Oral Oncology 47 (2011) 174–179

Contents lists available at ScienceDirect

Oral Oncology

journal homepage: www.elsevier .com/locate /ora loncology

Compartmental tongue surgery: Long term oncologic results in the treatmentof tongue cancer

Luca Calabrese a,⇑, Roberto Bruschini a, Gioacchino Giugliano a, Angelo Ostuni a, Fausto Maffini b,Maria Angela Massaro a, Luigi Santoro c, Valeria Navach a, Lorenzo Preda d, Daniela Alterio e,Mohssen Ansarin a, Fausto Chiesa a

a Division of Head and Neck Surgery, European Institute of Oncology, Milan, Italyb Division of Pathology and Laboratory Medicine, European Institute of Oncology, Milan, Italyc Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italyd Division of Radiology, European Institute of Oncology, Milan, Italye Division of Radiotherapy, European Institute of Oncology, Milan, Italy

a r t i c l e i n f o s u m m a r y

Article history:Received 28 July 2010Received in revised form 17 December 2010Accepted 21 December 2010Available online 22 January 2011

Keywords:Tongue cancerCompartmental surgeryOncological results

1368-8375/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.oraloncology.2010.12.006

⇑ Corresponding author. Address: European Institut435, 20141 Milano, Italy. Tel.: +39 02 57489490; fax:

E-mail address: luca.calabrese@ieo.it (L. Calabrese

Compartmental tongue surgery (CTS) is a surgical technique that removes the compartments (anatomo-functional units) containing the primary tumor, eliminating the disease and potential muscular, vascular,glandular and lymphatic pathways of spread and recurrence. Compartment boundaries are defined aseach hemi-tongue bounded by the lingual septum, the stylohyoid ligament and muscle, and the mylohy-oid muscle. In this non-randomized retrospective study we evaluated the oncologic efficacy of CTS inpatients with squamous cell carcinoma (SCCA) of the tongue treated from 1995 to 2008. We evaluated193 patients with primary, previously untreated cT2-4a, cN0, cN+, M0 SCCA with no contraindicationto anesthesia and able to give informed consent. Fifty patients treated between October 1995 and July1999 received standard surgery (resection margin >1 cm); 143 patients treated between July 1999 andJanuary 2008 received CTS. Study endpoints were: 5-year local disease-free, locoregional disease-freeand overall survival. After 5 years, local disease control was achieved in 88.4% of CTS patients (16.8%improvement on standard surgery); locoregional disease control in 83.5% (24.4% improvement) and over-all survival was 70.7% (27.3% improvement). The markedly improved outcomes in CTS patients, comparedto those treated by standard surgery, suggest CTS as an important new approach in the surgical manage-ment of tongue cancer.

� 2010 Elsevier Ltd. All rights reserved.

Introduction

Surgery remains the primary therapeutic approach to earlystage tongue cancer, and is often part of a multi-modal approachto advanced disease. Overall survival for advanced tongue canceris about 50% and has not changed substantially over the last threedecades.1,2 Surgery with curative intent aims to remove theprimary tumor with a wide margin of normal tissue – a 1–2 cm cir-cumferential macroscopic margin is usually considered sufficient.3

However the rational bases for such a margin are unclear,4 actualmargins vary,5 and portion of tongue musculature to be removedin accordance with the ‘wide margin’ approach is not alwayswell-defined pre-operatively.

Wide resection is the common surgical approach to solid tu-mors in various parts of the body. However for musculoskeletal

ll rights reserved.

e of Oncology, Via Ripamonti+39 02 94379216.

).

or parenchymal sarcomas, application of the notions of anatomicalcompartments has revolutionized the concept of radicality.6–10

Since cancer progresses following the path of least resistance, itis possible to define an anatomical compartment that confinesthe tumor, and whose boundaries (layers of fascia) form a barrierto tumor spread. Complete removal of such a compartment istherefore the goal of compartmental surgery and has resulted insignificant improvements in local disease control compared towide resection.6–10 In one of the key experiences on 471 patients,Azzarelli7 found that local disease control was 76% in sarcoma pa-tients treated with compartmental surgery, compared to 53% inthose receiving wide excision (P < 0.001) although other ‘tradi-tional’ variables also significantly influenced outcome.

The tongue is a symmetrical organ whose extreme boundariesare the periosteum (lateral, inferior, and anterior) and the midlinelingual septum. It is composed of intrinsic and extrinsic musclesthat function in a highly coordinated manner to perform the funda-mental functions of speech and swallowing. Anatomically, thetongue consists of paired symmetrical compartments, without

Table 1Preoperative characteristics, pathological findings and postoperative treatments ofpatients undergoing standard and compartmental tongue surgery (CTS).

Characteristic Conventional(N = 50)

CTS(N = 143)

P value

Median age (interquartile range) 55 (47–66) 53 (45–63) 0.40<40 years 7 (14.0%) 25 (17.5%)40–49 years 11 (22.0%) 32 (22.4%) 0.9250–59 years 13 (26.0%) 38 (26.6%)P60 years 19 (38.0%) 48 (33.6%)

Sex 0.16Men 33 (66.0%) 109 (76.2%)Women 17 (34.0%) 34 (23.8%)

Alcohol consumption 0.35No 18 (36.7%) 59 (41.3%)Yes 29 (58.0%) 81 (56.6%)Former 3 (6.0%) 3 (2.1%)

Smoking 0.37No 19 (38.0%) 40 (28.0%)Yes 15 (30.0%) 55 (38.5%)Former 16 (32.0%) 48 (33.6%)

No. of comorbiditiesa 0.191 28 (56.0%) 84 (58.7%)2 17 (34.0%) 33 (23.1%)3 5 (10.0%) 26 (18.2%)

Neck involvement 0.13Bilateral 25 (50.0%) 89 (62.2%)Ipsilateral 25 (50.0%) 54 (37.8%)

Tumor site 0.73Base of the tongue 12 (24.0%) 31 (21.7%)Tongue 38 (76.0%) 112 (78.3%)

L. Calabrese et al. / Oral Oncology 47 (2011) 174–179 175

distinction between the body and the base of the tongue. The lin-gual septum is the medial margin separating adjacent compart-ments in each hemi-tongue. The periosteum along the lingualsurface of the mandible constitutes the lateral and anterior mar-gins. The stylohyoid muscle and ligament form the posteriorboundary. The mylohyoid muscle forms the floor of the compart-ment along its entire length. The compartments contain the tonguemusculature (intrinsic and extrinsic) together with stromal, glan-dular (sublingual) and neuro-vascular structures (lingual nerve, ar-tery and vein, terminal portion of hypoglossal nerve).

We recently reported on technical aspects of compartmentaltongue surgery (CTS).11 In essence it involves: (a) an anatomicalapproach to the disease involving removal of the primary lesionand all of its potential pathways of progression – muscular, lym-phatic and vascular; and (b) identification of the territory at riskfor metastatic representation: the muscular (mylohyoid), neuro-vascular (lingual nerve and vein) and glandular (sublingual andsubmandibular) tissues within the compartment, that form abridge between the primary tumor and the cervical lymphaticchain.

In this retrospective study we assessed the long-term oncologicefficacy of compartmental tongue surgery in comparison to stan-dard surgery in 193 patients treated from October 1995 to January2008 at the European Institute of Oncology (IEO) in Milan. Patientstreated from 1995 to 1999 received standard surgery; patientstreated after 1999 received CTS.

Reconstruction flap 0.0003None 5 (10.0%) 5 (3.5%)Chinese 10 (20.0%) 38 (26.6%)ALT 6 (12.0%) 62 (43.4%)Pectoralis major 21 (42.0%) 27 (18.9%)Platysma 7 (14.0%) 6 (4.2%)Others 1 (2.0%) 5 (3.5%)

Tumor grade 0.03G1 7 (14.0%) 9 (6.6%)G2 26 (52.0%) 56 (39.2%)G3 17 (34.0%) 78 (54.5%)

Metastatic nodes (N) 0.980 16 (32.0%) 47 (32.9%)1 14 (28.0%) 35 (24.5%)2 5 (10.0%) 20 (14.0%)3 7 (14.0%) 19 (13.3%)4 3 (6.0%) 8 (5.6%)>4 5 (10.0%) 14 (9.8%)

Surgical margin (microscopic) 0.62Negative 43 (86.0%) 130 (90.9%)Positive 1 (2.0%) 2 (1.4%)Close (down to 1 mm) 6 (12.0%) 11 (7.7%)

T status 0.07T2 25 (50.0%) 46 (32.2%)T3 4 (8.0%) 12 (8.4%)T4a 21 (42.0%) 85 (59.4%)

Materials and methods

Patient selection

We considered consecutive patients treated for stages II–IV(AJCC 7th edition12) squamous cell carcinoma (SCCA) of the oraltongue or base of the tongue who underwent major ablative sur-gery followed by reconstruction. Data on the cases was archivedin the prospective database of the Division of Head and Neck Sur-gery, European Institute of Oncology. Eligible patients were>18 years of age, of any sex, able to give written informed consent(standard surgery or CTS), who had previously untreated cT3-4a,cN0, cN+, M0 SCCA of the tongue or the tongue base, or patientswith cT2, cN0, cN+, M0 SCCA and clinical or radiologic suspicionof infiltration of the deeper muscle planes suggesting higher path-ological than clinical stage. We excluded patients with stage ycT1,cT2 (without clinical or radiological evidence of infiltration of theextrinsic musculature), cN0, cM0 disease amenable to trans-oralsurgery, and patients with co-morbidities compromising the safetyof surgery.

N status 0.87N0 16 (32.0%) 46 (32.2%)N1 11 (22.0%) 31 (21.7%)N2 (a,b,c) 23 (46.0%) 65 (45.4%)N3 0 (-) 1 (0.7%)

Stage 0.42II 9 (18.0%) 24 (16.8%)III 8 (16.0%) 12 (8.4%)IVa 33 (66.0%) 106 (74.1%)IVb 0 (-) 1 (0.7%)

Postoperative treatment 0.015None 21 (42.0%) 46 (32.2%)RT 24 (48.0%) 53 (37.0%)RT + chemotherapy 5 (10.0%) 44 (30.8%)

Follow-up duration, yearsMedian (range) 2.4 (0–9.8) 2.6 (0.1–9.1)

a From among the following: cardiovascular disease, lung disease, gastrointesti-nal disease, renal disease, metabolic disease, central nervous system disease, viraldiseases.

Pre-operative evaluation

Patients who presented for initial evaluation were interviewed,given a physical examination together with a lifestyle question-naire and assessed by fiber-optic-video pharyngolaryngoscopy(FOVPL). Lesions were biopsied with a conchotome or punch toavoid distortion of lesion architecture and also avoid subsequentunderestimation of tumor size and possibly stage. Staging wascompleted with CT or MRI with contrast, and PET if considered nec-essary. MRI is the preferred imaging modality as it allows identifi-cation of individual muscle involvement and provides estimates oflesion thicknesses that strongly correlate with histologic tumorthicknesses (although MRI estimates are usually greater than his-tologic findings).13 However, not all patients received MRI, as aconsiderable proportion come from other parts of Italy and hadvariable preoperative work-up (often CT rather than MRI). Imagingstudies performed outside the Institute were reviewed with radiol-

ogists as part of surgical planning and repeat scans were obtainedif considered necessary.

176 L. Calabrese et al. / Oral Oncology 47 (2011) 174–179

Surgical techniques

All operations were performed under general anesthesia. Pro-phylactic tracheotomy (surgical or percutaneous), separate fromthe neck dissection field, was performed in all cases. Lateral neckdissection was performed depending on the primary lesion andclinical/radiological status of the neck, and was always en blockwith the primary. Glossectomy employed either standard surgery(margins >1 cm) or CTS.11

Post-operative pathologic staging

The removed surgical specimen was oriented and mounted on asponge block to prevent tissue retraction prior to fixation (10%formalin). Specimen were examined to determine histological iso-type, grade, presence or absence of vascular invasion, and presenceor absence of lymph node metastases.

Postsurgical treatments

In accord with the policy in force at our institute before 2005,patients with disease infiltrating a surgical margin, inadequate sur-gical margins, residual disease, advanced tumors (pT4 and somepT3) or metastatic disease in removed lymph nodes, received post-operative RT. After 2005, patients with positive margins and extra-capsular spread in lymph nodes were considered high risk12–16 andwere given chemotherapy in addition to standard postoperativeRT.

Follow-up

Clinical examination with FOVPL were performed every3 months in the first 2 years, every 6 months from years three tofive and annually thereafter. USG, CT, MRI, or PET-CT images wereobtained if considered necessary.

Statistical analysis

Disease-free survival, local disease-free survival and locore-gional disease-free survival were calculated from date of surgeryto the date of event or latest follow-up. In these analyses death

Table 2Univariate (Kaplan–Meier) and multivariate (Cox proportional hazards model) analyses of

Characteristic N Recurrence Kapla

N (%) Estim

Overall surgery 193 24 (12.4) 84.2%Standard 50 12 (24.0) 71.6%CTSs 143 12 (8.4) 88.4%

Neck involvementBilateral 114 11 (9.7) 86.8%Ipsilateral 79 13 (16.5) 80.7%

GradeG1 16 2 (12.5) 89.4%G2 82 14 (17.1) 82.4%G3 95 8 (8.4) 85.6%

T statusT2 71 6 (8.4) 90.2%T3–T4a 122 18 (14.7) 79.3%

N statusN0 62 4 (6.4) 93.3%N+ 131 20 (15.3) 78.6%

Postoperative therapyNone 67 10 (14.9) 82.2%RT 77 11 (14.3) 80.9%RT + chemotherapy 49 3 (6.1) 93.9%

a Log-rank test.b Cox model with stepwise selection at P = 0.20 for entry. The table only shows variab

was considered as a competing event.17 Overall survival wascalculated from date of surgery to date of death or latest follow-up. Survival experience was represented by the Kaplan–Meierapproach18 with differences between groups evaluated by thelog-rank test. The analysis focused on the influence of type of sur-gery (standard vs. CTS) on the main endpoints of disease-free andoverall survival. Since patients were not randomized to type of sur-gery, multivariate proportional hazard Cox modeling was neces-sary to take account of differences in patient, tumor and surgeryvariables between the two groups.19 Covariates candidate forinclusion in the multivariate models were selected from: surgery(standard vs. CTS), T status (T2 vs. T3–T4), N status (N0 vs. N+),neck involvement (bilateral vs. ipsilateral), tumor grade (G1 + G2vs. G3), postoperative treatment (none vs. RT vs. RT + chemother-apy), alcohol consumption (no vs. yes) and BMI (<25 vs. P25 kg/m2). Predictors for inclusion in the multivariable models were cho-sen stepwise, if the Wald-test P value was 60.20. The results fromthe Cox models were presented as hazard ratios (HR) with 95%confidence intervals (CI) along with the Wald test P value. The Graytest,20 which takes competing risks into account, was also used tosupport the Cox model results. The groups defined by type of sur-gery were also characterized by descriptive statistics (medianswith range for continuous variables; proportions for categoricalvariables) with the significance of differences assessed by Wilco-xon’s rank-sum test for continuous variables and Pearson’s chi-square for categorical variables. The analyses were performed withSAS statistical software (SAS Institute, Cary, NC). All tests weretwo-sided.

Results

As the study was non-randomized it is unsurprising that pa-tients in the two surgical groups differed considerably. Specificallythe two series differed in terms of sex (P = 0.036), and type ofreconstruction (P = 0.0003) (Table 1). Tumor grade was worse inthe CTS group (P = 0.03), and RT plus chemotherapy more often ap-plied (P = 0.015); CTS patients also tended to have more advanceddisease stage (P = 0.07) (Table 1).

Table 3 tabulates events, and shows that deaths, local recur-rences and regional recurrences were less common in the CTSgroup, while the proportion who developed metastases in the

factors related to 5-year local recurrence.

n–Meier Coxb

ate P valuea HR (95% CI) P value

0.003 10.31 (0.13–0.72) 0.006

0.10 1 0.0073.37 (1.40–8.12)

0.12 0.044(G1 + G2) 10.40 (0.16–0.98)

0.13 1 0.0025.44 (1.89–15.69)

0.04 1 0.023.95 (1.29–12.09)

0.36 10.48 (0.20–1.19) 0.110.26 (0.07–1.02) 0.054

les entered in the final model.

L. Calabrese et al. / Oral Oncology 47 (2011) 174–179 177

CTS group was more than double that in the standard group. Tables2 and 3 present the results of univariate (Kaplan–Meier) and mul-tivariate (Cox) analyses of the influences of risk factors onoutcomes.

CTS was associated with significantly decreased risk oflocal (HR: 0.31; 95%-CI: 0.13–0.72, P = 0.006, Wald test) andlocoregional recurrence at 5 years (HR: 0.36; 95%-CI: 0.17–0.80,P = 0.011, Wald test), whereas there was no difference in distantrelapse risk at 5 years (HR: 1.90; 95%-CI: 0.55–6.51, P = 0.31, Waldtest). For all these outcomes, the Gray test provided similar results(data not shown). Overall survival was also better in the CTS group,

Figure 1 Tongue cancer: tumor progression. (A) Cadaver specimen showing the orientatalong the orientation of the muscle fibers. (C) Surgical specimen: progression along the

Table 3Univariate (Kaplan–Meier) and multivariate (Cox proportional hazards model) analyses of fasurvival.

Characteristic Events (%) Kaplan–Meier

Estimate (%)

Locoregional recurrence 38 (19.7) 75.9Standard 17 (34.0) 59.1CTS 21 (14.7) 82.0

Metastasis 22 (11.4) 86.5Standard 3 (6.0) 93.0CTS 19 (13.3) 84.4

Any recurrence 60 (31.1) 64.3Standard 20 (40.0) 54.0CTS 40 (28.0) 68.1

Overall survival 61 (31.4) 62.6Standard 27 (54.0) 43.4CTS 34 (23.8) 70.7

a Log-rank test.b Cox models with stepwise selection at P = 0.20.c Final model included alcohol consumption, type of surgery (standard vs. CTS), T stad Final model included CTS, T status (0.05 < P 6 0.20), N status (0.05 < P 6 0.20) and ne Final model included type of surgery (standard vs. CTS), T status, N status, neck invf Final model included type of surgery (standard vs. CTS), neck involvement, T status

with risk of dying (any cause) within 5 years about one third ofthat in the standard surgery group (HR: 0.41; 95%-CI: 0.24–0.70,P = 0.0012, Wald test). Survival analyses were performed with cen-soring at 5 years since most events occurred within the first 5 years(local 24 of 26; regional 15 of 15; metastases 24 of 25; deaths 63 of68).

Discussion

Cancer of the tongue is aggressive, as evidenced by its locallyinvasive behavior and propensity to metastasize to cervical lymph

ion of the tongue musculature. (B) Gross specimen of a tongue tumor: progressiongenioglossus muscle. (D) Pre-operative MRI images of the patient in C.

ctors related to 5-year locoregional recurrence, metastasis, any recurrence and overall

Coxb

P valuea HR (95% CI) P value

10.002 0.36 (0.17–0.80) 0.011c

10.22 1.90 (0.55–6.51) 0.31d

10.06 0.57 (0.32–0.99) 0.047e

10.0004 0.41 (0.24–0.70) 0.0012f

tus, N status, neck involvement (0.05 < P 6 0.20) and BMI (0.05 < P 6 0.20).eck involvement.olvement, and tumor grade., N status, tumor grade and postoperative therapy.

Figure 2 Kaplan–Meier curves showing local disease-free and overall survival inthe standard surgery and CTS groups.

178 L. Calabrese et al. / Oral Oncology 47 (2011) 174–179

nodes. The current goal of surgery with curative intent is removalof the primary together with as much surrounding tissue asnecessary to ensure that all macroscopic disease is eradicated; thisis thought to provide the best chance that microscopic disease –potential source of recurrence – is also removed. According to this‘more-is-better’ approach, it is widely accepted that >1 cm ofapparently healthy tissue beyond the macroscopic margin of thedisease should be removed.4,21

Better disease-free survival is in fact associated with wide clearresection margins and negative nodes; and obtaining wide musclemargins is also considered important because the most commoncause of tumor-related death is lack of disease control at theprimary site. However, 9–25% of patients with negative marginsdevelop recurrence at the primary site.22–24 For T1 and T2 disease,without clinical or radiological evidence of neck metastases,disease control is 80–90% and 60–70%, respectively, while 5-yearsurvival for patients with stage III or IV disease is in the range18–55%.25,26

The surgical approach to tongue cancer is influenced heavily bythe current TNM classification12: cancers are primarily categorizedaccording to the two-dimensional surface diameter approach (T),and the body and base of the tongue are considered distinct loca-tions, notwithstanding the absence of any anatomical differencebetween them. Tumor size and invasion of the extrinsic musclesare the most important variables for the pathologic staging of thedisease, regardless of location. Microscopic disease presence inthe extrinsic muscles is one of the criteria for classifying a tumoras pT4 rather than pT2.12 The limitations of this approach to stag-ing have been extensively discussed in the literature and indicatethat surface diameter on its own has insufficient prognostic util-ity.4,5,25–27 Other prognostic indicators not contemplated by theTNM classification that are important are histopathologic riskscore,24,28 depth of infiltration,29–33 the presence or absence of per-ineural34,35 and lymphovascular36 infiltration (permeation orembolization present in almost 50% of cases5,37,38), tumor-inducedmodification of myofibroblasts37,39,40 and the spread to the areolarfibro-fatty connective tissue that links the tongue (oral and mobile)to the sublingual and submandibular glands, and the cervical lym-phatic chain.4,5,25

Furthermore, while the radiological41 and histological24,37,42

features of disease progression are well recognized, there is a fun-damental lack of understanding of the anatomy of tumor progres-sion. Steinhart and Kleinsasser43 described patterns of initialinfiltration in floor of the mouth and tongue cancer: infiltrationof the sublingual gland, infiltration between the tongue intrinsicmuscles and the genioglossus, and direct infiltration of the intrinsicmusculature. What they did not assess was macroscopic progres-sion along those structures and how these features impact surgicaltherapy.

We developed CTS based on an anatomical study of macro-scopic specimens from tongue cancer patients (Fig. 1). We foundthat in the tongue, and contrary to some reports,4 the tumor be-haved in an identical manner to tumors in musculoskeletal com-partments of the extremities: tumor cell migration occurredlongitudinally from the primary extending along and betweenthe intrinsic and extrinsic muscle fibers, as noted by others5,43

and progression was deflected by the anatomical boundaries ofthe compartment. In agreement with Shaheen,44 we also foundthat tumor progression occurred inferiorly (in depth) as well asposteriorly following the sharp bend from the base of the tongueto the floor of the mouth. In this district the course of nerves andvessels runs parallel to the muscle fibers and these structures arealso facile pathways for tumor progression. Anatomical relation-ships therefore force tumor progression to take place longitudi-nally, de-emphasizing the circumferential approach to removingprimary lesions.

From these considerations it is evident that CTS introduces arationale to what constitutes an adequate margin: completeremoval of the involved muscle(s) affords oncologic radicality asit removes the most likely path of disease spread and potentialsites of residual disease foci.

These ideas are illustrated by our findings of statistically signif-icant improvements of 16.8% in local disease control, 24.4% in loco-regional disease control, and 27.3% in overall survival at 5 yearsfollow-up, compared to our historical series undergoing standardsurgery. We note, however, that CTS patients received significantlymore chemotherapy plus radiotherapy than the standard groupand this may have contributed to our good control rates. It is alsonoteworthy that results for the standard surgery group are consis-tent with those of the literature25,26 and control rate differencesbetween the two groups were maintained throughout the 5 yearsof follow-up (Fig. 2).

To conclude, we propose that the change from circumferentialto longitudinal compartmental resection amounts to a paradigmshift in the surgical approach to locally advanced SCCA of the ton-gue. Our clinical experience also indicates that compartmental sur-gery permits a rational and functionally conservative approach todemolition and we are seeking to demonstrate this in a prospectiveevaluation of quality of life and functional rehabilitation in patientswith locally advanced disease. At this point we make a simpleobservation to indicate how this might be possible: muscles lose

L. Calabrese et al. / Oral Oncology 47 (2011) 174–179 179

function even if only partially cut, so complete removal of extrinsicmuscles from their bony insertions cannot worsen the functionaldefect compared to partial removal.

Conflict of interest statement

None declared.

References

1. Haddadin KJ, Soutar DS, Webster MHC, Robertson AG, Oliver RJ, MacDonald DG.Natural history and patterns of recurrence of tongue tumors. Br J Plast Surg2000;53(4):279–85.

2. Sessions DG, Spector GJ, Lenox J, Haughey B, Chao C, Marks J. Analysis oftreatment results for oral tongue cancer. Laryngoscope 2002;112(4):616–25.

3. Chong V. Oral cavity cancer. Cancer Imaging 2005;5(Spec No A):S49–52.4. Prince S, Bailey BM. Squamous carcinoma of the tongue: review. Br J Oral

Maxillofac Surg 1999;37:164–74.5. Woolgar JA. Histopathological prognosticators in oral and oropharyngeal

squamous cell carcinoma. Oral Oncol 2006;42(3):229–39.6. Enneking WF, Spanier SS, Goodman MA. A system for the surgical staging of

musculoskeletal sarcoma. Clin Orthop 1980;153:106–20.7. Azzarelli A. Surgery in soft tissue sarcomas. Eur J Cancer 1993;29A(4):618–23.8. Enneking WF, Spanier SS, Malawer MM. The effect of the anatomic setting on

the results of surgical procedures for soft parts sarcoma of the thigh. Cancer1981;47(5):1005–22.

9. Karakousis CP, Kontzoglou K, Driscoll DL. Anterior compartment resection ofthe thigh in soft-tissue sarcomas. Eur J Surg Oncol 1998;24(4):308–12.

10. Mendenhall WM, Indelicato DJ, Scarborough MT, et al. The management ofadult soft tissue sarcomas. Am J Clin Oncol 2009;32(4):436–42.

11. Calabrese L, Giugliano G, Bruschini R, Ansarin M, Navach V, Grosso E, et al.Compartmental surgery in tongue tumours: description of a new surgicaltechnique. Acta Otorhinolaryngol Ital 2009;29(5):259–64.

12. AJCC. Cancer staging manual. TNM classification of malignant tumours. 7th ed.New York: Springer; 2010.

13. Preda L, Chiesa F, Calabrese L, Latronico A, Bruschini R, Leon M, et al.Relationship between histologic thickness of tongue carcinoma and thicknessestimated from preoperative. MRI Eur Radiol 2006;16(10):2242–8.

14. Cooper JS, Pajak TF, Forastiere A, et al. Post-operative concurrent radiotherapyand chemotherapy for higher risk squamous cell carcinoma of the head andneck. NEJM 2004;350(19):1937–44.

15. Bernier J, Domenge C, Ozsahin EM, et al. Post operative irradiation with orwithout concomitant chemotherapy for locally advanced head and neck cancer.NEJM 2004;350(19):1945–52.

16. Bernier J, Cooper JS, Pajak TF, van Glabbeke M, Bourhis J, Forastiere A, et al.Defining risk levels in locally advanced head and neck cancers: a comparativeanalysis of concurrent postoperative radiation plus chemotherapy trials of theEORTC (#22931) and RTOG (#9501). Head Neck 2005;27(10):843–50.

17. Marubini E, Valsecchi MG. Analysing survival data from clinical trials andobservational studies. Chichester: John Wiley and Sons; 1995.

18. Kaplan E, Meier P. Non-parametric estimation from incomplete observations. JAm Stat Assoc 1958;53:457–81.

19. Cox DR. Regression models and life-tables (with discussion). J R Stat Soc, Ser B1972;34:187–220.

20. Gray RJ. A class of K-sample tests form comparing the cumulative incidence of acompeting risk. Ann Stat 1988;16:1141–54.

21. Huang TY, Hsu LP, Wen UH, Huang TT, Chou YF, Lee CF, et al. Predictors oflocoregional recurrence in early stage oral cavity cancer with free surgicalmargins. Oral Oncol 2010;46(1):49–55.

22. Cook JA, Jones AS, Phillips DE, Soler Lluch E. Implications of tumour in resectionmargins following surgical treatment of squamous cell carcinoma of the headand neck. Clin Otolaryngol Allied Sci 1993;18(1):37–41.

23. Laramore GE, Scott CB, Schuller DE, Haselow RE, Ervin TJ, Wheeler R, et al. Is asurgical resection leaving positive margins of benefit to the patient with locallyadvanced squamous cell carcinoma of the head and neck: a comparative studyusing the intergroup study 0034 and the Radiation Therapy Oncology Grouphead and neck database. Int J Radiat Oncol Biol Phys 1993;27(5):1011–6.

24. Brandwein-Gensler M, Teixeira MS, Lewins CM, Lee B, Rolnitzky L, Hille JJ, et al.Oral squamous cell carcinoma. Histologic risk assessment, but not marginstatus, is strongly predictive of local disease-free and overall survival. Am J SurgPathol 2005;29(2):167–78.

25. Rogers SN, Brown JS, Woolgar JA, Lowe D, Magennis P, Shaw RJ, et al. Survivalfollowing primary surgery for oral cancer. Oral Oncol 2009;45(1):201–11.

26. Nallet E, Ameline E, Moulonguet L, Barry B, Guédon C, Depondt J. Gehanno P T3and T3 cancer of the oral cavity, surgical treatment with oral tongue resection.Ann Otolaryngol Chir Cervicofac 2001;118(2):74–9.

27. Woolgar JA. T2 carcinoma of the tongue: the histopathologist’s perspective. Br JOral Maxillofac Surg 1999;37(3):187–93.

28. Vered M, Dayan D, Dobriyan A, Yahalom R, Shalmon B, Barshack I, et al. Oraltongue squamous cell carcinoma: recurrent disease is associated withhistopathologic risk score and young age. J Cancer Res Clin Oncol2010;136(7):1039–48.

29. Clark JR, Naranjo N, Franklin JH, de Almeida J, Gullane PJ. Established prognosticvariables in N0 oral carcinoma. Otolaryngol Head Neck Surg2006;135(5):748–53.

30. Fukano H, Matsuura H, Hasegawa Y, Nakamura S. Depth of invasion as apredictive factor for cervical lymph node metastasis in tongue carcinoma. HeadNeck 1997;19(3):205–10.

31. Mohit-Tabatabai MA, Sobel HJ, Rush BF, Mashberg A. Relation of thickness offloor of mouth stage I and II cancers to regional metastasis. Am J Surg1986;152(4):351–3.

32. Shah JP, Lydiatt W. Treatment of the cancer of the head and neck. CA Cancer JClin 1995;45(6):352–68.

33. Spiro RH, Huvos AG, Wong GY, Spiro JD, Gnecco CA, Strong EW. Predictive valueof tumor thickness in squamous carcinoma confined to the tongue and floor ofthe mouth. Am J Surg 1986;152(4):345–50.

34. Fagan JJ, Collins B, Barnes L, D’Amico F, Myers EN, Johnson JT. Perineuralinvasion in squamous cell carcinoma of the head and neck. Arch OtolaryngolHead Neck Surg 1998;124(6):637–40.

35. Soo KC, Carter RL, O’Brien CJ, Bar L, Bliss JM, Shaw HJ. Prognostic implications ofperineural spread in squamous carcinomas of the head and neck. Laryngoscope1986;96(10):1145–8.

36. Jones HB, Sykes A, Bayman N, Sloan P, Swindell R, Patel M, et al. The impact oflymphovascular invasion on survival in oral carcinoma. Oral Oncol2009;45(1):10–5.

37. Vered M, Dobriyan A, Dayan D, Yahalom R, Talmi YP, Bedrin L, et al. Tumor hosthistopathologic variables, stromal myofibroblasts and risk score, aresignificantly associated with recurrent disease in tongue cancer. Cancer Sci2010;101(1):274–80.

38. Woolgar JA. Histological distribution of cervical lymph node metastases fromintraoral/oropharyngeal squamous cell carcinomas. Br J Oral Maxillofac Surg1999;37(3):175–80.

39. De Wever O, Mareel M. Role of tissue stroma in cancer cell invasion. J Pathol2003;200(4):429–47.

40. Lacina L, Dvorankova B, Smetana K Chovanec M, Plzák J, Tachezy R, et al. Markerprofiling of normal keratinocytes identifies the stroma from squamous cellcarcinoma of the oral cavity as a modulatory microenvironment in co-culture.Int J Radiat Biol 2007;83(11–12):837–48.

41. Stambuk HE, Karimi S, Lee N, Patel SG. Oral cavity and oropharynx tumors.Radiol Clin North Am 2007;45(1):1–20.

42. Spiro RH, Guillamondegui OJ, Paulino AF, Huvos AG. Pattern of invasion andmargin assessment in patients with oral tongue cancer. Head Neck1999;21(5):408–13.

43. Steinhart H, Kleinsasser O. Growth and spread of squamous cell carcinoma ofthe floor of the mouth. Eur Arch Otorhinolaryngol 1993;250(6):358–61.

44. Shaheen OH. Carcinoma of the floor of the mouth. J Laryngol Otol1985;99:881–6.