1 7%) 4 (1 3%) 1 (3%) 1 75 (76-298)

20 (80%) 5 (1 7%)

Table 2 Patient characteristics. Age, weight, anastomosis level and operating time are expressed as mean values, min imum and maximum between brackets. * from the prom. Laryngea. ** from the anal verge

Technical detai ls of measurements The oximeter system that was used cons isted of a vis i b le l ight spectroscopy (VLS) ox imeter (T-Stat"' , model 303 lschemia Detection System ; Spectros Corp. , Porto la Val ley, CA, USA) . The system is both FDA and CE approved . The ox imeter em its wh ite l ight from a hand held probe placed on or near the t issue , col lecting any l ig ht retu rn ing to the probe from the tissue. The l ight trave l l i ng through biolog ical t i ssues i s absorbed by mu lt iple molecules and scattered by cel l u lar and anatomical structu res. Each type of molecule has a characteristic absorbance s pectrum. As oxygenated and deoxygenated haemog lobin have a d ifferent absorption spectrum and haemoglobin is cons idered a critical derivative of perfus ion , t issue m icroperfusion can be assessed us ing a techn ique that detects these d ifferences. The abu ndant haemog lobin i n gastro intesti nal cap i l lary beds accounts for the majority of absorption i n the mucosa! layer. The port ion of the vis ible spectrum from 476 to 5 84 nm (blue-to-ye l low) is

each of the major forms of haemog lobin (deoxyhaemog lobin , oxyhaemoglobin) u s ing d ifferential spectroscopy and least squares fitting to known hemoglobin spectra. In th is range of wave length the l ig ht penetrates to a depth of approximately 1 m m. On the bas is of an algorithm , concentrations of oxy- and deoxyhaemoglobin were calcu lated ; haemog lobin oxygen satu ration (St02) was used as readout. A major advantage of vis ib le l ight i s that it penetrates capi l laries we l l , but not larger vesse ls , resu lting in a cap i l lary-weighted measurement in vivo, most c losely correlated to t issue oxygen satu ration (St02)27,28 . A steri le hand he ld probe (6 mm d iameter x 2 0 cm) placed in the operative fie ld was used in each procedure, ti ssue contact i s not requ i red 91 for measu rement as the instrument corrects for an uneven base l ine and analyses a fu l l spectrum.

Measurements were performed at the intended local isat ion of the future anastomosis j ust after start of the operation , before transpos it ion of the gastric condu it or the descending colon and after completion of the anastomosis (Figu re 1 ). To measure St02 the probe was either held a few m i l l imetres above gauze c leaned serosal or adventitial su rface unti l val ues stabi l ized with in a 5% marg in. At each St02-record ing , non-invasive b lood pressu re , heart rate, temperatu re , use of vasoactive drugs and peripheral oxygen satu ration measu red with a pu lse-ox imeter at the finger or earlobe were s imu ltaneous ly recorded. Probes were re-steri l i sed accord ing to hospital procedure and instructions by the manufactu rer us ing ethylene-oxide steri l i sation .

Data col lection and statistical analysis Data on the occurrence of postoperative comp l ications were prospectively col lected. Fo l low-up was completed 30 days after hospital d ischarge. Serious adverse events (defined as events lead ing to hospital ization , undu ly pro longation of hospital ization , danger of l ife , necess ity of intervention or death) due to the use of the measurement protocol and the VLS-system or probes were recorded. Data were analyzed with SPSS 1 2 .0. 1 software. Differences in d istri bution of data were analyzed us ing one sample Kol mogorov-Smirnov test for continuous variables and X2

test for nom inal variables . Comparison of St02 between groups was tested by Student-T- test and Mann-Wh itney U test when appropriate. Wi l coxon s igned ranks test was used for re lated variables. Un ivariate analys is was used to determine explanatory variables for the independent variable St02 . Statistical s ign ificance was set at the 5% leve l .

,, (D Q) Vl

92

n ::::, Sll � ro � en

"Tl (1) Sll Vl

5r.

�-0 -,.,

� iiJ 0

"O ro � � <" ro <

A

B

left gastroepiploic artery

right gastroepiploic artery

Figure 1 Measurement sites (grey dots) during oesophagectomy with gastric conduit reconstruction

and oesophagogastrostomy (A) and during rectosigmoidal resection (B).

Resu lts

Surg ical outcome

Morbid ity rate in patients who u nderwent oesophageal resect ion was h igher than i n those that u nderwent colorectal resect ion . Especial ly, the numbers o f wou nd i nfections , pneumon ias and card iac arrhythm ias were h igher after oesophageal resection and this trans lated in a longer ICU and hospital stay. Further detai l s with regard to postoperative compl ications are out l ined in Table 3 .

Tissue oxygenation saturation measurements

The StO2 record ings with the VLS-oximeter us i ng a steri le handheld probe on serosal or adventitial t issue were easy to perform during the operation and took in general 5 to 1 0 seconds for each measu rement. Pos ition ing of the probe was performed by the operat ing su rgeon and data were recorded by the i nvestigators (AK and AvdS). No adverse event related to the use of the device was recorded in any of the operat ions .

With regard to the oesophageal anastomoses mean StO2 measurements of the gastric serosa and oesophageal adventitia are shown in F igure 2a. During construction of the gastric conduit oesophageal adventitia was not avai lable for measurements as the thoracic cavity was temporari ly closed during the abdominal phase of the operation . Throughout the operation no s ign ificant changes in StO2 occurred and standard deviation values were between 4.6% and 8.6% (Figure 2b) . With regard to the colorectal anastomoses, mean StO2 val ues are represented in Figu re 3a and median values with the i nterquart i le range and 5th and 95th percent i les in the resection just before position ing of the stapled end in the position of the anastomosis

oesophageal resection colorectal resection

ICU stay (days) 4.5 (0-1 1 ) 0.3 (0-7) hospital stay (days) 21 (9-37) 1 4 (4-43) reoperation 2 (6%}* wound infections 2 (1 4%) pulmonary complications

none 8 (61 %) 30 (91 %) pneumonia 4 (31 %) 2 (6%) pleural effusion 1 (8%) 1 (3%)

cardiac complications none 9 (69%) 32 (94%) ischemia 1 (8%) decompensation 2 (6%) arrythmia 3 (23%)

Table 3 Postoperative complications and follow up in patients after oesophageal and colorectal resection. *Two patients were reoperated. In one case anastomotic leakage was suspected based on sepsis, but no intra-abdominal focus was found. The other patient was reoperated as fascial dehiscence occurred.

93

94

"Tl (D PJ Ill

A

B

80

60

50

le:::= 66.0 (sd 4.6)

73. 7 (sd 3.3)

prox. gastrfo -oph. oonduit

r

68.3 (sd 6.0)

.:::::::;:. �

) 69.8 (sd 8.0)

70.6 (sd 8.6)

proX. IIIIIDSf. dist. IIIIIDSf.

Figure 2 St02 (% oxygenated Hb) during esophageal resection. A: Mean values (standard deviation) for the oesophagus proximal of the resection l ine after thoracotomy, for the proximal part of the gastric conduit in abdominal position and for the completed anastomosis are shown. B: Boxes represent first and third quartile with the med ian central. The whiskers represent 51h and 951h percenti les.

box-whi sker p lot in figure 3B. Mean St02 of the prox imal colon was measured after Mean St02 of the serosa in the proximal part i ncreased s ign ificantly du ring resect ion (from 72 . 3 to 76.6%).

There was no re lation of St02 i n both types of h igh-ri sk gastro-enteral anastomoses with any of the hemodynamic parameters or with co-morbid ity in un ivariate analys is .

Discuss ion

I n general , vi s ib le l ight spectrometry (VLS) for eval uation of bowel saturation i s a safe, fast and easy to perform real-time intraoperative procedu re. The methodo log ical stabi l ity and adequacy of VLS measure ments for ischemia detection has been demonstrated in previous stud ies for mucosal St0228-34 , but its i ntraoperative su itab i l ity in the upper and lower ends of the gastroi ntesti nal tract has not been

A

B 90

80

11170

60

71.9 (sci 8.7)

• L04

72.7 (sci S.3

Figure 3 StOz (96 oxygenated Hb) during rectos igmoidal resection. A: Mean values (standard deviation) for d istal colon/ rectum and for proximal colon during the resection and for the completed anastomos is are shown. B: Boxes represent first and th i rd quartile with the med ian central. The whiskers represent 5th and 95th percentiles. Significant differences (p< 0,05, Student-T-test) are indicated

prox. colon dist. colon prox. anast. dist. anast. *.

descri bed yet. I n this feas ib i l ity study we tested this techn ique during h igh risk gastro-i ntesti nal anastomoses and fou nd out that even in tedious procedures such as oesophageal and colorectal resections serosal haemoglobin satu ration can be determined accurately i ntraoperative ly with the use of VLS. Serosal St02 as measu red with VLS showed low standard deviations and as such appears to be most su itable for assessment of m icroperfusion at both anastomotic s ites.

With regard to the VLS measurements in oesophageal and gastric mucosa at the start of the operat ion, the mean val ues and standard deviations in the cu rrent study corroborate with data found du ring gastroscopy as measured by others2s , confirming the absence of an effect of anesthes ia on m icroperfusion. After creat ion of the gastric condu it and the anastomosis , mean val ues are sti l l comparable with those at the start of the operation but standard deviations are h igher i n our study. This m ight be explained by changes due to resection of feed i ng vessels , hand l ing and

95

96

man ipulation of the gastric condu it du ring operation . I n contrast to s im i lar stud ies evaluating i ntraoperative perfus ion of the gastric condu it serosa us ing d ifferent methods (scann i ng laser Doppler flowmetryl 4, laser Doppler flowmetry1 1 ) , we did not find a decrease in StO2 after formation of the gastric condu it. On the other hand , the i ncreas ing standard deviat ions duri ng the procedure as fou nd by others were also detected in the cu rrent study1 4 .

There are no stud ies reporti ng the i ntraoperative use of VLS on rectosigmoidal serosa. However, VLS has been used endoscopical ly i n the rectos igmoid. I n two previous stud ies , s im i lar mean val ues for mucosa saturation but lower standard deviations were fou nd (3 . 5 -4%27,29 vs 6-8% in our study), whi l st in two other studies lower mean StO2 (36 .8% 30 and 4 1 . 9%35) but s im i lar standard deviation val ues were noticed . From the pub l i shed resu lts no defi n ite conclus ions to the differences in mean StO2 can be derived so far without conduct ing a comparative study us ing both devices .

Besides VLS, several other new techn iques analyz ing m icroperfus ion have been used . Compared to other techn iques , VLS uses relative ly short wavelengths (476 to 584 nm, blue-to-ye l low) for oximetry in smal l tissue vo l umes . Th is resu lts in a more d isti nct d ifference i n the spectrum of the major forms of haemoglobin compared to techn iques based on (near) i nfrared wavelengths s uch as N IRS and pu lse ox i metry, which use larger range of wave lengths i n the infrared spectrum (630 - 9 1 0 nm) . Th is broader band of wave lengths resu lts i n re latively good transm iss ion of the l i ght i n t issue and thus deeper t i s sue penetration of the l i ght e m itted and reflected. The consequence of th is deeper t issue penetration is a measu rement in a larger t issue vol ume with consequently less d i sti nctive hemoglob in bands due to scatteri ng i n other t issue com ponents and in l ess precise point measu rements compared to VLS27. Thus VLS, when compared to i n struments eval uat ing longer wavelengths , measures shal low, smal l t issue vol umes usual ly not penetrat ing blood vessels36 .

Techn iques for mon itori ng systemic ischemia such as m ixed venous saturat ion, card iac i ndex and i ntramucosal tonometry requ i re indwe l l i ng catheters . Early local i schemia as duri ng a gastro intesti nal resection may not be detected in t ime by systemic mon ito ri ng and thu s the use for detecti ng local gastroi ntesti nal ischem ia i s very l im ited32 . However, as poi nted out i n a recent study compari ng techn iques for m icroci rcu latory eval uation s uch as vis i ble l ight s pectroscopy, Clark-type e lectrodes and i ntravital m icroscopy to systemic c i rcu latory parameters , tissue oxygenation is i nfl uenced by systemic hypoxia and hypovolemia and therefore systemic parameters should be eval uated for correct i nterpretation of t issue oxygenation 37. I n our c l i n ical study we could not demonstrate a corre lat ion between t issue oxygenation and systemic parameters, poss ibly due to the fact that in a s u rg i cal procedure resection of vascu lar supply of the bowel , local i schemia m ig ht play a more prominent role than systemic parameters. Near-i nfrared s pectroscopy (N I RS) showed comparable mean values and s l ightly lower standard deviat ions in a s im i lar study des ign as ours (7 1 % sd 4.2), but was on ly used in a small number of pat ients35 . In two other studies us ing (scanni ng) laser Doppler flowmetry1 7,36 mean val ues were not comparable due to measurement val ues in d ifferent un its , both stud ies however demonstrated a clear but

not s ignificant decrease i n saturation dur ing resection of the rectos igmoid . One study1 7 showed an i ncrease i n standard deviat ion du ring resection s imi lar to our find ings .

I n our study populat ion we cou ld not demonstrate decreas ing oxygenation of the serosa as found i n some but not al l aforementioned studies 1 7,3 5 ,3 8 • I n colorectal resection we even found a s ign ificant rise in oxygenation of the proximal part of the anastomosis. Th is might be attributed to our se lection of uncompl icated cases and was not related to other variables. Agai n , an i ncreas ing standard deviation du ri ng the resection seems to be a more common observation . It should be stressed, however, 97 that comparison with other publ i shed studies shou ld be made with caut ion due to cons iderable d ifferences in methodology or study des ign.

I ncreas ing standard deviat ions dur ing both surg ical ly and anaesthesiolog ical ly demanding oesophageal and rectosigmoidal resections can poss ibly be explained by reactive hyperaemia due to man ipu lation and resection of arteries. Difficulty to obtai n stable measurements , and thus a certain variabi l ity, can be due to movement artefacts of the probe (heart beat or breath i ng) , contami nation of the serosa with b lood (especial ly low in the pelvic cavity and h igh in the thoracic cavity), and d ifficult ies to obtai n access to restrained space at the measurement s ite. However, pre l im inary data of a diagnostic accu racy study for the pred ictive value of VLS for anastomotic leakage, currently carried out at our center, point towards the observation that patients with anastomotic leakage tend to have a decreased St02 during the procedu re at the serosal s ite. So far, no fi rm concl us ions can be d rawn as these patients were not i nc luded i n the study evaluating stabi l ity and feas ib i l ity of vis ib le l i ght spectroscopy measurements i ntraoperatively.

Th i s feas ib i l ity study has shown that i ntraoperative evaluation of anastomotic microperfus ion by VLS is feas ible and an easy to perform procedu re in oesophageal and colorectal surgery. Moreover, i ntraoperative record ings of serosal St02 can be obtained with smal l variat ion both at the upper and lower gastrointesti nal tract and is the recommended l ayer for evaluation of microperfus ion . Futu re studies wi l l g ive an answer whether these record ings are pred ictive for anastomotic l eakage both for oesophageal and colorectal anastomoses.

98

.,, (D llJ V,

CT

;:::;: -< 0 ......

;:?. ;;1 0 "C (I)

<' (I) < r Vl

References

1 . Eri ksen MT, Wibe A, Norste in J, et al . ; Norwegian Rectal Cancer Group. Anastomotic leakage fo l lowing routi ne mesorectal excis ion for rectal cancer in a national cohort of patients. Colorectal Dis 2005 ; 7 : 5 1 - 57.

2 . Karl RC, Schreiber R , Bou lware D , e t al. Factors affecting morbidity, mortality and survival i n patients undergoing Ivor Lewis esophagogastrectomy. Ann Surg 2000; 5: 635-643 .

3 . Whooley BP, Law S , Murthy SC, e t a l . Analys is o f reduced death and compl ication rates after esophageal resection. Ann Surg 200 1 ; 2 3 3 : 338-344.

4 . Alves A, Panis Y, Trancart D, et al . Factors associated with c l in ically s ign ificant anastomotic leakage after large bowel resection : mu ltivariate analys is of 707 patients. World J Surg 2002; 26: 499-502 .

5. Vik lund P, Lindblad M, Lu M, et al . Risk factors for compl ications after esophageal cancer resection : a prospective popu lation-based study. Ann Surg 2006; 243 : 204-2 1 1 .

6 . Vig nal i A, Fazio VW, Lavery IC, et al . Factors associated with the occurrence of leaks i n stapled rectal anastomoses: a review of l 0 1 4 patients . J Am Col l Surg l 997; l 8 5 : 1 1 3- 1 2 1 .

7. Makela JT, Kiviniemi H, Lait inen S. Risk factors for anastomotic leakage after left sided colorectal resection with rectal anastomosis. Dis Colon Rectum 2003 ; 46: 6 53-660.

8 . Sorensen LT, Jorgenson T, Ki rkeby LT, et al . Smoking and alcohol abuse are major risk factors for anastomotic leakage in colorectal su rgery. Br J Surg 1 999; 86: 92 7-93 1 .

9 . G ray H et al. Gray's Anatomy. Vessels of the esophagus . 892. Running Press Book Publ ishers, Phi lade lphia 1 974.

1 0. Dworki n MJ, Allen-Mersh TG. Effect of inferior mesenteric artery l igation on blood flow in the marginal artery-dependent sigmoid colon. J Am Col l Surg 1 996; 1 83 : 3 5 7-360.

1 1 . Ambrosetti P, Robert J , Mathey P, et al. Left-sided colon and colorectal anastomoses : doppler u ltrasound as an aid to assess bowel vascu larization. Int J Colorectal Dis 1 994; 9: 2 1 1 -2 1 4.

1 2 . Redae l l i CA, Sch i l l i ng MK, and Carrel TP. l ntraoperative assessment of intestinal viabi l i ty by laser Doppler flowmetry for surgery of ruptured abdominal aortic aneurysms. World J Surg 1 998; 22 : 283-289.

1 3. Pierie JP, De Graaf PW, Poen H , et al . Impaired heal ing of cervical oesophagogastrostomies can be pred icted by estimation of gastric serosal blood perfusion by laser Doppler flowmetry. Eur J Surg 1 994; 1 60: 5 99-603.

1 4 . Krohg-Sorensen K, Line PD, Horn RS, et al . l ntraoperative pred iction of ischaemic injury of the bowel : a comparison of laser Doppler flowmetry and t i ssue oximetry to histolog ical analys is . Eu r J Vase Surg 1 992 ; 6: 5 1 8-524.

1 5. Miyazaki T, Kuwana H , Kato H , et al . Predictive value of b lood flow in the gastric tube in anastomotic i nsufficiency after thoracic esophagectomy. World J Surg 2002; 26: 1 3 1 9- 1 323 .

1 6 . Boyle NH , Pearce A , Hunter D, et a l . l ntraoperative scann ing laser Doppler flowmetry in the assessment of gastric tube perfusion during esophageal resection. J Am Col l Surg 1 999 May; 1 88: 498-502.

1 7 . Boyle NH, Manifold D, Jordan MH, et al. l ntraoperative assessment of colon ic perfusion us ing scann ing laser Doppler flowmetry during colonic resection . J Am Col l Surg 2000; 1 9 1 : 504-5 1 0 .

1 8 . Hajivass i l iou CA, Greer K, F isher A, et a l . Non-invasive measurement of colonic blood flow d istribution us ing laser doppler imaging. Br J Surg 1 998; 8 5 : 52-55 .

1 9 . Toens C , Krones CJ , B l um U , et a l . Val idation of IC-VIEW fl uorescence videography in a rabbit model of mesenteric ischaemia and reperfusion. lnt J Colorectal Dis 2006; 2 1 : 332-338 .

20 . Mi l lan M , Garcia-Granero E, F lor B, et al . Early pred iction of anastomotic leak in colorectal cancer surgery by intramucosal pH. Dis Colon Rectum 2006; 45: 5 95-60 1 .

2 1 . Senagore A, Mi Isom JW, Wais haw RK, et al. I ntramural pH: a q uantitative measurement for predicti ng colorectal anastomotic heal ing . Dis Colon Rectum 1 990; 33: 1 75- 1 79.

22. Tarui T, Murata A, Watanabe Y, et al . Earl ier prediction of anastomotic i nsufficiency after thoracic esophagectomy by intramucosal pH. Crit Care Med 1 999; 2 7: 1 824- 1 83 1 .

2 3 . Krohg-Sorensen K, Lunde OC. Perfusion of the human d istal colon and rectum evaluated with endoscopic laser Doppler flowmetry. Methodologic aspects. Scand J Gastroenterol 1 993; 28 :

1 04-1 08. 24. Bro l in RE, Semmlow JL, Sehonanda A, et al . Comparison of five methods of assessment of i ntestinal

viabil ity. Surg Gynaecol Obstet 1 989; 1 68: 6-1 2 . 2 5 . Line PD , Mowincke l P , Lien B, e t a l . Repeated measurement variation and precision o f laser Doppler

flowmetry measurements . M icrovasc Res 1 992 ; 43 : 285-293 . 26 . Benaron DA, Parach ikov IH , Cheong WF, et al . Des ign of a visible- l ight spectroscopy c l in ical t issue

oximeter. J B iomed Opt 2005 ; 1 0: 44005- 1 -44005-9. 2 7 . Benaron DA, Parach ikov IH , Cheong WF, et al . Quantitative cl i n ical non-pulsati le and local ized vis i l ble

l ight oximeter: Design of the T-Stat™ Tissue Oximeter. SPIE Optical Tomography Spectroscopy Tissue V 2003; 495 5 : 3 5 5-68.

28. Benaron DA, Parach ikov IH , Friedland S, et al . Continuous, noni nvasive and localised microvascu lar 99

tissue oximetry using vis ible l ight spectroscopy. Anesthesiology 2004; 1 00: 1 469- 1 475 . 29 . Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosa! perfusion in

the gastrointestinal tract. Gastrointest Endosc Cl in N Am 2004; 1 4 : 5 39-5 5 5 . 30 . Friedland S, Benaron D , Parachi kov I , e t a l . Measurement o f mucosal hemoglobin oxygen saturation i n

t h e colon by reflectance spectrophotometry. Gastrointest Endosc 2003; 5 7: 493-497. 3 1 . Ramirez FC, Sukhdeep P, Medlin S , et al . Reflectance spectrophotometry in the gastrointestinal tract:

l imitations and new appl ications. Am J Gastroenterol 2002 ; 97: 2 780-2 784. 3 2 . Temmesfeld-Wol lbruck B, Szalay A, Mayer K, et al . Abnormal ities of gastric mucosal oxygenation in

septic shock: partial responsiveness to Dopexamine. Am J Respir Crit Care Med 1 998; 1 5 7: 1 586- 1 592.

33 . Fournell A, Schwarte LA, Scheeren TWL, et al. Cl inical evaluation of reflectance spectrophotometry for the measurement of gastric microvascu lar oxygen saturation in patients undergoing card iopulmonary bypass. J Card iothorac Vase Anesth 2002; 1 6 : 576-58 1 .

34 . Hirano Y, Omura K, Tatsuzawa Y, et al. Tissue Oxygen Saturation during colorectal surgery measured by near-infrared spectroscopy: pi lot study to pred ict anastomotic complications. World J Surg 2006; 30: 45 7-46 1 .

3 5 . Kon ish i K, Akita Y, Yosh ikawa N, et al . Rectal mucosal hemodynamics, evaluated by reflectance spectrophotometry, in patients with chronic hepatitis. J Gastroenterol 1 998; 33 : 477-48 1 .

36 . Liu H , Chance B, H ie lscher AH, et al. I nfluence of b lood vesse ls on the measurement of hemog lobin oxygenation as determined by t ime-resolved reflectance spectroscopy. Med Phys 1 995 ; 22 : 1 209- 1 2 1 7.

3 7. Duchs R, Foitzik T. Poss ible pitfal ls in the interpretation of m icrocirculatory measurements. Eur Surg Res 2008; 40: 47-54.

38. Vignal i A, Gianotti L, Braga M , et al . Altered microperfus ion at the rectal stump is predictive for rectal anastomotic leak. Dis Colon Rectum 2000; 43: 76-82 .

l ntraoperative assessment of m icroperfu sion with

v is ible l ig ht spectroscopy for pred iction of

anastomotic leakage i n colorectal anastomoses

A. Karliczek 1•2

, D.A. Benaron3 , P.C. Baas2, C.J. Zeebregts 1 , T. Wiggers 1 ,

G.M. van Dam 1•4

•

'Department of Surgery, University Medical Center Gron ingen , G ron ingen , The Netherlands 2Department of Surgery, Mart in i Hospital Groningen, Gron ingen, The Netherlands

3Stanford Un iversity School of Medicine, Palo Alto, CA, USA 4Biooptical Imaging Center Gron ingen (BICG), Univers ity of G roningen, The Netherlands

Submitted

Background: Anastomotic leakage in colorectal surgery i s associated with i ncreased morbid ity and mortal ity. There is, however, no accu rate tool to predict its occurrence. We evaluated the predict ive value of Vis ible Lig ht Spectroscopy (VLS), a nove l method to measure t issue oxygenation (St02) for anastomotic leakage of the colon. Methods: In 77 colorectal resections with anastomosis between 2-30 cm from the anal verge St02 was measured in the colon and rectu m before and after construction of the anastomosis , together with a reference measurement at the level of the caecum. Data on the occu rrence of postoperative compl ications were col lected. Results : I n 1 4 patients ( 1 8%) anastomotic leakage was seen. During the operation St02 i ncreased i n the proximal part of the anastomoses without leakage but not in anastomoses showing leakage (mean St02 72 . 1 sd 9 .0 - 76. 7 sd 8.0 (p:::; 0 . 0 5) vs 73 .9 sd 7.9 - 73. 1 sd 7 .4). I n anastomoses showing l eakage , s ign ificantly lower StOs was recorded i n the coecum (73.6 sd 5. 7 in non-leakage vs 69 .6 sd 5 .6 , p:::; 0.05) and showed to be a predictor of anastomotic leakage. Conclusion: St02 as measured by VLS seems a potential ly usefu l tool in predict ing anastomotic leakage after colorectal anastomosis. Fu rther studies wi l l set its defi n ite role in c l i n ical practice.

Introduction Anastomotic leakage is a major c l i n ical problem i n gastrointestinal surgery. The reported incidence in general ranges between 2% and 1 9% 1 ,2 . Occurrence of anastomotic leakage is associated with cons iderable early morbid ity and mortal ity3.4, and also with a worse oncological outcome i n the longer term 5 • I n several stud iesG,7, it was hypothes ized that ischemia and increased strain on the colon anastomosis play important ro les i n the development of anastomotic leakage. Furthermore, several c l i n ical studies us ing uni- and mu ltivariate analys is identified ri sk factors for 103 anastomotic leakage primari ly associated with card iovascu lar d isease , an i nd i rect ind icator of ischemia8- 1 2 . Apart from compromised systemic b lood supply, the surg ical ly induced altered vascu lar supply due to resection of the d iseased colorectal segment and its s upplyi ng arteries compromises the m icroci rcu lat ion at both ends of the anastomosis. Th is alteration in blood supply is main ly respons ib le for the h igher rate of leakage of pe lvic anastomoses compared to smal l and other large bowel anastomoses 1 3.

l ntraoperative i ntesti nal m icroci rcu lation is estimated during the operation by the surgeon by means of subjective assessment of the colour of the t issue and pu lsating blood flow of the i ntesti nal marg i n s at the d issected end of the bowel segments. Currently, th is practice is sti l l cons idered the gold standard to dete rm ine viab i l ity of bowel ends. However, i n a recent study performed at our institution on the predictive val ue of the su rgeons ' r isk assessment of anastomotic leakage, it was demonstrated that this j udgement is un re l iable (unpubl i shed data).

In several cl i n ical stud ies various methods to assess mucosal perfus ion as a pred ictor for anastomotic dehiscence have been described, inc lud ing Doppler u ltrasound 1 4- 1 s, laser Doppler flowmetryl ?-20, scann i ng laser flowmetry2 1 , 22 , fl uorescence videography23 , near- infrared spectroscopy24 and intramucosal pH measurements25 ,26 • However, none of these methods became widely accepted due to h igh variab i l ity of measurements (laser Doppler flowmetry), long duration of measurements (scann i ng flowmetry), d ifficult access when measuring i ns ide the pe lvic cavity (Doppler u ltrasound, scann ing laser flowmetry and fluorescence videography) or the necessity of leaving catheters i n s itu ( intramucosal pH) 1 S, 27, 2s .

During the operation preventive measures can be taken to reduce the potential hazardous effects of anastomotic leakage by defunction ing the anastomosis by plac ing a proximal stoma 1 . Therefore it is of paramount importance to predict anastomotic leakage during the i n it ial operation. In th is study we evaluated a new method to measure t issue oxygenation i ntraoperatively, known as vis i ble l i ght spectroscopy (VLS). Compared to near-i nfrared spectroscopy (N IRS) and pu lse oximetry, VLS oximetry u ses shal low-penetrating vis ib le l i ght to measure haemoglobin oxygen satu ration i n th i n , smal l t issue vol umes (denominated in StO2). In previous studies vis ib le l ight spectroscopy has been val idated and proven to be usefu l for the detection of t issue i schemia in hu man gastric and colon ic mucosa29-33

ro llJ 7' llJ

lD ro

1 04

and in murine serosa of the colon34 • A pi lot study, evaluating the feasibi l ity of the technique, has been conducted in our center, showing that intraoperative VLS is feasible and easy to perform with low standard deviation of measurements35 • The current human study was conducted to evaluate the predictive value of VLS for the occurrence of anastomotic leakage in colorectal anastomoses.

Materials and methods Design of the study Between Apri l 2006 and September 2007 al l consecutive patients undergoing elective colorectal resection with planned subsequent anastomosis in the Martini Hospital Groningen and in the University Medical Center Groningen were included. The study protocol was approved by the med ical ethical committees of both hospitals.

Technical details of measurements The oximeter system that was used consisted of a visible l ight spectroscopy (VLS) oximeter (T-Star, model 303 lschemia Detection System; Spectros Corp. , Portola Val ley, CA, USA). The system is both FDA and CE approved. The oximeter emits white l ight from a probe placed on or near the tissue, col lecting any l ight returning to the probe from the tissue. The absorption spectra of oxygenated and deoxygenated haemoglobin are different, and as haemoglobin is considered a critica l derivative of perfusion, tissue microperfusion can be assessed by the use of the visible l ight oximetry technique which detects these differences. Haemoglobin, abundantly present in gastrointestinal cap i l lary beds, is responsible for the majority of absorption in the mucosa of the bowel wal l . An important advantage of visible l ight is that it penetrates capi l laries wel l , but not larger vessels, resulting in vivo in a capi l lary-weighted measurement, most closely correlated to tissue oxygen saturation (St02). No tissue contact is required for measurements, as an uneven basel ine is corrected for by the instrument and a ful l spectrum is analysed. Based on an algorithm, concentrations of oxy- and deoxyhaemoglobin were calculated ; haemoglobin oxygen saturation (St02) was used as readout.

Measurements were performed at the intended local isation of the future anastomosis just after start of the operation, before transposition of the descending colon into the pelvic space and after completion of the anastomosis (Figure 1 ). To measure St02, a steri le handheld probe (6 mm diameter x 20 cm) was placed in the operative field and either held a few mi l l imetres above gauze cleaned serosal or adventitial surface unt i l values stabi l ized within a 5% margin. Measurements were taken approximately 1, 5 cm away from the cutting edge of the proximal and d istal colon after resection of the d iseased specimen. At each St02-record ing non-invasive blood pressure, heart rate, temperature, use of vasoactive drugs and

Figure l Various levels at which VLS measurements were recorded during colorectal resection.

peripheral oxygen saturation measured with a pulse-oximeter at the finger or earlobe was simultaneously recorded.

Data col lection and statistical analysis

Data on the occurrence of postoperative complications were prospectively collected. Anastomotic leakage was considered to be present when described as such at relaparotomy or endoscopy, and/or a postoperative CT scan showing air- or fluid collections or an infiltrate surrounding the anastomosis. Follow-up was completed 3 months after hospital discharge.

Data were analyzed with SPSS 1 5 .0. Categorical variables were compared using X2-test. Differences in distribution of continuous variables were analyzed using one sample Kolmogorov-Smirnov test. Comparison of St02 between groups was tested by Student-T-test and Mann-Whitney-U test when appropriate. Wilcoxon signed­ranks test was used for related variables. Univariate analysis was used to determine explanatory variables for the independent variable St02. Predictive variables in the univariate analysis (p< 0.05) were entered into a logistic regression model to determine independent predictors. Statistical significance was set at the 5% level.

Results General outcome

In 80 procedures VLS measurements were performed. Two patients were excluded as no anastomosis was performed and in one procedure measurements

1 05

t" o·

-PJ ::l PJ V'I 8 3 � n· ro Ill 7' Ill

\0 (I)

� ;:;: ::::, < r (J")

106

(I)

!=). a· 0 -,.,

� :::s � VI

8 3 g_ n

(I) � " fl)

I.Cl (I)

;:;: ::::, < ' VI

number of patients n type of suture sutured 32 (42%)

male sex 42 (55%) stapled 43 (56%) age (years) 66 (33-87) missing 2(3%) Body Mass Index 27 (20-35) level of anastomosis (cm) 13 (2-30)* ASA classification >10 cm** 50 (65%)

I 23 (30%) 5-10 cm** 21 (27%) I I 42 (55%) <5 cm** 6(8%) I l l 10 (13%) disease

missing 2(3%) cancer 49 (64%) cardiovascular disease diverticular disease 15 (20%)

yes 30 (39%) continuity 8 (10%) no 43 (56%) other 5(6%)

missing 4(5%) operating time (minutes) 183 (76-396) diabetes mellitus neoadjuvant treatment

yes 10 (13%) none 53 (69%) no 63 (82%) 5x5 Gy 14 (18%)

missing 4(5%) chemoradiation 3(4%) missing 7(9%)

Table 1 Patient characteristics. Age, weight, anastomosis level and operating time are expressed as mean values and range between brackets. * range. **measured from the anal verge.

could not be performed due to technical problems with the VLS-system. The remaining 77 procedures were further analysed. Epidemiological data are shown in Table 1. In the postoperative course one patient died after colorectal resection. In 14 patients (18%) anastomotic leakage was seen. Other complications, duration of ICU­and hospital admission and reoperations are shown in Table 2. There were no significant differences in patient characteristics between the patients with and

ICU stay (days) 1 .3 (0-50) hospital stay (days) 16 (3-96) reoperation 11 (14%) wound complications 9 (12%) pulmonary complications none 70 (91%) pneumonia 5(7%) pleural effusion 2(3%) cardiac complications none 72 (94%) ischemia 2(3%) decompensation 2(3%) arrythmia 1 (1%)

Table 2 Postoperative compl ications and follow up in patients after colorectal resection.

without anastomotic leakage in the postoperative cou rse. Also, no s ign ificant pred ictors of anastomotic leakage cou ld be found by u n ivariate analys is.

St02 measurements

The changes in mean St02 during the resection are shown in Figu re 2. The refe rence measurements performed at the level of the coecum corre lated wel l with measurements on the serosa of the proximal and d istal part of the anastomosis and showed no s ign ificant changes du ring the resection. Compari ng St02 in anastomoses with and without leakage a s ign ificant rise i n t issue saturat ion after complet ion of the anastomosis i n the prox imal part of the anastomosis is seen i n anastomoses without leakage, wh ich was absent i n anastomoses showing anastomotic leakage (mean St02 72. 1 sd 9.0 - 76. 7 sd 8.0 (p::; 0.05) vs 73.9 sd 7 .9 -73. 1 sd 7.4) A s im i lar rise in St02 recorded i n the d istal part of the anastomosis was seen , but there was no statistical s ign ificance (mean St02 7 1 .4 sd 7.9 - 73.8 sd 8.6 i n non- leakage vs 73.5 sd 6. 7 - 72.4 sd 1 0.4 i n leaki ng anastomoses, n . s .). In anastomoses showing leakage a s ign ificantly lower St02 was recorded in the coecum (73.6 sd 5 .7 i n non- leakage vs 69.6 sd 5 .6 , p::; 0.05). An i ncrease of St02 appeared to be a s ign ificant predictor of anastomotic leakage at u n ivariate analys is , whereas no other predictors cou ld be found.

90

80

C'I 70 0

60

50

*

A

Proximal

B

*

#

A B A B

Distal Refe re nce

Figure 2 Tissue saturation (St02) proximal and d istal of the anastomosis and reference measurements on the caecum performed before (A) and after (B) construction of the anastomosis. The dark bars show values of anastomoses that showed leakage, l ight bars represent anastomoses without leakage. Boxes represent first and th ird quartile with the median central. The whiskers represent 5 th and 95th

percentiles. * Represent sign ificant differences p5 0.05, # p5 0 . 1 .

1 07

�-o· 0 -,., AJ :::s AJ Vl l'""1'

3 g_

ro AJ 7-AJ

IQ (D

;:::;: :::; < ' Vl

1 08

Discussion In colorectal surgery, anastomotic leakage remains a major clinical challenge often leading to increased morbidity and mortality. Reliable intraoperative preventive measures are not available, besides the judgement of perfusion by the operating surgeon, which often lacks predictive power. Among others, predominantly cardiovascular disease, representing systemic atherosclerosis and decreased adaptor mechanisms of the microcirculation upon reperfusion37 , seems to be related to anastomotic leakage. Current systems for detection of microperfusion lack predictive power, or are either not feasible or impractical to handle in an intraoperative setting. Recently, visible light spectroscopy (VLS) has emerged as a novel promising technique to investigate intraoperatively non-invasively the level of microperfusion through measurement of oxygenated and deoxygenated haemoglobin29-32. As such, VLS might be a predictor of anastomotic leakage in colorectal surgery, providing the surgeon with a practical and prognostic tool intraoperatively.

In this study, we found St02 levels, as measured by VLS in human colon tissue during surgery, to be fairly stable and reproducible. This confirms our earlier results in murine colon anastomoses34 and a human pilot study3 5 •

In order to evaluate the predictive value of VLS for anastomotic leakage, St02 was measured at different levels of the colon before and after anastomosis. In patients without anastomotic leakage, a significant rise in tissue oxygenation in colon tissue immediate adjacent to the anastomosis was detected, which was absent in those anastomoses that leaked. This might indicate a significant local refractory response to manipulation and clamping (i.e. ischemia) of the bowel, apparently less effective in patients with anastomotic leakage. This different response can be explained by the inherent physiological refractory mechanisms of the collateral circulation38 • The extent of the response seems to be related to the presence of cardiovascular disease (i.e. atherosclerosis). Atherosclerosis leads to diminished microperfusion caused by impaired endothelial function and early inflammatory events, leading to a decreased (collateral) perfusion and a decreased vasodilatory capacity, as described by Stokes and Granger36 •

Furthermore, we found a decreased St02 in remote colon tissue as measured in the caecum to be predictive for anastomotic leakage, which indicates that systemic factors are related to anastomotic leakage. As described by Carden37 , risk factors for cardiovascular disease (hypercholesterolaemia, hypertension, and diabetes) enhance microvascular alterations caused by ischaemia and reperfusion (1 / R), in colorectal surgery induced by manipulation and clamping. The inflammatory mediators released as a consequence of 1 / R activate endothelial cells in remote sites that are not exposed to the initial ischaemic insult, in our case the caecum. Adaptational responses to 1 / R injury allow for protection of acute ischaemia against the detrimental effects of subsequent, prolonged ischaemia, a phenomenon called ischaemic preconditioning, as demonstrated in an animal model39• So far, the effects

of (remote) ischemia and reperfus ion i n the i ntesti ne have been described i n an imal stud ies on molecular and h istolog ical leve l40 • However, these m icrophys iolog ical and h isto log ical changes have been re lated to t issue satu ration i n on ly one animal study, describ ing reduced t issue satu ration after ischem ia-reperfus ion i nj u ry, which returns to pre- ischemic leve ls after ischem ic precond ition ing i s appl ied4 1 • Our find ings of decreas ing St02 i n the caecum but not the anastomos i s might be explained by such an i schemia-reperfus ion i nj u ry. Th is m ight explain the d ifferential find ings of an impaired 1 / R response in patients with anastomotic leakage and re lates card iovascu lar ri s k factors to impairment of m icroci rcu lat ion in patients with 1 09 colorectal su rgery.

Based on our study resu lts (assuming an incidence of l 0% anastomotic leakage , a St02 decrease of 6% in the caecu m and St02 increase of 6% i n the prox imal anastomosis) , a number of 1 60 patients wou ld be needed to have sufficient power (95%) to demonstrate a c lear effect of St02 being a predictor of anastomotic leakage (alpha = 0 .05). Therefore , th i s study lacks s ufficient power to draw defi n it ive conc lus ions , although. As a consequence , add it ional patient accrual is taking p lace to answer the hypothes ized effect of St02 as pred ictor of anastomotic leakage.

I n conclus ion , an i ncreased St02 i n the colon adjacent to the anastomosis and a decreased St02 i n a colon area remote to the anastomosis as measu red by VLS seems to be a prom is ing method to predict anastomotic leakage in colorectal surgery. A mu lticenter study is needed to answer with sufficient power and s ign ificance the pred ictive value of VLS for anastomotic leakage in colon surgery.

(I) 0.. ;:;· 5·

0 -SlJ ::s SlJ V, ,..,. 3 0 ;:;·

ro ll) 7' Ill

lO (I)

;:;: ::, < r Vl

1 1 0

n :::r PJ

ro --,

ro 0..

0 :::s 0 ....., PJ :::s PJ l/l 8 3 g_ ro PJ 7' PJ LCl (1)

< ' Vl

References

l . Matthiessen P, Hallback 0, Rutegard J, et al. Defunctioning stoma reduces symptomatic

anastomotic leakage after low anterior resection of the rectum for cancer, a randomized

multicenter trial. Ann Surg 2007; 246: 207-21 4.

2. Katory M, Tang CL, Koh WL, et al. A 6-year review of surgical morbidity and oncological outcome

after high resection for colorectal malignancy with and without splenic flexure mobilization.

Colorectal Dis 2007; 10:165-169.

3. Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic

leakage after large bowel resection : multivariate analysis of 707 patients. World J Surg 2002; 26:

499-502.

4. Eriksen MT, Wibe A, Norstein J, et al. ; Norwegian Rectal Cancer Group. Anastomotic leakage

following routine mesorectal excision for rectal cancer in a national cohort of patients. Colorectal

Dis 2005; 7: 51-57.

5. Ptak H, Marusch F, Meyer F, et al. Impact of anastomotic leakage on oncological outcome after

rectal cancer resection. Br J Surg 2007; 94: 1 548-1554.

6. Shikata J I , Shida T. Effects of tension on local blood flow in experimental anastomoses. J Surg Res

1986; 40: 1 05-111 .

7. Vignali A, Gianotti L, Braga M, et al. Altered microperfusion at the rectal stump is predictive for

rectal anastomotic leak. Dis Colon Rectum 2000; 43: 76-82.

8 . Benoist S, Panis Y, Alves A, et al. Impact of obesity on surgical outcomes after colorectal

resection. Am J Surg 2000; 1 79: 275-28 1 .

9 . Kruschewski M, Rieger H, Pohlen U, et al. Risk factors for clinical anastomotic leakage and

postoperative mortality in elective surgery for rectal cancer. Int J Colorectal Dis 2007; 22: 9 1 9-

927.

l 0. Makela JT, Kiviniemi H, Laitinen S. Risk factors for anastomotic leakage after left sided colorectal

resection with rectal anastomosis. Dis Colon Rectum 2003; 46: 653-660.

11. Kasperk R, Philipps B, Vahrmeyer M, Willis S, Schumpelick V. Risk factors for anastomosis

dehiscence after very deep colorectal and coloanal anastomosis. Chirurg 2000; 71: 1 365-1 369.

l 2. Sorensen LT, Jorgensen T, Kirkeby LT, et al. Smoking and alcohol abuse are major risk factors for

anastomotic leakage in colorectal surgery. Br J Surg 1 999; 86: 927-931.

13. Dworkin MJ, Allen-Mersh TG. Effect of inferior mesenteric artery ligation on blood flow in the

marginal artery-dependent sigmoid colon. J Am Coll Surg 1996; 183:357-360.

14. Krohg-Sorensen K, Line PD, Horn RS, et al. lntraoperative prediction of ischaemic injury of the

bowel: a comparison of laser Doppler flowmetry and tissue oximetry to histological analysis. Eur J

Vase Surg 1 992; 6: 51 8-524.

l 5. Brolin RE, Semmlow JL, Sehonanda A, et al. Comparison of five methods of assessment of

intestinal viability. Surg Gynaecol Obstet 1989; 168: 6-12.

16. Miyazaki T, Kuwana H, Kato H, et al. Predictive value of blood flow in the gastric tube in

anastomotic insufficiency after thoracic esophagectomy. World J Surg 2002; 26: 1319-1323.

l 7. Ambrosetti P, Robert J, Mathey P, et al. Left-sided colon and colorectal anastomoses: doppler

ultrasound as an aid to assess bowel vascularization. Int J Colorectal Dis 1994; 9: 2 1 1-2 1 4.

l 8. Redaelli CA, Schilling MK, Carrel TP. lntraoperative assessment of intestinal viability by laser

Doppler flowmetry for surgery of ruptured abdominal aortic aneurysms. World J Surg 1 998; 22:

283-289.

l 9. Kashiwaga H. The lower limit of tissue blood flow for safe colonic anastomosis: an experimental

study using laser Doppler velocimetry. Surg Today 1 993; 23: 430-438.

20. Attard JAP, Raval M, Martin GR, et al. The effects of systemic hypoxia on colon anastomotic

healing: an animal model. Dis Colon Rect 2005; 48: 1460-1 470.

21. Boyle NH, Manifold D, Jordan MH, et al. lntraoperative assessment of colonic perfusion using

scanning laser Doppler flowmetry during colonic resection. J Am Coll Surg 2000; 191 : 504-510.

22. Hajivassiliou CA, Greer K, Fisher A, et al. Non-invasive measurement of colonic blood flow

distribution us ing laser doppler imaging. Br J Surg 1 998; 85: 52-55.

23. Toens C, Krones CJ, Blum U, et al. Validation of IC-VIEW fluorescence videography in a rabbit

model of mesenteric ischaemia and reperfusion. lnt J Colorectal Dis 2006; 2 1 : 3 32-3 3 8.

24. Hirano Y, Omura K, Tatsuzawa Y, et al. Tissue Oxygen Saturation during colorectal s u rgery

measured by near-infrared spectroscopy: pilot study to predict anastomotic complications. World

J Surg 2006; 30: 457-461 .

25. Millan M, Garcia-Granero E, Flor B, et al. Early prediction of anastomotic leak i n colorectal cancer

surgery by intramucosal pH. Dis Colon Rectum 2006; 45: 595-60 1 .

26. Senagore A, Milsom JW, Walshaw RK, et al. Intramural pH: a quantitative measurement for

predicting colorectal anastomotic healing. Dis Colon Rectum 1 990; 33: 1 75- 1 79. 1 1 1

27. Krohg-Sorensen K, Lunde OC. Perfusion of the human distal colon and rectu m evaluated with

endoscopic laser Doppler flowmetry. Methodologic aspects. Scand J Gastroenterol 1 993; 28: l 04-

1 08.

28. Line PD, Mowinckel P, Lien B, et al. Repeated measurement variation and precision of laser

Doppler flowmetry measurements. Microvasc Res 1 992; 43: 285-293.

29. Benaron, D. A., Parachikov IH, Cheong WF, et al. Design of a visible-light spectroscopy clinical

tissue oximeter. J Biomed Opt 2005; 1 0: 44005-1 -44005-9.

30. Benaron DA, Parachikov IH, Friedland S, et al. Continuous, noninvasive and localised

microvascular tissue oximetry using visible light spectroscopy. Anesthesiology 2004; 1 00: 1 469-

1 475.

3 1 . Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosa! perfusion

in the gastrointestinal tract. Gastrointest Endosc Clin N Am 2004; 1 4: 539-555.

32. Friedland S, Benaron D, Parachikov I, et al. Measu rement of mucosa! hemoglobin oxygen

satu ration in the colon by reflectance spectrophotometry. Gastrointest Endosc 2003; 57: 493-

497.

3 3. Ramirez FC, Sukhdeep P, Medlin S, et al. Reflectance spectrophotometry in the gastrointestinal

tract: l imitations and new applications. Am J Gastroenterol 2002; 97: 2780-2784.

34. Karliczek A, Benaron DA, Zeebregts CJ, et al. lntraoperative ischemia of the distal end of colon

anastomoses as detected with vis ible light spectroscopy causes reduction of anastomotic strength.

J Surg Res 2008, E-pub May 8.

3 5. Karliczek A, Benaron DA, Baas PC, et al. lntraoperative assessment of microperfusion with vis ible

light spectroscopy in esophageal and colorectal anastomoses. Eur Surg Res 2008; 41 : 303-3 1 1 .

36. Stokes KY, Granger DN. The microcirculation: a motor for the systemic inflammatory response

and large vessel disease induced by hypercholesterolaemia? J Physiol 2005; 562: 647-653.

37. Carden DL, Granger DN. Pathophysiology of ischaemia-reperfusion inju ry. J Pathol 2000, 1 90:

255-266.

3 8. Power N, Atri M, Ryan S, et al. CT assessment of anastomotic bowel leak. Clin Radiol. 2007;

62:37-42.

39. Colak T, Turkmenoglu 0, Dag A, et al. The effect of remote ischemic preconditioning on healing

of colonic anastomoses. J Surg Res 2007; 1 43: 200-205.

40. Mallick I, Yang W, Winslet M, et al. lschemia-Reperfusion injury of the intestine and protective

stategies. Dig Dis Sci 2004; 49: 1 359-1 377.

4 1 . Mallick I, Yang W, Winslet M, et al. lschemic preconditioning improves microvascular perfusion and oxygenation following reperfus ion injury of the intestine. Br J Surg 2005; 92: 1 1 69-1 1 76.

Anastomotic leakage-

Cl i n i cal observat ions on mechan ical forces and review

of the l iterature

A. Karl iczek 1 , K. Havenga 1 , W. Bleeker2 , T. Wiggers 1 •

' Departm ent of Surgery, University Medical Center Groningen, Groningen, The Netherlands 2Department of Surgery, Wi lhelm ina Hospital , Assen, The Netherlands

Submitted

clin ical observations

Case I

A SO-year o ld male was operated for T3N l Mx rectal carci noma, preoperative ly confi rmed by biopsy, located approx imately 6 cm from the anal verge. A s hort course of rad iotherapy of SxS Gray through 3 fie lds was appl ied one week before an u ncompl icated colorectal resection with a stapled anastomosis 3 cm from the anal verge was performed . On the fi rst postoperative day the patient deve loped fever and leucocytosis , but no abdominal tenderness . On the th ird postoperative day a CT-scan with contrast 1 1 5 enema was performed , as anastomotic leakage was suspected (Figu re 1 ). In the first scan no contrast outs ide of the rectum cou ld be seen , but free air and fl u id around the anastomosis was present. The patient was subsequently asked to perform a val salva manoeuvre and the scan was repeated, reveal i ng substantial leakage of contrast med ium from ins ide the rectum i nto the pe lvic space .

Figure l Computed tomography of a patient showing anastomotic leakage. A shows a CT s l ice of the patient described in case I in rest after application of rectal and intravenous contrast. Indirect signs of anastomotic leakage (ai r and flu id in the perianastomotic area) are present, but no leakage of contrast outside of the anastomosis . B shows the same level shortly after A, but the patient has performed a valsalva manoeuvre between the two scans. The leakage of contrast medium out of the anastomosis into the pelvic space is clearly visible (white arrow).

Case I I

A 7 5 -year o ld m a l e was operated for a T2 N 0 s i gm oid carc i noma. A t co lonoscopy the tu mor was located at 1 8 cm from the anal ve rge . I n a laparoscopic proced u re , t he s i g m oid with i t s m esocolo n was mob i l ized . D ista l ly t he bowe l was transected at the l eve l of the peritoneal refl ection . A stapled anastomos i s was made . Th is pat ient consented to part ic i pate i n a p i lot study i n wh ich a b io-deg radab le

1 16

n ::r Pl

(0

00

Q :::I /"'I � 0 CJ'" Vl (0 < �-0 :::I Vl

polyurethane endoluminal sheath (C-seal) was tested (Figure 2) . This anvil of the

circular stapler was put into this sheath, the open end of the sheath was glued

to the anvi l . After making the stapled anastomosis in the standard way the

sheath was pul led through the anus with the removal of the circular stapler. The

sheath was cut from the circular stapler, leaving approximately S to 6 cm

outside of the anus. We hypothesize that in this way smal l defects in the

anastomosis wil l not lead to leakage. On the second postoperative day the

sheath was no longer visible. Rectal examination was performed at which the

sheath was found in the ampul la and taken out again. Apparently, movements of

the anastomosis had occurred, sufficiently to retract the C-seal into the rectum.

Figure 2 Principle of the (-seal. A bio-degradable polyurethane endoluminal sheath (C-seal) is glued to the head of a usual ci rcu lar stapl ing device and stapled i nto the anastomosis. The sheath is pul led through the anus with the removal of the circular stapler and is cut from the ci rcular stapler, leaving approximately 5 to 6 cm outside of the anus.

Review of the literature In the presented cases a clear influence of mechanical factors working on the

anastomosis seems to be present. We demonstrated the occurrence of anastomotic

leakage during the performance of a valsalva manoeuvre. The fluid and air col lection

around the anastomosis, together with clinical symptoms suggestive for

anastomotic leakage, indicate that leakage had also occurred prior to the date of

performance of the CT-scan in this case. This in itself is not remarkable, as

approximately 1 7% of left-sided colon anastomoses show radiological leakage,

whereas only 30% of those with radiological leakage show clinical signs and

symptoms such as peritonitis, leucocytosis, fever or sepsis. Furthermore, 5% of the

patients with normal contrast enemas showed clinical signs of leakage1 • The second

observation re lates to the d i sappearance of a protective s leeve i nto the anus , stapled into the anastomosis in the course of a pi lot study protocol evaluat ing C-seal , apparently by peristaltic movements of the colon. Th is m ight cause subsequently a strong traction on the rectal stump. These two observations caused us to review the l iteratu re for articles descri b ing mechan ical forces on pe lvic anastomoses in the human.

The l iteratu re on this s ubject can be d ivided i nto two main categories. F i rst we review articles descri b ing research performed on phys io log ical forces i n the i ntact colon and on i ntact anastomoses, and second we evaluate trials descri b ing i nterventions a imed at reduc ing the forces appl ied on the anastomosis.

Physiolog ical forces

Mechanical forces appl ied on an anastomosis are usua l ly descri bed as traction i n the long itudinal axis of the anastomosis (breaking strength) and pressu re from the i ns ide (bursti ng pressu re). Th is way of descri b ing forces on anastomoses orig i nates main ly in animal stud ies, where bursting pressu re and breaking strength are val idated measures to compare anastomoses2-4• As the anastomosis breaks du ring the test the method is not appl icable i n l ivi ng humans.

A study conducted to measure pressu re i n the normal , unprepared colon 5 ,6

showed that i n the descend ing colon between the sp len ic flexure and the s igmoid colon the pressure i s h ighest when compared to the rest of the co lon, reach ing max imum val ues of 1 1 7 mmHg. After bowel preparation the pressure i n this part of the colon rises cons iderably to maximum pressure of 1 60 mmHg, as described in a study7 us ing a s imi lar design. On ly two studies descri be the pressu re i n the colon prox imal and d istal of the anastomosis after colorectal resection and subsequent anastomosis. The fi rst of these studies descri bes pressu re recorded d istal of the anastomosis8 • Duri ng the fi rst postoperative week the max imum pressu re recorded was 90 mmHg. In the second study a pressu re transducer was left in p lace measuring the pressure both proximal and d i stal of the anastomosis in 1 9 patients. Start ing on the fi rst postoperative day the pressu re rose from i n itial ly 3 7 mmHg to l 02 mmHg on the th ird postoperative day9• The on ly pub l i shed study descri b ing bursting pressure i n the human colon tested i n vitro constructed anastomoses i n colon segments . Anastomoses showed leakage at a mean pressure of 88 mmHg and d isru ption at 1 22 mmHg1 0. S imi lar val ues are described for the colon of dogs 1 1 and pigs 1 2 .

To our knowledge there are no stud ies pub l i shed systematical ly describ ing the phys io logical forces i n longitud inal axis of the human colon or s igmoid. However, an early study by Ritchie et al. describes very concise the normal movements of colon contents. I n th is study the observation of a contrast-fi l led d iverticu l um moving in the long itud inal ax is of the colon towards the bowel content i s descri bed , provid ing casu i stic evidence of such a long itud inal movement of the colon wal l 1 3 •

1 1 7

00

Reduction of forces on the anastomosis Some trials have been conducted to evaluate measures to reduce mechan ical

forces on a new anastomosis . One of these measures i s the mobi l i sation of the s plen ic flexure i n order to reduce long itud i nal traction on the anastomos is . A recent study 1 4 descri bes routi ne mobi l i sation of the splen ic flexure compared to no mob i l i sation in 1 00 anastomoses. No d ifference between mobi l i sation and no mob i l i sation was demonstrated . I n order to reduce mechan ical stra in by passage of stool through an anastomosis d i rectly after the operation it is argued that the colon

1 1 8 shou ld be cleansed of faeces by bowel preparation and a defunction ing stoma can be placed proximal to the anastomosi s . Several comparative studies were conducted 1 5 , but no con nection between bowel preparation and anastomotic leakage

00

Q

0 O'" Vl (0

< � o· ::i Vl

cou ld be demonstrated . However, a randomised c l in ical trial showed a considerably h ig her leakage rate i n low pelvic anastomoses without a stoma 1 6. Cons idering the intral u minal pressure app l ied on the anastomos i s , caused by the c losed anal sph incters , it is argued by some authors that this pressu re should be reduced in the first postoperative phase by insert ion of an anal stent preventi ng the sphi ncter to c lose. Two randomised c l i n ical trials were conducted 1 7, 1 s , but none of them fou nd a reduction in anastomotic leakage after insertion of transana l stents .

Discussion In th is c l in ical observation we describe anastomotic leakage, demonstrated after raised i ntra-abdominal pressure duri ng a CT-study and a patient showing evidence of considerable movement of a rectal anastomosis in long itud inal d i rection of the rectum. Reviewing the sparse l iteratu re it can be concl uded that l itt le is known about the longitud inal , axial forces on a colorectal anastomosis. However, phys iolog ical pressure val ues in the colon may exceed early bursting pressure of an anastomosis . I n combination with other factors weaken ing the anastomos is , such as ischem ia, impai red wound heal i ng , i rrad iaton and seps is 1 9-2 1 , th is m ight contri bute lead to the occu rrence of c l in ical anastomotic leakage .

Reviewing the l iterature on mechan ical forces i n the colon it i s clear that there are pressures present that m ight exceed the pressure an anastomosis can withstand . However, none of the present studies addresses pressure i n the co lon in re lationsh ip to anastomotic leakage. As few stud ies on pressure i n and around the human anastomosis were performed concl u s ions shou ld be drawn with caution , but th is m ight form an interesti ng field for fu rther research for bas ic research as we l l as i nterventional stud ies .

References

l . Akyol AM, McGregor JR, Galloway DJ, et al. Early postoperative contrast radiology in the assessment of colorectal anastomotic integrity. Int J Colorectal Dis 1992; 7: 141-143.

2. Jiborn H, Ahonen J, Zederfeldt B. Healing of experimental colonic anastomoses; II Breaking strength of the colon after left colon resection and anastomosis. Am J Surg 1978; 136: 595-599.

3. Jiborn H, Ahonen J, Zederfeldt B. Healing of experimental colonic anastomoses; I Bursting strength of the colon after left colon resection and anastomosis. Am J Surg 1978; 136: 587-594.

4. Hendriks T, Mastboom WJB. Healing of experimental intestinal anastomoses, parameters for repair. Dis Colon Rectum 1990; 33: 891-901.

5. Rampton PA, Dinning PG, Kennedy ML, et al. Spatial and temporal organization of pressure patterns throughout the unprepared colon during spontaneous defecation. Am J Gastroenterol 2000; 95: l 028-1035.

6. Rampton PA, Dinning PG, Kennedy ML, et al. Prolonged multi-point recording of colonic manometry in the unprepared human colon: providing insight into potentially relevant pressure wave parameters. Am J Gastroenterol 2001; 96: 1838-1848.

7. Narducci F, Basotti G, Gaburri M, et al. Twenty four hour manometric recording of colonic motor activity in healthy man. Gut 1987; 28: 17-25.

8. Burkitt DS, Donovan IA. lntraluminal pressure adjacent to left colonic anastomoses. Br J Surg 1990; 77: 1288-1290.

9. Huge A, Kreis ME, Zittel TT, Becker HD, et al. Postoperative colonic motility and tone in patients after colorectal surgery. Dis Colon Rectum 2000; 43:932-939.

l 0. Schwab R, Wessendorf S, Gutcke A, et al. Early bursting strength of human colon anastomoses- an in vitro study comparing current anastomotic techniques. Langenbeck's Arch Surg 2002; 386: 507-511.

11. Shikata JI, Shida T. Effects of tension on local blood flow in experimental anastomoses. J Surg Res 1986; 40:l 05-111.

12. Goto T, Kawasaki K, Fujino Y, et al. Evaluation of the mechanical strength of functional end-to-end anastomoses. Surg Endosc 2006; 21: l 508-1 511.

13. Ritchie JA. Colonic motor activity and bowel function. Gut 1968; 9: 442-456. 14. Brennan DJ, Moynagh M, Brannigan AE, et al. Routine mobilisation of the splenic flexure is not

necessary during anterior resection for rectal cancer. Dis Colon Rectum 2007; 50: 302-307. l 5. Guenaga KF, Matos D, Castro AA, et al. Mechanical bowel preparation for elective colorectal surgery.

Cochrane Database Syst Rev. 2005; 25: CD00l 544. 1 6. Matthiessen P, Hallbook 0, Rutegard J, et al. Defunctioning stoma reduces symptomatic anastomotic

leakage after low anterior resection of the rectum for cancer: a randomized multicenter trial. Ann Surg, 2007; 246: 207-214.

17. Bulow S, Bulut 0, Christensen IJ, et al. Transanal stent in anterior resection does not prevent anastomotic leakage. Colorectal Dis 2006; 8: 494-496.

18. Amin Al, Ramalingam T, Sexton R, et al. Comparison of transanal stent with defunctioning stoma in low anterior resection for rectal cancer. Br J Surg 2003; 90: 581-582.

19. Eriksen MT, Wibe A, Norstein J, et al. Norwegian Rectal Cancer Group. Anastomotic leakage following routine mesorectal excision for rectal cancer in a national cohort of patients. Colorectal Dis 2005; 7: 51-57.

20. Folkesson J, Nilsson J, Pahlman L, et al. The circular stapling device as a risk factor for anastomotic leakage. Colorectal Dis 2004; 6: 275-279.

21. Kruschewski M, Rieger H, Pohlen U, et al. Risk factors for clinical anastomotic leakage and postoperative mortality in elective surgery for rectal cancer. lntJ Colorect Dis 2007; 22: 919-27.

119

Sum mary, concl us ions and futu re perspectives

Summary I n gastroi ntesti nal surgery anastomotic leakage has a major impact on morbid ity and even mortal ity in patients u ndergoing surgery. Besides the impact on patients' recovery after surgery and q ual ity of l ife, health cost are dramatical ly increased because of repetit ive su rg ical procedures, pro longed ICU and hospital stay and a s lower recovery phase after d i smissal from the hospital. Futhermore, even the oncolog ical outcome may be i nfl uenced negative ly. We therefore sought to get more ins ight in poss ibl e causat ive mechan isms, predict ion and i ntervent ions i n both animal and patient studies.

Chapter 1 g ives a short i ntroduction to the impact and d ifferent c l i n ical and pathophysio logical aspects of anastomotic leakage in gastroi ntesti nal surgery. It outl i nes the aims of this thes i s : i) to i nvestigate the contri bution of m icrocircu latory factors in anastomotic leakage, and i i) to deve lop more objective measures to identify patients at risk of developing anastomotic leakage. The two main hypotheses analysed were : i ) drainage of flu id col lect ions near the anastomosis can i nfluence anastomotic leakage, and i i) m icroci rcu latory d isturbances play an important role in anastomotic heal ing .

I n Chapter 2 the resu lts of a systematic review on the use of prophylactic drainage of colorectal anastomoses are described. We searched C INAHL, EMBASE, LILACS, MEDLINE, the Contro l led C l in ical Trials Database, the Trials Reg i ster of the Cochrane Colorectal Cancer Group and reference l i sts for random ized control led trials compari ng any drai nage with non-drai nage reg imes after anastomoses in colorectal surgery. The primary outcome measure was defi ned as c l i n i cal anastomotic deh iscence. Secondary outcome measures were mortal ity, rad io log ical anastomotic dehiscence, wou nd infection, reoperation, length of hospital stay and extra-abdominal compl ications. The methodological qual ity of each trial was assessed and the resu lts of each RCT were summarized and stratified based on the experimental group .

Of n i ne random ized contro l led trials identified, three were excluded based on i nadequate conceal ment of randomization or inc lus ion of emergent operation. Thus, 1 1 40 patients were analysed, reveal i ng no s ign ificant d ifferences between routi ne drainage for prevention of anastomotic leakage (primary outcome measure) nor any of the secondary outcome measures (mortal ity, rad io log ical anastomotic dehiscence, wou nd infection, reoperat ion, length of hospital stay and extra-abdominal comp l ications).

As a pre lude to the stud ies described in Chapter 4-7 we fi rst evaluated the accu racy of the surgeons ' i ntraoperative pred iction for the probabi l ity of anastomotic leakage. In Chapter 3 the resu lts of this study are descri bed. In 247 patients u ndergoi ng gastro i ntesti nal resection with anastomosis the operat ing surgeon ind icated the estimated ri sk for anastomotic leakage on a visual analogue scale. Th is prediction was s ubsequently compared to the occurrence of anastomotic leakage in the postoperative course. We concluded that the su rgeons ' i ntraoperative

1 23

lO

124

Vl C

3 3

n 0 ::l

� C l/1

0 ::l l/1

Ill ::l 0.. --t,

"Cl (D .., l/1

"Cl (D � <'

cl in ical r isk assessment has low predictive val ue for the occurrence of anastomotic leakage, due to a low a-priori ri sk of leakage and lack of sensit ivity . These resu lts clearly ind icate the need for addit ional research efforts for the deve lopment and eva luat ion of a more predictive tool of anastomotic leakage .

The stud ies descri bed in the fol lowing chapters focus on the infl uence of m icroci rcu latory d i stu rbances on anastomotic heal ing as evaluated with a method for eval uat ion of t issue oxygenation wel l descri bed in other fie lds , denominated as vis i bl e l ight spectroscopy (VLS). In Chapter 4 we conducted measurements in an experimental model for i schemic murine co lon anastomoses. Thi rty-two rats were d ivided into two groups . In the fi rst g roup an anastomosis in the d istal descending co lon was constructed . In the second group i schemia of the d i stal part of the descend ing colon was induced by l igat ion of mesenteric vesse l s unti l t issue satu ration d ropped below 5 5%, as measured by vis i b le l ight spectroscopy, fo l lowed by construction of an anastomosis in the i schem ic colon . In order to evaluate early and late phases of anastomotic heal ing , the an i mals were sacrificed on the th i rd or seventh postoperative day and the mechan i cal strength of the anastomos is was evaluated together with macroscopic integrity and adhes ion format ion .

Partial ischemic anastomoses resu lted in more adhesions and d im in ished res istance to long itud inal forces (breaking strength) in the early phases of heal ing, res istance to intral um inal pressure (bu rsting pressure) however was not affected . It was therefore concl uded that low t issue oxygenation is a ri sk factor for anastomotic deh iscence at the th i rd postoperative day and beyond in murine co lon ic anastomos i s .

I n Chapter 5 we explore the effect of preoperative i rrad iation (5 days of 5 Gray i rrad iat ion) on t issue oxygenation and anastom otic strength in mu rine colon . The an imal model resembles the human situat ion as closely as poss ib le with respect to i rrad iation fractionation and local ization of i rrad iation of rectal cancer. To ach ieve th is model design, we developed a surg i cal techn ique to exteriorize a loop of the descend ing colon creating an abdominal wal l hern ia which was subsequently i rrad iated with SxS Gy. After i rrad iat ion an anastomosis in the i rrad iated bowel loop was constructed . As descri bed in Chapter 4 , anastomotic strength was evaluated by testing res istance to long itud inal force (breaking strength), intral um inal pressure (bursting pressure) as wel l as a so-cal led ad hes ion score. To study the effects of both i rrad iation and the hernia operation itse lf, three g roups of an imals were stud ied : a g roup with an i rrad iated hern ia, a g roup with sham i rrad iated hern ia and a contro l grou p without hern ia nor i rrad iat ion .

Preoperative i rrad iat ion of the exteriorized co lon loop (d i stal part of a colon anastomosis) was successfu l ly adm in istered by combination of a new surg ical techn ique and a special des igned led cover. After i rrad iat ion and anastomos is s ide effects were acceptable and no red uced t issue oxygenation was seen after i rrad iat ion nor c l in ical s igns of anastomotic weaken ing, objective reduction in bursting pressu re or breaking strength .

After appl ication of vis ib le l ight spectroscopy in an i mal models we evaluated the techn ical feas ib i l ity and the stabi l ity of the measurements in an intraoperative

setting in humans. Chapter 6 describes the results of a pilot study evaluating tissue oxygenation of uncomplicated anastomoses in oesophageal and colorectal resection by applying VLS. We found visible light spectroscopy to be a feasible and fast method for intraoperative evaluation of tissue oxygenation of the serosa (or adventitia) of the anastomoses. Apart from increasing tissue oxygenation in the proximal part of the colon during colorectal resections, tissue oxygenation remained stable during the operation with acceptable variability in measurements.

In Chapter 7 we report a prospective clinical study to evaluate the predictive value of tissue oxygenation, measured by visible light spectroscopy, on the 125 occurrence of anastomotic leakage in colorectal resection. We evaluated 80 procedures with construction of anastomoses within 20 cm from the anal verge, measuring tissue saturation before and after resection, as well as after construction of the anastomosis. Anastomotic leakage appeared to occur in 14 ( 18%) of all procedures. Although measurements were reproducible and fairly stable throughout the study (mean 71 .8-76.2 SD 7. 1-8.0), the relationship between tissue oxygenation and anastomotic leakage we found was controversial and could not be explained based on findings in other studies. Possible explanations for the rise in St02 in the colon around the anastomoses and decreasing St02 in the coecum in patients that later on showed anastomotic leakage might be explained by ischemia-reperfusion after release of the bowel clamps temporarily obstructing blood flow.

Based on examples of two recent clinical observations, Chapter 8 explores the presence of mechanical forces in human colorectal anastomoses described in the literature. It appeared that there is sparse literature describing the normal pressure in vivo in the human, unaffected colon. Two small studies describe the intraluminal pressure characteristics proximal and distal of colon anastomoses. As bursting pressure in the human colon can not be evaluated in vivo we must rely on in vitro measurements and on animal studies to determine the maximal intraluminal pressure an anastomosis can sustain. Based on such studies it can be concluded that physiological intraluminal pressure might exceed bursting pressure of colorectal anastomoses. Moreover, there is no body of evidence based on the current literature on longitudinal forces present in either unoperated colon or on anastomoses in humans. Our 'real-time' observations during a CT study in a patient with a colorectal anastomosis and monitoring of the impact of mechanical longitudinal forces on the occurrence of anastomotic leakage opens a new perspective to the field of colorectal surgery. Additionally, in our hospital we recently started a phase I study using a new surgical technique for protection of the anastomosis by using a biodegradable seal (i.e C-Seal™). Post-operative observations of the length of the C-seal indicate strong longitudinal forces due to peristalsis, pointing towards a different causative mechanism of mechanical forces.

l.O

Vl C: 3 3

C: ., "O (1) ., VI

"O (1) n ,-+ <" (1)

1 26

l./) C

3 3

Futu re perspectives Currently the most important measures to reduce the consequences of anastomotic leakage are in the first place construction of a defunctioning stoma proximal to the anastomosis 1 and in the second place to be highly alert for the presence of symptoms and signs inidicative of leakage in the postoperative course in order to intervene in time in case anastomotic leakage occurs2 •

Considering the findings of our study described in Chapter 3 of this thesis it must be concluded that the operating surgeon can not predict anastomotic leakage in most cases. Continuing this line of though one could argue, taking into account the finding of the Swedish study describing significant reduction of anastomotic leakage in a group of patients randomised for construction of defunctioning stoma 1 , that every patient with a low colorectal anastomosis should be defunctioned. The disadvantage of this procedure is that many of these stomas never will be closed and in these cases only the risks of leakage are handled without any long-term benefit for the patient3• So the main challenge for future studies is to identify those patients particularly being at risk of developing anastomotic leakage by more accurate risk assessment. This could be achieved by preoperative evaluation based on constitutional factors and comorbidity, intraoperatively by better evaluation of tissue factors (i.e. ischemia, tension) and in the postoperative phase by faster and more accurate diagnostic procedures evaluating patients at risk.

Considering the consequences of a defunctioning stoma on the one hand and the consequences of leakage of a non-defunctioning anastomosis on the other hand, the surgeons' judgement of the risk of anastomotic leakage is of crucial importance. In this perspective the low accuracy of the clinical judgement is worrying, but opens opportunities for further investigations of the clinical decision making process. More importantly, it would be helpful to assess risk for anastomotic leakage during preoperative evaluation. As described in earlier studies evaluating risk factors for anastomotic leakage4 ,5 there is a clear correlation between a higher number of risk factors present preoperatively and the occurrence of anastomotic leakage. Better awareness of the presence and number of risk factors, particularly those concerning anesthesiological parameters and/or concomitant disease, might improve the accuracy of the surgeons' preoperative risk assessment of anastomotic leakage if explicitly presented before and during the operation. As such, prospective studies are needed to evaluate multivariate clinical risk models for anastomotic leakage in multiple data sets from different institutions and geographical areas.

It might be possible to predict increased risk for anastomotic leakage by considering biomechanical properties of the bowel wall itself, ideally as early as during preoperative evaluation. In normal anastomotic healing a strong upregulation of matrix metalloproteinase 9 (MMP-9) was detected around anastomoses6, more in particular in the immediate presence around sutures7 • This upregulation is the highest in the first three postoperative days leading to increased collagenolytic activity. Furthermore, there are distinct changes in remodelling of collagen by matrix

metal loproteinases (in particluar MMP-1 , -2 and -9) in leaking and non-leaking anastomoses8 • This m ight explain the decreased sutu re-hold ing capacity of anastomoses in the first few days after operation. Considering th is , representing the old surg ical knowledge that a wound heals between the sutu res and not through them, it m ight be interesting to repeat the study by Agren 7 us ing staplers , as i t m ight be hypothes ized that a more pronounced upregu lation of matrix metal loproteinase 9 m ight be found due to more 'sutu repoints ' appl ied by staplers. If the presence and activity of MMPs is correlated with anastomotic leakage early in the course of anastomosis , it might be of interest to use innovative scanning techniques l i ke near-infrared fluorescence imag ing to detect the presence of MMPs intraoperatively. Recent developments in both camera systems and so-cal led 'smart activatable probes ' l i ke MMP sens itive fl uorescent optical contrast agents (i.e MMPSense™) m ight reveal a complete new field of molecu lar imaging in gastrointestinal anastomosis to detect non- invas ively biomarkers of leakage9 •

As shown in a study evaluating the effect of preoperative i rrad iation with Sx S Gy in humans 1 0 an upregu lation of MMP 2 and 9 in colonic tissue as wel l as more wound deh iscence and fistu la formation associated with MMP 2-upregu lation was found after i rrad iation. Furthermore, in patients with Marbus Crohn, a d isease known for its h igher incidence of fistu la formation , increased express ion of MMP 1 was observed1 1 . It could therefore be argued that profi les of matrix metal loproteinases present in the bowel wal l in pre- and postoperative biopsies and maybe other extracel lu lar matrix components , such as c luster of d ifferentiation (CD) 1 1 7 , transform ing growth factor (TGF)-� and plasminogen activator inh ibitor (PAI) scores 1 2, 1 3 might represent clear constitutional d ifferences in bowel wal l structure and explain an elevated risk for anastomotic leakage.

There are strong arguments to support ischemia as a cause of anastomotic leakage. I t can therefore be argued that detecting local ischem ia during the operation should be a powerfu l pred ictor of elevated risk for anastomotic leakage. Furthermore, selecting bowel segments with optimal oxygenation cou ld contri bute in lowering anastomotic leakage rates. Current methods to do so have both practical and methodolog ical drawbacks as described in Chapter 6 of this thes is. Vis ible l ight spectroscopy, the method for intraoperative assessment of tissue ischemia of the colon evaluated in this thesis , shows a clear correlation between rising tissue satu ration during the operation and anastomotic leakage in hu mans, although in colon of the rat we demonstrated significantly more compl ications in anastomoses with low tissue oxygenation as determined by VLS. Furthermore, we found decreased tissue satu ration in an area of the colon , remote to the anastomotic s ite to be significantly correlated to anastomotic leakage. These observations in humans are oppos ite to the hypothesized effect, as based on l iterature, the an imal stud ies and p i lot studies. We expected to find decreased tissue satu ration during the operation to be pred ictive of anastomotic leakage. Although StO2 seems to be able to pred ict anastomotic leakage in a fast and cl in ical ly feas ible way, these find ings make it necessary to conduct fu rther stud ies to compare the find ings of different

1 27

i nvas ive and non-invasive methods to measure t issue oxygenation in order to fi nd the physiological mechan ism caus ing our observations. Th is shou ld be done in exper imental studies to e lucidate the effect of (remote) reperfus ion- ischemia i nju ry o n t issue oxygenation in re lat ion to StO2. Furthermore , c l in ical stud ies compari ng i nvas ive and non-invasive methods of measuring t issue oxygenation i n the human co lon shou ld be compared to eval uate the background of our fi ndi ngs.

I t m ight be argued that vis ib le l ight spectroscopy, measuring poi nt measures i n smal l , thi n t issue areas, does not predict anastomotic leakage accurately because it

128 was measured i n the colon serosa. The serosal s urface m ight not be the part of the colon ic wal l most sensitive to i schem ia, so that the mucosa might in fact be a representative of early t issue deoxygenat ion. This i s supported by stud ies showing

V') C: 3 3

ll.J ::J Cl.

C: ' (D

"O (D ' 1./l

"O (D � <" (D 1./l

exce l lent corre lation between StO2 of mucosa and actual t issue i schemia1 4- 1 6 • This was confi rmed by our own observat ions i n rat colon where the measured t issue sample was actual ly cons iderable th i nner compared to humans (approximately 0. 5 vs 2-4 mm), therefore measuring the ent ire bowel wal l thickness i n rats. Th is is also supported by the cl i n ical observation that macroscopic changes due to ischemia are fi rst observed in colon mucosa even if the serosa looks macroscopical ly normal. Neverthe less, du ring colorectal resection it i s often not poss ib le to measure the colon mucosa, as after resection of the d iseased part the remain i ng co lon frequently is c losed by staples to avoid contamination of the operative fie ld. In p i lot-studies lead i ng to the cl i n ical stud ies described i n th is thes i s we performed measurements of the human colon mucosa, but these showed high standard deviat ions due to contamination of the mucosa by fecal material and , more important, artefacts due to manipu lation of the colon (unpub l i shed resu lts). Measurement of mucosa might prove a better pred ictor, but the measurements are presumably more d ifficu lt and time-consuming , making the method less su itable for surg ical i ntraoperative use for th is particu lar purpose. As mentioned earl ier, new i nnovative camera systems which scan a larger area and with deeper ti ssue penetration of the anastomosis real-time might be a better tool for measu rement of ti ssue oxygenation.

As demonstrated in an animal model there m ight be a considerable effect of ischem ia-reperfusion on anastomotic leakage 1 7• l schemia of the anastomosis or parts of the anastomosis m ight not on ly be present during the operat ion, but also i n the early postoperative phase, a s oedema, hypotens ion and mechan ical strain i nduces a re lative low-flow state of the splanchn ic c i rculat ion i n the d i rect postoperative period. Furthermore , peristals is wi l l commence during the fi rst postoperative day1 8 lead ing to higher i ntral um inal and long itud i nal forces. A study relat ing CT-angiographic fi nd ing of perianastomotic vascu lature to hemodynamic parameters , as we l l as to postoperative anastomotic leakage, m ight further e l ucidate th is mechan ism. As shown by a study assess ing the effect of tens ion on t issue oxygenation i n anastomoses i n a can ine mode l , t issue i schemia occu rs if tens i le stress on colon anastomoses rises 1 9• The e levated tens ion phys iolog ical ly present due to peristals is in the colon , poss ibly even ri s ing after colorectal resection 1 7, 20-23 m ight thus induce i schemia after the actual operation. Taken together, this m ight

contri bute to anastomotic leakage through d i rect mechan ical tract ion on the colon ic wal l , and through ischemia induced by e levated tens ion on the bowel wal l . A re latively short mesentery left in place in many low colorectal anastomoses in order to preserve the inferior mesenteric artery for perfus ion of the prox imal part of the anastomosis m ight play a role i n this mechan ism through increased traction on the anastomosis. These mechan isms are confi rmed by c l in ical observations. When cons ideri ng mechan ical strai n on the anastomosis the controvers ial role of bowel preparation prior to construct ion of l eft-sided colon anastomoses shou ld be mentioned. Although a n umber of large studies and trials show no benefit from 1 29 colon ic preparation preoperative to colon resection , many surgeons prefer colon ic cleans ing , argu ing that passage of stool might damage the newly made anastomosis even in the presence of a stoma. Based on the l iterature avai lable on the subject there is no bas is for th is practice as there are ind ications that colon preparation leads to even h igher peak pressu res 1 B, 2 1 , 22 . Conti nu i ng this l ine of thought, i nterventional strategies a imed at reduction of pressure from with i n the anastomos is such as the C-seal described i n Chapter 8 m ight prove successfu l i n prevention of anastomotic leakage.

Cons idering the smal l amount of stud ies analys ing pressu re before and after construction of anastomoses and more importantly, the absence of stud ies i nvestigat ing pressure and traction i n re lationsh ip to anastomotic leakage, there i s an interesting fie ld for future research. Furthermore , the connection between ischemia, anastomotic leakage and mechan ical forces on the anastomosis cou ld be explored by the use of i schemia markers such as p imondazol , ideal ly combined with near- infrared fl uorescence markers for appl icat ions with near- infrared fl uorescence camera systems. Other factors i nfl uenci ng the q ual ity of the bowel wal l such as d ifferences in col lagen metabol i sm (MMPs) m ight fu rther e lucidate causes of anastomotic leakage and could ideal ly be i nvestigated in animal studies using smart activatable near- infrared fl uorescent optical contrast agents, as fol low-up of anastomotic heal i ng cou ld be done daily without sacrifici ng the animal. Further studies on the effects of preoperative irrad iation us i ng MMP sens itive probes (e.g . MMP 2 , 8 and 9) combined with the isolated loop model cou ld be conducted , determ in ing optimal i rrad iation t iming and dosage as wel l a s the effects of rad ioprotectors such as Tempol24.

Other factors playi ng a role in the (early) postoperative period are adhes ion formation and drai nage of pelvic anastomoses. D im in ished heal ing with subsequent weaken ing of the anastomosis might i n the c l in ical s ituation be compensated by other mechan isms such as adhesion formation around a compromised anastomosis. We demonstrated in ischemic anastomoses i n rats a h igher degree of adhes ion formation (Chapte r 4) . Based on interventional studies to prevent adhes ion formation, showing more anastomotic leakage if adhesion formation is effective ly prevented25 ,2 6 , a s im i lar mechan ism m ight play a ro le in human anastomoses . I n contrast with our fi nd ings of the meta-analys is it might be usefu l to drain the pelvic space as no peritoneum i s present in the cavity left beh ind after TME

resect ions , thus fl u i d col lect ions m ight not be resorbed adequately and are associated with a h igher rate of anastomotic leakage27 • There are no sufficiently powered randomised c l i n ical trials i nvestigat ing the use of drai ns in pelvic anastomoses, but one retrospective study shows a c lear benefit of drai nage28 • Furthermore, drai ns m ight be usefu l i n the pred ict ion of anastomotic leakage i n the d i rect postoperative course through analys i s of b iomarkers in the peritoneal fl u id29 •

There is strong evidence that early i n the postoperative cou rse a systemic i nflammatory react ion i s present in patients who develop an anastomotic leakage ,

1 30 for wh ich C-reactive prote i n levels might be a good pred ictor30 . Currently a Dutch study i s be ing carried out to identify both phys iolog ical and b iochemical predictors of anastomotic leakage early i n the postoperative course (Den Du lk, personal

V'l C:

3 3

communication). Taken together, the use of drai ns m ight faci l itate identification of patients at risk m uch earl ier i n the postoperative course by carefu l monitori ng of phys iological and biochem ical parameters , wh i ch cou ld resu l t i n less frequent and more se lective placement of a defunction i ng stoma.

In conclu s ion , anastomotic leakage i n gastroi ntest inal surgery remains a major c l i n ical chal lenge in respect to bas ic research, predictive statistical mode ls , biomarker research, non- i nvas ive imag ing tool and i ntraoperative r i sk assessment for the next five to ten years to come.

References

l . Matthiessen P, Hallbook 0, Rutegard J, et al. Defunctioning stoma reduces symptomatic

anastomotic leakage after low anterior resection of the rectum for cancer, a randomized

multicenter trial. Ann Surg 2007; 246: 207-214.

2. Doeksen A, Tanis PJ , Vrouwenraets BC, et al. Factors determining delay in relaparotomy for

anastomotic leakage after colorectal resection. World J Gastroenterol 2007; 13: 3721-3725. 3. den Dulk M, Smit M, Peeters KC, et al. A multivariate analysis of limiting factors for stoma reversal

in patients with rectal cancer entered into the total mesorectal excision (TME) trial: a retrospective

study. Lancet Oneal. 2007; 8: 297-303. 4. Makela JT, Kiviniemi H, Laitinen S. Risk factors for anastomotic leakage after left sided colorectal

resection with rectal anastomosis. Dis Colon Rectum 2003; 46: 653-660. 5 . Alves A, Panis Y, Trancart D, et al. Factors associated with clinically significant anastomotic leakage

after large bowel resection : multivariate analysis of 707 patients. World J Surg 2002; 26: 499-502.

6. De Hingh I, Lamme R, van Goar H, et al. Changes in gelatinase activity in the gastrointestinal tract

after anastomotic construction in the ileum or colon. Dis Colon Rect 2005; 48: 2133-2141. 7. Agren MS, Andersen TL, Mirastschijski U, et al. Action of matrix metalloproteinases at restricted

sites in colon anastomosis repair: an immunohistochemical and biochemical study. Surgery 2006;

140: 72-82. 8 . Stumpf M , Klinge U, Wilms A . et al. Changes of the extracellular matrix a s a risk factor for

anastomotic leakage after large bowel surgery. Surgery 2005; 1 37: 229-234.

9. Weissleder R, Pittet MJ. Imaging in the era of molecular oncology. Nature 2008; 452:580-589.

1 0. Angenete E, Langenskjold M, Falk P, et al. Matrix metalloproteinases in rectal mucosa, tumour and

plasma: response after preoperative irradiation. lnt J Colorectal Dis 2007; 22: 667-674. 11. Stumpf M, Cao W, Klinge U, et al. Reduced expression of collagen type I and increased expression

of matrix metalloproteinases 1 in patients with Crohn's Disease. J Invest Surg 2005; 18: 33-38.

12. Krones CJ, Klinge U, Butz N, et al. The rare epidemiological coincidence of diverticular disease and

advanced colonic neoplasis. lnt J Colorectal Dis 2006; 21: 18-24.

13. Klinge U, Rosch R, Junge K, et al. Different matrix micro-environments in colon cancer and

diverticular disease. lnt J Colorectal Dis 2007; 22: 515-520.

14. Benaron DA, Parachikov IH, Cheong WF, et al. Quantitative clinical non-pulsatile and localized

visilble light oximeter: Design of the T-Stat™ Tissue Oximeter. SPIE Optical Tomography

Spectroscopy Tissue V 2003; 4955: 355-68.

15. Benaron DA, Parachikov IH, Friedland S, et al. Continuous, noninvasive and localised microvascular

tissue oximetry using visible light spectroscopy. Anesthesiology 2004; 100: 1469-1475.

16. Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosa! perfusion

in the gastrointestinal tract. Gastrointest Endosc Clin N Am 2004; 14: 539-555.

17. Colak T, Turkmenoglu 0, Dag A et al. The effect of remote ischemic preconditioning on healing of

colonic anastomoses. J Surg Res 2007; 143: 200-205.

18. Huge A, Kreis ME, Zittel TT, et al. Postoperative colonic motility and tone in patients after colorectal

surgery. Dis Colon Rectum 2000; 43:932-939.

19. Shikata JI, Shida T. Effects of tension on local blood flow in experimental anastomoses. J Surg Res

1986; 40:105-111.

20. Burkitt DS, Donovan IA. lntraluminal pressure adjacent to left colonic anastomoses. Br J Surg 1 990;

77: 1288-1290.

21. Bampton PA, Dinning PG, Kennedy ML, et al. Spatial and temporal organization of pressure

patterns throughout the unprepared colon during spontaneous defecation. Am J Gastroenterol

2000; 95: l 028-1035.

22. Bampton PA, Dinning PG, Kennedy ML, et al. Prolonged multi-point recording of colonic

manometry in the unprepared human colon: providing insight into potentially relevant pressure

wave parameters. Am J Gastroenterol 2001; 96: 1838-1848.

1 31

23. Narducci F, Basotti G, Gaburri M, et al. Twenty four hour manometric recording of colonic motor

activity in healthy man. Gut 1987; 28: 1 7-25.

24. Vitolo JM, Cotrim AP, Sowers AL, et al. The stable nitroxide tempol facilitates salivary gland

protection during head and neck irradiation in a mouse model. Clin Cancer Res 1 0: 1 807- 1 8 1 2. 25. Dine S, Ozaslan C, Kuru B, et al. Methylene blue prevents surgery-induced peritonela adhesions but

impairs the early phase of anastomotic wound healing. Can J Surg 2006; 49: 321 -328.

26. Beck DE, Cohen Z, Fleshman JW, et al. A prospective, randomized, multicenter, controlled study of

the safety of seprafilm adhesion barrier in abdominopelvic surgery. Dis Colon Rectum 2003; 46:

1310-131 9.

27. Power N, Atri M, Ryan S, et al. CT assessment of anastomotic bowel leak. Clin Radial. 2007; 62:37-

1 32 42.

l.O

Vl C:

3 3 SlJ :< ("\ 0 ::l !l C: Vl

0 ::l Vl

SlJ ::l Cl.

..., C:

ro "O (l) -, Vl

"O (l) � <" (l) Vl

2 8. Peeters KC, Tollenaar RA, Marijnen CA, et al. Risk factors for anastomotic failure after total

mesorectal excision of rectal cancer. Br J Surg, 2005; 92: 21 1 -2 1 6.

29. Komen N, de Bruin R, Klein rensink G, et al. Anastomotic leakage, the search for a reliable

biomarker. A review of the literature. Colorectal Dis 2008; 1 0: 1 09-1 1 7.

30. Matthiessen P, Henriksson M, Hallbook 0, et al. Increase of serum C-reactive protein is an early

indicator of subsequent symptomatic anastomotic leakage after anterior resection. Colorectal Dis

2008; 1 0: 75-80.

M M �

Chapter 9 Summary, conclusions and future perspectives

Samenvatting, conclusies en toekomstperspectief

Samenvatting Lekkage van darmnaden is een in de gastroi ntesti nale chi rurg ie gevreesde compl icatie vanwege de hoge morbid iteit en mortal iteit d ie h iermee gepaard gaan. Naast een d u idel ij k langere herstelperiode heeft naadlekkage een ongunst ige invloed op de kwal iteit van leven na de operatie en z ijn de kosten van de behandel ing aanzien l ij k hoger door reoperaties, I ntensive Care behandel i ng en een verlengd postoperatief traject. Ook is een hogere kans op lokaal recid ief aangetoond wanneer de primaire operatie resectie van een mal ign iteit betrof. De kl i n i sche- en dierstud ies, d ie i n d it proefschrift beschreven warden , hebben dan ook tot doel meer i nzicht i n de oorzaken en voorspelbaarheid van naad lekkage te krijgen.

Hoofdstuk 1 bevat een korte i ntroductie tot verschi l lende pathofysio log i sche mechan ismen bij het ontstaan van naad lekkage. Van weefsel i schemie wordt i n het algemeen aangenomen dat d it de darmnaad verzwakt en tot naad lekkage leidt. De belangrijkste doelstel l i ngen van d it proefschrift zij n dan ook: i) het onderzoeken van de b ijdrage van veranderingen in weefseldoorbloed ing op het ontstaan van naad lekkage, i i) het ontwikkelen van een objectieve methode om naad lekkage tijdens de i n it ie le operatie te voorspel len. H iertoe werden de volgende hypotheses geformuleerd : I) drainage van vochtcol l ecties rond de darmnaad be·invloedt de kans op het ontstaan van naad lekkage, en 1 1 ) veranderi ngen i n weefseldoorbloed i ng spelen een belangrijke rol bij het ontstaan naad lekkage.

In Hoofdstuk 2 hebben we een l iteratuurstudie naar de effecten van het plaatsen van drai ns op naad lekkage verricht. H iertoe hebben we in de databestanden van CINAHL, EMBASE, LILACS, MEDLINE, the Contro l led C l in ical Trials Database, het Tria ls Reg ister van Cochrane Colorectal Cancer Group en referentie l ij sten naar randomized control led trials (RCT's) gezocht, d ie het wel en n iet p laatsen van dra ins bij colorectale darmnaden vergeleken. H ierbij was het optreden van naad lekkage de belangrijkste u itkomstmaat, gevolgd door mortal iteit, rad iolog ische naad lekkage, wond infectie, reoperatie, verbl ijfsduur in het z iekenhuis en extra-abdominale compl icaties. Van elke gevonden studie werd de methodologische kwal iteit beoordeeld , de resu ltaten ervan samengevat en gestratificeerd aan hand van de i nterventiegroep (wel /geen drain).

Er werden negen gerandomiseerde gecontroleerde trials gevonden , waarvan er drie vanwege onvoldoende b l i nderi ng van de randomisatieprocedure of vanwege inc lus ie van acute operaties werden geexcl udeerd. Aldus werden 1 1 40 patienten geanalyseerd. H ierbij vonden we geen s ign ificante versch i l len tussen pat ienten , waarbij routinematig drai ns bij colorectale anastomoses werden geplaatst en degenen zonder drai n . Dit gold zowel voor de belangrijkste u itkomstmaat, namel ij k naad lekkage, als d e andere u itkomstmaten.

Om later, zoals beschreven in hoofdstuk 4-7, te trachten meer objectieve methoden voor het voorspel len van naad lekkage te kunnen ontwikkelen , hebben we i n Hoofdstuk 3 de betrouwbaarheid van het oordeel van de chi rurg over de waarsch ijn l ijkheid van het optreden van naad lekkage beschreven. Dit oordeel vormt

1 37

0

bij het voorspellen en eventueel trachten te voorkomen van naadlekkage vooralsnog de gouden standaard. Hiertoe gaf de chirurg, na het voltooien van de operatie, op een zgn. visual analogue scale (VAS) zijn inschatting van de kans op het ontstaan van naadlekkage weer. Vervolgens werd de voorspelde waarde met het daadwerkelijke optreden van naadlekkage vergeleken. In totaal werden 247 operaties geanalyseerd. We concluderen dat het klinische, intraoperatieve oordeel van de chirurg een lage predictieve waarde voor het daadwerkelijke optreden van naadlekkage heeft. Dit wordt veroorzaakt door een lage a-priori kans op het

138 optreden van naadlekkage en gebrek aan sensitiviteit. Deze resultaten onderstrepen de noodzaak van het vinden van een objectieve methode om tijdens de operatie de kans op naadlekkage beter te kunnen evalueren.

n :::; /l)

/1) .., 0

Vl /l)

3 /1) ::i <

8. ::i

LO

n 0 ::i

C!.. C VI

;;;· VI

/1) ::i

A° 0 3 VI

VI "Cl /1)

� ;;;·

In de volgende hoofdstukken richten we ons op de invloed van veranderingen in weefselperfusie rond de darmnaad op naadlekkage. Hierbij gebruiken we een methode die op vele andere terreinen betrouwbaar voor het beoordelen van weefseloxygenatie is gebleken, namelijk visible light spectroscopy (VLS). In Hoofdstuk 4 beschrijven we een diermodel om de invloed van ischemie op de sterkte van colonanastomoses bij ratten te evalueren. Hiertoe werden 32 ratten in twee groepen verdeeld. Bij de eerste groep werd -zonder verdere interventie- een anastomose in het distale colon gelegd. Bij de tweede groep werd door het successievelijk ligeren van mesenteriaalvaten ischemie van het distale deel van het colon ge·induceerd totdat de weefseloxygenatie (gemeten met VLS) lager dan 5 5% was. Vervolgens werd een anastomose in het deels ischemisch colon aangelegd. De dieren werden op cruciale fasen in de genezing van de darmnaad (d.w.z. de 3e en 7e dag postoperatief) gereopereerd, waarbij opnieuw de weefseloxygenatie werd gemeten en de adhesievorming werd beoordeeld. Aansluitend werden de dieren getermineerd, vervolgens de anastomose gereseceerd en op mechanische sterkte getest. Rond ischemische anastomoses trad significant meer adhesievorming op; ook bleken deze anastomoses in de vroege fase van wondgenezing (dag 3) minder bestand tegen longitudinale krachtinwerking dan normale anastomoses. Er was geen verschil in weerstand tegen intraluminale druk tussen ischemische en niet­ischemische anastomoses. We concluderen dan ook dat weefselischemie een risicofactor vormt voor naadlekkage, met name in de eerste dagen van naadgenezing.

In Hoofdstuk 5 wordt het effect van preoperatieve bestraling (SxS Gy) op weefseloxygenatie en op sterkte van colonanastomoses in een diermodel geevalueerd. Hierbij hebben we geprobeerd om de humane situatie met betrekking tot bestralingsfractionering en timing ten opzichte van de operatie zo goed mogelijk te simuleren. Hiertoe hebben we in een diermodel een chirurgische techniek ontwikkeld, waarbij we het colon descendens mobiliseren en in een breuk buiten de buikwand plaatsen. Aldus kan, met zo min mogelijk strooistraling, een ge·isoleerde colonlis gefractioneerd warden bestraald. De bestraalde lis werd een week later uit de buikwandbreuk geprepareerd en er werd een anastomose aangelegd, die proximaal uit gezond colon en distaal uit bestraald colon bestaat. Zoals in

Hoofdstuk 4 beschreven , werden de anastomoses op de 3e en 7e postoperatieve dag op weerstand tegen long itudi nale krachten en tegen intral um inale druk getest. Tevens werd bij e lke operatie de weefse loxygenatie met VLS gemeten en werden bij terminatie de ernst van adhes ievorm ing gescoord. Aangezien het aan leggen van de hern ia zelf ook effect op adhes ies en darmnaadgenezing kan hebben , werden d rie graepen bestudeerd : een g raep d ieren waarbij de hern ia met 5x5 Gy werd bestraa ld , een g roep met hern ia en sham-bestral i ng en een contraleg raep zonder hern ia en een normale anastomose. Bestral ing van de colon l i s i n de hern ia kon met succes warden u itgevoerd met behu lp van een speciaal voor d it doel ontworpen 1 39 loodafscherm ing. Tijdens de tweede operatie, d ie na bestral ing werd u itgevoerd.,_

zagen we geen veranderi ng in weefse loxygenatie ten opzicht van voor de bestral i ng. K l i n isch waren er geen tekenen van toegenomen naad lekkage. De mechan ische sterkte van de bestraalde anastomoses was ge l ij k aan d ie i n de sham-bestraalde g raep, maar grater dan de contralegraep zonder hern ia.

Naast toepass ing van VLS i n d iermode l len hebben we de techn ische haalbaarheid en stabi l iteit van de meti ngen in een i ntraoperatieve sett ing bij mensen bestudeerd. I n Hoofdstuk 6 warden de resu ltaten van een p i lot-stud ie beschreven , waarbij we weefseloxygenatie rand anastomoses op versch i l lende plaatsen en tijdstippen gedurende de operatie i n oesofagus- en rectos igmoidresecties eval ueren . De metingen ze lf b leken eenvoudig u itvoerbaar te zij n ; metingen i n de serasa (dan wel adventitia) waren stabie l , terwijl metingen van de m ucosa een vrij hoge standaarddeviatie vertoonden. Daarnaast constateerden we b ij rectos igmoidresecties gedu rende de operatie toenemende weefseloxygenatie in het praximale deel van de anastomose. Voor het overige traden er geen veranderi ngen i n weefse loxygenatie op.

In Hoofdstuk 7 wordt een praspectieve k l in ische studie besch reven , waarin we de voorspe l lende waarde van weefseloxygenatie , gemeten met VLS, voor het optreden van lekkage van anastomoses evalueren. Bij 80 rectosigmoidresecties , waarbij een anastomose b innen 20 cm vanaf de anus werd aangelegd werd t ijden s de operaties -voor en na resectie van het rectos igmoid en na aan leg van de anastomose- de oxygenatie van de serasa rand de (toekomstige) anastomose gemeten. Bij 1 4 ( 1 8%) van de patienten trad naad lekkage op. De metingen waren gedurende de studie repraduceerbaar en rede l ijk stabie l (mean 7 1 .8-76.2 SD 7. 1 -8 .0) . B ij anastomoses, d ie later naad lekkage vertoonden , steeg de weefseloxygenatie in het colon proximaal van de anastomose en daalde deze in een ver van de anastomose verwijderd deel van het colon (caecum). Deze s ign ificante veranderingen zij n n iet eerder i n humane studies beschreven en kunnen ook door experimentele studies , s lechts gedeelte l ij k warden verklaard . Moge l ijkerwijs spee lt h ierbij een ischemie-reperfus ie fenomeen bij het afklemmen van mesenteriaalvaten tijdens de resectie een ral .

Aan hand van twee k l i n i sche casu ist i sche observaties wordt i n Hoofdstuk 8 de l iteratuur met betrekki ng tot de ral van mechan ische krachten op humane colorectale anastomoses beschreven . H ierbij valt op dat er we in ig stud ies zijn , d i e

0

Vl Ill 3 ro :::I <

8. n 0 :::I !:l. C Vl iii' ro :::I

ct> A° 0 3 Vl �

"O ro ""' Vl

"O ro n �. ro .....,

140

0

Vl PJ 3 ro :J <

8. :J lO n 0 :J !l s::: V,

ro VI

ro :J

ro � 0 3 V,

ro -, VI "C ro � ro·

de fysiologische intraluminale druk in het colon beschrijven. Twee kleine studies beschrijven karakteristieken van intraluminale druk praximaal en distaal van ongecompliceerde colorectale anastomoses bij mensen. Aangezien het meten van de maximale weerstand tegen intraluminale druk in humane anastomoses in vivo op ethische en methodologische bezwaren stuit, moeten we ans op experimenteel onderzoek baseren. Uitgaande van dergelijke studies is het aannemelijk dat de fysiologische druk onder omstandigheden de maximale weerstand van de anastomose overschrijdt. Over krachten, die in longitudinale richting op de anastomose inwerken, is in de literatuur niets beschreven. De in dit hoofdstuk beschreven gevallen zetten in ieder geval vraagtekens bij de ral van mechanische krachten bij het ontstaan van naadlekkage.

Conclusie en toekomstperspectieven Op dit moment is het plaatsen van een stoma ter bescherming van de anastomose de belangrijkste maatregel om de gevolgen van naadlekkage te beperken 1 • Daarnaast is het belangrijk om in de postoperatieve fase tijdig tekenen van naadlekkage te onderkennen en daarnaar te handelen2 •

Aan hand van de resultaten van het onderzoek, beschreven in hoofdstuk 3 van dit praefschrift, kunnen we concluderen dat de operateur zelden kan voorspellen welke anastomose zal gaan lekken. Als we dit combineren met de bevindingen van een grate gerandomiseerde Zweedse studie, die significant meer naadlekkage vond bij lage colorectale naden zonder devierend stoma 1 , moeten we overwegen uit voorzorg bij elke distale colonnaad een devierend stoma aan te leggen. Het nadeel van een zodanig beleid is dat een relatief groot deel van deze stoma's nooit opgeheven zal worden3, waardoor alleen het risico op naadlekkage verlaagd wordt zonder duidelijk voordeel voor de patient. Voor toekomstig onderzoek is het dan ook een grate uitdaging om nauwkeuriger te kunnen voorspellen welke patienten een verhoogd risico op naadlekkage zullen hebben door het beter formuleren van risicoprafielen. Door preoperatieve evaluatie van constitutionele factoren en co­morbiditeit, intraoperatief door evaluatie van weefselfactoren (zoals ischemie, mechanische tractie) en postoperatief door snellere en nauwkeurigere diagnostiek naar het ontstaan van naadlekkage, zouden patienten met een verhoogd risico kunnen worden opgespoord.

Als we de gevolgen van enerzijds een devierend stoma en anderzijds naadlekkage beschouwen, valt op hoe belangrijk het oordeel van de chi rurg over de kans op naadlekkage is. Tijdens de initiele operatie kan immers de belangrijkste maatregel, namelijk de aanleg van een devierend stoma, warden genomen. In dit opzicht opent de lage predictieve waarde van dit oordeel mogelijkheden om het klinische besluitvormingspraces verder te onderzoeken en te verbeteren. Om deze reden zou het nuttig zijn voor en tijdens de operatie een risico-inschatting te kunnen maken. In eerdere studies werd een duideli jke relatie tussen het aantal preoperatieve risicofactoren en het optreden van naadlekkage beschreven4,5 •

Daarom zou het nadrukkelijker meewegen van de pre- en intraoperatieve risicofactoren bij het besluit om wel of niet een devierend stoma aan te leggen de risico-inschatting van de chirurg kunnen verbeteren. In een prospectieve studie, waarin de operateur voor en tijdens de operatie expliciet met de risicofactoren wordt geconfronteerd, zou dan kunnen worden nagegaan of de predicatieve waarde van het oordeel van de chirurg verbeterd kan worden.

Er zijn aanwijzingen dat biomechanische eigenschappen van de darmwand een verhoogd risico voor naadlekkage kunnen definieren. Wellicht kun nen deze eigenschappen voorspeld warden, idealiter al tijdens preoperatieve onderzoeken. 141 Tijdens normale genezing van colonanastomoses treedt een duidelijke verhoging van activiteit van matrix metalloproteinase 9 (MMP-9) rond de anastomose op6,

speciaal rond hechtingen in de anastomose7 • Deze MMP-9 activiteit is het hoogst tijdens de eerste 3 dagen van het genezingsproces en leidt tot toegenomen collagenolyse. In vergelijking met normaal genezende anastomoses treden bij naadlekkage duidelijk veranderingen in collageen remodellering door matrix metalloproteinases (met name MMP-1, -2 en -9) op8• Hierdoor hebben hechtingen in de eerste dagen van genezing minder houvast in het darmweefsel. Deze bevindingen samen met het oude chirurgische adagium "de wond geneest niet door de hechtingen, maar tussen de hechtingen", bieden uitzicht op een interessante vervolgstudie. Zoals Agren 7 stelt, vindt meer collagenolyse plaats rond de hechtingen. Daarom zou deze studie herhaald moeten worden met gebruik van staplers. Dit zou dan tot meer uitgesproken verhoging van matrix metalloproteinase activiteit moeten leiden, omdat staplers meer hechtpunten en perforaties in de darmwand veroorzaken. Er van uitgaande dat bij naadlekkage verhoogde aanwezigheid en activiteit van MMPs in de vroege fase van genezing bestaat, is het interessant na te gaan of innovatieve scanning-technieken zoals near-infrared fluorescence imaging intraoperatief de aanwezigheid van MMPs kan aantonen. Recente ontwikkelingen op het gebied van zowel camerasystemen als zgn. "smart activable probes" zoals MMP-gevoelige fluorescente optische contrast agentia (b.v. MMPSense™) zouden een nieuw terrein voor onderzoek naar non-invasieve detectie van biomarkers voor naadlekkage door middel van moleculaire imaging kunnen vormen9.

In een humane studie naar de effecten van preoperatieve bestraling met 5x5 Gy werd verhoogde activiteit van MMP-2 en -9 in colonweefsel gevonden. Verhoogde MMP-2 activiteit werd hierbij geassocieerd met meer wondgenezingsstoornissen en fistelvorming1 0 • I n patienten met Marbus Crohn, een ziekte die berucht is om fistelvorming in de tractus digestivus, werd verhoogde activiteit van MMP-1 aangetroffen 1 1 • Er zou dan ook een verband kunnen zijn tussen een matrix metalloproteinase profiel in preoperatieve colonbiopsieen en naadlekkage. Ook andere extracellulaire matrix componenten zoals cluster of differentiation (CD) 1 1 7, transforming growth factor (TGF)-� an plasminogen activator inhibitor (PAI)

scores 1 2 , 1 3 zouden rol kunnen spelen bij een constitutioneel verhoogd risico op naadlekkage.

0

n 0 ::I � s:::: �-11) VI 11) ::I

0 ro ;::-0

-0 � VI "C 11)

ro .....

Er zij n du idel ij ke aanwijz ingen dat i schemie een oorzake l ij ke ra l speelt bij het ontstaan van naad lekkage . Het l igt dan oak voor de hand dat detectie van i schemie gedurende de operatie een goede voorspel ler voor een verhoogd ris ico op naad lekkage zou moeten zijn . Verder zou se lectie van goed geoxygeneerde co lonsegmenten de incidentie van naad lekkage verder kunnen terugdringen . De gangbare techn ieken ter detectie van ischem ie, d ie i n de l iteratuu r beschreven zij n (zie hoofdstuk 6), hebben versch i l lende methodologische en /of praktische bezwaren. I n d it proefschrift wordt VLS als i ntraoperatieve methode voor het beoordelen van

1 42 ischemie in het colon geevalueerd. Bij mensen zien we een verband tussen stijg ing van weefseloxygenatie rand de anastomose en naad lekkage , terwij l we in d ierexperimente le stud ies een verband tussen verlaagde weefseloxygenatie en

0

� llJ

3 rt> ::s <

8. ::s

LC

/"I 0 ::s � C Vl

rt> Vl

(D ::s

rt> A° 0 3 Vl

-8 � Vl

-a rt>

rt> .....,

gecompl iceerde naadgenezi ng zagen. Verder vonden we bij patienten met naad lekkage s ign ificant verlaagde weefse loxygenatie in een deel van het colon , dat zich ver verwijderd van de darmnaad bevi ndt. Deze observaties zij n tegenovergesteld aan het verwachte effect op bas is van zowel l iteratuurgegevens als eigen experimentele stud ies en p i lotstud ies. Hypothetisch zou namel ijk een afname van weefseloxygenatie rand de anastomose te verwachten z ijn en geen effect op het caecum moeten hebben , aangezien de c i rcu latie van dat deel van het colon i n pri nc ipe onveranderd b l ijft. Hoewel St02 i n de experimentele stud ies een sne l le en k l i n i sch goed toepasbare parameter voor het voorspe l len van naad lekkage zou ku nnen zijn , nopen onze bevi nd i ngen tot verdere studies . I n experimentele studies zou moeten warden nagegaan wat de ra l van ischemie en reperfus ie (oak op afstand van de e ige l ijke anastomose) bij weefse loxygenatie in het colon i s . H ierbij moeten tevens versch i l lende invas ieve en non- i nvasieve methoden ter evaluatie van weefseloxygenatie van het colon warden verge leken om de onderl iggende fys io logische mechan ismen van onze bevi nd i ngen te kunnen verklaren.

Het i s moge l ijk dat VLS geen accu rate voorspel ler van naad lekkage is aangezien er puntmeti ngen i n re lat ief kle i ne , oppervlakkige weefse lmonsters van de serasa warden u itgevoerd. Moge l ij kerwijs is de serasa n iet het meest i schemie-gevoel ige deel van de colonwand en zou deoxygenatie veel eerder i n de mucosa kunnen warden gedetecteerd . Deze aanname wordt ondersteund door versch i l lende studies d ie een u itstekende corre latie tussen St02 van colonmucosa, gemeten met VLS, en fe ite l ijke ischemie aantonen 1 4-1 6 • Ook onze e igen dierexperimente le studies laten d it verband zien bij metingen in de co lonserasa, waarbij moet warden opgemerkt dat het gemeten weefse lmonster bij ratten aanzien l ijk du nner is i n ve rge l ij king met mense l ij k colon (ca. 0. 5 vs 2-4 m m). Het i s aannemel ijk dat meti ngen met VLS bij ratten daardoor fe ite l ij k een meti ng van de vo l ledige wanddikte van het colon representeren . Deze aanname wordt ondersteund door k l in i sche observaties bij patienten , waarbij in colon met op het oog normale serasa macrascopische ischemische veranderi ngen u its l u itend i n de mucosa warden waargenomen. Helaas i s het, zoals we tijdens de k l in i sche p i lotstud ie hebben gezien , vaak n iet moge l ij k om tijdens de operatie meti ngen in de m ucosa u it t e voeren , aangezien tijdens resectie het colon vaak met staples wordt gesloten om contam inatie van het

operatieterre in te voorkomen. De metingen van colonmucosa, d ie we tijden s de pi lotstudies door m iddel van verbl ijfsprobes i n rectosigmoid en oesofagus en door puntmetingen van de mucosa ( ind ien mogel ijk) hebben u itgevoerd, b leken onbru ikbaar door hoge standaarddeviaties ten gevolge van bewegingsartefacten door man ipu latie tijdens operatie en door contam inatie met fecaal materiaal (ongepubl iceerde resu ltaten). Mucosa-metingen zouden dus een betere voorspel ler van naadlekkage kunnen zijn , maar goede meti ngen zij n waarschijn l ij k tijdrovend en daardoor m inder gesch ikt voor i ntraoperatief gebru i k voor d it doe l . De a l eerder genoemde n ieuwe camerasystemen, waarbij een g roter weefselgebied rond de 1 43 anastomose met g rotere penetratied iepte real-time gescand kan warden, zouden een betere methode voor i ntraoperatieve ischemiemeti ng kunnen zijn.

Zoals aangetoond in een recente d ierexperimente le studie is er een aanz ien l ij k effect van ischemie en reperfus ie i n andere delen van het colon op darmnaadgenezing 1 7• Mogel ij k treedt n iet al leen tijdens, maar ook pas na het beeind igen van de e igen l ij ke operatie s ign ificante ischemie van de anastomose of andere delen van het colon op. Dit zou kunnen warden veroorzaakt doordat oedeem van de darmwand en tractie op de naad een re latieve afname van de splanchn icus c i rcu latie i n de vroege postoperatieve periode veroorzaken. Daarnaast beg int de peristaltiek van het colon gedurende de eerste postoperatieve dag 1 8, waardoor een toename van mechan ische i ntral um inale en long itudi nale krachten op de anastomose kan worden veronderste ld . Een stud ie, waarbij door middel van CT­ang iografie gegevens over perianastomotische vaatvoorz ien ing worden gekoppeld aan zowel hemodynamische parameters als het optreden van naad lekkage, zou h ieromtrent waardevol le informatie kunnen opleveren. In een d iermode l met honden wordt een du ide l ij ke verlag ing van weefseloxygenatie bij stijgende mechan i sche longitudi nale krachti nwerking op colonanastomoses beschreven 1 9 • Een derge l ijke mechan ische krachti nwerki ng, d ie fys iolog isch aanwezig is in het colon en waarschij n l ij k toeneemt na rectos igmoid resectie 1 7, 20-23, zou daarom ook pas na de i n it ie le operatie ischemie kunnen induceren. In veel rectos igmoidresecties wordt een re latief kart mesenteri um van het aanvoerende colondeel achtergelaten , bedoe ld om de arteria mesenteria inferior als bloedvoorz ien ing voor de anastomose te behouden. Met het oog op mechan ische krachten, die op de anastomose inwerken, zou d it echter contraproductief kunnen werken

Bij het overwegen van de ro l van mechan ische krachten op de anastomose mag de rol van darmvoorbere id ing van het colon voor operatie n iet bu iten beschouwing b l ijven. Hoewel in een aantal grote studies en trials geen voordeel door darmvoorbere id ing wordt aangetoond, wordt d it door veel ch i rurgen toegepast vanwege de veronderste l l i ng dat passage van feces de n ieuwe anastomose beschad igt, ook ind ien een devierend stoma geplaatst wordt. Deze praktij k kan n iet door gegevens u it de l iteratuur warden ondersteund. Er zij n zelfs aanwijz ingen dat het voorbereiden van het colon tot hogere piekdrukken in het colon descendens le idt 1 s, 2 1 ,2 2 • Mogel ij k bl ij ken dan ook interventies, gericht op het reduceren van

0

V'l PJ

:::s < PJ r-t !:!. :::s

lO

intraluminale druk op de darmnaad (bijvoorbeeld C-Seal, hoofdstuk 8) succesvol bij het voorkomen van naadlekkage.

Gezien het kleine aantal studies, dat de intraluminale druk voor en na aanleg van colonanastomoses beschrijft, en de afwezigheid van studies, die zowel druk als tractie op humane anastomoses in relatie tot naadlekkage beschrijven, vormt dit een interessant onderwerp van toekomstige studies. Ook de relatie tussen mechanische krachten, ischemie en naadlekkage zou verder geevalueerd moeten warden, bijvoorbeeld door gebruik te maken van ischemiemarkers zoals pimonidazol,

144 idealiter gecombineerd met near-infrared markers in combinatie met near-infrared camerasystemen. De rol van andere factoren, die de kwaliteit van de darmwand kunnen be"invloeden, zoals MMPs, zouden door middel van dierexperimentele

0

V'I PJ 3 (D :::::l <

8. n 0 :::::l � C Ill

iii" Ill

(D :::::l

(D � 0 3

Ill "O (D

(D ....,

studies verder in kaart kunnen warden gebracht door zgn. smart activatable near­infrared fluorescent optical contrast agents, aangezien de fasen in darmnaadgenezing op deze wijze afgebeeld kunnen warden zonder het dier te hoeven termineren. Verdere studies naar de effecten van preoperatieve bestraling kunnen met behulp van het door ons ontwikkelde hernia-diermodel en gebruik van MMP-gevoelige probes (bijvoorbeeld MMP-2, -8 en -9) warden uitgevoerd. Hierbij kunnen zowel vragen naar optimale timing, dosis en fractionering van bestraling als effecten van radioprotectoren zoals Tempol24 warden nagegaan.

In de vroege postoperatieve fase spelen een tweetal andere factoren een rol, nameli jk adhesievorming rand de anastomose en drainage. Bij genezingsstoornissen en beginnende naadlekkage zou in de klinische situatie afdekking met adhesies plaats kunnen vinden, zoals we dit bij ratten in hoofdstuk 4 van dit proefschrift hebben beschreven. Bij klinische studies naar methodes voor preventie van adhesievorming wordt dan ook bij effectieve adhesievorming meer naadlekkage gevonden2s , 26.

Ondanks de uitkomsten van de meta-analyse beschreven in hoofdstuk 2 zou het zinvol kunnen zijn preventief drains te plaatsen bij een lage colorectale naad, aangezien meestal door het uitvoeren van TME-resecties een holte achterblijft in een deel van de buikholte waar geen peritoneum aanwezig is. Achtergebleven vochtcollecties kunnen daardoor onvoldoende geresorbeerd warden, wat geassocieerd wordt met meer naadlekkage27 • Er zijn geen gerandomiseerde klinische trials met voldoende power beschreven om deze vraag te beantwoorden, maar een retrospectieve studie laat een duidelijk voordeel van drainage van lage naden na TME zien28• Verder zouden drains in het postoperatieve traject zinvol kunnen bijdragen aan vroegere diagnostiek van naadlekkage door analyse van biomarkers van het drainvocht29• Er zijn duidelijke aanwijzingen dat patienten met naadlekkage al zeer vroeg in het postoperatieve beloop een inflammatoire reactie vertonen, die aan hand van C-reactive protein (CRP) voorspeld kan worden30• Op dit moment wordt in Nederland een multicenter study uitgevoerd, waarbij vroege, zowel biochemische als fysiologische, voorspellers van naadlekkage ge"identificeerd warden (Den Dulk, persoonlijke communicatie). Samenvattend zou het plaatsen van drains en het zorgvuldiger observeren van fysiologische en biochemische

parameters een vroegere identificatie van patienten met naad lekkage moge l ij k kunnen maken , waardoor m inder vaak en meer selectief e e n devierend stoma aangelegd kan warden.

Tens lotte kunnen we ste l len dat op versch i l lende terre inen vervolgonderzoek zou kunnen bijdragen aan een beter begrip van naadlekkage. Toekomstige stud ies op het gebied van predict iemode l len , b iomarkers en weefsel karakteristieken , non­invas ieve imag ing-techn ieken en i ntraoperatieve risico- inschatt ing zouden i n de komende vijf tot t i en jaar kunnen bijdragen aan een beter begrip van het ontstaan 1 45 van naad lekkage en kun nen he lpen bij het voorkomen daarvan.

0

Vl SlJ 3 (D ::l <

8. ::l

lO n 0 ::l n C V, (D V, (D ::l r-1' 0 (D .,_. 0 3 V, r-1'

"'O (D ...... V,

"'O (D n r-1' ro· -t,

1 46

0

Vl $l.l

3 (D ::, <

8. ::i

I.O

n 0 ::i

C V,

(D V,

(D ::,

(D ;,:::-0 3 V,

(D

V, -0 (D

(D .....,

Referenties

1 . Matthiessen P, Hallb66k 0, Rutegard J , et al. Defunctioning stoma reduces symptomatic

anastomotic leakage after low anterior resection of the rectum for cancer, a randomized

multicenter trial. Ann Surg 200 7; 246: 207-2 1 4.

2. Doeksen A, Tanis PJ, Vrouwenraets BC, et al. Factors determining delay in relaparotomy for

anastomotic leakage after colorectal resection. World J Gastroenterol 2007; 1 3: 3 72 1 -3725.

3. den Dulk M, Smit M, Peeters KC, et al. A multivariate analysis of limiting factors for stoma reversal

in patients with rectal cancer entered into the total mesorectal excision (TME) trial: a retrospective

study. Lancet Oneal. 2007; 8: 297-303.

4. Makela JT, Kiviniemi H , Laitinen S. Risk factors for anastomotic leakage after left sided colorectal

resection with rectal anastomosis. Dis Colon Rectum 2003; 46: 653-660.

5 . Alves A, Panis Y, Trancart D, et al . Factors associated with clinically significant anastomotic leakage

after large bowel resection : multivariate analysis of 707 patients. World J Surg 2002; 26: 499-502.

6. De Hingh I, Lamme R, van Goar H, et al. Changes in gelatinase activity in the gastrointestinal tract

after anastomotic construction in the ileum or colon. Dis Colon Rect 2005; 48: 2133-2 1 41 .

7 . Agren MS, Andersen TL, Mirastschijski U , et al. Action of matrix metalloproteinases at restricted

sites in colon anastomosis repair: an immunohistochemical and biochemical study. Surgery 2006;

140: 72-82.

8. Stumpf M, Klinge U, Wilms A. et al. Changes of the extracellular matrix as a risk factor for

anastomotic leakage after large bowel surgery. Surgery 2005; 1 37: 229-234.

9. Weissleder R, Pittet MJ. Imaging in the era of molecular oncology. Nature 2008; 452:580-589.

1 0. Angenete E, Langenskjold M, Falk P, et al. Matrix metalloproteinases in rectal mucosa, tumour and

plasma: response after preoperative irradiation. lnt J Colorectal Dis 2007; 22: 667-674.

1 1 . Stumpf M, Cao W, Klinge U, et al. Reduced expression of collagen type I and increased expression

of matrix metalloproteinases 1 in patients with Crohn's Disease. J Invest Surg 2005; 1 8: 33-38. 1 2 . Krones CJ, Klinge U, Butz N, et al . The rare epidemiological coincidence of diverticular disease and

advanced colonic neoplasis. Int J Colorectal Dis 2006; 21 : 1 8-24.

1 3. Klinge U, Rosch R, Junge K, et al . Different matrix micro-environments in colon cancer and

diverticular disease. Int J Colorectal Dis 2007; 22: 51 5-520.

1 4. Benaron DA, Parachikov IH , Cheong WF, et al. Quantitative clinical non-pulsatile and localized

visilble light oximeter: Design of the T-Stat™ Tissue Oximeter. SPIE Optical Tomography

Spectroscopy Tissue V 2003; 4955: 355-68. 1 5. Benaron DA, Parachikov IH , Friedland S, et al. Continuous, noninvasive and localised microvascular

tissue oximetry using visible light spectroscopy. Anesthesiology 2004; 1 00: 1 469-1475.

1 6. Friedland SR, Benaron D. Reflectance spectrophotometry for the assessment of mucosal perfusion

in the gastrointestinal tract. Gastrointest Endosc Clin N Am 2004; 1 4: 539-555.

17. Colak T, Turkmenoglu 0, Dag A et al. The effect of remote ischemic preconditioning on healing of

colonic anastomoses. J Surg Res 2007; 1 43: 200-205.

1 8. Huge A, Kreis ME, Zittel TT, et al. Postoperative colonic motility and tone in patients after colorectal

surgery. Dis Colon Rectum 2000; 43:932-939.

1 9. Shikata J I , Shida T. Effects of tension on local blood flow in experimental anastomoses. J Surg Res

1 986; 40: 1 05-11 1 .

20. Burkitt DS, Donovan IA. lntraluminal pressure adjacent to left colonic anastomoses. Br J Surg 1 990;

77: 1288-1290.

2 1 . Bampton PA, Dinning PG, Kennedy ML, et al. Spatial and temporal organization of p ressure

patterns throughout the unprepared colon during spontaneous defecation. Am J Gastroenterol

2000; 95: 1 028-1 035.

22. Bampton PA, Dinning PG , Kennedy ML, et al. Prolonged multi-point recording of colonic

manometry in the unprepared human colon: providing insight into potential ly relevant pressure

wave parameters. Am J Gastroenterol 2001; 96: 1 838- 1 848.

2 3 . Narducci F, Basotti G, Gaburri M, et al. Twenty four hour manometric recording of colonic motor

activity in healthy man. Gut 1987; 28: 17-25.

24. Vitolo JM, Cotrim AP, Sowers AL, et al. The stable nitroxide tempol facilitates salivary gland

protection during head and neck irradiation in a mouse model. Clin Cancer Res 10:1807-1812 . 25. Dine S , Ozaslan C , Kuru B , e t al . Methylene blue prevents surgery-induced peritonela adhesions but

impairs the early phase of anastomotic wound healing. Can J Surg 2006; 49: 321-328.

26. Beck DE, Cohen Z, Fleshman JW, et al. A prospective, randomized, multicenter, controlled study of

the safety of seprafilm adhesion barrier in abdominopelvic surgery. Dis Colon Rectum 2003; 46:

1310-1319.

27. Power N, Atri M, Ryan S, et al. CT assessment of anastomotic bowel leak. Clin Radial. 2007; 62:37-

42. 1 47 28 . Peeters KC, Tollenaar RA, Marijnen CA, et al. Risk factors for anastomotic failure after total

mesorectal excision of rectal cancer. Br J Surg, 2005; 92: 211-216.

29. Kamen N, de Bruin R, Kleinrensink G, et al. Anastomotic leakage, the search for a reliable

biomarker. A review of the literature. Colorectal Dis 2008; 10: 1 09-11 7.

30. Matthiessen P, Henriksson M, Hallbook 0, et al. Increase of serum (-reactive protein is an early

indicator of subsequent symptomatic anastomotic leakage after anterior resection. Colorectal Dis

2008; 10: 75-80.

Dankwoord

Dankwoord

Sinds 2004 heb ik -meestal met plezier- aan d it proefschrift gewerkt. Hoewel het schrijven van een proefschrift doorgaans een eenzame aangelegenheid is wi l ik op deze plaats meerdere personen bedanken voor bijd rages van verschi l lende aard .

Al lereerst prof. dr. T. Wiggers. Beste Theo, j ij was degene d ie de handschoen voor m ijn voeten wierp. Jouw u itspraak "als je een cabaret kan maken , dan kan je ook een proefschrift schrijven" heeft m ij over de streep getrokken om de promotie door te zetten. Daarna heb je me een poos wei n ig gezien en wist je waarschijn l ij k n iet echt wat ik al lemaal u itspookte in dat lab en het Martin i z iekenhuis , maar kreeg je vooral reken i ngen. Ook hebben je ongetwijfeld berichten bere ikt over m ijn avontu ren i n besneeuwde bergen over de he le wereld. l k kan me voorstel len dat het soms moe i l ijk was om er in te bl ijven ge loven dat de arti ke len er echt g i ngen komen. Hoe dan ook , je was altijd bereid om de u itkomsten van de studies en de verdere plann i ng met me door te nemen, samen met meer algemene zaken zoals ch i ru rg zij n en warden. En we lke promotor komt 's avonds bij storm en regen n og naar het d ieren lab te komen om kadaverstud ies bij ratten te verrichten? Theo, is ben je zeer dankbaar voor deze plezierige samenwerki ng.

De tweede essentiele persoon voor d it proefschrift is m ijn copromotor, dr. G.M. van Dam. Beste Go. Er zij n wein ig ch iru rgen die op biomolecu lair n iveau toepass ing voor al ledaagse ch irurg ische problemen proberen te vi nden. De manier waarop jij d it doet is enorm inspirerend geweest, en ik vond het een un ieke ervari ng om via jou over het hek van de ch i ru rg ie i n onderzoekstu i ntjes van al le moge l ij ke special ismen te kij ken. Van jouw enthousiasme, je bege le id ing bij het opzetten van het onderzoek en het schrijven van de stukken heb i k vee l geleerd , hiervoor m ijn dank!

Dr. C.J. Zeebregts. Beste Clark, je hebt op twee belangrijke fronten d it proefschrift begele id. Ten eerste door je gedu ld ige correcties van de soms e i ndeloze stroom van conceptart ike len . Ten tweede door je betrokkenheid en je nooit aflatende optimisme wat betreft dead l i nes en de moge l ijkheden om die toch te halen. Jammer dat we je n iet vroeger "gevonden" hebben , dan had i k twee copromotoren !

De l eden van de leescommiss ie , prof. d r. R.P. Ble ichrodt, Prof. d r. J.A. Langendijk en Prof. dr. R.J . Ploeg dank ik voor de aandacht voor het manuscript en ben ik zeer erkente l ij k voor het zitt ing nemen in de leescommissie. Dr. Rob Coppes en Piet Wiersma van de afde l i ng Rad iobiolog ie ("d ie afdel ing met d ie onmoge l ij k lange naam") voor de vakkundige hu lp bij het opzetten van de bestral i ngsstud ie .

Anne van der Stoe l , N ie ls Harlaar en Klaas Kuperus , d ie a ls keuzeco-assistenten in het Marti n i Ziekenhuis vee l van de ReSpect-metingen op de OK hebben u itgevoerd

151

hebben daarmee een fl i nke b ijd rage aan d it proefschrift geleverd. M ijn harte lijke dank hiervoor!

Arie N ijmeijer van het CDL wi l i k heel hartel ij k bedanken voor de hu lp bij de log istiek rond de d ierstud ies. Zonder jouw vakkund ige hu lp en je kl i n i sche b l ik bij het beoordelen van het wel en wee van mijn labratten was het al lemaal wel wat saaier geweest ("Arie, waarom braken ratten n iet gewoon als ze een i leus hebben?") . Ook de andere medewerkers en biotechnici van het CDL wi l i k bedanken voor de

1 52 hu lp , het improvisatievermogen (0, dat vervoer ook altij d !) maar ook de gezel l igheid bij het u itvoeren van deze stud ies. Met name Angela, Natasja, Maurice, Yvonne, Sylvia en Ar wi l ik hier speciaal bedanken.

Verder wi l i k graag David Benaron bedanken voor de starthu lp en het advies bij het gebru ik van de T-Stat.

De ch i rurgen en ass istenten u i t het Mart in i Ziekenhu i s en het UMCG wi l ik bedanken voor het gedu ld bij het u itvoeren van de metingen van ReSpect studie en het steeds weer i nvu l len van de VAS voor de Lekstud ie. Annette Ol ieman, Adri Stael , Gerard Beerthu izen hebben het leeuwendeel van de meti ngen verricht. Ook Peter Baas wi l ik hier als opleider speciaal noemen. Oak j ij hebt veel van de meti ngen gedu ld ig doorstaan , maar je hebt ook aan d it proefschrift bijgedragen door er als op leider achter te staan. Dat is best wat waard .

De operatieass istenten in het Marti n i Ziekenhu i s , d ie altijd bereid waren om met d ie "vreemde" apparatuu r op de OK te helpen en waarden te noteren, ook al was het n iet j u l l ie taak.

Esther Kompagne, d ie als secretaresse in het Marti n i Ziekenhu is aan m ij gekoppeld was , heeft met engelengeduld het leeuwendeel van de statussen vers leept, die ik steeds weer nodig had. Ook de andere secretaresses in het Marti n i ben ik dankbaar voor hun d i recte en ind i recte hu lp !

Astrid Hei nstra, maar i k heb het contact en je hu lp bij de soms bijna onmogel ijke taak om m ijn promotor te pakken te krijgen altijd als bijzonder positief ervaren. Dank je wel l

El len Kl i nkert en Evelien Lemstra. "When wi l l we three meet agai n , i n thunder l ightn ing or in rain . . . " Er zij n van die periodes dat het l ij kt alsof er van d ie meetings nooit meer wat terecht komt ("d rukd rukdruk") , en ik vind het geweldig dat "de drie­eenheid" van het eerste jaar Geneeskunde desondanks bl ijft bestaan. Fantastisch dat j u l l ie deze keer m ijn paran i mfen z ij n !

En tenslotte Bert, die de wereld heldhaftig voor mijn interpunctiefouten en anakoloeten heeft behoed: je bent een wereldvent, duizend bommen en granaten.

1 51

Curricu lum Vitae Anne Karliczek was born on March 2nd, 1971 in Hannover, Germany. After

graduating from secondary school in 1990 at the Sint Maartenscollege in Groningen,

the Netherlands, she attended Medical School at the University of Groningen.

Between September 1998 and March 2001 she worked as a resident medical officer

at the cardiothoracic surgical department (prof. dr. T. Ebels) and the intensive care

department (dr. J .H. Zwaveling) of the University Medical Center Groningen. Surgical

training took place between March 2001 and August 2007 at the Martini Hospital

Groningen (dr. P.C. Baas) and at the University Medical Center Groningen (prof. dr.

H.J. ten Duis), interrupted between September 2006 and January 2007 to perform

part of the experiments described in this thesis at the Animal Laboratory Facility of

the University of Groningen. Between July 2004 and October 2007 the experiments

and clinical studies described in this thesis were carried out. Between November

2007 and April 2008 she worked as a general surgeon in the Martini Hospital

Groningen. Currently she is subspecialising in gastro-intestinal surgery in the

University Hospital Bergen, Norway (dr. G. Baatrup). The author is married to Bert de

Bruin.

155