© 2007. Elsevier Masson SAS. Tous droits réservés Gastroenterol Clin Biol 2007;31:286-291

286

Place of multidisciplinary consulting meetings and clinical trials in the management of colorectal cancer

in France in 2000

Anne-Marie BOUVIER (1, 2), Éric BAUVIN (2), Arlette DANZON (2), Pascale GROSCLAUDE (2), Patricia DELAFOSSE (2), Antoine BUÉMI (2), Brigitte TRÉTARRE (2), Nicole RAVERDY (2), Nabil MAAROUF (2),

Michel VELTEN (2), Guy LAUNOY (2), Jean FAIVRE (1, 2)(1) Registre des Cancers Digestifs, Faculté de Médecine, Dijon, INSERM EPI 0106 ; (2) FRANCIM, Réseau français des registres de cancers.

SUMMARY

Aim — The 1998 consensus conference dealing with colon cancer,and the 2003 Cancer Plan underlined the need for multidisciplinarymeetings and for including patients in therapeutic trials. The aim ofthis study, which pooled data from the French Cancer Registriesoperating within the Francim network, was to report on diagnosticand therapeutic practices in the general French population beforeimplementation of the Cancer Plan.

Methods — The study population was composed of 2935 patientswith colorectal cancer diagnosed in 2000 in twelve French adminis-trative districts accounting for 15% of the geographical area ofFrance. Data were collected using a standardized procedure. Threecategories of place of diagnosis were defined: public university hos-pitals, public non-university hospitals, and private clinics.

Results — Overall, multidisciplinary meeting was conducted for32.2% of patients with colorectal cancer. This proportion variedfrom 6.4% to 76.9%, depending on the geographical area(P < 0.001). The place of diagnosis affected this practice: 52% inpublic university hospitals, 31% in public non-university hospitalsand 29% in private clinics (P < 0.001). In multivariate analysis, age(OR> 75 years: 0.71, P < 0.001), site (ORrectum: 1.80, P < 0.001)and health care facilities (ORpublic non-university vs public university: 0.36,P < 0.001, ORprivate vs public university: 0.40, P < 0.001) affected theuse of multidisciplinary meeting.

Overall, 4.3% of patients were included in a therapeutic trial.This concerned 6.2% of patients aged under 75 and 1.0% of thoseaged over 75 (P < 0.001). The proportion of inclusions, taking intoaccount the trials proposed in 2000 and 2001, varied from 0.7% to16.4% according to geographical area (P < 0.001). This proportionwas 10.3% if there had been multidisciplinary meeting and 5.1% ifnot (P < 0.001). Neither cancer site, gender, nor healthcare facilityresponsible for diagnosis influenced trial inclusion.

Conclusion — This population-based study underlines geographicalvariations in the management of colorectal cancer in France.In 2000, multidisciplinary meeting was conducted for an insufficientproportion of patients, and an insufficient number of patients wereincluded in therapeutic trials. Repeating the same survey in 2005will provide information on the effects of the Cancer Plan and thediffusion of these recommendations.

RÉSUMÉ

Anne-Marie BOUVIER, Éric BAUVIN, Arlette DANZON, Pascale GROSCLAUDE, Patricia DELAFOSSE, Antoine BUÉMI, Brigitte TRÉTARRE, Nicole RAVERDY, Nabil MAAROUF, Michel VELTEN, Guy LAUNOY, Jean FAIVRE

But — La Conférence de Consensus consacrée au côlon en 1998puis le Plan Cancer en 2003 ont souligné la nécessité de mettreen place des Réunions de Concertation Pluridisciplinaire (RCP) etd’inclure les malades dans les essais thérapeutiques. L’objectif de cetravail était, par la mise en commun des données des registresdes cancers français, au sein du réseau FRANCIM, de faire un étatdes lieux de la situation dans la population française avant la miseen œuvre du plan cancer.

Méthode — L’étude a porté sur 2 935 cas de cancers colorectauxdiagnostiqués en 2000 dans 12 départements français couvrant15 % du territoire national. Les données ont été recueillies selon uneprocédure standardisée. Le lieu de diagnostic était réparti en3 classes : CHU-CLCC, CH et secteur privé.

Résultats — Globalement, 32,2 % des cas de cancers colorectauxont été présentés en RCP. Ces proportions variaient selon le dépar-tement, entre 6,4 % et 76,9 % (P < 0,001). Le lieu de diagnosticinfluençait cette pratique : 52 % dans les structures publiques uni-versitaires, 31 % dans les structures publiques non universitaires et29 % dans le secteur privé (P < 0.001). En analyse multivariée,l’âge (OR> 75ans :0,71, P < 0,001), la localisation (ORrectum : 1,83,P < 0,001) et la filière (ORpublic non universitaire vs public universitaire :0,36, P < 0.001, ORpivé vs public universitaire : 0,40, P < 0,001) influen-çaient la discussion des cas en RCP.

Globalement, 4,3 % des malades ont été inclus dans un essai théra-peutique. Cette pratique concernait 6,2 % des cas âgés de moins de75 ans et 1,0 % des plus de 75 ans (P < 0,001). La proportiond’inclusion en tenant compte des essais proposés en 2000 et 2001variait de 0,7 % à 16,4 % selon le département (P < 0,001).Elle était de 10,3 % pour les cas présentés en RCP et de 5,1 % pourles non présentés (P < 0,001). Ni la localisation, ni le sexe ni lafilière de diagnostique n’influençaient l’inclusion dans un essai.En analyse multivariée, le stade et le passage en RCP étaient signifi-cativement associés à l’inclusion dans un essai thérapeutique.

Conclusion — Ce travail, qui a l’intérêt d’avoir été réalisé sur unebase de population, met en évidence des variations géographiquesdans la prise en charge du cancer colorectal en France. La propor-tion de dossiers de malades présentés en RCP ou inclus dans unessai thérapeutique était en 2000 encore insuffisante. La répétitiond’une étude comparable en 2005 permettra de connaître l’influencedu plan cancer dans la diffusion de ces recommandations.

Reprints: A.-M. BOUVIER, Registre des Cancers Digestifs, Faculté de Médecine, BP 87900, 21079 Dijon Cedex.E-mail : anne-marie.bouvier@u-bourgogne.fr

Place des Réunions de Concertation Pluridisciplinaire et des essais thérapeutiques dans la prise en charge du cancer colorectal en France en 2000

(Gastroenterol Clin Biol 2007;31:286-291)

Multidisciplinary meeting, therapeutic trials and colorectal cancer

287

Introduction

Cancer is a major public health issue in France, with specialconcern about colorectal cancer because of its frequency andseverity. The number of cases diagnosed annually increased by50% between 1978 and 2000, a consequence both of popula-tion aging and increased disease prevalence [1]. It has been esti-mated that in 2000, a new diagnosis of colorectal cancer wasestablished in 36 000 people living in France. At diagnosis, theirchances of survival at five years was about 55%, a rate whichhas slightly improved over the last decade [2]. Dedicated surveysare conducted regularly by the Francim network pooling datafrom French Cancer Registries. Their data can be used to obtaina population-based analysis of clinical practices in France.

Analysis of available epidemiological data contributed to theelaboration of public health policies which led to a nationwide“Cancer Plan” implemented progressively from 2003 through2007. A circular issued in February 2005 by the French Ministryof Health described the regional organization of the plan andspecific measures to be undertaken. One of the objectives wasto improve coordination of healthcare delivery: according tomeasure n°31, multidisciplinary meeting meetings (Réunions deConcertation Pluridisciplinaire, RCP) should be held for 100% ofpatients with a new diagnosis of cancer. A second objective(measure n°67) was to include new patients in clinical researchtrials with a goal of 10% included in regional cancer centers and5% in other healthcare facilities. Assessment of the usefulnessand efficacy of these measures requires a precise evaluation ofclinical practices. The cancer registries available in France pro-vide an ideal tool for determining whether these new strategieshave become routine practices in the general population, avoi-ding the problem of recruitment bias inherent in reports fromspecialized centers.

The purpose of this study was to report on the situation in theFrench population in 2000, before implementation of the CancerPlan. The goal was to determine the proportion and characteris-tics of patients with colorectal cancer whose cases was discussedin a RCP and the proportion of these patients included in rando-mized therapeutic trials.

Patients and Methods

Study populationThis study concerned all new cases of colorectal cancer (codes C18,

C19, C20 of the IDC-10th edition [3]) diagnosed in 2000 in twelveadministrative districts of France covered by cancer registries. The admi-nistrative districts were Bas-Rhin, Calvados, Côte d’Or, Doubs, Haut-Rhin, Hérault, Isère, Loire Atlantique, Manche, Saône-et-Loire, Somme,and Tarn. The present survey was conducted within the framework of theFrancim network of cancer registries which cover 15% of the Frenchpopulation (8,737,410 inhabitants according to the 1999) census.

Data collectedIn addition to data collected regularly by the registries, this survey

also included specific information items collected from specialists (gas-troenterologists, surgeons, cancer specialists) concerning the circumstan-ces of diagnosis, the type of management, and treatment delivered.Patients were divided into two age groups, < 75 years and ≥ 75 years.The stage at diagnosis was defined using all available data: pathologyreports, operative findings, disease extension. Six stages were definedusing the TNM classification [4]: cancer limited to the wall (stage I,19.1%), cancer invading the serosa (stage II, 28.6%), resected cancerwith lymph node metastasis (stage III, 25.3%), cancer with visceralmetastasis (stage IV, 22.2%). Non-resectable cancers without patent vis-ceral metastasis were classified as “advanced stage” (3.0%). There werealso 1.8% of resected cancers of unknown stage. There were threecategories for place of diagnosis: public university hospitals (including

university hospitals and regional cancer centers), public non-universityhospitals (public general hospitals, mutual hospitals and not-for-profitprivate facilities participating in the public hospital service), and privateclinic. Available medical records were used to determine whether or nota RCP had been held for the patient and whether or not the patient wasincluded in a randomized therapeutic trial (drugs, radiotherapy).

Statistical analysisUnivariate analysis was conducted with the chi-square test for dis-

crete variables. For multivariate analysis, a logistic regression model wasretained to determine significant variables independently linked to thepresence or not of a RCP and inclusion or not in a clinical trial. Resultsare expressed as odds ratio with 95% confidence interval. Models wereadjusted for age, gender, and district of residence.

Results

In all, 2935 cases of cancer were identified in 1659 men(56.5%) and 1276 women (43.5%). These patients had cancer ofthe colon (including the rectosigmoid junction) (N = 2187,74.5%) and cancer of the rectal ampulla (N = 748, 25.5%).There were 1137 patients (38.7%) aged 75 years or older.

Multidisciplinary consultation meetings (RCP)Whether or not a RCP had been held for the patient was

known for 93.4% of cases and whether the patient had beenincluded in a clinical trial was known for 99.1%. Globally,32.2% of the files of patients with colorectal cancer mentioned aRCP. This proportion varied from 6.4% to 76.9% (P < 0.001)(figure 1) depending on the administrative district. The characte-ristic features of patients whose case was discussed in a RCP arepresented in table I. Univariate analysis revealed that male gen-der, rectal cancer, and age less than 75 years were more fre-quent than female gender, colorectal cancer and age 75 yearsor over respectively. A RCP was held more often (37.1%) forpatients diagnosed with advanced stage disease (stage III, IV)than those with early-stage disease (27.5%) (P < 0.001). In all, aRCP was held for 51.7% of patients whose diagnosis was esta-blished in a public university hospitals versus 31.6% in publicnon-university hospitals and 28.9% in private clinics (P < 0.001).

After adjustment for gender and district of residence, age,stage at diagnosis and cancer localization were independently

district

8060

40%20

0

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

average of the 12 districts 95% Cl of mean

Fig. 1 – Proportion of medical files discussed in a “multidisciplinarymeetings” (RCP Réunion de Concertation Pluridisciplinaire)according to patient’s area of residence.Proportion de dossiers présentés en Réunions de ConcertationPluridisciplinaire selon le département de résidence des patients.

A.-M. Bouvier et al.

288

and significantly associated with a RCP (table II). The place ofdiagnosis had a significant effect on multidisciplinary meeting:compared with public university hospitals (OR = 1), a RCP was

three times less likely in other facilities (ORpublic non-universitary vspublic university = 0.36 [0.26-0.50], ORprivate vs public universitary= 0.40 [0.31-0.54]).

Table I. – Characteristic features of patients whose medical file was discussed during a “multidisciplinarymeeting” (RCP, Réunion de Concertation Pluridisciplinaire).Caractéristiques des patients dont les dossiers ont été présentés en Réunion de concertationpluridisciplinaire (RCP).

N Percent P(a)

Gender Male 1552 34.90%

Female 1189 28.70% 0.001

Age < 75 years 1683 37.50%

≥ 75 years 1058 23.80% < 0.001

Localization Colon 2034 29.90%

Rectum 707 38.90% < 0.001

Stage(b) TNM I 520 24.00%

TNM II 762 29.20%

TNM III 700 36.40%

TNM IV 622 40.20%

Advanced(c) 84 20.20% 0.001

Place of diagnosis(d) Public university hospital 309 51.70%

Public non-university hospital

720 31.60%

Private clinic 1687 28.80% 0.001

(a) Chi-square test; (b) stage unknown for 53 patients who are not presented; (c) non-resectable cancer without patentvisceral metastasis; (d) place of diagnosis unknown for 25 patients.

Table II. – Factors associated with presentation of medical files at a “multidisciplinary meeting” (RCP,Réunion de Concertation Pluridisciplinaire).Facteurs associés à la présentation des dossiers en Réunion de concertation pluridisciplinaire(RCP).

OR [ ] 95% P

< 75 years 1

75 years 0.71 [0.64-0.79] < 0.001

Public university hospital 1

Public non-university hospital 0.36 [0.26-0.50] < 0.001

Private 0.40 [0.31-0.54] < 0.001

Colon 1

Rectum 1.83 [1.48-2.25] < 0.001

TNM I 1

TNM II 1.43 [1.07-1.91] 0.007

TNM III 1.99 [1.49-2.65] < 0.001

TNM IV 2.29 [1.71-3.08] < 0.001

Advanced(a) 0.57 [0.43-1.59] 0.194

Adjusted for geographic area and gender.(a) Those who had no evidence of metastatic disease and did not undergo resection (unknown stages excluded).

Multidisciplinary meeting, therapeutic trials and colorectal cancer

289

Clinical trialsThe proportion of patients included in a therapeutic trial was

4.3%. This proportion was 6.1% for patients aged less than75 years and 1.0% for their older counterparts (P < 0.001).The proportion varied from 0.7 to 9.3% among the geographicdistricts (P < 0.001) (figure 2). During 2000 and 2001, openrandomized protocols were available for patients with stage II,III, and IV or non-resectable cancer and for patients aged lessthan 75 years. Considering only patients who met these criteria(and thus were eligible for trial inclusion) the overall proportionof patients effectively included was 7.3%, ranging from 0.7 to16.4% depending on the geographic district (P < 0.001). Investi-gators had requested enrolment of eleven patients (1.2%) agedover 75 years in protocols with an inclusion criterion of age lessthan 75 years (table III). Univariate analysis retained presence ofa RCP and diagnostic stage as factors significantly associatedwith inclusion in a trial while the place of diagnosis or the locali-zation of the cancer did not affect inclusion in a therapeutic trial.The proportion of patients included in clinical trials was 9.5%,8.1% and 6.6% in public university hospitals, public non-university hospitals, and private clinics (P = 0.589).

After adjustment for district of residence, multivariate analy-sis demonstrated a significant link between presence of a RCPand inclusion in a trial: ORRCP+ vs RCP- = 1.99 [1.23-3.21](table IV). Stage at diagnosis was also an independent factorsignificantly associated with inclusion in a therapeutic trial,patients with stage III and IV cancer being included more fre-quently than the others.

Discussion

In order to better coordinate care for cancer patients andimprove the quality of care, the consensus conference devoted tocancer of the colon held in 1998 [5] and the French Cancer Plan(2003-2007) emphasized the need for multidisciplinary mee-tings (Réunions de Concertation Pluridisciplinaire, RCP) and theimportance of including patients in clinical trials. Cancer regis-tries operating within the framework of the Francim network con-duct specific population-based surveys to obtain precise data on

medical practices in the general population in France, thus avoi-ding the recruitment bias of information issuing from specializedcenters. Pooling data from the French Cancer Registries provideda report of routine practices in the general population in France in2000 which can be considered as the “baseline” situation beforethe implementation of the Cancer Plan for colorectal cancer.

Globally, a RCP was held for about one-third of patients withcolorectal cancer diagnosed in 2000. The present surveydemonstrates that practices varied greatly: the proportion ofpatients with a RCP ranged from 1 to 10-fold depending on thedistrict of residence. Age was a significant factor associated withmultidisciplinary meeting. Earlier studies have emphasized thattreatment delivered to older patients complies less well with con-sensus conference guidelines than that delivered to youngerpatients [6-9]. Holding a RCP for a greater proportion of oldersubjects might have an impact on such practices. The criteriaconcerning operability, respectability, and management optionsfor rectal cancer are technically more complex than for coloniccancer. This might explain why the proportion of RCP was loweramong patients with colonic cancer than those with rectal cancer.The stage at diagnosis also affected the proportion of patientswith a RCP, the proportion being lower for stage I disease. This isundoubtedly related to the fact that there is a simple recommen-ded therapeutic protocol for stage I cancer which does notrequire specific complementary treatment. The proportion ofpatients discussed in a RCP varied considerably depending onthe place of diagnosis. In 2000, multidisciplinary care wasundoubtedly already a well-established practice in most univer-sity centers, particularly in cancer centers. In other facilities, themultidisciplinary approach may have become a routine practicelater. This point emphasizes the measures announced bythe Cancer Plan: with the institution for each individual patient ofa personalized plan for coordinated care. Repeating such cross-sectional surveys will be helpful in better understanding howthese measures are being implemented in the different categoriesof healthcare facilities. Taking into consideration the organizatio-nal implications of RCPs, where a quorum of at least three spe-cialists, including a cancer specialist, is required to reach aconsensus, it would not be reasonable to expect to reach the goalof the Cancer Plan (a RCP for 100% of patients) in the very nearfuture. For this reason, the ministerial circular of February 2005(N°DHOS/SDO/2005/101) devoted to the organization ofcancer care introduced the notion of computer-assisted patientfile management, with or without a RCP, for patients with a clini-cal presentation corresponding to a standardized treatment pro-tocol. The development of videoconferences might also helpimprove practices.

It is well known that as many patients as possible should beincluded in therapeutic trials. The consensus conference held in1998 on colonic cancer established specific guidelines [5]. In2000, 7.3% of patients who were potentially eligible for inclu-sion in ongoing trials were actually enrolled. This proportion isnot too far from the goal of 10% set by the 2003 Cancer Plan.Considering the figures by category of place of diagnosis, thesituation is even more favorable since nearly 7% of patients dia-gnosed in public non-university clinics and private clinics wereincluded in trials on colorectal cancer (goal 5%) and more than9% of those diagnosed in university clinics (goal 10%). It wasnevertheless noteworthy that by administrative district, the pro-portions ranged from 1% to 13%. This geographical variabilityemphasizes the importance of measures announced by theCancer Plan nationwide since the rate of inclusion in trials is verylow in certain areas. The development of mobile teams with clini-cal research assistants should be helpful in diffusing the informa-tion and should contribute to reduced geographic variation.

There were a small number of persons aged over 75 yearswho were included in therapeutic trials at the request of theinvestigators despite the fact that, theoretically, in 2000 and

3020

10

%

district

0

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

average of the 12 districts 95% Cl of mean

Fig. 2 – Proportion of patients included in a therapeutic labelled trialaccording to their area of residence (among potentially eligiblesubjects in 2000 or 2001. i.e. excluding subjects aged 75 years ormore and TNM stage I).Proportion de patients inclus dans un essai thérapeutique labelliséselon le département de résidence (parmi les sujets potentiellementéligibles en 2000 ou 2001 c’est-à-dire en excluant les sujets âgés deplus de 75 ans et les stade TNM I).

A.-M. Bouvier et al.

290

2001 there were no trials available for this category of patients.This situation was related to the undertreatment of older subjects.It has been known, for example, since 1990 that chemotherapyis beneficial for patients with stage III colonic cancer and shouldbe delivered whenever possible. Despite this, in the generalpopulation, 8.7% of patients aged 80-84 years and 4.8% aged85 or older were given chemotherapy for this indication in 1997[10]. This observation, together with longer life expectancy andbroadening therapeutic options in oncogeriatrics, has led to twotrials specifically devoted to subjects aged over 75 years withcancer of the rectum. These trials were initiated in 2003 and2004 by the FFCD (Fédération Francophone de CancérologieDigestive) [11]. This illustrates that presentation of a patient’s fileto a RCP favors inclusion in therapeutic trials. Even if the patientsselected for discussion at a RCP would be potentially more eligi-ble for inclusion than others, it would be reasonable to assumethat the RCP would have a favorable effect on the number ofpatients included in trials. The stage at diagnosis also affectedpatient inclusion, both at univariate analysis and after adjust-ment for possible confounding factors.

The health care setting at the time of diagnosis had no signi-ficant effect on the proportion of patients included in therapeutictrials: 9.5%, 8.1% and 6.6% for public university hospitals,public non-university hospitals, and private clinics, respectively.For FFCD trials, 37% of patients were included by university hos-pitals, 32% by general hospitals, 20% by cancer centers, andonly 11% by private clinics [12]. Knowing that colorectal canceris more often treated in private clinics, large-scale clinical trialsdevoted to adjuvant treatment of colorectal cancer cannot beconducted without the participation of physicians practicing inprivate clinics. This implies allocation of resources for clinicalresearch (ARC, TEC) in oncology in each region.

The data collected in this survey reflect routine practices con-cerning the use of multidisciplinary meeting meetings and inclu-sion in therapeutic trials for colorectal cancer patients before theannouncement of the Cancer Plan. A survey of diagnostic andtherapeutic practices for colorectal cancer in 2005 is being con-ducted within the Francim network. The results of this survey will

Table III. – Characteristics of patients included in therapeutic trials among those potentially eligible in2000 or 2001; i.e. excluding patients aged 75 years or over and TNM stage I(a).Caractéristiques des cas inclus dans les essais thérapeutiques parmi les sujets potentiellementéligibles en 2000 ou 2001 c’est-à-dire en excluant les sujets âgés de plus de 75 ans et les stadesTNM I(a).

Percent P(b)

Gender Male 7.3%

Female 7.3% 0.948

Age < 75 years 7.3%

Site Colon 6.8%

Rectum 8.8% 0.201

RCP Yes 10.3%

No 5.5% 0.001

Stage TNM II 3.5%

TNM III 10.6%

TNM IV 8.4%

Advanced(c) 4.8% 0.001

Place of diagnosis Public university hospital 9.5%

Public non-university hospital

8.1%

Private clinic 6.6% 0.359

(a) 1.2% of subjects aged 75 or over were included in trials at the request of investigators;(b) chi-square test; (c) thosewho had no evidence of metastatic disease and did not undergo resection (unknown stages excluded).

Table IV. – Factors associated with the inclusion in therapeutic trials.Logistic regression model among potentially eligible patientsin 2000 or 2001; i.e. excluding patients aged 75 years or moreand TNM stage I.Facteurs associés à l’inclusion dans un essai thérapeutique.Régression logistique réalisée parmi les sujets potentiellementéligibles en 2000 ou 2001 c’est-à-dire en excluant les sujetsâgés de plus de 75 ans et les stade TNM I.

OR [ ] 95% P

RCP - 1

RCP + 1.99 [1.23 - 3.21] 0.005

TNM II 1

TNM III 2.95 [1.62 - 5.37] < 0.001

TNM IV 2.38 [1.28 - 4.41] 0.006

Advanced(a) 1.17 [0.18 - 12.01] 0.706

Adjusted for geographic area.(a) Those who had no evidence of metastatic disease and did not undergo resection(unknown stages excluded)RCP (Réunion de Concertation Pluridisciplinaire) multidisciplinary meeting.

Multidisciplinary meeting, therapeutic trials and colorectal cancer

291

enable a mid-term assessment of the diffusion of the measuresannounced by the Cancer Plan.

REFERENCES

1. Remontet L, Esteve J, Bouvier AM, Grosclaude P, Launoy G,Menegoz F, et al. Cancer incidence and mortality in France over theperiod 1978-2000. Rev Epidemiol Sante Publique 2003;51:3-30.

2. Mitry E, Bouvier AM, Esteve J, Faivre J. Benefit of operativemortality reduction on colorectal cancer survival. Br J Surg 2002;89:1557-62.

3. ICD-10 (ed). International classification of disease for oncology.10th revision, ed 2. Geneva: World Health Organization, 1990.

4. Sobin L, Wittekind C. TNM Classification of Malignant TumorsUnion Internationale Contre le Cancer and the American JointCommittee on Cancer, 5th ed. New York: John Wiley and Sons Inc,1997.

5. Conférence de Consensus. Prévention, dépistage et prise en chargedes cancers du côlon. Gastroenterol Clin Biol 1998;22:S275-88.

6. Surgery for colorectal cancer in elderly patients: a systematic review.Colorectal Cancer Collaborative Group. Lancet 2000;356:968-74.

7. de Rijke JM, Schouten LJ, Schouten HC, Jager JJ, Koppejan AG,van den Brandt PA. Age-specific differences in the diagnostics andtreatment of cancer patients aged 50 years and older in the province ofLimburg, The Netherlands. Ann Oncol 1996;7:677-85.

8. Faivre-Finn C, Bouvier AM, Mitry E, Rassiat E, Clinard F, Faivre J.Chemotherapy for colon cancer in a well-defined French population: isit under- or over-prescribed? Aliment Pharmacol Ther 2002;16:353-9.

9. Sargent DJ, Goldberg RM, Jacobson SD, Macdonald JS, Labianca R,Haller DG, et al. A pooled analysis of adjuvant chemotherapy forresected colon cancer in elderly patients. N Engl J Med 2001;345:1091-7.

10. Bouvier AM, Launoy G, Lepage C, Faivre J. Trends in the manage-ment and survival of digestive tract cancers among patients aged over80 years. Aliment Pharmacol Ther 2005;22:233-41.

11. Fédération Francophone de Cancérologie Digestive, http://www.ffcd.fr/.

12. Fournet J, Dhumeaux D. Le livre blanc de l’hépato-gastroentérologie.Paris : Masson, 2001.