AJR:168, May 1997 1209

Anatomic Variants of MesentericVeins: Depiction with Helical CTVenography

Oswald Graf1GilesW. Boland1John A. Kaufman1Andrew L Warshaw2Carlos Fernandez del Castillo2Peter A. Mueller1

OBJECTIVE. The purpose of this study was to describe the variable anatomy of mesentericveins on axial CT images and on volume-rendered CT venograms that use maximum intensity

projection and shaded-surface display.

SUBJECTS AND METHODS. Fifty-seven patients undergoing helical CT ofthe pancreas

were included in the study. The mesenteric venous system was analyzed in 54 patients. Three

patients were excluded because the helical CT data were unsatisfactory.

RESULTS. On helical CT with maximum intensity projection and shaded-surface display,

the superior mesenteric vein (SMV) was seen as a single trunk of variable length in 4() patients.

In seven other patients. two mesenteric trunks merged separately with the splenic vein. In the

remaining seven patients. the SMV was occluded by tumor. The inferior mesenteric vein drained

into the splenic vein in 28 patients (56%). into the SMV in 14 patients (26%). and into the sple-

nomesenteric angle in nine patients (18%).

CONCLUSION. Both axial and volume-rendered CT venograms accurately reveal the vari-

able mesenteric venous anatomy. CT venograms may replace conventional angiography in pre-

surgical planning.

Received June 11, 1996; accepted afterrevisionOctober 21, 1996.

tDepartment of Radiolo9y-WHT 220, MassachusettsGeneral Hospital, 32 Fruit St, P.0. Box 9657, Boston,MA 02114. Address correspondence to G. W. Boland.

2Oepartment of Surgery, Massachusetts General Hospital,

Boston, MA 02114.

AJR1997;i68:1209-1213

0361-803X/97/1685-1 209

American Roentgen Ray Society

C onventional and helical CT are thecurrent imaging methods of choice

to evaluate the pancreas and the

peripancreatic region. Axial images provide

most of the essential information. especially

when determining the potential resectability of

pancreatic neopla.sms [1 , 2]. To date. however.

many surgeons still request preoperative con-

ventional angiography. Their stated reasons

include accurate depiction of vascular anatomic

variants and depiction of the disease process

and its relationship to splanchnic vasculature on

anterior-posterior and lateral views. which give

a more pictorial quality to spatial information

than do axial (1 images 13-5 1-

More recently. contrast-enhanced helical CT

with three-dimensional image reconstruction

(CT angiography) has become increasingly

important to evaluate the vascular systeni. This

method permits diagnostic vascular imaging

with less patient morbidity and at a lower cost

than conventional angiography [6. 7J. The

application of this new technique has concen-

trated on the arterial system in various regions

of the body [8. 91. and few reports describe its

applications in the venous system. Previous

studies have evaluated the portal vein and its

intrahepatic segmental branches with three-

dimensional helical CT rendering techniques

I 10. 1 1]. However. the superior mesenteric vein

(SMV) and its tributaries. which represent criti-

cal anatomic structures, especially in pancreatic

disease, have not been evaluated with volume-

rendering techniques. The purpose of this study

was to show the ability of helical CT venogra-

phy to accurately depict the mesenteric venous

system. For this pur)se. we analyzed the nor-

mal anatomy and anatomic variants of the supe-

nor and inferior mesenteric vein on axialimages and on volume-rendered CT venograms

using maximum intensity projection (MIP) and

shaded-surface display (SSD). Furthermore. we

correlated the anatomy shown on CT veno-

grams with the anatomy seen during conven-

tional angiography or surgery.

Subjects and Methods

Fifty-seven consecutive patients (37 men, 20

women: 3()-X3 years old: mean age. 64 years old)with known r suspected disease of the pancreas

were referred for CT and were included in thestud. Forty-one patients had neoplastic disease of

the pancreas (36 adenocarcinomas, two islet celltumors. two ampullary carcinomas, one metastaticdisease ). Eight patients had chronic pancreatitis.

Another eight patients were scanned for suspected

Graf et al.

1210 AJR:168, May 1997

disease of the pancreas and subsequently were

tOtlfld flot tO have pancreatic disease.

ihe pancreas and the Piril)itcreatic region were

exaniined in detail using a dual-phase helical CT

prt(cl in both the arterial and the portal venous

Phases. Fur the purpose of this study. mesentericveins were aiialvzed from the portal venous phasedat:t set. CT scans were performed on a 1-liSpeed

Advantage scatiner (General Electric Medical Sys-

tenis. s1ilv,aukee. \Vl ). Initial Iocaliting Iosv-milli-

ampere-second precontrast axial images were

performed from the diaphragm to [.3. The cephalad

i1argin of the helical scan volume was chosen 2 cm

abt)\e the origin of the celiac trunk. The caudal

slice was chosen 2 cm below the uncinate process.

This positioning equated to i scan range of I ()- I I

cut in the z-directioii in the 57 patients. One hundred

sixty ailliliters of nonionic contrast riiediuni

(Omnipaque 3(X): Nycomed. New York. NY) was

injected at a llO\ rate of 4 ml/ec into an ainecubital

\,eiil ( I -(- to 20-gauge needle) by a power injector

( Niedrad. Pittsburgh. PA . The first helical CTsequence was started I l sec aller the initiation of the

injection. and the secoiid helical scan for venousphase inlaging was started fit) sec after initiation ofthe infection. Helical scanning paraiaeters for venous

phase scanning included l4() kVp. 2(X)-22() mAs, 3-i,ilifl collinlatI(in. i pitch of I .3. and I -naii overlap-ping reconstri.iction. Patients held their breath forapproximately 27-30 sec. A narrow field ofview (20cia ) centered over the origin of the superior niesen-

teric arteri svis used.

Reconstnicted data were transferred to a worksta-

tiOfl (Advantage \Vorkstation: General Electric Mcd-

cal S stems I for pstprocessing to show the nonnal

5,eiR)tis anatomy ind variants with three-diitiensional

rendenng techniques. A radiologist performed all

volume-rendering maneuvers. MIP generation

required remos al of the aorta and txine posterior to

the splanclinic vessels using cutting techniques.

Generation of SSD images required measurement of

Hounsfield units by regions of interest placed on the

axial images at the splenoportal junction. in the

SMV. 3 cm below the splenoportal junction. in distal

intestinal branches of the SMV. and in the proximal

and distal intiior mesenteric vein (IMV). Advanced

fx)stprocessing algorithnis (cut and remove func-

tions. filter floaters. erosion) were applied to erase

structures outside the regions of interest (e.g.. spinal

coluiiin) or overlapping structures (e.g.. aorta. opaci-

lied pancreatic parenchyma. mesentery).

In 54 patients axial scans and CT venograms were

considered to be of sufficient quality for evaluation of

the mesentenc venous system. Helical CT data (axial

and three-dimensional reconstructions) were unsatis-

factory in three of 57 examinations because of eitherbody habitus (one patient) or poor cooperation with

breath-holding (two patients). Anatomic details of the

SMV and IMV were analyzed on MIP and SSD pro-jections of all axial source images by four radiolo-

gists. CT venograms were correlated with findings at

conventional angiography in 15 patients. In 19

patients who underwent surgery. axial images and CT

venograms were additionally reviewed by two sur-

geons aid correlated with findings at surgery.

Results

The time to generate MIP images was

approximately 10 mm, and generation of SSD

images required approximately 15-20 mm.

The anatomy shown on volume-rendered CT

venograiTis corresponded to the anatomy

shown on axial source images. In the 15

patients who also underwent conventional

angiography, the CT venograms correlated

with the venous anatomy shown on the con-

ventional venograms. In the 19 patients who

also underwent surgical exploration, surgical

findings confirmed the anatomic pattern seen

with CT venography.

SMV

In 28 of 54 patients a main trunk of variable

length (5-50 mm) was observed before its

division into two intestinal branches (Figs. I

and 2). In 12 of 54 patients a division of the

main trunk into right and left intestinal

branches was not observed within the scan-

ning volume (6.5 cm below the splenoportal

confluence) (Fig. 3). In seven of 54 patients

the main trunk ofthe SMV was absent and two

large mesenteric branches drained directly into

the splenic vein (Figs. 4 and 5). In the remain-

ing seven of 54 patients the main trunk of the

SMV was occluded by tumor and venous

drainage was by collateral pathways (Fig. 6).

The gastrocolic trunk, which is a tributary of

the SMV, was revealed in 47 of 54 patients; in

the other seven patients the gastrocolic trunk

could not be seen because oftumor invasion. Of

these 47 patients the gastrocolic trunk drained

into the main trunk ofthe SMV in 25 (53%) andinto the right intestinal branch of the SMV in

the other 22 (47%). The distance from the ter-

mination of the gastrocolic trunk into the SMV

or the right intestinal branch to the splenoportal

confluence ranged from I I to 39 mm (mean, 24

mm). Although the gastrocolic trunk was visi-

ble on the anteroposterior MIP and SSD

images. anatomic details were best visualized

on the axial source images and on axial MIP

images (Figs. 3B and 6B). On anteroposterior

MIP images tributaries to the gastrocolic trunk

(anterior superior pancreaticoduodenal vein,

right gastroepiploic vein, right colic vein) were

frequently superimposed by the SMV. On SSD

images the tributaries were frequently erased by

a threshold effect.

The first jejunal branch of the SMV drain-

ing the duodenojejunal flexure and the first

jejunal loop was observed in 41 patients. This

vessel drained either into the main trunk of the

SMV (22 patients) or into the left intestinal

branch (19 patients).

!MV

The IMV could be seen in 51 of 54 patients

on axial and MIP images. On SSD images the

proximal segment was occasionally erased by

Fig. 1.-Normal mesenteric veins in 56-year-old man ex-amined because of abdominal pain with no evidence ofpancreatic disease.A, Restricted anterior maximum intensity projection ofvenous phase of contrast-enhanced helical CT scanshows superior mesenteric vein (smv)formed by right(R)and left(L) intestinal branches. Splenic vein Isv) and por-tal vein (pv) are clearly seen. Inferior mesenteric vein(open arrow) joins smv just below confluence withsplenic vein. Left colic vein (curvedarrow) drains into in-ferior mesenteric vein. Gastrocolic trunk (large solidarrow) drains into right intestinal branch. Superior me-senteric artery (smailsolid arrow) is also seen.B, Anterior shaded-surface display of edited data from Ashows portal vein (pv), splenic vein Isv), superior mesen-teric artery (smv), inferior mesenteric vein (open arrow),right (RI and left (LI branches of smv, and gastrocolictrunk (solid arrow).

Fig. 2.-55-year old man with pancreatic carcinoma in-vading superior mesenteric vein (confirmed at surgery).A. Source axial image from contrast-enhanced helicalCT scan at level 2 cm below splenoportal confluenceshows right intestinal branch (straightsolid arrow) of su-penor mesenteric vein (SMV) adjacent to tumor. Largerleft intestinal branch of SMV (open arrow) and superior

mesentenc artery (SMA) (cuivedarrow) are normal.B, Restricted anterior maximum intensity projection ofvenous phase of contrast-enhanced helical CT scanshows narrowing of right branch of SMV (small solidstraight arrow) and main trunk of SMV (large solidstraight arrow) because of tumor invasion. Left branchof SMV (open straight arrow) is normal. Short segmentof SMA (solid cuived arrow) is seen, as well as inferiormesenteric vein (IMV) (open curved arrow).C,Anterior shaded-surface displayofedited data from Bshows narrowing of right branch of SMV (small solidstraight arrow) and main trunk of SMV (large solidstraight arrow) because of tumor invasion. Also seenare left branch of SMV (open straight arrow) and IMV(open curved arrow).D, Portal phase from conventional cut-film SMA angio-gram (left posterior oblique) shows invasion of lateralwall of SMV (arrow) by tumor.

Fig. 3.-Normal mesenteric veins in 68-year-old womanwith colon cancer metastatic to liver but no evidence ofpancreatic disease.A, Restricted anterior maximum intensity projection(MIP) of venous phase of contrast-enhanced helical CTscan shows single large superior mesentenc vein (SMV)(large straight arrow), gastrocolic trunk (small straightarrow), and jejunal vein (caned arrow).B, Restricted axial MIP from same study shows jejunalvein (solid caned arrow) and gastrocolic trunk (smallsolid straight arrow) joining SMV (large solid straightarrow). Right colic vein (open straightarrow) drains intogastrocolic trunk. Portion of middle colic vein (opencaned arrow) is seen anterior to SMV.

Fig. 4.-Variant but otherwise normal superior mesen-teric vein (SMV) anatomy in 64-year-old woman withpancreatic carcinoma.A, Restricted anterior maximum intensity projection ofvenous phase of contrast-enhanced helical CT scanshows right (RI and left (L) intestinal branches of SMVmerging directly with splenic vein. Inferior mesentericvein (IMV) (open arrow) drains into splenic vein. Portionof superior mesenteric artery (SMA)(solid arrow) is seenbetween branches of SMV.B, Anterior shaded-surface display of edited data from Ashows right (RI and left (LI intestinal branches of SMV,IMV (open arrow), and SMA (solid arrow).

Fig. 6.-Occlusion of superior mesenteric vein (SMV) by pancreatic carcinoma in 73-year-old woman.A, Restricted anterior maximum intensity projection (MIP) of venous phase of contrast-enhanced helical CT scanshows occlusion of SMV by tumor (black arrow) and dilatation of gastrocolic trunk (white arrow).B, Restricted axial MIP at level of gastrocolic trunk (straight white arrow) and middle colic vein (cuived white arrow)shows tumor invasion of SMV (black arrow).

Although most radiologists depend on axial phy, a noninvasive technique, has shown

source data when interpreting CT scans, many promise in its ability to reveal the arterial sys-

Graf et al.

1212 AJR:168, May 1997

Fig. 5.-Variant but otherwise normal superior mesenteric vein (SMV) anatomy in53-year-old man with pancreatic carcinoma.A, Restricted anterior maximum intensity projection of venous phase of contrast-enhanced helical CT scan shows two large intestinal branches of SMV forming single confluencewith splenic vein.Inferiormesenteric vein (straight arrow) joins leftbranch ofSMV. Artifacts from plastic biliary stent (curved arrow) are presentB, Anterior shaded-surface display of edited data from A.C, Portalphase from conventional cut-film angiogram (rightposterioroblique)ofsuperior mesenteric artery shows inflow of unopacifled blood from splenic vein (arrow) at level ofconfluence with SMV branches.

postprocessing maneuvers. The IMV drained

into the splenic vein in 28 patients (Figs. 2 and

4), into the SMV in 14 patients (Fig. 5), and

into the angle of the splenoportal confluence in

nine patients (Fig. 1). The left colic vein,

which is a tributary of the IMV draining the

region of the left colic flexure, was visualized

in 45 patients (Fig. 1).

Discussion

surgeons find it essential to preoperatively eval-

uate the disease process and its relationship to

surrounding structures in a three-dimensional

manner. This need has led to an increasing

demand for three-dimensional image recon-

structions in various regions of the body [12,

l3J. In vascular, thoracic, and abdominal sur-

gery, many surgeons still require conventional

angiography for obtaining spatial information

about the vasculature and its relationship to dis-

ease processes. Recently, helical CT angiogra-

tem, including the celiac trunk and the SMA.

This study establishes the ability of CT venog-

raphy to show the mesenteric venous system.Helical CT angiography requires matching

the helical scan to maximal vessel opacification

to obtain optimal three-dimensional images. To

reveal the mesenteric venous system using

helical CT venography, we chose a scan delayof 60 sec alter initiation of a relatively high-

dose and high-flow contrast injection before

starting the helical Cl scan. Although we did

not study lower contrast injection rates, we

believe that a high rate is necessary to optimize

vessel opacification. Furthermore, an injection

rate of4 mI/sec was readily accomplished in all

our patients. We limited the duration of the

helical scan and the breath-hold to 30 sec. Heli-

cal scanning parameters including collimation,

pitch, overlapping reconstructions, and field of

view were chosen specifically to optimize

reconstruction of axial images into three-

dimensional images. With this technique, high-

quality venous-phase axial images and recon-

structed CT venograms were obtained for most

patients. Poor image quality in one patient was

likely due to low photon statistics because of

excessive weight. The technique is unlikely to

be successful in uncooperative patients, as evi-

denced by two patients in this study.The axial images and both volume-render-

ing techniques, MIP and SSD, clearly revealed

the anatomic variants of the mesenteric veins.

Additionally, the reconstructed CT venograms

accurately depicted the spatial anatomy of the

mesentenc vessels and their relationship to

Helical CT of Mesenteric Veins

AJR:168, May 1997 1213

surrounding structures, and the CT venograms

correlated well with findings at angiographyor surgery. This study suggested that CT

venography is now able to replace angiogra-phy as the primary technique to visualize

these structures before surgery. The primary

purpose of this study was not to evaluate thistechnique as a method to determine resectabil-

ity. Further studies are needed to determinewhether CT venograms are as accurate as

conventional angiography in detecting infor-

mation about disease.

Postmortem studies [14-16] and other CF

studies [17-2 1] have reported that the SMV

was represented as a single common tnrnk of

variable length formed by its chief tributaries,

including the ileocolic, gastrocolic, right colic,

and middle colic veins. In this study, a single

common superior mesenteric trunk was

observed in most patients, but in seven of 54

patients (13%) the main tnmk ofthe SMV was

not present and a large right and left mesenteric

branch merged separately with the splenic vein

to form the portal vein. This variation has been

recognized in the angiographic literature [22]

but to our knowledge has not been described in

CT studies. Misregistration of this variant in

prior CT studies was most likely due to the use

of a wide collimation, a longer scan duration,

and lower IV contrast flow rates and therefore

to poorer opacification of mesenteric veins. The

left mesenteric branch, which drains separately

into the splenic vein, may have been interpreted

in previous reports as being the IMV [2, 13].The IMV itself, which under normal conditions

has a relatively small diameter, may not have

been identified at all.

Previous CT reports have documented that

the gastrocolic trunk uniformly drained into the

anterior right lateral wall of the main trunk of

the SMV [20, 21]. In this study we found thatthe gastrocolic trunk drained into the right lat-

eral wall of the main trunk in approximately

only half the patients and in the other half

drained into the right intestinal branch of the

SMV. Similarly, the firstjejunal branch drained

into the main trunk of the SMV in approxi-

mately half the patients and in the other half

drained into the left intestinal branch.

CT venography is an elegant method to

depict the course of the IMV in the left

paraduodenal space. Usually, its course can be

visualized only with conventional angiogra-

phy after catheterization of the inferior mesen-

teric artery. Its major tributaries are the

superior hemorrhoidal vein, the sigmoid vein,

and the left colic vein. The former two veins

usually unite to form a common ascending

trunk beforejoining the left colic vein [14-16].

On cross-sectional imaging the IMV was visu-

alized running in the left paraduodenal space

and forming an arc cephalad to the duodenoje-

junal junction before termination [23, 24].

This anatomic pattern was also clearly shown

on MIP images in most patients. On SSD

images the IMV was occasionally erased

because of postprocessing maneuvers. The

IMV terminated into the splenic vein in 56%

of the patients in this study. In 18% of the

patients the IMV drained into the splenoportal

angle, and in the remaining 26% the IMV

drained into the SMV.

In conclusion, helical CT can produce

high-quality axial images using a narrow colli-

mation and overlapping reconstructions in com-

bination with a high-volume and high-flow

injection of contrast media. Furthermore, vol-ume-rendered CT venograms generated from

the axial data sets depict the spatial anatomy of

the splanchnic venous system well-an advan-

tage that may aid presurgical planning. This

information is generated from the latter half of

the dual-phase helical scanning protocol. Cor-

responding arterial anatomy is obtained by

generating volume-rendered images from the

first helical scan. A limitation of this scanning

protocol is that only a reduced volume is

scanned and no information is obtained from

organs and structures outside this volume,

such as when large parts ofthe liver lie outside

the scan range. Information about organs out-

side the scan volume is provided by a further

CT study or other imaging techniques. How-

ever, the intention of this specific examination

is the trade-offof noninvasive CT angiography

versus conventional angiography, which is

performed in many institutions in those

patients who will undergo surgery.

References

1. Freeny PC, Marks WM, Ryan JA, Traverso LW.

Pancreatic ductal adenocarcinoma: diagnosis andstaging with dynamic CF. Radiology 1988:166:125-133

2. Zeman RK, Silverman PM, Ascher SM, Pan RH,

Cooper C. Al-Kawas F. Helical (spiral) CT of the

pancreas and biliary tract. Radio! Clin North Am1995:33:887-902

3. Wardsaw AL, Gu Z-Y, Wittenberg J, Waitman AC.Preoperative staging and assessment of resectability

ofpancreatic cancer. Arch Surg 1999:125:230-2334. Warshaw AL. Fernandez-del Castillo C. Pancreatic

carcinoma. N EnglJMed 1992:326:455-463

5. Biehl TR, Traverso LW, Hauptmann E, Ryan JA.Preoperative angiography alters intraoperative strat-egy during the Whipple procedure. Am J Surg

1993:165:600-606

6. Napel S. Mark.s MP, Ruhin GD, et al. Cf angiogra-

phy with helical CT and maximum intensity projec-tion. Radiology 1992: I 85:607-610

7. Rubin GD, Dake MD. Semba CP. Current status of

three-dimensional helical CF scanning for imaging

the vasculature. Radio! C/in North Am 1995:33:

5 1-70

8. Rubin GD, Dake MD, Napel SA. Mctkrnnell CH,

Jeffrey RB Jr. Ihree-dimensional helical CTangiography of the abdomen: initial clinical experi-

ence. Radiologs 1993; I 86: 147- I 529. Rubin GD. Three-dimensional helical CT angiogra-

phy. RadioGraphics 1994:14:905-912

10. Soyer P, Bluemke DA, Bliss DF, et al. Surgical seg-

mental anatomy of the liver: demonstration withhelical CT during arterial portography and multi-

planar reconstruction. AiR 1994:163:99-103I I. van Leeuwen MS. Noordzij J, Femandez MA. et al.

Portal venous and segmental anatomy of the right

hemiliver: observations based on three-dimensionalhelical CT renderings. AiR l994;163: I 395-1404

12. Harris JH. Spinal imaging. In: Browner BD, ed.

Skeletal trauma. Philadelphia: Saunders, 1992:

605-613

I 3. Zeman RK. Davros Wi, Berman P. et al. Three-

dimensional models of the abdominal vasculature

based on helical CT: usefulness in patients with

pancreatic neoplasms. AiR 1994:162:1425-142914. Douglass BE. Baggenstoss AH. Hollinshead WH.

The anatomy of the portal vein and its tributaries.

Surg Gvnecol Obstet 1950:91:562-576

15. Falconer CWA, Griffith E. The anatomy of the

blood vessels in the region of the pancreas. Br JSurg 1950:37:334-344

16. Reichardt W, Cameron R. Anatomy of the pancre-

atic veins: a postmortem and clinical phlebographicinvestigation. Aeta Radio! 1980:21:33-41

17. Zerin JM, DiPietro MA. Mesenteric vascular anat-omy at CT: normal and abnormal appearances.Radiology 1991:179:739-742

18. Mam CM, Glazer GM, Williams DM, Francis IR.

CT-angiographic correlation of collateral scnouspathways: new observations. Radiology 1990:175:

375-38019. Mori H, McGrath FR Malone DE. Stevenson GW.

Dilated posterior superior pancreaticoduodenal

vein: recognition with CT and clinical significance

in patients with pancreaticobiliary carcinomas.Radiolog 1991:181:793-800

20. Mon H. McGrath FR Malone DE, Stevenson GW.

The ga.strocolic trunk and its tributaries: CT evalua-

tion. Radiology 1992:182:871-87721. Crabo LG. Conley DM. Graney DO. Freeny PC.

Venous anatomy of the pancreatic head: normal

CT appearance in cadavers and patients. AiR

1993; 160:1039-1045

22. Reuter SR. Redman HC, Cho KJ. Gastrointestinalangiography. In: Saunders Monographies in C/mi.

ca! Radiologs 3rd ed. Philadelphia: Saunders.1986: 33-77

23. Shapir J, Rubin J. CT appearance of the inferiormesenteric vein. J Conna Assist Tomogr 1984:8:

877-880

24. Chamsangavej C, DuBrow RA, Varma DG, Her-

ron DH, Robinson TJ, Whitley NO. CT of themesocolon. I . Anatomic considerations. Radio-

Graphics 1993:13:1035-1045