Embed Size (px)
Citation preview
ÉVOLUTION DE LA PRISE EN CHARGE PALLIATIVE DU CANCER DU PANCREAS
AVANCE EN RELATION AVEC L’ARRIVEE DE NOUVEAUX MEDICAMENTS
IMPLICATIONS SUR LA SURVIE ET
CONSIDERATIONS MEDICO -ECONOMIQUES
Thèse
présentée à la Faculté de Médecine de l’Université de Genève
pour obtenir le grade de Docteur en médecine par
Sébastien MAZZURI de
Lausanne (VD)
Thèse no 10452
Genève
2005
UNIVERSITÉ DE GENÈVE FACULTÉ DE MÉDECINE
Section de médecine Clinique
Département de Chirurgie
Oncochirurgie
Thèse préparée sous la direction du Dr. Arnaud D. ROTH , Privat-Docent et du Professeur Philippe MOREL
i
Le présent travail a fait l’objet d’une publication :
Mazzuri, S., Roth, A. (2005). Modern systemic therapy of advanced pancreatic cancer: impact
on overall survival and medico economical aspects. Rev Med Suisse 1(24): 1616-20. French.
ii
iii
SYNOPSIS
Ce travail présente les résultats médico-économiques d’une étude rétrospective conduite sur la
cohorte des cas d’adénocarcinomes pancréatiques avancés suivis par la clinique d’Oncochirurgie
des Hôpitaux Universitaires de Genève ces dix dernières années. Le protocole comparait les
périodes 1994-98 et 1999-2003 (distinguées sur la base de la disponibilité de molécules modernes
telles que la gemcitabine).
Survie : les patients de la seconde période ont survécu significativement plus longtemps que
ceux de la période précédente (médiane de 8.7 versus 7.1 mois, probabilité de survie à 1 an de
40% versus 14% ; log-rank p=0.012), sans augmentation significative des épisodes
d’hospitalisation ;
Coûts : ce bénéfice de survie s’est accompagné d’une augmentation significative des dépenses
ambulatoires (médiane de 8'537 versus 5'005 CHF ; p=0.006).
Ces résultats sont à interpréter avec réserve au vu des limitations méthodologiques de l’étude.
Ils sont cependant corroborés dans la littérature actuelle par les données récentes de plusieurs
essais cliniques randomisés.
iv
REMERCIEMENTS
Un très grand merci aux personnes suivantes, qui m’ont toutes été d’une aide inestimable dans
l’élaboration de ce travail et m’ont accordé beaucoup de leur temps malgré un emploi du temps
chargé :
- Merci en particulier à Arnaud Roth, de m’avoir proposé un sujet de thèse aussi intéres-
sant et comportant d’aussi multiples facettes, qui m’ont permis d’appréhender des
aspects pratiques et notamment statistiques à peine effleurés dans mes études. Merci
encore davantage pour sa sympathie, sa bonne humeur, son accessibilité et sa disponi-
bilité durant toute l’élaboration de ce travail ;
- Merci également à Nadine K’bourch pour son accueil, son dévouement et son aide
dans les méandres organisationnels du service d’Oncochirurgie ;
- Merci à toute l’équipe d’Oncochirurgie des H.U.G. pour leur accueil, leur sympathie,
leur cohésion et l’espoir qu’ils savent insuffler chaque jour à leurs patients par leur sou-
tien et leur attachement ;
- Merci à Bernadette Mermillod, de m’avoir permis d’éviter certains écueils statistiques
et d’aboutir à une présentation plus claire et solide de mes données ;
- Merci à Delphine Jaquier et Patrice Maricot pour leur maîtrise du système de
comptabilité analytique des H.U.G., rendant possible l’analyse médico-économique qui
donne une dimension plus originale car moins strictement médicale à ce travail ;
- Merci à Mariette Lapalud pour avoir su trouver et me faire parvenir aussi rapidement
les références nécessaires à la documentation de ce travail ;
- Merci à Isabelle Neyroud-Caspar de m’avoir fourni les informations épidémiologi-
ques relatives au Registre Genevois des Tumeurs ;
Ce travail, qui compte beaucoup pour moi, n’aurait sans doute pas pu aboutir sans votre aide.
Je vous souhaite à toutes et tous un parcours des plus enrichissant dans l’avenir !
Sébastien
- 1 -
TABLE DES MATIERES
Table des matières __________________________________________________________ - 1 -
Résumé ___________________________________________________________________ - 3 -
Méthodologie ____________________________________________________________________- 5 -
Résultats ________________________________________________________________________- 7 -
Discussion_______________________________________________________________________- 9 -
Plan du travail __________________________________________________________________- 10 -
Introduction to pancreatic cancer _____________________________________________ - 11 -
Epidemiological considerations ____________________________________________________- 11 -
Pathology ______________________________________________________________________- 12 -
Development and genetics of pancreatic cancer_______________________________________- 13 -
Diagnosis and Staging____________________________________________________________- 14 -
Treatment overview and Survival __________________________________________________- 16 - Resectable pancreatic cancer _____________________________________________________________ - 16 - Locally advanced and metastatic pancreatic cancer ___________________________________________ - 18 -
Chemotherapy in advanced pancreatic carcinoma: literature review _________________ - 20 -
Methodological considerations_____________________________________________________- 20 - Performance status_____________________________________________________________________ - 20 - Evaluation of response in pancreatic cancer _________________________________________________ - 20 -
5-fluorouracil ___________________________________________________________________- 22 -
Other single-agent modalities [55]__________________________________________________- 23 -
Gemcitabine____________________________________________________________________- 23 - Initial phase I-II single-agent studies_______________________________________________________ - 23 - Phase III single-agent studies ____________________________________________________________ - 24 - Association regimens___________________________________________________________________ - 24 -
• Gemcitabine and 5-FU ___________________________________________________________ - 24 - • Gemcitabine and cisplatin ________________________________________________________ - 24 - • Gemcitabine and oxaliplatin (GemOx regimen) _______________________________________ - 25 - • Gemcitabine and irinotecan (IrinoGem regimen)_______________________________________ - 25 - • Gemcitabine and docetaxel________________________________________________________ - 25 - • Gemcitabine and epirubicin _______________________________________________________ - 25 -
Gemcitabine in radio-chemotherapeutic modalities ___________________________________________ - 25 -
Summary ______________________________________________________________________- 26 -
Retrospective study on a Geneva cohort ________________________________________ - 29 -
Introduction____________________________________________________________________- 29 -
Methodology ___________________________________________________________________- 30 - Patients______________________________________________________________________________ - 30 - Data mining __________________________________________________________________________ - 30 - Data analysis _________________________________________________________________________ - 32 -
• Cost analysis system in Geneva Cantonal Hospital _____________________________________ - 34 -
- 2 -
• Conversion into standard units _____________________________________________________ - 35 - • Extra-hospital costs______________________________________________________________ - 36 -
Results ________________________________________________________________________- 37 - Epidemiological data ___________________________________________________________________ - 37 - Patients and Groups ____________________________________________________________________ - 38 - Treatment____________________________________________________________________________ - 45 - Survival analyses ______________________________________________________________________ - 47 - Hospitalization days____________________________________________________________________ - 49 - Economic analyses_____________________________________________________________________ - 53 -
Discussion______________________________________________________________________- 58 - Methodology _________________________________________________________________________ - 58 - Clinical Results _______________________________________________________________________ - 60 - Medico-economic Results _______________________________________________________________ - 63 -
Conclusion _______________________________________________________________ - 65 -
References________________________________________________________________ - 67 -
- 3 -
RESUME
L’adénocarcinome pancréatique représente environ 90% des cas d’atteinte maligne du pan-
créas. Avec une incidence moyenne de 11.7 cas pour 100'000 de 1995 à 1998 à Genève, il repré-
sente environ 2 à 3% des diagnostics de néoplasies. Tous stades confondus, les taux de survie à 1
an et à 5 ans sont respectivement de 20% et de moins de 5%, tandis que la survie médiane après
résection n’est que de 13 à 20 mois. Pire encore, la survie médiane des cas métastatiques ainsi
que des cas loco régionalement avancés au diagnostic ne dépasse pas 3 à 6 mois et 6 à 10 mois,
respectivement. Ce pronostic catastrophique explique que la mortalité annuelle du carcinome du
pancréas se rapproche fortement de son incidence, puisque moins de 20% des patients se présen-
tent avec une maladie macroscopiquement limitée à l’organe lors du diagnostic initial : approxi-
mativement 40% souffrent d’une maladie loco régionalement dépassée, 40% présentent des
métastases viscérales –principalement hépatiques–, et 35% des implants tumoraux péritonéaux
(certains patients appartenant à deux, voire à ces trois catégories mutuellement non exclusives).
D’un point de vue thérapeutique, le seul traitement curatif repose sur une pancréatectomie, la
plus souvent partielle, qui ne peut être raisonnablement proposée qu’aux patients présentant une
maladie macroscopiquement limitée (stades I et II). Une radiothérapie adjuvante est parfois
offerte à distance de la chirurgie dans l’espoir d’un meilleur contrôle locorégional. Malheureuse-
ment, et en raison d’une symptomatologie frustre et souvent tardive, la majorité des patients sont
diagnostiqués à des stades avancés. Les possibilités thérapeutiques se limitent alors à des inter-
ventions palliatives de dérivation biliaire et/ou digestive si nécessaire, et à des traitements de
chimiothérapie/radiothérapie. Concernant la chimiothérapie, le 5-fluorouracile était considéré
comme le traitement de choix jusqu’au milieu des années 90, même si quelques études randomi-
sées contrôlées n’avaient pu mettre en évidence de bénéfice significatif par rapport à un traite-
ment de soutien seul en termes de survie.
L’introduction de la gemcitabine dans cette indication, en Suisse au deuxième semestre de
1998, puis de l’oxaliplatine, apporta un second souffle à la prise en charge palliative de ces
patients, comme peut en témoigner l’augmentation exponentielle des études cliniques de phase I
et II ces dernières années, cherchant à définir les régimes de drogues modernes les plus efficaces
et surtout les moins toxiques. D’une manière générale, ces innovations thérapeutiques ne se sont
pas traduites de manière fulgurante en termes d’indicateurs classiques que sont la survie médiane
et la survie à 1 an, n’améliorant la situation que de manière marginale quoique cependant statisti-
- 4 -
quement significative. En réalité, la véritable amélioration a plutôt concerné la symptomatologie
de ces patients et l’impact de la maladie sur leur qualité de vie en termes d’index de performance
et de soulagement des douleurs. En effet, les essais randomisés contrôlés comparant des
protocoles modernes à base de gemcitabine au précédent traitement de référence à base de 5-fluo-
rouracile ont mis en évidence la supériorité significative des premiers sur le second en termes de
bénéfice clinique (Clinical Benefit Response) sans augmentation significative des effets secondai-
res indésirables sévères. Ainsi, l’attitude de la majorité des oncologues a passablement évolué
depuis l’introduction de la gemcitabine, passant d’une tendance à l’abstention thérapeutique ou à
la prescription peu convaincue de 5-fluorouracile, à la prescription de molécules modernes plus
fréquente car reconnue comme efficace en termes de bénéfice clinique. D’un point de vue plus
classique et quantitatif, et selon les données de la littérature actuelle que nous résumons en pre-
mière partie de notre travail, les patients atteints d’un carcinome du pancréas dépassé pouvaient
espérer une survie médiane d’environ 5 à 7 mois sous traitement de 5-fluorouracile seul. Cette
espérance est actuellement d’environ 7 à 10 mois sous traitement de gemcitabine. Concernant les
associations de gemcitabine à d’autres agents chimiothérapeutiques classiques tels que le cispla-
tine ou l’oxaliplatine, les espoirs fondés sur les résultats de nombreuses études de phase II ont été
très récemment mis à mal par la présentation au congrès de l’ASCO 2004 des conclusions déce-
vantes de plusieurs essais cliniques randomisés de phase III, ne parvenant pas à démontrer de
supériorité significative des régimes d’association sur les protocoles à base de gemcitabine seule
en termes de survie. Les résultats définitifs de deux études de phase III comparant une association
de gemcitabine et de Xeloda (formulation orale du 5-fluorouracile se rapprochant d’une infusion
continue) à la gemcitabine seule devraient être disponibles dans un avenir proche (protocoles
SAKK et MRC).
Dans ce contexte, il nous a paru intéressant de mener une étude rétrospective1 sur le sujet en
analysant l’ensemble des cas de carcinomes du pancréas ayant été suivis par le service
d’oncochirurgie des Hôpitaux Universitaires de Genève depuis octobre 1993 jusqu’en octobre
2003, soit sur la dernière décennie. Notre intention était davantage de comparer deux périodes
que deux traitements précis, c’est-à-dire une période de 1993 à 1998 avant l’introduction de la
gemcitabine, et la période suivante de 1999 à 2003, en posant l’hypothèse initiale selon laquelle
le changement d’attitude des oncologues relatif à la prescription de chimiothérapie dans cette
1 Protocole de recherche formellement approuvé par le Comité d’éthique des Hôpitaux Universitaires de Genève
- 5 -
indication se traduirait par une différence significative en termes de survie et d’indicateurs gros-
siers de qualité de vie tels que la durée d’hospitalisation. De plus, au vu des considérations politi-
ques, assécurologiques et économiques actuelles qui remettent en question les pratiques
médicales modernes de la plupart des pays développés, nous avons voulu étendre l’analyse à une
dimension médico-économique et comparer les deux périodes en termes de dépenses pour
l’hôpital liées au cancer du pancréas avancé.
METHODOLOGIE
Pour ce faire, nous avons tout d’abord obtenu du Registre Genevois des Tumeurs les informa-
tions épidémiologiques relatives à la décennie analysée. Puis nous avons répertorié l’ensemble
des cas de carcinome du pancréas ayant bénéficié d’une consultation par le service
d’oncochirurgie, et inclus les patients suivis auprès de ce dernier durant une proportion significa-
tive de la maladie. Nous avons ensuite constitué une base de données comprenant des informa-
tions démographiques ainsi que les caractéristiques et jalons de la maladie : date de diagnostic
(définie comme la date d’entrée du séjour hospitalier ayant mené à ou confirmé le diagnostic) ;
facteurs pronostiques (âge, stade, taille de la tumeur, site primitif et index de performance au dia-
gnostic) ; nombre et durée d’hospitalisations (catégorisées en traitement, complication du traite-
ment, complication de la maladie et autres) ; détails du traitement (opérations, radiothérapie,
nombre de lignes, de cycles et doses de chimiothérapie). En ce qui concerne la définition des
groupes, nous avons classé chronologiquement les patients retenus en fonction de la date du
diagnostic de la maladie oncologique, puis considéré la première ligne de chimiothérapie reçue,
en excluant les modalités combinées de radio- et chimiothérapie. Nous avons alors retenu comme
date charnière pour la définition des deux périodes la date du premier cycle de gemcitabine pour
le premier patient recevant ce médicament en première intention. En outre, avec l’aide du l’Unité
d’Information Médico-Economique de l’hôpital, nous avons calculé sur la base du système de
comptabilité analytique des H.U.G. le coût de tous les séjours hospitaliers (H.U.G., Beau-Séjour
et CESCO) et des visites ambulatoires en oncochirurgie pour chacun des patients inclus dans
l’étude. De plus, nous avons calculé un coût moyen par jour d’hospitalisation pour les séjours
dans des établissements de convalescence (Joli-Mont, Montana, Loëx, Génolier etc.) afin de les
inclure dans notre estimation du total des coûts. En raison de la période de temps étendue recou-
verte par notre étude, l’ensemble des coûts a été annualisé en unités 2002 au moyen de facteurs
de pondération spécifiques afin de prendre en compte l’augmentation des coûts de la santé et de
permettre une comparaison entre les deux groupes. La quasi-totalité des patients considérés ayant
- 6 -
été suivis intégralement par le service d’oncochirurgie de l’Hôpital Cantonal en ambulatoire, il
est peu probable que des dépenses importantes aient été encourues en dehors de cette
consultation. En conséquence, et étant donné que leur détermination aurait été largement aléatoire
compte tenu du design rétrospectif du protocole, les coûts extrahospitaliers (à ne pas confondre
avec les coûts ambulatoires) n’ont pas été considérés dans l’analyse.
Des analyses statistiques classiques ont alors été conduites sur la base des données récoltées,
avec un seuil de signification fixé à 0.05 : l’objectif primaire consistait en une analyse de survie,
effectuée selon la méthode de Kaplan et Meier afin de tenir compte des patients perdus de vue et
des patients toujours vivants au moment de l’analyse. Les objectifs secondaires comprenaient une
analyse de durée d’hospitalisation, ainsi qu’une comparaison économique des coûts pour l’hôpital
au cours des deux périodes considérées.
A noter que si notre intention initiale était de considérer les cas de carcinome pancréatique
tous stades confondus, nous avons finalement décidé de restreindre l’analyse aux stades III et IV,
respectivement loco régionalement avancés et métastatiques, pour des raisons d’homogénéité : les
deux groupes issus de la cohorte globale se sont en effet révélés lors de l’analyse des facteurs
pronostiques significativement différents en termes de proportion de stades I et II versus III et IV,
introduisant un important biais potentiel pour l’analyse de survie : il est bien démontré que les
patients bénéficiant d’une résection survivent significativement plus longtemps que les patients
diagnostiqués directement à un stade palliatif, et que ces deux catégories correspondent à deux
profils différents et incomparables de biologie tumorale. Une solution potentielle eût été de défi-
nir pour les patients réséqués une date de rechute, et de ne considérer que leur survie en situation
palliative. Cependant, la définition d’une telle date s’avère extrêmement hasardeuse étant donné
l’absence de protocole de suivi précis et l’absence de corrélation entre les marqueurs biologiques
et la charge tumorale en termes pronostiques. D’une manière générale, l’établissement d’un stade
pronostique en situation de récurrence étant un non-sens en oncologie, il aurait été difficile
d’apprécier l’homogénéité de groupes ainsi définis. Dans ce contexte, nous avons en définitive
conduit les analyses de survie ainsi que les calculs médico-économiques sur une population res-
treinte aux stades III et IV uniquement.
- 7 -
RESULTATS
Avec un total de 84 malades, notre échantillon représentait environ 16% de l’ensemble des cas
de carcinome du pancréas ayant été déclarés dans le canton durant la période 1994-2001 (données
épidémiologiques non disponibles pour 2002 et 2003). La date pivot retenue correspondant au 22
décembre 1998, nous avons initialement déterminé deux groupes comprenant 39 patients pour la
première période (Groupe A) et 45 patients pour la seconde période (Groupe B). L’analyse de ces
patients en termes de facteurs démographiques et pronostiques n’a pas révélé de différences
significatives pour le sexe, la taille de la tumeur, sa localisation et l’index de performance au dia-
gnostic initial. L’âge au diagnostic était par contre significativement plus jeune dans le groupe A.
Mais surtout les deux groupes se sont avérés significativement différents en termes de stade au
diagnostic, les proportions respectives de stades I et II (patients résécables) et de stades III et IV
(maladies dépassées) se montant à 44%/56% versus 22%/78%, respectivement pour les groupes
A et B (p=0.037). Pour les raisons citées précédemment, nous avons alors décidé de conduire les
analyses ultérieures sur une cohorte restreinte aux cas loco régionalement avancés ou métastati-
ques au diagnostic, dont l’évolution clinique peut être considérée comme homogène. Les groupes
finaux comprenaient 22 patients pour le groupe A et 35 pour le groupe B. Une comparaison sta-
tistique en termes de facteurs démographiques et pronostiques n’a pas révélé de différence statis-
tiquement significative excepté la persistance d’un écart d’âge au diagnostic initial (médiane de
59 ans pour le groupe A contre 66 pour le groupe B, p=0.005).
En ce qui concerne l’objectif primaire de l’étude, une analyse de survie a révélé une différence
significative entre les deux groupes, avec une médiane de survie de 7.1 et 8.7 mois et une proba-
bilité de survie à un an de 14% et 40% pour le groupe A et le groupe B, respectivement (log-rank
p=0.012, hazard ratio=0.52, intervalle de confiance à 95% 0.23-0.84). Les patients ayant survécu
plus d’une année, soit les « survivants à long terme », sont majoritairement responsables de cette
observation, comme en témoigne l’écartement des courbes au-delà de 300 jours. Cependant, un
recouvrement des intervalles de confiance à 95%, expliqué par la petite taille des effectifs, nous
pousse à interpréter ces résultats plutôt comme une tendance qu’une différence formellement
significative (intervalles de confiance à 95% pour la probabilité de survie à un an : 3-35% contre
24-58%, respectivement). Ces valeurs sont compatibles avec les chiffres moyens publiés dans la
littérature. Une analyse de la quantité de chimiothérapie administrée aux patients des deux grou-
pes a également révélé une différence significative, avec une médiane de 1 dose administrée par
patient pour le groupe A contre 14 pour le groupe B (p=0.0002).
- 8 -
Une analyse du nombre total de journées d’hospitalisation n’a pas révélé de différences signi-
ficatives entre les deux groupes. Dès lors, une analyse comparant les deux groupes en termes de
survie ambulatoire (durée des séjours en hôpital retranchée de la survie totale) a logiquement
confirmé le bénéfice de survie initialement observé pour le groupe B (médiane de 144 jours hors
hospitalisation pour le groupe A contre 199 pour le groupe B, P=0.013, hazard ratio=0.52, inter-
valle de confiance à 95% 0.24-0.84). Si l’on accepte ce paramètre comme une appréciation gros-
sière de la qualité de vie, nous concluons prudemment que les gains observés en termes de survie
avec les traitements actuels ne semblent pas être grevés d’une morbidité excessive : en effet, la
majorité des séjours hospitaliers pour les patients inclus dans l’étude correspondaient à des situa-
tions de complications sévères liées à la maladie, telles que des obstructions biliaire ou digestive
sanctionnées chirurgicalement, ou à des situations de soins de confort en fin de vie.
L’analyse du nombre de consultations en ambulatoire au service d’oncochirurgie pour traite-
ment ou suivi a elle aussi révélé une différence significative entre les deux groupes, les patients
du groupe B affichant une médiane de 28 visites contre 21 pour le groupe A (p=0.007). Ces
résultats sont à mettre en relation avec les observations médico-économiques correspondantes :
une comparaison globale des deux groupes n’a pas révélé de différence significative en termes de
total des coûts hospitaliers. Par contre, une stratification plus fine en coûts reliés aux hospitalisa-
tions et coûts reliés aux visites ambulatoires a corroboré les résultats précédemment exposés : les
deux groupes n’ont pas présenté de différence significative en termes de coûts reliés aux hospita-
lisations, mais les patients du groupe B ont coûté significativement plus cher en visites
ambulatoires (médiane de 5'005 CHF pour le groupe A versus 8'537 pour le groupe B, p=0.006).
La faible proportion du total des coûts représentée par ces dernières (6% pour le groupe A versus
15% pour le groupe B) peut expliquer que cette différence ne se reflète pas dans l’analyse glo-
bale. De même, l’augmentation de cette proportion entre les deux périodes est à mettre en relation
avec l’augmentation de la durée de survie et du nombre de chimiothérapies administrées, de
même qu’avec l’augmentation du coût moyen des drogues correspondantes.
- 9 -
DISCUSSION
Si nos résultats sont indéniablement sujets aux biais classiques inhérents aux analyses rétros-
pectives et à l’importance de la période de temps considérée, ils n’en suggèrent pas moins une
conclusion intéressante, dont la portée locale et exploratoire mériterait d’être confirmée prospec-
tivement sur une échelle plus large : considérées dans leur globalité, nos observations suggèrent
qu’il existe une faible proportion de patients atteints de carcinome pancréatique avancé dont la
biologie tumorale plus favorable pourrait se traduire par une meilleure susceptibilité à la chimio-
thérapie. Ce sont ces patients qui semblent bénéficier majoritairement des nouvelles molécules, et
il n’est désormais plus rare de suivre des survivants à long terme pendant parfois 2 à 3 ans, voire
même davantage. Étant donné que la différence de survie observée est une différence de survie
ambulatoire, on peut supposer à première vue que ce bénéfice s’accompagne d’un certain main-
tien de la qualité de vie. De plus, et par voie de conséquence, à cette prolongation correspond
également une multiplication des consultations de suivi et de traitement loco régionalement,
entraînant un surcoût ambulatoire certain quoique modeste. Si dans le cas précis du carcinome
pancréatique avancé ce surcoût n’est pas très impressionnant en raison de la survie médiane très
limitée des patients concernés (quoique certains survivants à long terme inclus dans notre étude
aient atteint une fourchette de coûts ambulatoires allant de 20'000 à 50'000 CHF), une extrapola-
tion à la population suisse dans son ensemble, et surtout à d’autres pathologies malignes bien plus
prévalentes, permet de replacer cette problématique au centre des préoccupations économiques et
politiques actuelles. A cet égard, le cas du carcinome colorectal métastatique est édifiant, puisque
la survie médiane observée a quasiment doublé sur les dix dernières années : estimée entre 10 et
12 mois au début des années 90, elle dépasse actuellement les 20 mois avec les associations chi-
miothérapeutiques les plus modernes. A ce doublement de l’espérance de vie médiane des
patients s’est associée une augmentation de près de 340 fois du prix moyen des médicaments,
passant d’environ 100-300 dollars pour le 5-fluorouracile à environ 20’000-30'000 dollars pour
les nouvelles thérapies ciblées (bevacizumab, cetuximab), et ce en considérant uniquement les
huit premières semaines de traitement !
L’évolution de l’oncologie moderne illustre donc particulièrement bien le prix à payer pour
repousser davantage les limites de la maladie : si le progrès médical n’est certainement pas à lui
tout seul responsable de l’augmentation des coûts de la santé, il en participe indéniablement, tant
par la mise à disposition de molécules toujours plus ciblées, plus complexes à développer et pres-
que invariablement plus chères que par l’augmentation de survie qu’elle rendent possible pour
- 10 -
des patients nécessitant des soins –soit une consommation de ressources médicales– réguliers.
Compte tenu de l’impact possible qu’auront les mesures politiques actuellement en chantier en
Suisse sur la disponibilité de ce progrès au plus grand nombre à moyen terme, il est de notre point
de vue nécessaire d’impliquer au plus vite tous les acteurs du secteur –législateur, fournisseurs de
soins, payeurs, et avant tout, patients– dans une réflexion plus transparente et consensuelle.
PLAN DU TRAVAIL
Après une introduction épidémiologique et clinique générale à la pathologie maligne du pan-
créas, nous mènerons une revue systématique de la littérature concernant la prise en charge loco
régionalement du cancer pancréatique avancé et présenterons les associations médicamenteuses
les plus prometteuses évaluées ces dernières années dans le cadre du traitement de cette patholo-
gie. Ensuite, nous introduirons la problématique de notre étude, présenterons notre méthodologie
ainsi que nos résultats, que nous commenterons enfin sur les plans méthodologique, clinique et
médico-économique.
- 11 -
INTRODUCTION TO PANCREATIC CANCER
EPIDEMIOLOGICAL CONSIDERATIONS
Pancreatic ductal adenocarcinoma is the most common malignant disease of the exocrine pan-
creas and accounts for approximately 90% of pancreatic cancer cases. There are about 30,000
newly diagnosed cases and approximately the same number of deaths from this disease annually
in the United States. Pancreas cancer accounts for 2% to 3% of all malignant neoplasms diag-
nosed worldwide [1]. All stages combined, 1-year and 5-year survival rates are 20% and less than
5%, respectively. Median survival time is 13 to 20 months after pancreatic resection with less
than 10% to 20% of patients being long-term survivors (i.e. surviving more than one year); it is 3
to 6 months and 6 to 10 months for patients with metastatic and locally advanced disease, respec-
tively. Unfortunately, at the time of presentation, less than 20% of patients have a disease macro-
scopically confined to the organ: approximately 40% have locally advanced disease, 40% show
visceral metastases and 35% show peritoneal implants [2] (the latter categories being not mutu-
ally exclusive). This explains why, whilst being the ninth cancer in men and the eighth in women
in terms of frequency, pancreatic cancer is the fourth cause of cancer deaths in both sex (US data)
[1], causing around 200,000 deaths yearly in the world. Data from the National Cancer Database
report a peak age of occurrence of 70 to 79 years [3].
The two biggest risk factors for pancreatic cancer are ageing and smoking. Other predisposing
factors include occupational exposure to leather tanning and machine shops [4]. Patients with
hereditary chronic pancreatitis are at increased risk. The estimated risk for this specific
population to age 70 may be as high as 40%, several decades after onset of the initial pancreatitis
[5]. This familial syndrome of autosomal dominant inheritance and of 80% penetrance rate is
characterized by recurrent episodes of pancreatitis occurring early in childhood and recurring
throughout life. Its incidence is uncertain, but as many as 5% of patients experiencing an episode
of pancreatitis may have the hereditary type of the disease. In hereditary pancreatitis, uninhibited
mutated trypsin [6] initiates a proteolytic cascade within the pancreas itself, leading to
autodigestion and acute pancreatitis. In patients with chronic pancreatitis from other causes, the
risk of pancreatic adenocarcinoma has been estimated between 2 and 15 times greater than in
control people [7, 8].
- 12 -
Familial pancreatic cancer is another autosomal dominant condition with an unidentified
causative mutation. Diagnostic criteria are the following: two or more first degree relatives with
pancreatic ductal adenocarcinoma, one first degree relative with early-onset disease (less than 50
years at the time of diagnosis) or two or more second degree relatives with the disease, one of
whom having experienced early-onset cancer. Other genetic causes of pancreatic cancer, associ-
ated with an increased risk of invasive skin melanoma, include mutations of the CDKN2 [9],
p16INK4 [10] and BRCA1/2 [11, 12] tumor suppressor genes.
PATHOLOGY
The TNM classification system for pancreatic tumors was modified in 2002 [13]:
TNM Classification and
Staging System Characteristics
Tx Primary tumor not assessable T0 No evidence of primary tumor Tis Carcinoma in situ T1 Tumor limited to the pancreas, <2cm in diameter T2 Tumor limited to the pancreas, >2cm in diameter
T3 Tumor extending beyond the pancreas but without involving the celiac axis or the superior mesenteric artery
T4 Tumor involving the celiac axis or the superior mesenteric artery (unresectable primary tumor)
Nx Regional lymph nodes not assessable N0 Regional lymph nodes not involved N1 Regional lymph nodes involved Mx Distant metastasis not assessable M0 No distant metastasis M1 Distant metastasis/es
Stage 0 Tis, N0, M0 Stage I T1-2, N0, M0 Stage II T3, N0-1, M0 Stage III T4, any N, M0 Stage IV Any T, any N, M1
This revised classification is clinically more coherent in terms of prognostic: contrasting with
previous iterations, stages I and II now correspond to resectable tumors and stages III and IV to
unresectable ones, respectively locally advanced and metastatic.
Based on National Cancer Database registry data, 78% of pancreatic adenocarcinomas arise in
the head region, while 11% arise in the body and 11% in the tail [3]. Because of their proximity
to the common bile duct, patients with a tumor arising in the pancreatic head commonly present
- 13 -
at earlier stages (longest axis usually 2 to 3.5 cm) with painless jaundice caused by bile duct
obstruction. Conversely, patients with a tumor arising in the body and/or tail region present at a
later stage (longest axis usually 5 to 7 cm), typically with pain caused by malignant infiltration of
neural and retroperitoneal structures. Accordingly, the ratio of limited to advanced disease (stage
I/stage IV) is 0.7 for tumors arising in the head region, but only 0.24 for body tumors and 0.1 for
tail tumors [3].
From a practical viewpoint, locally advanced pancreatic cancer is defined as an unresectable
tumor without evidence of distant metastases. Generally, one of the following features accounts
for surgical unresectability [14]:
- Extensive peripancreatic lymph node involvement; - Encasement or occlusion of the superior mesenteric vein or superior mesenteric vein/portal
vein confluence; - Direct involvement of the superior mesenteric artery, inferior vena cava, aorta or celiac
axis.
DEVELOPMENT AND GENETICS OF PANCREATIC CANCER
Regarding the pancreatic carcinoma precursor lesion, some clinical studies have demonstrated
a chronological sequence between pancreatic duct lesions and invasive carcinomas and thus
established that the former were truly precursors to the latter [15, 16]. The Pancreatic Intraepithe-
lial Neoplasia (PanIN) nomenclature for early lesions was standardized in 1999 based on this
sequence. It grades lesions on a 1 to 3 scale, grade 1 showing the least architectural and cytologi-
cal atypia and grade 3, equivalent to carcinoma in situ, showing the most [17].
From a molecular genetic perspective, activating point mutations in proto-oncogene K-ras
(seen in 90% of pancreatic adenocarcinoma and in roughly 30% of all cancers, including colon,
lung and bladder carcinoma) and over expression of proto-oncogene Her2/neu (a member of the
epidermal growth factor receptor family) are seen in the earliest precursor lesions and are thought
to be early genetic events in the development of pancreatic carcinoma [8]. K-ras protein normally
goes through a cycle of GTP-GDP exchange and is involved in signal transduction of extra cel-
lular mitogenic stimuli, thus promoting cell growth and proliferation. Oncogenic K-ras is unable
to hydrolyze bound GTP, resulting in constant K-ras signaling that cannot be switched off [18].
This is followed by papillary hyperplasic lesions. Inactivation of the p16 tumor suppressor gene,
which stops cellular proliferation, seems to occur at this stage (PanIN 2 and 3), and the p53 (seen
in 50-70% of pancreatic tumors [19]), DPC4 (or SMAD4) and BRCA2 genes are apparently lost
in even later stages of the neoplasic process (almost exclusively in PanIN 3) [15, 16, 20]. The
- 14 -
timing of other tumorigenic activities (DNA methylation, telomerase and matrix metalloprotein-
ase activation) and a number of low-frequency genetic alterations (inactivation of the MKK4,
STK11, ALK5 and TGFBR2 tumor suppressor genes) has not been described yet.
DIAGNOSIS AND STAGING
Initial symptoms of pancreatic carcinoma include pain, jaundice, anorexia, early satiety,
weight loss, depression and new-onset diabetes. Among them, painless jaundice offers the best
prognosis, with more than 50% of patients being eligible for resection based on preoperative
assessment. Conversely, 80% to 90% of patients presenting with pain are declared unresectable,
since this almost invariably results from celiac plexus invasion. Weight loss is due on the one
hand to interruption of gastrointestinal flow of bile and pancreatic juice, and on the other hand to
the typical catabolic state accompanying aggressive malignancies. Jaundice is most commonly
obstructive as a result of a tumor arising in the head of pancreas and obstructing the intra pancre-
atic bile duct. Other clinical signs of the disease include hepatomegaly, a palpable gallbladder
(Courvoisier’s sign, seen in 40% of cases), cachexia, Troisier’s sign (involved Virchow’s node)
and ascites [18].
Currently, no serum marker offers enough sensitivity or specificity to be employed as an
effective early diagnosis or screening tool. CA 19-9 serum level is the most sensitive and specific
one, but it is almost never positive in small, i.e. curable tumors [21]. Furthermore, benign causes
of bile duct obstruction and other gastrointestinal cancers may also raise serum CA 19-9 levels.
On the other hand, CA 19-9 dosages are commonly used in the follow-up of patients.
As already mentioned, a number of genetic lesions are associated with pancreatic adenocarci-
noma, including mutations of the K-ras (90%), p16 (80%), p53 (50%) and DPC4 (50%) genes
[22]. Among them, however, none has demonstrated enough sensitivity nor specificity to serve as
a discriminating diagnostic tool, even the most extensively studied K-ras mutation [17]. Potential
for gene therapy is being currently intensively explored, for instance by replacement of tumor
suppressor genes, use of antisense oligonucleotides, immunotherapy or gene-directed enzyme
prodrug therapies [23]. So far, results have been rather disappointing: for example, a combination
regimen of gemcitabine plus tipifarnib (farnesyltransferase inhibitor) has not been shown superior
to gemcitabine alone in terms of overall survival time in a recent randomized phase III trial [24].
The role of diagnostic imaging in the assessment of a suspected pancreatic carcinoma is, first,
to establish the diagnosis of a malignant lesion, and second to assess resectability. CT scanning is
- 15 -
the first-choice imaging technique, since it assesses pancreatic parenchymal lesions and provides
information about the ductal system, local extension, the presence or absence of biliary dilation,
regional lymph nodes status and the status of other target organs of distant metastases [4]. Small
tumors, i.e. curable lesions, may fall below the detection level of this standard imaging technique.
However, the sensitivity of dual-phase spiral CT scanning for identifying unresectable tumors
may be as high as 90% [25], although less optimistic results have been published in a recent
review [26] (i.e. about 75%). Thus, in patients with radiological absence of the above mentioned
unresectability criteria, surgeons may still be confronted to advanced tumors at the time of
laparotomy [27, 28].
Magnetic resonance imaging does not seem to be more sensitive than CT scanning for the
diagnosis of pancreatic cancer, but its specificity may be superior, allowing for the distinction of
a clear mass in patients whose CT assessment can only evidence indeterminate enlargement of
the pancreatic head. However, its ability to discriminate resectable lesions from unresectable
tumors has not been proven superior to CT scan imaging [29].
ERCP adds generally little to diagnostic staging, but it remains indicated for patients present-
ing with biliary obstruction when an ampullary lesion is suspected. On the other hand, endoscopic
ultrasound may be the most accurate imaging test for the diagnosis of pancreatic cancer. This
technique seems especially effective in identifying tumors less than 3 cm in diameter, which may
be missed on CT scans [30]. It is also a highly effective modality to obtain a tissue diagnosis with
a very low complication rate, minimizing the risk of tumor dissemination (which is occasionally
attributed to percutaneous needle biopsy).
Finally, laparoscopic staging is performed in some centers, since it may prevent unnecessary
laparotomy by detecting occult micro metastasis in 30% of patients [26, 31]. However, when
laparoscopy is performed on patients thought to have resectable disease according to high-quality
CT criteria, the proportion of unresectability findings is in the range of 4% to 13% only, as
reviewed by Pisters et al. [26]. This observation does not support the use of laparoscopy as a rou-
tine staging procedure.
- 16 -
Nevertheless, the condition for defining an adequate treatment strategy relies on an accurate
tissue diagnosis, which is unfortunately often difficult to obtain, despite clinical or radiological
evidence of pancreatic malignancy. Indeed, paucity of cancer cells in the diagnostic sample as
well as intense local fibrosis associated with the disease often challenge histological diagnosis
[17]. Hence, many patients enter treatment without a formal tissue diagnosis (depending on cen-
ters). Also, presentation and radiological findings in chronic pancreatitis and pancreatic cancer
overlap in a significant proportion of cases. Thus, as much as 5 % to 10% of patients with sus-
pected pancreatic cancer may turn out to suffer from chronic pancreatitis upon examination of the
resection specimen. Conversely, around 5% of patients with presumed chronic pancreatitis who
undergo resection end up with a final histological diagnosis of pancreatic cancer. This diagnostic
dilemma places a significant degree of stress on surgical decision-making, since pancreatic resec-
tion may have a mortality rate of as much as a 5% and a morbidity rate of about 40% [18].
TREATMENT OVERVIEW AND SURVIVAL
RESECTABLE PANCREATIC CANCER
The only curative treatment of pancreatic adenocarcinoma is surgical resection, i.e. either a
pancreaticoduodenectomy for lesions arising in the head region or a left splenopancreatectomy or
sometimes a total pancreatectomy for lesions arising in the body and tail regions. Surgery is not
proposed if liver or peritoneal metastases are present at the time of diagnosis or if metastases are
identified in lymph nodes that are not routinely removed as part of the procedure. In the absence
of distant metastases, resectability depends on major vascular invasion by the tumor. However,
advances in surgical technique now allow for en bloc resection of the superior mesenteric
vein/portal vein confluence for isolated tumor extension although it is not clear whether such a
procedure shall have a positive influence on the overall survival time [32]. Data support a current
mortality rate for pancreaticoduodenectomy in the range of 1% to 5% [33, 34], strongly contrast-
ing with the 30-day mortality rate of this procedure in the 1960s and 1970s, which approached
25%. Unfortunately, despite evident progresses in diagnostic and staging modalities and
improved surgical management, there has not been any dramatic improvement in overall survival
time for patients diagnosed with pancreatic carcinoma over the past two decades [35].
- 17 -
Adjuvant therapy
Patterns of failure after surgical resection are well established: loco regional recurrence
develop from persistent intra abdominal microscopic disease as well as hepatic or peritoneal
metastases from persistent micro metastatic disease [36]. The principles of adjuvant therapy sug-
gest trying and eradicating this persistent microscopic disease before additional proliferation into
clinically evident recurrence.
In a 2003 review, Abrams et al. point out recent evidence regarding adjuvant therapy in
resectable pancreatic cancer [37]: in 1985, an important although controversial study by the
Gastrointestinal Tumor Study Group prospectively randomized post-pancreatectomy patients to
either observation (control arm) or to split course radiotherapy (40 Gy in 2 20-Gy sequences, 20
fractions and 6 weeks) with bolus 5-FU (500 mg/m2 intravenously daily on each of the first three
radiation days of each 20 Gy sequence), followed by bolus 5-FU once weekly for 2 years after
completion of radiotherapy [38]. This study evidenced a positive effect this adjuvant protocol,
since median and long-term survival time were both significantly improved in the latter group of
patients (20 months versus 11 months, P=0.03, and 19% versus 5% 5-year survival probability,
respectively). Yet, the results of this study were contested since less than fifty patients entered the
trial. A follow-up GITSG study registered an additional 30 patients to receive adjuvant 5-FU and
radiation therapy [39]. This group of patients showed a median survival time of 18 months and a
5-year survival probability of 17%.
The European Organization for Research and Treatment of Cancer (EORTC) recently com-
pleted a randomized phase III trial evaluating adjuvant chemo radiation versus observation in 218
patients having undergone curative-intent resection of the pancreatic head and carcinomas of the
periampullar region [40]. The chemo radiation regimen was similar to the one given in the
GITSG trial, but without maintenance therapy. Survival data showed improvements in the chemo
radiation arm, with a median survival time of 24.5 months versus 19 months and a 5-year survival
probability of 28% versus 22%, however not reaching statistical significance. Although the
authors concluded that routine use of adjuvant chemo radiotherapy should not become a standard
treatment in resectable pancreatic cancer, their radiation schedule (40 Gy in 2 20-Gy sequences)
was not the most effective one. Overall however, chemo radiotherapy as an adjuvant treatment
after resection of pancreatic cancer still remains controversial in terms of survival time benefit
[41]. A randomized trial by the European Study Group for Pancreatic Cancer even evidenced
potential deleterious effects for chemo radiotherapy versus chemotherapy [42].
- 18 -
Neoadjuvant therapy
As much as 26% of patients with a tumor considered as resectable suffer from metastatic evo-
lution 3 months after surgery [43]. In this context, the rationale for proposing preoperative chemo
radiation regimens in potentially resectable pancreatic cancer is three-fold [44]: first, neoadjuvant
treatment would eliminate the delay of adjuvant treatment due to postoperative recovery; second,
neoadjuvant treatment might downstage tumors, thus increasing the likelihood of negative surgi-
cal margins; and third, neoadjuvant treatment could help prevent peritoneal tumor cell implanta-
tion and dissemination caused by manipulation during surgery. Preoperative chemo radiation for
pancreatic cancer was pioneered by Evans et al. in a phase II trial administering preoperative
radiation (50.4 Gy) and concurrent continuous infusion 5-FU 300 mg/m2/day in 28 patients with
cytological or histological proof of pancreatic head adenocarcinoma [45]. The ECOG then con-
ducted a trial testing preoperative continuous 5-FU 1000 mg/m2/day on days 2 to 5 and 29 to 32
and mitomycin-C 10 mg/m2 on day 2 of a 50.4 Gy radiation schedule [46]. Of 53 patients, 41
underwent surgery; 17 did not benefit from resection because of local extension or metastases,
while 24 underwent resection and showed a median survival time of 15.7 months compared with
9.7 for the entire group.
More controversial is the ultimate role of downstaging neoadjuvant chemo radiation followed
by surgery for patients with initially unresectable tumors. Indeed, interpretation of the literature is
complicated by differences in respectability criteria. Moreover, it is argued that surgical compli-
cations may occur more frequently secondary to inflammation and fibrosis following chemo
radiation [44].
LOCALLY ADVANCED AND METASTATIC PANCREATIC CANCER
Being the main focus of the present report, this section will be further developed with a review
of the use of recent chemotherapy modalities in this largely predominant subgroup of pancreatic
cancer patients. General considerations about their management are however introduced first.
Advanced pancreatic carcinoma is a major challenge to oncologists because of its chemo
resistance. However, observation has long suggested that chemotherapy could improve survival
time in inoperable pancreatic carcinoma [47, 48]. Especially the first trial, by Mallinson et al., led
clinicians to test many chemotherapeutic drugs as single agents or in combination therapies in
advanced pancreatic tumors. Response rates were highly variable and often not reproducible.
Among chemotherapeutic agents, 5-fluorouracil (5-FU) has certainly been the most studied.
Observed response rates were variable and did not show any consistent benefit on either survival
- 19 -
time or symptoms relief. Despite this lack of evidence in advanced pancreatic carcinoma man-
agement, 5-FU was the first chemotherapeutic agent considered as the treatment of choice for this
affection, since other molecules or combinations had failed to demonstrate any improvement in
response rates or survival time. Over the last decade however, gemcitabine, a pyrimidine antime-
tabolite, was shown to somewhat improve disease stabilization, clinical benefit response and sur-
vival time. After approval for use in the late 1990s, this agent is now considered as first choice
for treatment of advanced pancreatic cancer. These observations tended to boost clinical research
on combinations with other chemotherapeutic agents as well as radiation therapy. It is obvious
now that the introduction of gemcitabine did not fundamentally change the prognosis of advanced
pancreatic cancer in terms of classical endpoints such as survival time, and that its quantitative
benefit is only minor. However, before the approval of gemcitabine in this indication, chemother-
apy was considered as not very efficient and was therefore not systematically offered to patients.
Pioneer studies with gemcitabine, demonstrating significant benefits in terms of symptoms
improvements, changed the attitude of oncologists, as reflected by the exponential increase in the
literature about palliative management of advanced pancreatic cancer over the last decade. Che-
motherapy is now considered as an important therapeutic modality in this management, and an
intense clinical research activity is underway to determine the most efficient regimen.
In addition to chemotherapy, palliation by surgical or medical means is of primary importance:
when advanced disease is diagnosed. For example, biliary bypass may be indicated in patients
with obstructive jaundice, and duodenal bypass when gastric outlet obstruction is pending. For
the numerous patients whose predominant symptom is pain, use of long-acting opiate analgesics
may be indicated. When ineffective or not tolerated, intra operative, percutaneous or EUS-
directed celiac block may provide adequate and efficient pain relief [49]. Also, to compensate for
pancreatic exocrine failure, patients can be supplemented with enteric-coated pancreatic enzymes.
- 20 -
CHEMOTHERAPY IN ADVANCED PANCREATIC CARCINOMA: LITERATURE REVIEW
This section systematically reviews the results of recently published studies regarding man-
agement of advanced pancreatic carcinoma. The respective references were obtained on Medline
or by consulting review articles [50-58]. General considerations about clinical endpoints are pre-
sented first.
METHODOLOGICAL CONSIDERATIONS
PERFORMANCE STATUS
The Karnofsky index [59] rates performance on a 0 to 100 scale in steps of 10 depending on
patient’s ability to perform normal activities and to do active work as well as on need for
assistance:
- 100: Normal, no complaints, no evidence of disease - 90: Able to carry on normal activities, minor signs of symptoms of disease - 80: Normal activity with effort, some signs or symptoms of disease - 70: Cares for self, unable to carry on normal activity or to do active work - 60: Requires occasional assistance, but able to care for most of his needs - 50: Requires considerable assistance and frequent medical care - 40: Disabled, requires special care and assistance - 30: Severely disabled, hospitalization indicated though death not imminent - 20: Very sick, hospitalization necessary, active supportive treatment necessary - 10: Moribund, fatal processes progressing rapidly - 0: Dead
An alternate categorization of performance status was then developed by the ECOG [60], pro-
viding a five-point scale easier to use and more precise and narrow in its message:
- 0: Normal activity - 1: Symptoms, but nearly ambulatory - 2: Some bed time, but needs to be in bed less than 50% of normal daytime - 3: Needs to be in bed more than 50% of normal daytime - 4: Unable to get out of bed
Intra- and inter observer correlations for the two scales were found to be very high [61] and
some authors even developed an equivalency table between them [62].
EVALUATION OF RESPONSE IN PANCREATIC CANCER
Response to chemotherapy is highly dependent on performance status and disease stage at ini-
tial diagnosis [63]. Thus, comparison between studies is difficult due to the variability in such
prognostic factors at the time of inclusion. Moreover, despite modern imaging techniques, it is
frequently impossible to get accurate localization and measurement of tumor extension within
- 21 -
adjacent tissues because of the desmoplastic and local inflammatory reactions induced by the dis-
ease itself, making it particularly difficult to differentiate between tumor and normal surrounding
tissue [64]. Consequently, tumor response may be underestimated because reactive tissue would
not readily shrink during successful chemotherapy, or by contrast overestimated in case of faster
decrease of inflammatory reactions. Endoscopic ultrasound has been shown to be more sensitive
than CT scan for the assessment of the initial tumor and regional lymph nodes (i.e. staging), but
its performance is not better when it comes to the follow-up assessment of pancreatic parenchyma
[65, 66], a situation that might change with recent developments in three-dimensional linear
endoscopic ultrasound [67]. Hence, clinicians are facing two problems of tumor size definition at
the time of diagnosis and of imprecise estimation of tumor response after chemotherapy. There-
fore, conventional endpoints such as tumor response rates may well inadequately evaluate
response to treatment [64]. Unfortunately, the vast majority of studies considered stabilization of
disease as the primary goal of palliative treatment and focused on such outcomes as complete
response (disappearance of all known disease), partial response (50% or more decrease in tumor
size) and stable disease (between 50% decrease and 25% increase in tumor size), according to the
reduction, stabilization or increase of the primary tumor and/or its metastases as evidenced by
radiological findings (World Health Organization reporting principles for results of cancer treat-
ment [68]). From that perspective, it might be worth paying more attention to survival data,
namely median survival time and 1-year survival probability, the latter characterizing patients
who showed a prolonged treatment benefit and providing superior information about the
responder minority.
Importantly, specific attention was given in recent years to pain relief and improvement of
other disease-related symptoms, leading to the development of the Clinical Benefit Response
(CBR) notion. CBR is a measure of clinical improvement in terms of pain intensity and analgesic
consumption, performance status and weight change: a patient is considered as a responder if
rated positive for either of the primary measures of pain (i.e. reduction of more than 50% in pain
intensity or analgesics consumption) or performance status (i.e. improvement of KPS by more
than 20 points) for at least 4 weeks without a negative rating in the other parameters within 12
weeks, or if there is a weight gain of 7% above pretreatment baseline while the two primary
parameters remain stable [69]. CBR is now considered as a better indicator of therapeutic
efficacy in palliative settings in pancreatic cancer, as clinical experience has shown that small
reductions of tumor size, corresponding to stable disease from radiological criteria, may often
- 22 -
lead to spectacular improvements of pain and abdominal discomfort symptoms. Indeed, because
of its proximity to the celiac ganglion and its propensity for early invasion of this structure, small
changes in size of a pancreatic adenocarcinoma can lead to significant relief of symptoms. Evi-
dence of the impact of gemcitabine on CBR is the main rationale, which supports its current
position as a first-choice treatment modality in advanced pancreatic cancer.
Regarding serum markers, the prognostic importance of CA 19-9 has been demonstrated for
adjuvant treatment of operable pancreatic cancer [70]. Stemmler et al. [71] evaluated patients
with advanced pancreatic carcinoma and initially elevated CA 19-9 levels, who were subse-
quently treated with a gemcitabine-cisplatin combination regimen. Interestingly, independent of
CT response evaluation, CA 19-9 responders survived significantly longer than CA 19-9 non-
responders (9.8 versus 5.8 months, P=0.022), suggesting that CA 19-9 serum kinetics may serve
as an early and reliable indicator of response and warranting further evaluation.
In the future, Positron Emission Tomography (PET) may be more sensitive than traditional
imaging as a technique to measure treatment response in pancreatic cancer [50]. Indeed, uptake
and phosphorylation of glucose is higher in tumor cells, making it possible to tell apart malignant
and nonmalignant tissue in the tumoral region.
5-FLUOROURACIL
Although many oncologists considered the thymidylate synthase inhibitor 5-FU as a standard
treatment prior to the introduction of gemcitabine, there was no consistent evidence to support its
benefit in patients with advanced pancreatic cancer in terms of disease response, survival time or
quality of life. Indeed, drug efficacy (bolus 5-FU alone or biomodulated by folinic acid or inter-
feron) measured by objective response rate was highly variable, as well as median survival time,
reported to range between 3.5 and 6.2 months in a recent review [55]. However, recent assess-
ment of high dose 24-hour infusion 5-FU 2600 mg/m2 and of capecitabine 2500 mg/m2 (oral
equivalent to 5-FU, which simulates a continuous infusion) showed median survival times of 8.3
months and 6.1 months, respectively, with low toxicity, suggesting some efficacy of 5-FU
administered intravenously at continuous high doses [72, 73].
- 23 -
OTHER SINGLE-AGENT MODALITIES [55]
Topoisomerase I inhibitors: Irinotecan (or CPT-11) was associated with response rates of 9%
and a median survival time of 5.2 months [74]. Rubitecan or 9NC showed activity in previously
untreated patients with a median survival time of 7.3 months [75]. Exatecan showed an encourag-
ing activity with a median survival time of 10.6 months, and warrants further clinical evaluation
alone or in combination. Myelosuppression is the principal side effect of these drugs.
Taxanes: Both docetaxel (Taxotere®) and paclitaxel showed very modest activity in advanced
pancreatic cancer with low response rates and a median survival time of approximately 5 months
[76-79]. Jaundice and liver dysfunction is a clear limit in treatment with taxanes.
Platins: Cisplatin and oxaliplatin used alone are only modestly active, with published median
survival times of 3.4 to 4 months and response rates of 21% [80]. Combination with 5-FU yields
better results, with a median survival time between 4 and 11 months according to the regimen.
GEMCITABINE
INITIAL PHASE I-II SINGLE-AGENT STUDIES
Several initial phase I-II studies established gemcitabine as a promising molecule in the che-
motherapy of advanced pancreatic cancer, encouraging its subsequent evaluation in methodologi-
cally more robust trials. These protocols used 800-1500mg/m2 of gemcitabine on a weekly basis,
in either the conventional 4-week schedule (treatment on days 1, 8 and 15) or the initial 7-week
regimen (treatment continuously for 7 weeks with one week of rest, followed by the regular 4-
week schedule). Table 1 at the end of this section summarizes the details of these studies.
Although observed objective response rates were relatively low (4.3% to 18.2%), there were a
substantial number of patients who exhibited stable disease (18.8% to 50%). Excluding the trial
of Rothenberg et al. [81], which enrolled patients who were experiencing disease progression
under 5-FU therapy, the median overall survival time was 4.8 to 9.8 months, i.e. higher than the
typical 3-6 months survival time without treatment. 1-year survival ranged between 14.3% and
39%. Importantly, CBR was achieved in 26.6% to 48% of the patients, and toxicities were gener-
ally mild and manageable, with grade 3-4 mostly reversible hematological toxicities ranging from
6.1% to 27.3% (leucopenia), 6.6% to 36.4% (neutropenia), 0% to 18% (thrombocytopenia) and
3% to 22% (anemia). Other toxicities were mild, with nausea and vomiting, fever, rash and
transaminase elevation most commonly reported.
- 24 -
Following these encouraging results, other administration modalities of gemcitabine were
evaluated, for instance a high-dose infusion (2200 mg/m2) [82] and a fixed dose-rate infusion
regimen (10 mg/min/m2) [83]. These alternatives to the conventional gemcitabine administration
modalities are still undergoing clinical validation and are not considered as standard yet.
PHASE III SINGLE-AGENT STUDIES
Subsequently, single-agent gemcitabine regimens were evaluated in randomized phase III trials,
six of which are detailed in Table 2 at the end of this section, as reviewed by Fung et al. [51].
The pioneer trial by Burris et al. [84] demonstrated superiority of gemcitabine over 5-FU as a
first-line therapy in patients with locally advanced or metastatic pancreatic carcinoma: 23.8% of
patients in the gemcitabine arm achieved CBR compared to 4.8% in the 5-FU arm (P=0.002).
Median survival time was 5.65 months versus 4.41 (P=0.003), and 1-year survival probability
18% versus 2%. The rather low values for median survival time have to be considered in the
context of unfavorable eligibility criteria.
Several other randomized trials later compared gemcitabine to newer therapeutic classes in
pancreatic cancer treatment: the matrix metallo proteinase inhibitor marimastat [85] and Bay 12-
9566 [86]; ZD9331, a novel antifolate inhibitor of thymidylate synthase [87]; more recently the
farnesyl transferase inhibitor tipifarnib [88]; and exatecan [89]. Single-agent gemcitabine
systematically demonstrated its superiority to the other molecules.
ASSOCIATION REGIMENS
Considering that bi-therapies are often more effective than single-agent regimens in oncology,
gemcitabine was also evaluated in combination with other chemotherapeutic molecules:
• Gemcitabine and 5-FU
After in vitro demonstration of synergistic cytotoxicity [90] and encouraging initial evaluation
in phase I-II trials [91, 92] (reviewed by Oettle et al. [93]), a randomized phase III study by the
ECOG recently failed to show any significant advantage of regimens combining gemcitabine and
5-FU over gemcitabine alone [94, 95].
• Gemcitabine and cisplatin
After in vitro observation of synergy in several other solid tumors [96] and encouraging early-
phase trials [97, 98], a recent randomized phase III study failed to show any significant difference
in terms of survival time between gemcitabine alone and in combination with cisplatin [99], how-
ever being of small statistical power (97 patients).
- 25 -
• Gemcitabine and oxaliplatin (GemOx regimen)
After in vitro observation of a sequence-dependent synergy with gemcitabine [100] and
encouraging initial clinical trials [101, 102], a recent GERCOR/GISCAD study presented at the
2004 ASCO meeting failed to reach statistical significance for the GemOx arm in comparison to
single-agent gemcitabine in terms of overall survival time [103]. However, the GemOx arm evi-
denced significantly better response rate, clinical benefit and progression-free survival time, thus
warranting further evaluation in trials designed with a higher statistical power.
• Gemcitabine and irinotecan (IrinoGem regimen)
Early-phase evaluation of this regimen evidenced only poor results [104, 105].
• Gemcitabine and docetaxel
Early-phase evaluation of this regimen evidenced only poor results [106-110].
• Gemcitabine and epirubicin
Early-phase evaluation of this regimen evidenced only poor results [111, 112].
During the last annual meeting of the American Society of Clinical Oncology, disappointing
results were also presented for gemcitabine in combination with exatecan [113] or pemetrexed
[114]. Hence, none of the above-mentioned association regimens demonstrated its superior-
ity to single-agent gemcitabine in terms of overall survival time so far. The results of two
randomized phase III trials enrolling more than 300 patients and evaluating gemcitabine in com-
bination with capecitabine versus gemcitabine alone are expected in the early future (SAKK and
MRC protocols).
GEMCITABINE IN RADIO-CHEMOTHERAPEUTIC MODALITIES
Gemcitabine demonstrated very potent radio sensitizing properties in preclinical studies [115,
116], and radiation enhancement is achieved by using much lower doses than those resulting in
cell kill. Gemcitabine-mediated radio sensitization is probably related to a concurrent disruption
of deoxynucleotide pools (inducing an inhibition of radio-induced DNA damage repair) and a
subsequent redistribution of cells into the radiosensitive S phase of the cell cycle [117, 118]. A
few disparate trials have been reported, with varying objectives from palliative to curative,
neoadjuvant or adjuvant use [51]. Among the variety of regimens that were evaluated, response
rates varied from 16.7% to 57.1%, while toxicities were generally manageable with appropriately
adapted doses of gemcitabine.
- 26 -
Several other trials evaluated more aggressive association regimens with radiation, for exam-
ple gemcitabine and 5-FU in protracted intravenous infusion [119, 120], gemcitabine and cis-
platin or gemcitabine and paclitaxel [121, 122]. Given the contrasting efficacy results of those
early-phase trials, larger phase III studies will be necessary to determine the precise role of
chemo radiation therapy in patients with locally advanced pancreatic cancer.
SUMMARY
Overall, these data show that the introduction of gemcitabine has only moderately benefited
patients in terms of survival time, but much more significantly in terms of CBR. Although
numerous combination regimens have undergone clinical testing, none has demonstrated its supe-
riority to gemcitabine alone in any randomized phase III study so far. Available evidence indi-
cates that remission rates of 10% to 30%, CBR of 40% to 60%, overall survival times of 7 to 10
months and 1-year survival probabilities of 20% to 30% may be currently expected from chemo-
therapy in advanced pancreatic carcinoma with regimens based on single-agent gemcitabine [53].
This chemotherapeutic modality should currently be considered as the standard treatment outside
any experimental protocol, until future randomized phase III trials establish the superiority of
another drug or combination regimen.
- 27 -
Table 1 – Single-agent studies of gemcitabine in advanced pancreatic cancer
Toxicity
Hematological (G3/4) (%)
First author [Ref]
Design of study
(Phase)
Number of patients
Regimen
(Drug dose)
(mg/m
2)
Partial response (%)
Stable disease (%)
Disease stabilization
(PR + SD)
Clinical Benefit
Response (%)
1-year survival (%)
Time to progression
(months)
Median survival time
(months)
Leucocytes
Neutrophils
Thrombocytes
Anemia
Others
Karasek 2003 [123] Prospective (observational)
100 1000 x q3 of 4 13 48 61 26 NA 3.2 7.5 NA 8 3 5 G3/4 not observed
Crino 2001 [124] Prospective (II) 33 1000 x 7 wk, then q3 of 4 wk 12.1 45.5 57.6 NA NA NA 7.7 9 NA 12 12
(G2/G3)
Nausea/vomiting G1/2 (24%/9%); G1/2 skin rash (9%); Peripheral edema (24%); Asthenia (39%); G1 alopecia
(3%)
Meyer 2001 [125] Prospective (II) 28 1000 x 7 wk, then q3 of 4 wk 7.2 39.2 46.4 NA 14.3 NA 9.8 11 NA 18 8 G3: Diarrhea (7%); Nausea (7%); Edema (4%); Alopecia
(20%)
Okada 2001 [126] Prospective (I) 11 (a) 1000 x q3 of 4 wk
(b) 1000 x 7 wk, then q3 of 4 wk
18.2 36.4 54.6 28.6 18.2 NA 6.4 27.3 36.4 0 9.1 G3/4 anorexia (27.3%); G3: Nausea/vomiting (9.1%); Fatigue (18.2%); GOT/GPT (9.1%); Weight loss (9.1%)
Aykan 2000 [127] Prospective (II) 14 1000 x 7 wk, then q3 of 4 wk 14.3 50 64.3 NA 14.3 6 6 NA NA NA NA No G3/4 toxicity
Kurtz 1999 [128] Prospective
(compassionate) 74 1000 x 7 wk, then q3 of 4 wk 4.3 39.1 43.4 48 NA NA 5 NA 27 14 22
G3/4: Asthenia (11.6%); Nausea/vomiting (9.3%); Anorexia (7%); (Partial) occlusion (7%); Edema (2.3%); Cardiac insufficiency (2.3%); Embolization of catheter (2.3%); Arterial hypotension (2.3%); Jaundice (2.3%);
Dehydration (2.3%); Abdominal pain (2.3%)
Storniolo 1999 [63] Prospective
(Investigational New Drug)
3023 1000 x 7 wk, then q3 of 4 wk 10.6 18.4 15 2.7 4.8 NA NA NA NA
Fever (7.3%); Pain (6.8&); Asthenia (6%); Abdominal pain (5.5%); Dyspnea (5%); Dehydration (4.5%); Nausea/vomiting (3.9%); Anorexia (3.6%); Deep thrombophlebitis (3.2%); Jaundice (3.2%); Ascites
(2.9%); Edema (2.9%); Sepsis (2.4%) Manzano 1998
[129] Prospective (II) 15 1000 x 7 wk, then q3 of 4 wk NA NA NA 26.6 39 NA NA NA 6.6 NA NA G4 asthenia (6.6%)
Petrovic 1998 [130] Prospective (II) 11 1000 x 7 wk, then q3 of 4 wk NA NA NA 33 26 NA 6.65 NA NA NA NA Well tolerated
Carmichael 1996 [131]
Prospective (II) 34 800 x q3 of 4 wk (max 1000) 6.3 18.8 25.1 NA NA NA 6.3 6.1 24.5 9.1 3
G3: Nausea/vomiting (26.7%); Diarrhea (3.3%); Pain (3.3%); Consciousness alteration (3.3%)
G2: Cutaneous (10%); Fever (20%); Infection (6.7%); Nausea/vomiting (20%); Consciousness alteration
(13.3%)
Rothenberg 1996 [81]
Prospective (II) 74 1000 x 7 wk, then q3 of 4 wk
(max 1250) 10.5 29.8 40.3 27 4 2.53 3.85 9.8 26.2 4.9 11.4
G4: Hemorrhage (1.6%); Vomiting (1.6%) G2/3: Constipation (4.8%/1.6%); Cutaneous (9.5%/0%); Fever (14.3%/1.6%); Alopecia (1.6%/0%); Infection (0%/3.2%); Nausea/vomiting (28.6%/6.3%); Oral mucositis (1.6%/0%); Pain (3.2%/1.6%); Peripheral neuropathy (1.6%/0%); Pulmonary (1.6%/0%);
Consciousness alteration (1.6%/3.2%)
Casper 1994 [132] Prospective (II) 44 800 x q3 of 4 wk (max 1500) 11 31.8 42.8 NA 23 3.7 5.6 NA NA NA NA Mild to moderate flu-like syndrome: 100% Mild hemolytic-uremic syndrome: 1pt
- 28 -
Table 2 – Randomized gemcitabine studies in advanced pancreatic cancer (underlined results correspond to a statistically significant difference between arms)
Toxicity
Hematological (G3/4) (%)
First author [Ref]
Design of study
(Phase)
Number of patients
Regimen
(Drug dose)
(mg/m
2)
Partial response (%)
(incl. complete resp.)
Stable disease (%)
Disease stabilization
(PR + SD)
Clinical Benefit
Response (%)
1-year survival (%)
Time to progression
(months)
Median survival time
(months)
Leucocytes
Neutrophils
Thrombocytes
Anemia
Others
139 Gemcitabine 1000 x 7 wk, then q3 of 4
wk 4.3 44.6 48.9 NA 25 3.5 6.6 NA 34.5 6.5 10.1 G3/4: Nausea (3.6%); Vomiting (5%); Lethargy (5.8%)
Moore 2003 [86]
Randomized (III)
138 Bay-12-9566 800 mg p.o. b.i.d. 0.7 22.5 23.2 NA 10 1.7 3.7 NA 2.2 0 10.1 G3/4: Nausea (8%); Vomiting (2.9%); Lethargy (6.5%)
163 Gemcitabine 1000 x q3 of 4 wk 5.6 NA NA NA NA 2.2 5.4 16 5 11 10 G3/4: Diarrhea (4%); Nausea/vomiting (19%)
Gemcitabine 1000 x q3 of 4 wk
Berlin 2002 [95]
Randomized (III)
164 + 5-FU 600 x q3 of 4 wk
6.9 NA NA NA NA 3.4 6.7 29 7 19 10 G3/4: Diarrhea (10%); Nausea/vomiting (15%)
44 Gemcitabine 1000 x 7 wk, then q3 of 4
wk 9.2 33.4 42.6 49 NA 1.9 4.7 4 5 2 4
G3/4: Mucositis (2%); Diarrhea (0%); Nausea/vomiting (2%)
Gemcitabine 1000 x 7 wk, then q3 of 4 wk
Colucci 2002 [99]
Randomized (III)
53 + Cisplatin 25 day 1 1h before gem.
26.4 30.2 56.6 52.6 NA 4.7 7 4 9 2 6 G3/4: Mucositis (0%); Diarrhea (4%); Nausea/vomiting
(2%)
103 Gemcitabine 1000 x 7 wk, then q3 of 4
wk 26 NA NA NA 19 3.8 5.6 2 7 5 2
G3/4: Cardiac dysfunction (1%); Constipation (1%); Fever without infection (1%); Infection (2%); Local (1%); Nausea (4%); Cortical (5%); Motor (4%);
Pulmonary (9%); Skin (3%)
104 Marimastat 5 mg b.i.d. 2.8 NA NA NA 14 1.8 3.7 0 1 3 4 G3/4: Cardiac dysrhythmias (1%); Diarrhea (1%);
Musculoskeletal (7%); Nausea (2%); Cerebellar (2%); Vision (1%); pulmonary (3%); weight gain or loss (1%)
105 Marimastat 10 mg b.i.d. 2.8 NA NA NA 14 2 3.5 0 0 0 3 G3/4: Musculoskeletal (7%); Nausea (1%); Motor (1%);
Vomiting (1%)
Bramhall 2001 [85]
Randomized (III)
102 Marimastat 25 mg b.i.d. 2.8 NA NA NA 20 1.9 4.2 0 0 2 2 G3/4: Musculoskeletal (12%); Nausea (1%); Motor
(1%); Pulmonary (1%)
54 Gemcitabine 1000 x q3 of 4 wk NA NA NA NA NA NA NA 5.5 NA 7.4 9.2 G3/4: Mucositis (3.7%); Diarrhea (7.4%)
G3: Nausea/vomiting (3.7%) Gemcitabine 1000 x q3 of 4 wk
Heinemann 2001 [133]
Randomized (III)
46 + Cisplatin 50 (day 1, 15) q4 wk
NA NA NA NA NA NA NA 13 NA 8.6 13 G3/4: Mucositis (4.3%); Diarrhea (10.8%)
G3: Nausea/vomiting (33%)
63 Gemcitabine 1000 x 7 wk, then q3 of 4
wk (max 1250) 5.4 39 44.4 23.8 18 2.1 5.65 9.7 25.9 9.7 9.7
G3/4: Nausea/vomiting (12-7%) G3: Diarrhea (1.6%); constipation (3.2%); consciousness
alteration (1.6%); Pain (1.6%)
Burris 1997 [84]
Randomized (III)
63 5-FU 600 weekly 0 19 19 4.8 2 0.9 4.41 1.6 4.9 1.6 0 G3: Nausea/vomiting (4.8%); Diarrhea (4.8%);
Constipation (1.6%); Consciousness alteration (1.6%)
- 29 -
RETROSPECTIVE STUDY ON A GENEVA COHORT
INTRODUCTION
Although not having revolutionarily improved classical therapeutic outcomes such as overall
survival time in patients suffering from advanced pancreatic cancer, gemcitabine significantly
improved more qualitative outcomes such as clinical benefit response, which is often physicians’
predominant concern in a palliative context. In parallel, it also markedly stimulated clinical
research.
Numerous combination regimens based on gemcitabine have been evaluated so far in early
clinical trials, and promising ones are now undergoing further assessment in randomized phase III
studies. However, for obvious methodological reasons, the experimental situation of those trials
is often rather different from daily clinical practice, both in terms of patient homogeneity and of
therapeutic algorithm rigidity. Therefore, and because introduction of gemcitabine in the mid-90s
was a cornerstone in chemotherapeutic management of advanced pancreatic cancer, it seemed to
us interesting to retrospectively analyze a Geneva cohort of patients suffering from this
malignancy who were followed in the cantonal hospital over the last decade, considering out-
comes such as overall survival time and 1-year survival probability. Our baseline hypothesis was
that the more organized and systematic delivery of chemotherapy after the above-mentioned
introduction of gemcitabine would have improved those outcomes in comparison to the pre-
ceding period, when chemotherapy prescription for advanced pancreatic cancer was poorly stan-
dardized. Besides, other efficient medications for the management of advanced pancreatic cancer,
such as oxaliplatin, were introduced following gemcitabine, and may also have contributed to
further improve those outcomes (e.g., in second or third-line therapies).
Economic concerns have a growing importance in healthcare-related discussions and decisions
in most developed countries as the sector’s share of their gross national product is rising:
clinicians are nowadays under increasing pressure to adopt more cost-effective treatment
practices as a result of initiatives taken by major third-party payers, government and business. For
example, in the Swiss hospital sector and following other European countries like Germany, there
has recently been a movement towards adopting prospective payment systems based on
diagnosis-related groups (DRGs) for the financial retribution of health care providers.
- 30 -
Although representing less than 1% of the population, physicians determine, through their
decisions about their patients’ care, how nearly 10% of gross national product is being spent.
Considerations of justice in macro allocation of scarce resources raise the question whether pro-
longing survival time by only a few months (e.g., in some oncology situations) can be considered
as ethically justified with regard to society as a whole when incurred costs to achieve such results
are enormously high. But because of methodological as well as practical hurdles, precise deter-
mination of incurred costs for a given diagnostic or therapeutic option in a given clinical setting
has historically proven to be considerably difficult. Accordingly, cost-effectiveness and cost-
benefit studies are still rare in comparison to the wealth of publications considering clinical
aspects of medicine. In particular, these studies were sometimes criticized for their low practical
pertinence, given their often strongly rigorous experimental context. Therefore, it also seemed
interesting to us to retrospectively conduct basic economical analyses in the same cohort of
patients, i.e. mainly an assessment of costs.
METHODOLOGY
PATIENTS
References of all patients eligible for this retrospective study were retrieved by data mining
the local electronic Oncosurgery database in the Geneva Cantonal Hospital. All pancreatic ade-
nocarcinoma cases that were evaluated at least once by a practitioner belonging to this service
from October 1st, 1993 to October 31st, 2003 were considered as being eligible. The final cohort
was determined after exclusion of those patients who did not undergo at least one antitumoral
therapy under the responsibility of the service, i.e. of all patients who were evaluated only and
who underwent therapy outside of the cantonal hospital.
DATA MINING
First, we gathered epidemiological data from the Geneva Cancer Registry, which systemati-
cally records every cancer case among residents and non-residents, whether it be diagnosed in the
public or the private setting, from hospital databases, state laboratories and independent practitio-
ner declarations as well as death certificates, among others. The total number of pancreatic carci-
nomas declared in the Geneva canton from 1993 to 2003 was thus determined, so as to appreciate
the representativeness of the studied sample.
- 31 -
For each patient included in the study, the following information –when available– were then
collected from paper and electronic medical records:
- Sex;
- Birth date;
- Death date or censoring date;
- Initial diagnosis date, defined so as to minimize as much as possible potential biases in
survival time estimates and to allow for the maximal homogeneity of included patients:
the date retained was the first day of the hospital stay leading to the diagnosis or its
confirmation;
- Known prognostic factors, including ECOG PS at the time of diagnosis, defined as the
score mentioned in the record of the first oncosurgery consultation; tumor stage (TNM
classification) at the time of diagnosis, re-interpreted according to the new 2002 classi-
fication system from initial radiological, surgical and pathological records; primary site
at the time of diagnosis (head versus body and/or tail); and tumor size at the time of
diagnosis, defined as the size in centimeters of the tumoral maximal axis mentioned on
CT-Scan reports;
- Treatment, including the first modality (curative surgery or chemotherapy), as well as
subsequent radiotherapy and/or chemotherapy modalities. For chemotherapeutic proce-
dures, the number of effectively administered lines and cycles of each drug or drug
association as well as the doses were collected;
- Hospitalizations, including dates of entry and discharge, department and nature of stay
(categorized as treatment, complication of treatment, complication of disease and
others);
- Economic data available from the hospital’s Medico-Economic Department and corre-
sponding to the costs for the institution of every hospitalization and ambulatory visit to
the cantonal hospital for every included patient from the diagnosis of the disease to
death or censoring (cf. further explanations about the detailed cost accounting proce-
dures of the Geneva Cantonal Hospital after the following section).
- 32 -
DATA ANALYSIS
Group definition
For each and every patient included, the following parameters were then calculated when the
underlying information was available:
- Age at diagnosis, calculated by subtracting the birth date from the retained diagnosis
date;
- Survival time, calculated by subtracting the retained diagnosis date from the death or
censoring date. The date of October 31, 2003 was chosen as censoring date for patients
still alive at the time of analysis.
Then, two groups were defined on a chronological basis: patients included in the study were
ranked according to the date of their initial diagnosis and their first-intent chemotherapeutic
treatment was then considered (generally following diagnosis for stage III-IV patients and fol-
lowing recurrence for stage I-II resected patients), namely gemcitabine-containing regimen ver-
sus any other regimen. The diagnosis date for the first patient who received gemcitabine as a
first-line therapy to manage his disease was used to distinguish between the two groups. Thus,
group A comprised patients who did not receive gemcitabine as a first-line therapy, whereas all
patients in group B received this drug first. In this process, we did not consider any chemotherapy
regimen administered in combination with radiotherapy. We expected this date to occur some-
where in late 1998 according to the introduction of gemcitabine in Switzerland for this indication
and its official reimbursement by health insurances. Although a few patients categorized in group
A would inevitably prove to have received gemcitabine at some point of their later therapeutic
management, by definition it was not as first-intent therapy.
Following this step, statistic analyses for various parameters and outcomes were conducted.
Chi-square tests were used for binomial comparisons (a Fisher’s exact test was employed when
the number of patients under a given condition was inferior to five), as well as nonparametric
Mann-Whitney tests for ordered numerical data such as age. Two-sided p-values were systemati-
cally calculated, and a level of 0.05 was considered as statistically significant for their interpreta-
tion. Concerning survival analyses as well as cost comparisons, graphs were constructed accord-
ing to the method developed by Kaplan and Meier [134], and log-rank tests were used to compare
groups. SPSS 11.0 and GraphPad Prism 3 were used to compute those calculations.
- 33 -
Medical analyses
The following analyses were conducted first:
- Comparisons for demographic factors, mainly sex;
- Comparisons for known prognostic factors, i.e. stage, tumor size and age at initial
diagnosis;
- Comparisons of treatment modalities and especially of the total number of chemother-
apy doses received.
At this point, our first intention was to compare the two groups in terms of total survival time
(i.e. from initial diagnosis to death or censoring), including patients diagnosed at early stages and
resected as well as patients diagnosed at later stages and directly treated in a palliative setting.
However, as will be presented in the results section, it turned out that groups A and B were sig-
nificantly inhomogeneous with respect to the proportion of stage I-II versus stage III-IV patients,
group A including significantly more early stages. Therefore, any global survival comparison
would have been inevitably biased by this difference in favor of group A, since it is well known
that resected patients have a longer survival time than patients diagnosed at later stages and have
different tumor biology. A potential solution to this issue would have been to compute survival
time in a palliative setting for resected patients, i.e. from the time of their recurrence. However,
precise definition of the date of recurrence is problematic: since survival time is so short for
patients suffering from advanced pancreatic cancer, it is crucial to be as systematic as possible in
such a definition to avoid important biases and analyze meaningful data. Since there was no sys-
tematic and standardized follow-up algorithm for these patients after resection in terms of radio-
logical examinations and biologic analyses in Geneva during the considered period, retaining as
the date of recurrence the date of a CT-Scan evidencing loco regional recurrence or metastases,
the date of a serum CA 19-9 elevation or the date of a consultation whose record concludes to the
recurrence would all have jeopardized our results, being far too imprecise. Therefore, we decided
to restrict the main analyses to stage III-IV patients only, i.e. to patients initially diagnosed at
advanced stages and directly treated in a palliative setting. From a clinical viewpoint, this restric-
tion makes much more sense in that we ended up comparing homogeneous populations, with
diagnosis dates as precisely defined as could be in a retrospective design. For indicative purposes,
demographic figures regarding the whole cohort will be provided in the results section.
- 34 -
The following analyses were then conducted in the cohort restricted to stage III-IV patients:
- Primary endpoint: survival analysis;
- Secondary endpoints: comparison of the two groups in terms of hospitalization days,
distinguishing hospitalizations for treatment (e.g. permanent venous access implanta-
tion, chemotherapy or radiotherapy), complications of treatment (e.g. post-chemother-
apy fever with or without infection, severe vomiting or other treatment toxicity), com-
plications of disease (e.g. obstructive jaundice or ileus requiring surgical or interven-
tional radiological derivation procedures, cholangitis episodes and ultimately terminal
care and death) and other causes. The first hospitalization, leading to or confirming the
diagnosis, as well as the subsequent convalescence, were systematically considered as
treatment. Finally, total incurred economic costs for the hospital were compared using
log-rank tests to account for censored patients.
Economical analyses
• Cost analysis system in Geneva Cantonal Hospital
As far as economic analyses and cost comparisons are concerned, some explanations about the
method employed for costs computations using the cost accounting system used in the Geneva
Cantonal Hospital are detailed below. The fundamental problem in cost accounting is to link
costs actually incurred to a given unit of measurement –generally a product or service–, such as a
particular stay for a particular patient due to a particular pathology. This process, known as allo-
cation, is far more complex than it seems at first glance, since costs such as administrative,
maintenance, electricity or research and teaching expenditures –also known as indirect costs–
need to be allocated to each unit of output according to specific criteria since they are not directly
attributable to any given particular product or service. Indeed, although less critical than the
direct costs of medications, medical consultations or radiological procedures, for example, which
can all be related quite easily to a particular service, those indirect costs are at least as important
to consider when the true consumption of resources driven by a particular output are to be deter-
mined. Thus, the problem is to choose the least bad allocation method, knowing that it will
inevitably be wrong. Actually, true costs can never be known, although every organization needs
to understand as precisely as possible the mechanisms of consumption of resources used to
deliver its products/services. The process in the Geneva Cantonal Hospital can be summarized as
follows:
- 35 -
1. Financial data is processed to obtain information about true expenses incurred by the
consumption of individual resources (e.g. salary costs, electricity expenses, medica-
tions etc.) during a particular period;
2. Those expenses are then related to specific basic activities of the hospital to yield spe-
cific costs (e.g. the specific overall cost of radiological procedures or of medical care).
This step involves building a kind of virtual hospital, whose structure corresponds to
its basic activities;
3. Direct costs are linked directly to each type of service provided, while indirect costs
need to be allocated to different activities to yield a complete unitary cost for each
service (e.g. laboratory services, radiological exams, nurse and medical care). The
complete unitary cost includes all direct costs for the service provided (e.g. a medical
consultation, a particular radiological investigation or medications) as well as a given
proportion of all support services, whose costs are indirect (e.g. maintenance, research
and teaching expenses, electricity etc.);
4. From the previous step, detailed unitary costs for various units of output are computed,
such as the cost of one hour of hospitalization, the cost of one ambulatory visit to a
particular department or the cost of one unit of laboratory or radiological procedure.
Some of those data (such as the cost of one hospitalization day in a particular depart-
ment) are used for benchmarking and financing purposes;
5. More importantly for our discussion, the detailed costs for each particular case can be
further inferred from the detailed services provided in each care situation (listed as a
result of the coding process at the end of each stay for each patient) as well as from
previously calculated unitary costs. For example, if a patient has undergone the
implantation of a permanent venous access device as well as the administration of two
cycles of chemotherapy during a given stay, those services can be linked to their corre-
sponding unitary costs to yield the total cost of the stay.
• Conversion into standard units
Our intention was to compare costs incurred for patients who were hospitalized at different
periods in time, spanning a decade, which is considerable from an economic perspective. Costs
incurred for one hour of medical consultation in 1995, for example, are in no way directly com-
parable to the costs incurred for the same service in 2003: this is partly due to general inflation,
- 36 -
but more importantly to increases in general health care costs over the last decade. To overcome
this problem, a weighting factor was calculated to express each cost in 2002 units. 2002 was
chosen because most of the data for 2003 were not yet available at the time of analysis. To com-
pute those multiplication factors, the average cost of one typical hospitalization day in a general
care unit of the Geneva Cantonal Hospital was considered, in each year previous to 2002, with
respect to this reference period and based on the same database available from the Medico-
Economic Department. The average cost was obtained by dividing the total expenses incurred for
general care units by the total number of hospitalization days in the same units. Comparison to
2002 yielded several weighting factors, by which every cost incurred previous to this reference
year was multiplied to obtain standardized values. This approach was deemed reasonable, since
most hospitalizations of patients suffering from a palliative-stage pancreatic cancer occur in
general care units.
• Extra-hospital costs
Costs incurred during rehabilitation stays outside of the cantonal hospital, i.e. mainly conva-
lescence stays, as well as costs of terminal care stays, were also included in the analysis. Consid-
ering that every convalescence stay is more or less comparable to another, in that it rarely
includes specific and expensive surgical or radiological procedures, but only basic clinical man-
agement, a typical cost was calculated for one day of hospitalization in a convalescence hospital
whose financial statements and activity report were available for 2002 (Clinique Genevoise de
Montana). This value was then extrapolated to stays in other hospitals after correction for the
year with respect to 2002. Total estimated costs for each stay were calculated simply by multi-
plying this unitary per diem value by the length of the stay.
Last, the majority of patients included were followed almost exclusively by the oncology
service during most of their disease. Hence, it is unlikely that major expenses were incurred out-
side this consultation. Consequently, and due to obvious methodological issues, extra-hospital
costs were not considered in our medico-economic analyses.
- 37 -
RESULTS
EPIDEMIOLOGICAL DATA
The following pathologic variants of pancreatic cancer were considered when retrieving
information from the Geneva Cancer Registry (reference [135] was used until 1994 and reference
[136] from 1995 to 2002):
- Neoplasm, malignant
- Carcinoma, NOS
- Carcinoma, anaplastic type, NOS
- Adenocarcinoma, NOS
- Tubular adenocarcinoma
- Papillary adenocarcinoma, NOS
- Mucoepidermoid carcinoma
- Cystadenocarcinoma, NOS
- Mucinous adenocarcinoma
- Mucin-producing adenocarcinoma
- Infiltrating duct carcinoma
- Acinar cell carcinoma
- Adenosquamous carcinoma
- No microscopic confirmation; clinically malignant tumor
According to information available from the Registry, the incidence of pancreatic carcinoma
in the canton of Geneva was 13.3/100,000 newly diagnosed cases per year by men and
12.3/100,000 by women (standardized rates, Geneva population), or 12.7/100,000 by men and
8.3/100,000 by women (standardized rates, European population) from 1993 to 2001. 529 cases
of pancreatic carcinoma were diagnosed between January 1st, 1993 and December 31st, 2001 in
the canton of Geneva (about 400’000 inhabitants). 462 cases were residents in the canton at the
time of diagnosis and 67 cases were non-residents. Data for 2002 and 2003 were not available at
the time of analysis. Among these 529 cases, 51.8% were classified as adenocarcinoma, not oth-
erwise specified, and 31.9% as clinically malignant tumors without microscopic confirmation.
The breakdown by year is shown in the following table:
- 38 -
Year Residents Non-Residents
1993 56 7 1994 62 12 1995 40 5 1996 47 4 1997 43 4 1998 57 5 1999 45 12 2000 56 9 2001 56 9 Total 462 67
Table 3: Number of pancreatic carcinoma cases diagnosed in Geneva from 1993 to 2001
PATIENTS AND GROUPS
Our research on the Oncosurgery database in the cantonal hospital yielded a total of 86
patients who were followed by the service for a pancreatic adenocarcinoma during at least one
treatment cycle following the diagnosis of their disease from October 1st, 1993 to October 31st,
2003. Of these, two patients had to be excluded, one because his pancreatic tumor never recurred
after initial curative-intent surgery (and we intended to consider treatment at a palliative stage),
and one because his tumor was actually not of pancreatic origin, as evidenced by pathology
reports (it was an ampulla of Vater carcinoma). Thus, the final cohort consisted of 84 patients,
who were subsequently ranked according to the date of diagnosis. The pivotal date retained to
distinguish between group A and B was December 22, 1998, corresponding approximately to the
generalization of gemcitabine use in Switzerland (second half of 1998) in this indication.
Accordingly, a Group A, consisting of 39 patients, and a Group B, consisting of 45 patients,
were defined. The following table details the percentage of newly diagnosed cases of pancreatic
carcinoma in Geneva followed in Oncosurgery in the cantonal hospital, and included in the pre-
sent study. 1993 was not included in the total percentage calculation, since the starting point for
inclusion was only October of this year:
- 39 -
Year
Number of cases diagnosed
and followed-up in
Oncosurgery
Corresponding percentage
of total diagnosed cases in
Geneva
1993 (October onwards) 1 2% 1994 11 15% 1995 6 13% 1996 5 10% 1997 7 15% 1998 9 15% 1999 8 14% 2000 12 18% 2001 16 25% 2002 6 NA 2003 2 NA
Total (1994-2001) 74 16% Table 4: Percentage of pancreatic cancer patients diagnosed in Geneva and followed-up in Oncosurgery in
the Cantonal Hospital
- 40 -
Demographic and prognostic characteristics of the two groups are shown below:
Parameter Value Group A (n=39) Group B (n=45) Statistical Test p-value
F 16 (41%) 15 (33%) Sex M 23 (59%) 30 (67%)
Chi-Square 0.47
I 3 (8%) - (0%) II 14 (36%) 9 (21%) III 7 (18%) 13 (30%) IV 15 (39%) 22 (50%)
Stage at
diagnosis
Missing - 1
Mann-Whitney 0.06
Head 25 (68%) 29 (64%) Body/Tail 12 (32%) 16 (36%) Primary Site
at diagnosis Missing 2 - Chi-Square 0.77
Range 1.5-6 1.5-7 Median 4 4 Tumor Size at
diagnosis (cm) Missing 12 7 Mann-Whitney 0.93
Range 41-78 44-79 Median 60 66 25th
Percentile 50 60
Age at
diagnosis 75th
Percentile 68 72
Mann-Whitney 0.03
0 10 (39%) 7 (27%) 1 11 (42%) 13 (50%) 2 5 (19%) 5 (19%) 3 - (0%) 1 (4%)
ECOG PS at
diagnosis
Missing 13 19
Mann-Whitney 0.41
Curative Surgery (Stages I-II)
17 (44%) 10 (22%) Initial
Treatment Palliative (Stages III-IV)
22 (56%) 35 (78%)
Chi-Square 0.04
Table 5: Patients characteristics
As evidenced in Table 5, the two groups were roughly comparable in terms of demographic
factors such as sex (p=0.47), but not in terms of prognostic factors: indeed, while primary site,
tumor size and ECOG PS at initial diagnosis were quite homogeneous between the two popula-
tions (p=0.78, 0.93 and 0.41, respectively), age as well as stage at initial diagnosis differed sig-
nificantly (p=0.03 and 0.06, respectively, this latter result being at the significance limit). Ageing
is usually considered as a general prognostic factor for oncologic diseases, in part according to
the assumption that co morbidities generally increase with advancing age. Histogram graphs with
a theoretical normal curve as well as box plots for age at diagnosis are reproduced below for the
two groups:
- 41 -
Figure 1: Histogram graph for age at diagnosis (Group A)
Figure 2: Histogram graph for age at diagnosis (Group B)
- 42 -
Figure 3: Box plot for age at diagnosis (total cohort)
More importantly, and as mentioned in the methodology section, it appeared that the propor-
tion of patients diagnosed at early stages, when a curative first-intent therapy could still be
offered (stages I and II), was higher in group A than in group B. Indeed, considering together
stages I-II versus stages III-IV, proportions were respectively 44%/56% for group A versus
22%/78% for group B (p=0.04). As evidenced in the preceding literature review, it is well known
that patients who can be offered a curative-intent pancreatic resection have a longer survival time
than those whose tumor is diagnosed at a palliative stage, and who can be offered chemothera-
peutic management only. Tumor biology, in terms of clinical evolution and response to treatment,
is clearly different between the two situations. Thus, the two groups initially defined were not
comparable because of this significant difference in proportions between early and late stages,
which would have introduced a strong bias in favor of group A in subsequent global survival
comparisons. Therefore, we excluded all patients who had benefited from a surgical resection, i.e.
all patients who were diagnosed with a stage I or II tumor, and considered stage III-IV patients
only, i.e. patients who were directly treated in a palliative setting. This exclusion yielded a
smaller cohort, consisting of 22 patients for group A and 35 for group B. Subsequent analyses
were conducted on this restricted cohort. Demographic and prognostic factors comparisons are
presented below:
- 43 -
Parameter Value Group A
(n=22)
Group B
(n=35)
Statistical
Test
p-
value
F 9 (41%) 12 (34%) Sex M 13 (59%) 23 (66%)
Chi-Square 0.61
III (locally advanced disease)
7 (32%) 13 (37%) Stage at diagnosis IV (metastatic
disease) 15 (68%) 22 (63%)
Chi-Square 0.68
Head 11 (55%) 19 (54%) Body/Tail 9 (45%) 16 (46%) Primary Site at
diagnosis Missing 2 - Chi-Square 0.96
Range 2-6 1.5-7 Median 4 4 Tumor Size at
diagnosis (cm) Missing 9 6
Mann-Whitney
0.65
Range 41-78 50-79 Median 59 66
25th Percentile 50 61 Age at diagnosis
75th Percentile 65 73
Mann-Whitney
0.005
0 6 (38%) 6 (30%) 1 6 (38%) 9 (45%) 2 4 (24%) 4 (20%) 3 - 1 (5%)
ECOG PS at
diagnosis
Missing 6 15
Mann-Whitney
0.74
Table 6: Patients characteristics (Stage III-IV at initial diagnosis)
The two groups were still comparable for sex (p=0.61), primary site, tumor size and ECOG PS
at initial diagnosis (p=0.96, 0.65 and 0.74, respectively). In contrast to the preceding comparison,
the two populations were also comparable in terms of stage at diagnosis, with roughly two thirds
of patients initially diagnosed with stage IV (i.e. metastatic) disease, and one third with stage III
(i.e. locally advanced) disease (p=0.68). However, age at diagnosis was still inhomogeneous
between the two populations (p=0.005), as illustrated by the following histograms with a theo-
retical normal curve as well as by the corresponding box plot:
- 44 -
Figure 4: Histogram graph for age at palliative diagnosis (Group A)
Figure 5: Histogram graph for age at palliative diagnosis (Group B)
- 45 -
Figure 6: Box plot for age at palliative diagnosis (stage III-IV patients only)
TREATMENT
Details of the surgical, radio therapeutic and chemotherapeutic management for the two
groups of patients are shown below. Figures for chemotherapeutic modalities represent the num-
ber of doses administered. For indicative purposes, data for the total cohort are also provided:
- 46 -
Total Cohort Stage III-IV patients Treatment Modality
Group A (n=39) Group B (n=45) Group A (n=22) Group B (n=35)
Curative-intent
Duodenopancreatectomy 15 10 - -
Curative-intent left
Splenopancreatectomy 2 - - -
Palliative-intent surgical
intestinal derivation 11 7 10 7
Palliative-intent biliary
derivation 8 7 8 7
Palliative-intent biliary
stent 3 9 3 9
Palliative-intent
intestinal stent - 1 - 1
Adjuvant Chemo
radiotherapy (45-60 Gy)
14 (6 with 5-FU 500 mg/m2 and 6 with Cisplatin 6 mg/m2)
9 (7 with 5-FU 500 mg/m2 and 2 with Cisplatin 6 mg/m2)
- -
Palliative-intent
Radiotherapy (30-60
Gy)
8 10 8 10
Gemcitabine 1000
mg/m2 11 (92 doses) 11 (107 doses) 6 (54 doses) 8 (71 doses)
Gemcitabine 1000
mg/m2 + Capecitabine
1300 mg/m2
2 (12 doses) 31 (350 doses) - 25 (331 doses)
Oxaliplatin 50-80 mg/m2
+ Irinotecan 50-80
mg/m2
2 (3 doses) 14 (56 doses) - 12 (52 doses)
Leucovorin 30 mg/m2 +
5-FU 2300 mg/m2/24h 14 (206 doses) 13 (122 doses) 7 (28 doses) 12 (118 doses)
Leucovorin 30 mg/m2 +
5-FU 500 mg/m2 iv 7 (93 doses) 1 (2 doses) 3 (14 doses) 1 (2 doses)
5-FU 1700-2000
mg/m2/wk 1 (1 dose) 2 (17 doses) - 2 (17 doses)
Docetaxel 25-30 mg/m2
+ CPT-11 50-90 mg/m2 1 (1 dose) 1 (6 doses) - 1 (6 doses)
5-FU 200 mg/m2 +
Cisplatin 60 mg/m2 +
Epirubicin 50 mg/m2
2 (4 doses) - 2 (4 doses) -
Epirubicin 120 mg/m2 1 (2 doses) - 1 (2 doses) - Total number of
effectively administered
doses
414 660 102 597
Range of per-patient
administered doses 0-124 0-80 0-19 0-80
Median per-patient
administered doses 2 11 1 14
Log-rank p-value 0.07 0.0002 Table 7: Treatment characteristics for the total cohort and for stage III-IV patients
5-FU and Cisplatin doses given as part of a chemo radio therapeutic regimen are not mentioned under chemotherapeutic regimen, and were not included in total cycles calculations.
Median number of total per-patient administered doses and p-value were computed using a log-rank test to account for censored patients
- 47 -
The following Kaplan-Meier graph illustrates the significant difference in per-patient admin-
istered doses for the two groups of stage III-IV patients:
Figure 7: Kaplan-Meier graph of total per-patient chemotherapy doses for stage III-IV patients. Figures indicated in parentheses correspond to (total patients; censored). A tick indicates censored patients.
Log-rank p=0.0002
This graph provides the proportion of patients having received a given amount of chemother-
apy doses from initial diagnosis to death or censoring.
SURVIVAL ANALYSES
While none of stage III-IV patients included in group A was still alive and none had been cen-
sored, four stage III-IV patients included in group B were still alive and undergoing therapy at the
time of analysis (October 31st, 2003). The total survival time Kaplan-Meier graph is shown
below:
- 48 -
Figure 8: Kaplan-Meier graph of palliative survival time for stage III-IV patients. Figures indicated in
parentheses correspond to (total patients; censored). A tick indicates censored patients. Median survival time was 214 days for group A and 262 days for group B.
Log-rank p=0.012. Hazard ratio=0.52 (95% CI=0.23 to 0.84)
This graph plots survival time in palliative stage along the x-axis and the proportion of patients
still alive at any point in time along the y-axis.
Statistics Group A (n=22) Group B (n=35)
Median Survival time 214 days (7.1 months) 262 days (8.7 months)
95% confidence interval 131-296 days 116-408 days
25th Percentile 265 days (8.8 months) 785 days (26.2 months)
75th Percentile 107 days (3.6 months) 112 days (3.7 months)
1-year Survival probability 14% 40%
95% percentage confidence
interval 3-35% 24-58%
Log-rank p-value 0.012
Table 8: Log-rank test results for survival comparison in stage III-IV patients
Thus, this analysis revealed a statistically significant difference in survival time between the
two groups: median survival time was 7.1 months and 1-year survival probability was 14% for
group A versus 8.7 months and 40% for group B, respectively (p=0.012, hazard ratio=0.52, 95%
- 49 -
CI 0.23-0.84). As evidenced by an increasing gap between the two curves after 300 days, patients
having survived more than one year mainly accounted for this difference, i.e. group B long-term
survivors survived significantly longer than group A ones.
Building on the previously observed significant difference between the two groups in terms of
total per-patient administered chemotherapy doses, this latter variable was plotted against
survival time:
Survival
16001400120010008006004002000
100
80
60
40
20
0
Group B
Dead
Group B
Alive
Group A
Dead
Figure 9: Total number of per-patient doses versus survival time graph for stage III-IV patients.
Non-parametric Spearman’s correlation coefficient ρρρρ=0.76
As expected, the longer patients survived, the more doses of chemotherapy they received.
Non-parametric analysis of this association yielded a Spearman’s correlation coefficient ρ of
0.76.
HOSPITALIZATION DAYS
Four basic categories of hospitalization were distinguished: hospitalization for treatment pur-
poses, hospitalization due to treatment complications, disease complications, and finally other
causes not related to treatment or disease. Detailed results corresponding to these categories are
provided below, without consideration of patients group:
- 50 -
Treatment Treatment
complications
Disease
complications Others
Category Number of
stays Category
Number of
stays Category
Number of
stays Category
Number of
stays
Diagnosis 51 (including 10 explorative laparotomies)
Fever following
chemotherapy 4 Terminal care 41 Unknown 11
Port-a-Cath implantation
38 Treatment toxicity
6 (including 1 allergy, 1 diarrhea, 2 nausea & vomiting)
Jaundice � Biliary stent
12 Unrelated digestive hemorrhage
1
Digestive derivation
16 Port-a-Cath change
2 Fever 14 (including 3 pneumonia)
Biliary derivation
8 Hypoglycemia 1 General decline
10
Chemotherapy 9
Anemia following anastomotic ulcer
1 Ileus 8 (including 6 derivations)
Convalescence 6 Thrombosis 5 (including 2 Trousseau’s syndromes)
Radiotherapy 1 Ascites 3 Pain 3 Hernia 3
Digestive hemorrhage
3
Bone
metastasis 2
CNS vascular
event 2
Convalescence 2
Inaugural diabetes
2
Table 9: Number of stays for different hospitalization categories Categories are not mutually exclusive, i.e. a single stay may have been justified by several categories.
Derivative procedures were considered as treatment when occurring during the same stay as the diagnosis
In terms of total hospitalization days, a log-rank test was conducted to account for censored
patients. The corresponding Kaplan-Meier graph is shown below:
- 51 -
Figure 10: Kaplan-Meier graph of total days of hospitalization for stage III-IV patients. Figures indicated
in parentheses correspond to (total patients; censored). A tick indicates censored patients. Log-rank p=0.50
This graph indicates the proportion of patients having reached a particular number of hospi-
talization days from diagnosis to death or censoring. With a median of 55 days in group A and 60
days in group B, the two populations were not significantly different (log-rank p=0.50).
Considering this result and the significant difference in terms of total survival, it was thus
expected that the ambulatory survival time (calculated as total survival time minus hospitalization
days) be significantly different between the two groups. The corresponding Kaplan-Meier graph
is shown below:
- 52 -
Figure 11: Kaplan-Meier graph of palliative ambulatory survival time (total survival minus hospitalization days) for stage III-IV patients. Figures indicated in parentheses correspond to (total patients; censored). A
tick indicates censored patients. Median ambulatory survival time was 144 days for group A and 199 days for group B.
Log-rank p=0.013. Hazard ratio=0.52 (95% CI=0.24 to 0.84)
Indeed, the two groups were shown to differ significantly in terms of ambulatory survival
time, with a median of 144 days for group A patients and 199 days for group B (p=0.013, hazard
ratio=0.52, 95% CI 0.24-0.84).
The two groups were then compared in terms of ambulatory visits to Oncosurgery in the can-
tonal hospital: assuming that patients who survived longer were also followed and treated for a
longer period of time, one could expect a higher number of consultations for group B patients.
The corresponding Kaplan-Meier graph is shown below:
- 53 -
Figure 12: Number of ambulatory visits in Oncosurgery for stage III-IV patients. Figures indicated in
parentheses correspond to (total patients; censored). A tick indicates censored patients. Median=21 (Group A) and 28 (Group B). Log-rank p=0.007
Thus, with a median of 28 visits, patients from group B were indeed seen significantly more in
ambulatory consultations than patients from group A (median of 21 visits; p=0.007).
ECONOMIC ANALYSES
Cost data were not available for two Group B patients. Also, reliable economic information
was only available from January 1st, 1995. Hence, all patients whose records showed a hospitali-
zation prior to this date were considered in economic analyses, which corresponds to 8 stage III-
IV patients in group A. Thus, economic analyses were finally conducted on 33 group B patients
and 14 group A patients only. Remaining patients were handled as missing values in statistical
tests.
The following table shows the weighting factors that were computed to express all costs in
2002-units:
Year
Weighting
Factor
2003 1 2002 1 2001 1.0893 2000 1.1148 1999 1.0810 1998 1.1964 1997 1.1964 1996 1.1964 1995 1.1964
Table 10: Weighting factors for costs conversion into 2002 units
- 54 -
A per diem cost for convalescence and terminal care stays was computed by dividing total
expenses incurred by a clinic whose financial data were available (CHF 10,661,562.5) by its total
hospitalization days over the corresponding period (23,700). This yielded an average per diem
cost of CHF 449.85, which was rounded to CHF 450 to compute the costs of all convalescence or
terminal care stays after conversion into 2002 units.
Economic data for one patient is given below, as an example of the level of detail available:
Table 11: Example of economic data for one patient Values in Swiss Francs (CHF), annualized to 2002 levels (cf. Methodology).
Reference numbers censored for confidentiality purposes
As evidenced in Table 11, it was possible to break down total costs into five subgroups of
expenses, namely anesthesia, surgery, clinical care, imaging, laboratory and intensive care costs.
The sum corresponds to the total in the fourth column. Medication costs were a further break-
down available; they are however already spread over the five above-mentioned categories.
Finally, it was possible to distinguish between costs incurred for hospitalization stays and costs
incurred for ambulatory visits to the hospital (i.e. mainly for chemotherapeutic treatment and
follow-up in the context of the present analysis, since treatment was mainly provided on an out-
patient basis). Several visits are usually included under a unique reference number because of the
hospital’s administrative process (i.e. the same reference number is used for every ambulatory
visit until the patient is hospitalized). Similarly, convalescence and terminal care stays in other
clinics were taken into account as explained by extrapolating a global cost from a theoretical per
diem.
Patient Reference Number
Stay Reference Number
Entry Date for the considered stay
Total Costs annualized to 2002
Anaesthesia Costs annualized to 2002
Surgery Costs annualized to 2002
Clin
ical Care costs annualized to 2002
Imaging Costs annualized to 2002
Laboratory Costs annualized to 2002
Intensive Care Costs annualized to 2002
Included Cost of Medications annualized
to 2002
Hospital Stay
Ambulatory Visit
XXX XXX XXX XXX 30/01/2003 12,479 999 428 9,848 - 1,205 - 200 x
XXX XXX XXX XXX 13/02/2003 2,934 - - 1,632 395 907 - 57 x
XXX XXX XXX XXX 13/03/2003 18,799 1,539 2,869 13,727 - 663 - 305 x
XXX XXX XXX XXX 23/04/2003 72 - - 72 - - - 1 x
XXX XXX XXX XXX 10/05/2003 13,526 - - 12,066 588 873 - 225 x
47,811 2,538 3,297 37,345 983 3,647 - 788 Total
- 55 -
The following Kaplan-Meier graph shows total incurred costs for the two stage III-IV patients
groups, i.e. the costs of all hospitalizations, all ambulatory visits to the cantonal hospital and all
convalescence and terminal care stays for each patient:
Figure 13: Kaplan-Meier graph of total costs incurred by the hospital for stage III-IV patients. Figures
indicated in parentheses correspond to (total patients; censored). A tick indicates censored patients. Cost data were not available for 8 group A patients and 2 group B, who were treated as missing values.
Monetary unit: CHF Log-rank p=0.76
This graph provides the proportion of patients having reached a given cost from diagnosis to
death or censoring. With a median of CHF 82,380 for group A and of CHF 76,750 for group B,
the two populations were not significantly different (p=0.76) with respect to total costs incurred
by the cantonal hospital, which include all care services provided both on an in-patient and out-
patient basis.
Since the cost accounting system of the cantonal hospital allowed to distinguish between hos-
pitalization-related costs and costs related to ambulatory visits, statistics for the two categories
were computed for indicative purposes using data available for deceased patients. Results are
provided in the following table:
- 56 -
Subgroup A (n=14) Subgroup B (n=29)
Statistic Cost
Category Total
patients Mean Median
Total
patients Mean Median
Hospitalization 1,054,067 75,291 78,960 2,071,276 62,766 50,120
Ambulatory 67,622 4,830 5,005 379,230 11,491 8,537 Total Costs
Total 1,121,689 80,121 83,964 2,450,506 74,257 57,609
Ratio Hosp:Amb 15.6:1 15.8:1 5.5:1 5.7:1
Hospitalization 94% 85% Proportion of
total Ambulatory 6% 15%
Hospitalization 21,933 – 168,873
3,864 – 173,732
Range
Ambulatory 418 – 11,513
510 – 42,877
Table 12: Hospitalization-related and ambulatory costs for stage III-IV deceased patients Monetary unit: CHF
Additional analyses were then conducted to further assess possible differences between the
two groups along this dimension: given the absence of any significant difference in terms of hos-
pitalization days, one could expect that hospitalization-related costs would be comparable. How-
ever, considering that patients who survived longer were also followed and treated for a longer
period of time, as evidenced by the number of ambulatory visits analysis, one could expect
ambulatory costs in group B to exceed ambulatory costs in group A. Given the low proportion of
total costs represented by ambulatory costs (approximately 6% in group A and 15% in group B),
it was possible that this difference was not significant enough in absolute figures to influence
previously exposed total costs results. The corresponding graphs are shown below:
- 57 -
Figure 14: Hospitalization-related total costs for stage III-IV patients. Figures indicated in parentheses
correspond to (total patients; censored). A tick indicates censored patients. Cost data were not available for 8 group A patients and 2 group B, who were treated as missing values.
Monetary unit: CHF Median=78,960 (Group A) and 50,120 (Group B). Log-rank p=0.93
Figure 15: Ambulatory total costs for stage III-IV patients. Figures indicated in parentheses correspond to (total patients; censored). A tick indicates censored patients. Cost data were not available for 8 group A
patients and 2 group B, who were treated as missing values. Monetary unit: CHF
Median=5,005 (Group A) and 8,537 (Group B). Log-rank p=0.006
Thus, corroborating the ambulatory visits analysis, the two groups were shown to differ sig-
nificantly in terms of ambulatory costs.
- 58 -
DISCUSSION
Two groups of patients suffering from advanced pancreatic cancer were chronologically
defined, according to the first chemotherapy regimen with respect to the introduction of gemcit-
abine in Switzerland. The two groups were shown to differ significantly in terms of total survival
time (median of 7.1 in group A versus 8.7 months in group B, p=0.012), ambulatory survival time
(median of 4.8 in group A versus 6.6 months in group B, P=0.013), median per-patient chemo-
therapy doses received (1 in group A versus 14 in group B, p=0.0002), ambulatory visits in
Oncosurgery (median of 21 in group A versus 28 in group B, p=0.007) and ambulatory costs
(median of CHF 5,005 in group A versus 8,537 in group B, p=0.006). Patients having survived
more than one year, i.e. “long-term survivors”, mainly accounted for the observed differences.
There were no significant differences in terms of total hospitalization days as well as total and
hospitalization-related costs. These results are discussed below in more details.
METHODOLOGY
The method employed to define a cut-off date and distinguish the two groups, according to the
date of the first-line gemcitabine treatment for the first patient, was based on the following
rationale: the availability of this drug and of subsequently introduced medications such as
oxaliplatin, together with the significant improvements in clinical benefit responses that those
new treatments made possible, was supposed to have changed the attitude of many oncologists
regarding advanced pancreatic cancer management. As it is well corroborated by an exponential
increase in the relevant medical literature since 1996 (which corresponds to the introduction of
gemcitabine in the U.S.), chemotherapy was, from then on, considered as an efficient modality in
advanced pancreatic cancer management, and was therefore more systematically prescribed. Our
intention was to assess this change on a local cohort over the last decade along clinical dimen-
sions –i.e. mainly survival time– as well as medico-economic dimensions –i.e. costs incurred by
the hospital.
As far as diagnosis dates are concerned, the first day of the hospitalization stay leading to or
confirming the diagnosis was systematically retained. This systematic approach was meant to
allow for maximal precision in the definition of this crucial milestone. Given the rather large time
span of the study, i.e. a decade, it can however not be totally excluded that evolution of diagnos-
tic procedures introduced some bias towards an artificially longer observed survival time for
group B patients, an effect known as Will Rogers’ phenomenon [137]. By demonstrating metasta-
ses that would have formerly been silent and unidentified, more sensitive diagnostic technology
- 59 -
can result in stage migration: patients originally classified in less advanced stages could be
assigned to a more advanced one. As a consequence, the prognosis of each stage cohort shall
improve artificially because the less advanced stage cohort looses some of its worst cases (with
silent more advanced disease) while the more advanced cohort shall be enriched by better prog-
nostic cases. The following figure illustrates this phenomenon:
Figure 17: Will Rogers phenomenon; during the period corresponding to group B, patients with
metastases that would have gone undiagnosed in group A could have been assigned to a stage III or IV due to progresses in diagnostic procedures, hereby increasing the overall prognosis in both stage I-II and stage III-IV
subgroups
Given the retrospective design of the present study, it is impossible to assess the importance of
this effect in survival time estimations. However, apart from strictly technological evolutions in
ultrasonography and CT-Scan techniques, there has not been any fundamental change of diag-
nostic strategy for pancreatic cancer in Geneva over the last decade, for example towards a more
aggressive diagnostic behavior such as a screening process: patients were typically referred for
jaundice or general alteration and subsequently underwent radiological procedures (typically
ultrasound followed by CT-Scan) that led to the suspicion of a malignant process affecting pan-
creas. Diagnosis was then formally confirmed by histological examination of a tissue sample.
Therefore, we believe the occurrence of a significant Will Rogers’ phenomenon is improbable in
this particular setting.
With respect to medico-economic analyses, several methodological limitations deserve special
comments. First, the hospitalization cost calculation process relied mainly on the accuracy of
medical coding at the end of each stay or ambulatory visit. Since coding was not yet assumed by
professionals in Geneva but by care providers themselves, it may have sometimes been incom-
- 60 -
plete or imprecise. Consequently, incurred costs may have been underestimated. However, eco-
nomic calculations can be considered as homogenous between the two groups, since they rely on
the same cost accounting system, whose procedures have not fundamentally changed over the
considered time period (1995-2003). Second, the method employed to express all cost in 2002
units was correct, financially speaking. However, 2003 costs were expressed using the same
weighting as 2002 ones, since economic data to compute the corresponding conversion factor
were not available at the time of analysis. Similarly, information to compute weighting factors
was not available prior to 1998, and a single factor was therefore used for years 1995 to 1997.
Furthermore, weighting factors were corrected between 1999 and 2000, and the apparent reduc-
tion between those two years (weighting factor 1.08 versus 1.11) does not truly represent the real
evolution of health care costs in Switzerland, which were rather steadily increasing from year to
year. Although those limitations could have been circumvented by computing a compound annual
growth rate between 1995 and 2002, we chose to use official factors, which reflect the apprecia-
tion by the hospital itself of its incurred costs. Last, ambulatory costs outside Oncosurgery were
not included at all in cost analyses because corresponding data were not available retrospectively.
However, Oncosurgery followed all included patients during their entire disease and it is unlikely
that any significant expense occurred outside the hospital. Given those limitations though, med-
ico-economic results should more be considered more reflecting local experience of a particular
hospital service than pretend to any generalization.
CLINICAL RESULTS
First, epidemiological information provided by the Geneva Cancer Registry showed that our
study cohort was including approximately 16% of the total number of cases of pancreatic carci-
noma having occurred in Geneva from 1994 to 2001. Since such percentages are generally not
published in the literature, it is difficult to comment on the representativeness of our sample
compared to other studies. Interestingly, 32% of cases were considered as clinically malignant
tumors without histological confirmation, which might reflect the difficulty in getting appropriate
tissue samples in this pathology.
Second, in terms of basic characteristics, the significant initial stage difference between the
two populations at the time of diagnosis has already been mentioned in the methodology section.
It was decided to restrict all subsequent analyses to stage III-IV patients to avoid biased survival
time and medico-economic results: indeed, from a survival time perspective, resected patients
have a clearly different biological evolution from those suffering from a more advanced disease
- 61 -
that precludes any surgical intervention. Also, from a medico-economic perspective, the two
populations present a different cost structure, with resected patients incurring, for example, major
expenses related to the initial surgery, while patients suffering from an advanced disease incur
major expenses related, for instance, to mechanical intestinal or biliary complications. As already
mentioned, a potential solution would have been to consider only the palliative phase of the dis-
ease for resected patients. However, starting point definition of this period was highly hazardous
given the absence of any rigorously standardized follow-up procedures, both radiological and
biological. Furthermore, on the biological side, it is well known that CA 19-9 level at the time of
recurrence gives a much poorer indication than the initial TNM staging in terms of tumor load
and prognosis. More generally, it does not make any clinical sense to stage patients at relapse for
prognosis assessment purposes, making it almost impossible to assess patient homogeneity along
this dimension.
As far as stages are concerned, the proportion of locally advanced and metastatic pancreatic
cancers between the two groups was comparable, and corresponds roughly to published results on
larger series (proportion of about 1/3 to 2/3, respectively). With respect to other prognostic fac-
tors, the two populations were also homogeneous, with values corroborating previously published
results (ECOG PS, median size of the primary tumor), except for a slightly higher proportion of
body and tail locations for the primary tumor (30% in the present cohort versus around 20% in
the literature).
The subsisting significant difference between the two groups of stage III-IV patients in terms
of age at initial diagnosis (median of 59 for group A and 66 for group B, p=0.005) was expected
to favor group A in survival analyses. Given the limited influence of this factor in comparison to
other prognostic variables and the survival results actually in favor of group B, stratified analyses
were not conducted. Interestingly, this difference can be interpreted as reflecting behavioral
changes among medical oncologists who, realizing the clinical benefit of newer chemotherapeutic
molecules in this indication, might have proposed such treatments to a greater range of patients.
Third and as expected from the retrospective design of the present study and from therapeutic
management of included patients (usually outside any protocol), treatment modalities were rather
inhomogeneous. For instance, due to the method employed for defining the two groups, it was
unavoidable that some patients belonging to group A had actually received a few cycles of gem-
citabine at some point of their management, since they sometimes survived several months after
1998, i.e. after this drug had been made available. However and by definition, patients in group A
- 62 -
never received gemcitabine as a first-intent therapy, while it was the case for every group B
patient. Overall, the evolution of chemotherapy prescribing practice in Geneva over the last dec-
ade can be summarized as follows: in group A, the most often administered treatment consisted
of 5-FU and leucovorin-based regimens (total of 42 doses). In group B, contrastingly, the most
often administered fist-line regimen consisted of an association of gemcitabine and capecitabine
(which simulates a continuous infusion of 5-FU; total of 331 doses administered), with second-
line regimens generally associating 5-FU and leucovorin (total of 120 doses administered) or
oxaliplatin and irinotecan (total of 52 doses administered). The total number of per-patient doses
was significantly different between the two groups, probably related to the corresponding differ-
ence in survival time (patients in group B enjoyed a prolonged survival time compared to patients
in group A, which allowed them to receive more cycles) and to behavioral changes among medi-
cal oncologists towards prescribing chemotherapy more systematically in this indication.
Fourth, in terms of survival analysis, we observed a statistically significant difference in favor
of group B patients (median survival time of 214 days or 7.1 months in group A versus 262 days
or 8.7 months in group B, p=0.012, hazard ratio=0.52, 95% CI 0.23-0.84), despite the age profile
discrepancy, which was favorable to group A. Large and overlapping 95% confidence intervals
reflect the small size of the samples. Therefore, the observed significant difference should more
be considered as a positive trend. However, recalling previously exposed data from several
reviews [50-58], this trend is in line with average values published so far in the literature, respec-
tively for 5-FU-based regimens (which were the main treatment in group A patients) and for
gemcitabine-5-FU combinations (which were the first-intent treatment for a vast majority of
patients in group B). Interestingly, the difference was mainly accounted for by the long-term sur-
vivor fraction of patients in group B, which can be better appreciated by considering 1-year sur-
vival probabilities (14% in group A versus 40% in group B; overlapping 95% confidence inter-
vals: [3-35%] in group A versus [24-58%] in group B). As it is well corroborated by the litera-
ture, gemcitabine as well as some other new molecules available for advanced pancreatic cancer
management seem to particularly benefit those patients whose tumor biology tends to respond to
chemotherapy. Patients belonging to this specific category now survive longer, sometimes even
much longer than the average 9 months mentioned in medical textbooks. It is however possible,
given the small groups in the context of our study, that a difference in proportion of different
tumor biology profiles included in groups A and B introduced an unidentified bias in survival
analyses and partly accounted for the observed difference. It is also possible, as is suggested by
- 63 -
the older age of patients from group B at the time of diagnosis, that the overall epidemiology of
pancreas cancer (i.e. its biological characteristics) changed over the past few years and accounted
for the observed difference. In absence of supporting evidence in today’s medical literature, this
would require further investigations.
Fifth, in terms of total hospitalization days, the two groups were comparable, indicating that
patients in group B, although having survived longer, did not experience significantly more hos-
pital stays than in group A. This was further evidenced by comparing the two groups in terms of
ambulatory survival (calculated as total survival time minus hospitalization days for each
patient), which was shown to be significantly different (median of 144 days or 4.8 months in
group A versus 199 days or 6.6 months in group B, p=0.013). According to the hospitalization
category analysis (table 11), patients suffering from advanced pancreatic cancer were mainly
hospitalized for serious disease-related complications as well as for terminal palliative care. In
this context, the absence of difference in hospitalization days between the two groups suggests
that the increase in total survival time was accompanied by a relative preservation in terms of
quality of life, patients having been able to spend more time in their usual environment, followed
on an outpatient basis. Consequently, on the ambulatory side, the significant difference between
the two groups in terms of visits to the outpatient clinic for treatment or check-up is probably
related to the difference in survival time: patients in group B, who survived longer, were also
followed periodically, and they received more chemotherapy cycles than in group A, as previ-
ously exposed. Intuitively, this pattern had to be similar in terms of medico-economic analyses.
MEDICO-ECONOMIC RESULTS
The two groups were not significantly different in terms of total costs incurred by the hospital.
However, when costs were broken down into hospitalization-related and ambulatory-related cate-
gories, a significant difference was evidenced for ambulatory costs, which corresponded for pan-
creatic cancer mainly to outpatient visits for treatment, check-up and general management. Those
expenditures were significantly higher for patients in group B than in group A (median of CHF
8,537 versus 5,005; p=0.006). Considering absolute figures, the low proportion of total costs rep-
resented by the ambulatory category probably explains why, given the small size of the samples,
this difference was not reflected in global analyses (this proportion was 6% in group A and
increased to 15% in group B). In line with the previously observed difference in per-patient
administered chemotherapy doses as well as in the number of ambulatory visits to Oncosurgery,
- 64 -
these results are probably related to the increased survival time of group B patients, who were
followed and treated longer than in group A, without being hospitalized more often.
We are aware that the low number of patients for whom economic data were available (par-
ticularly in group A, n=14) somewhat jeopardizes our results from a statistical standpoint, and
hence that the latter can at best be considered as indicating a tendency. They are however worth
mentioning because the trend they suggest could be further assessed prospectively, and because
they rely on a methodology that draws strongly upon the Geneva teaching Hospital’s cost
accounting system, which could be more extensively used for such medico-economic analyses in
the future.
- 65 -
CONCLUSION
Several biases related to the retrospective design of the present analyses might limit their
scope and preclude them to be considered as general. Therefore, this study should be interpreted
in its local and exploratory context. Results exposed, and particularly medico-economic observa-
tions, need to be validated in prospective studies of larger scope. But taken all together, our
results suggest the following interpretation: since late 1998, a more standardized and systematic
prescription of new chemotherapeutic drugs in advanced pancreatic cancer benefited a small per-
centage of responders in terms of survival time, although the quantitative absolute increase was
minor. Furthermore, if the number of hospitalization days is accepted as a rough proxy for the
quality of life, those responders seem not to have experienced significantly more debilitating
complications in absolute terms. On the contrary, they were managed for a relatively long period
on an ambulatory basis, periodically consulting the outpatient clinic for treatment and general
check-ups. These observations suggest that in the future, along with the development of more
effective molecules and their systematic assessment in combination regimens, some patients suf-
fering from advanced pancreatic cancer may experience a prolonged relief of symptoms.
From an economic perspective, however, every progress has its cost. In that sense, present
medico-economic results should not be underestimated: prolonging survival time in patients suf-
fering from an advanced malignant disease that cannot be cured inevitably leads to prolonged
treatment and follow-up, and repeated biologic and radiological investigations. In the case of
advanced pancreatic cancer, given the dismal prognosis, ambulatory costs may not sound impres-
sive in absolute figures, although some individuals reached a range of CHF 20,000 to 50,000 of
ambulatory costs in the present cohort. However, extrapolating those figures to the entire Swiss
population gives a much clearer idea of the financial implications of medical progresses that may
seem minor at passing glance. Also, considering other much more prevalent advanced malignant
diseases helps understand this dilemma from the centrality it currently occupies in politicians’
and economic stakeholders’ minds. A particularly good example is metastatic colorectal cancer,
where chemotherapy evolved significantly over the last decade, from 5-FU modulated by leuco-
vorin to poly-chemotherapy regimens (fluoropyrimidines with oxaliplatin or irinotecan) inte-
grated into multi-disciplinary therapeutic strategies. As a result, median survival time of patients
with metastatic colorectal cancer currently ranges between 17 and 22 months [138], during which
they have to be treated with expensive association regimens, together with check-ups and peri-
- 66 -
odic reassessments. Benefits to those patients are beyond contest compared to the situation ten
years ago, when median survival time figures were approximately half the current ones (between
10 and 12 months). But those benefits have a cost. It was recently estimated that the doubling of
median survival time achieved over the past decade has been accompanied by a staggering 340-
fold increase in drug costs –just for the initial eight weeks of treatment (approximately $100-300
for fluorouracil-based regimens to $20,000-30,000 for bevacizumab or cetuximab via $10,000 for
irinotecan or oxaliplatin) [139]. Thus, medication costs for the initial eight weeks of treatment for
the estimated 56,000 colorectal cancer patients diagnosed at stage IV or developing metastases in
2004 in the United States were estimated in the range of $666 million to $1.2 billion (with the
addition of monoclonal-antibody therapy).
Health care is fascinating in that it incorporates two crucial dimensions of quality of medical
services and economic efficiency, which are often considered as antagonist or incompatible. This
multidimensionality is probably the reason why it cannot obey a single liberal economic model
and why it is currently characterized by a complex and country-specific mixture of regulations
and market-openness, whose ideal proportions have yet to be defined. That health care-related
costs are steadily increasing in developed countries is a fact, as is the necessity of addressing this
situation. However, it is also a fact that people overall are more and more expecting and
demanding medical excellence. Ultimately, trade-off between quality and economic burden is a
choice each society as a whole has to make for itself. This choice implies communication and
concord between all involved stakeholders –regulators, payers, providers and above all patients.
- 67 -
REFERENCES
1. Jemal, A., Murray, T., Samuels, A., Ghafoor, A., Ward, E., and Thun, M. J. (2003). Cancer statistics, 2003. CA Cancer J Clin 53, 5-26.
2. Evans, A., Abbruzzese, J., and Willett, C. G. (2001). Cancer of the pancreas. In "Cancer: Principles and practice of oncology" (V. T. DeVita, S. Hellman, and S. A. Rosenberg, Eds.), Lippincott, Williams & Wilkins, Philadelphia.
3. Sener, S. F., Fremgen, A., Menck, H. R., and Winchester, D. P. (1999). Pancreatic cancer: a report of treatment and survival trends for 100,313 patients diagnosed from 1985-1995, using the National Cancer Database. J Am Coll Surg 189, 1-7.
4. Ginsberg, G. G. (2000). New developments in pancreatic cancer. Semin Gastrointest Dis 11, 162-7.
5. Lowenfels, A. B., Maisonneuve, P., DiMagno, E. P., Elitsur, Y., Gates, L. K., Jr., Perrault, J., and Whitcomb, D. C. (1997). Hereditary pancreatitis and the risk of pancreatic cancer. International Hereditary Pancreatitis Study Group. J Natl Cancer Inst 89, 442-6.
6. Whitcomb, D. C., Preston, R. A., Aston, C. E., Sossenheimer, M. J., Barua, P. S., Zhang, Y., Wong-Chong, A., White, G. J., Wood, P. G., Gates, L. K., Jr., Ulrich, C., Martin, S. P., Post, J. C., and Ehrlich, G. D. (1996). A gene for hereditary pancreatitis maps to chromosome 7q35. Gastroenterology 110, 1975-80.
7. Lowenfels, A. B., Maisonneuve, P., Cavallini, G., Ammann, R. W., Lankisch, P. G., Andersen, J. R., Dimagno, E. P., Andren-Sandberg, A., and Domellof, L. (1993). Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group. N Engl J Med 328, 1433-7.
8. Konner, J., and O'Reilly, E. (2002). Pancreatic cancer: epidemiology, genetics, and approaches to screening. Oncology (Williston Park) 16, 1615-22, 1631-2; discussion 1632-3, 1637-8.
9. Whelan, A. J., Bartsch, D., and Goodfellow, P. J. (1995). Brief report: a familial syndrome of pancreatic cancer and melanoma with a mutation in the CDKN2 tumor-suppressor gene. N Engl J Med 333, 975-7.
10. Goldstein, A. M., Fraser, M. C., Struewing, J. P., Hussussian, C. J., Ranade, K., Zametkin, D. P., Fontaine, L. S., Organic, S. M., Dracopoli, N. C., Clark, W. H., Jr., and et al. (1995). Increased risk of pancreatic cancer in melanoma-prone kindreds with p16INK4 mutations. N Engl J Med 333, 970-4.
11. Lorenzo Bermejo, J., and Hemminki, K. (2004). Risk of cancer at sites other than the breast in Swedish families eligible for BRCA1 or BRCA2 mutation testing. Ann Oncol 15, 1834-41.
- 68 -
12. Lynch, H. T., Deters, C. A., Snyder, C. L., Lynch, J. F., Villeneuve, P., Silberstein, J., Martin, H., Narod, S. A., and Brand, R. E. (2005). BRCA1 and pancreatic cancer: pedigree findings and their causal relationships. Cancer Genet Cytogenet 158, 119-25.
13. (2002). Definition of TNM (pancreatic cancer section). In "AJCC Cancer Staging Handbook" (F. L. Greene, D. L. Page, I. D. Fleming, A. G. Fritz, C. M. Balch, D. G. Haller, and M. Morrow, Eds.), Springer, Chicago.
14. Ryan, D. P., and Willett, C. G. (2002). Management of locally advanced adenocarcinoma of the pancreas. Hematol Oncol Clin North Am 16, 95-103.
15. Hruban, R. H., Goggins, M., Parsons, J., and Kern, S. E. (2000). Progression model for pancreatic cancer. Clin Cancer Res 6, 2969-72.
16. Hruban, R. H., Wilentz, R. E., and Kern, S. E. (2000). Genetic progression in the pancreatic ducts. Am J Pathol 156, 1821-5.
17. Rocha Lima, C. M., and Centeno, B. (2002). Update on pancreatic cancer. Curr Opin Oncol 14, 424-30.
18. Magee, C. J., Shekouh, A., Ghaneh, P., and Neoptolemos, J. P. (2002). Update on pancreatic cancer. Hosp Med 63, 200-6.
19. Li, D., and Jiao, L. (2003). Molecular epidemiology of pancreatic cancer. Int J Gastrointest Cancer 33, 3-14.
20. Sohn, T. A., and Yeo, C. J. (2000). The molecular genetics of pancreatic ductal carcinoma: a review. Surg Oncol 9, 95-101.
21. Kardamakis, D. (1996). Tumour serum markers: clinical and economical aspects. Anticancer Res 16, 2285-8.
22. Hahn, S. A., and Kern, S. E. (1995). Molecular genetics of exocrine pancreatic neoplasms. Surg Clin North Am 75, 857-69.
23. Humphreys, M. J., Greenhalf, W., Neoptolemos, J. P., and Ghaneh, P. (1999). The potential for gene therapy in pancreatic cancer. Int J Pancreatol 26, 5-21.
24. Van Cutsem, E., van de Velde, H., Karasek, P., Oettle, H., Vervenne, W. L., Szawlowski, A., Schoffski, P., Post, S., Verslype, C., Neumann, H., Safran, H., Humblet, Y., Perez Ruixo, J., Ma, Y., and Von Hoff, D. (2004). Phase III trial of gemcitabine plus tipifarnib compared with gemcitabine plus placebo in advanced pancreatic cancer. J Clin Oncol 22, 1430-8.
25. Diehl, S. J., Lehmann, K. J., Sadick, M., Lachmann, R., and Georgi, M. (1998). Pancreatic cancer: value of dual-phase helical CT in assessing resectability. Radiology 206, 373-8.
26. Pisters, P. W., Lee, J. E., Vauthey, J. N., Charnsangavej, C., and Evans, D. B. (2001). Laparoscopy in the staging of pancreatic cancer. Br J Surg 88, 325-37.
- 69 -
27. Bluemke, D. A., Cameron, J. L., Hruban, R. H., Pitt, H. A., Siegelman, S. S., Soyer, P., and Fishman, E. K. (1995). Potentially resectable pancreatic adenocarcinoma: spiral CT assessment with surgical and pathologic correlation. Radiology 197, 381-5.
28. Fuhrman, G. M., Charnsangavej, C., Abbruzzese, J. L., Cleary, K. R., Martin, R. G., Fenoglio, C. J., and Evans, D. B. (1994). Thin-section contrast-enhanced computed tomography accurately predicts the resectability of malignant pancreatic neoplasms. Am J Surg 167, 104-11; discussion 111-3.
29. Ahmad, N. A., Lewis, J. D., Siegelman, E. S., Rosato, E. F., Ginsberg, G. G., and Kochman, M. L. (2000). Role of endoscopic ultrasound and magnetic resonance imaging in the preoperative staging of pancreatic adenocarcinoma. Am J Gastroenterol 95, 1926-31.
30. Rosch, T., Lorenz, R., Braig, C., Feuerbach, S., Siewert, J. R., Schusdziarra, V., and Classen, M. (1991). Endoscopic ultrasound in pancreatic tumor diagnosis. Gastrointest Endosc 37, 347-52.
31. Jimenez, R. E., Warshaw, A. L., Rattner, D. W., Willett, C. G., McGrath, D., and Fernandez-del Castillo, C. (2000). Impact of laparoscopic staging in the treatment of pancreatic cancer. Arch Surg 135, 409-14; discussion 414-5.
32. Leach, S. D., Lee, J. E., Charnsangavej, C., Cleary, K. R., Lowy, A. M., Fenoglio, C. J., Pisters, P. W., and Evans, D. B. (1998). Survival following pancreaticoduodenectomy with resection of the superior mesenteric-portal vein confluence for adenocarcinoma of the pancreatic head. Br J Surg 85, 611-7.
33. Yeo, C. J., Cameron, J. L., Lillemoe, K. D., Sitzmann, J. V., Hruban, R. H., Goodman, S. N., Dooley, W. C., Coleman, J., and Pitt, H. A. (1995). Pancreaticoduodenectomy for cancer of the head of the pancreas. 201 patients. Ann Surg 221, 721-31; discussion 731-3.
34. Yeo, C. J., Cameron, J. L., Sohn, T. A., Lillemoe, K. D., Pitt, H. A., Talamini, M. A., Hruban, R. H., Ord, S. E., Sauter, P. K., Coleman, J., Zahurak, M. L., Grochow, L. B., and Abrams, R. A. (1997). Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 226, 248-57; discussion 257-60.
35. Tsiotos, G. G., Farnell, M. B., and Sarr, M. G. (1999). Are the results of pancreatectomy for pancreatic cancer improving? World J Surg 23, 913-9.
36. Griffin, J. F., Smalley, S. R., Jewell, W., Paradelo, J. C., Reymond, R. D., Hassanein, R. E., and Evans, R. G. (1990). Patterns of failure after curative resection of pancreatic carcinoma. Cancer 66, 56-61.
37. Abrams, R. A. (2003). Adjuvant therapy for pancreatic adenocarcinoma: what have we learned since 1985? Int J Radiat Oncol Biol Phys 56, 3-9.
38. Kalser, M. H., and Ellenberg, S. S. (1985). Pancreatic cancer. Adjuvant combined radiation and chemotherapy following curative resection. Arch Surg 120, 899-903.
- 70 -
39. (1987). Further evidence of effective adjuvant combined radiation and chemotherapy following curative resection of pancreatic cancer. Gastrointestinal Tumor Study Group. Cancer 59, 2006-10.
40. Klinkenbijl, J. H., Jeekel, J., Sahmoud, T., van Pel, R., Couvreur, M. L., Veenhof, C. H., Arnaud, J. P., Gonzalez, D. G., de Wit, L. T., Hennipman, A., and Wils, J. (1999). Adjuvant radiotherapy and 5-fluorouracil after curative resection of cancer of the pancreas and periampullary region: phase III trial of the EORTC gastrointestinal tract cancer cooperative group. Ann Surg 230, 776-82; discussion 782-4.
41. Choti, M. A. (2004). Adjuvant therapy for pancreatic cancer--the debate continues. N Engl J Med 350, 1249-51.
42. Neoptolemos, J. P., Stocken, D. D., Friess, H., Bassi, C., Dunn, J. A., Hickey, H., Beger, H., Fernandez-Cruz, L., Dervenis, C., Lacaine, F., Falconi, M., Pederzoli, P., Pap, A., Spooner, D., Kerr, D. J., and Buchler, M. W. (2004). A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer. N Engl J Med 350, 1200-10.
43. Spitz, F. R., Abbruzzese, J. L., Lee, J. E., Pisters, P. W., Lowy, A. M., Fenoglio, C. J., Cleary, K. R., Janjan, N. A., Goswitz, M. S., Rich, T. A., and Evans, D. B. (1997). Preoperative and postoperative chemoradiation strategies in patients treated with pancreaticoduodenectomy for adenocarcinoma of the pancreas. J Clin Oncol 15, 928-37.
44. Tsai, J. Y., Iannitti, D. A., and Safran, H. (2003). Combined modality therapy for pancreatic cancer. Semin Oncol 30, 71-9.
45. Evans, D. B., Rich, T. A., Byrd, D. R., Cleary, K. R., Connelly, J. H., Levin, B., Charnsangavej, C., Fenoglio, C. J., and Ames, F. C. (1992). Preoperative chemoradiation and pancreaticoduodenectomy for adenocarcinoma of the pancreas. Arch Surg 127, 1335-9.
46. Hoffman, J. P., Lipsitz, S., Pisansky, T., Weese, J. L., Solin, L., and Benson, A. B., 3rd (1998). Phase II trial of preoperative radiation therapy and chemotherapy for patients with localized, resectable adenocarcinoma of the pancreas: an Eastern Cooperative Oncology Group Study. J Clin Oncol 16, 317-23.
47. Mallinson, C. N., Rake, M. O., Cocking, J. B., Fox, C. A., Cwynarski, M. T., Diffey, B. L., Jackson, G. A., Hanley, J., and Wass, V. J. (1980). Chemotherapy in pancreatic cancer: results of a controlled, prospective, randomised, multicentre trial. Br Med J 281, 1589-91.
48. Palmer, K. R., Kerr, M., Knowles, G., Cull, A., Carter, D. C., and Leonard, R. C. (1994). Chemotherapy prolongs survival in inoperable pancreatic carcinoma. Br J Surg 81, 882-5.
49. Polati, E., Finco, G., Gottin, L., Bassi, C., Pederzoli, P., and Ischia, S. (1998). Prospective randomized double-blind trial of neurolytic coeliac plexus block in patients with pancreatic cancer. Br J Surg 85, 199-201.
50. el-Rayes, B. F., Shields, A. F., Vaitkevicius, V., and Philip, P. A. (2003). Developments in the systemic therapy of pancreatic cancer. Cancer Invest 21, 73-86.
- 71 -
51. Fung, M. C., and Sakata, T. (2002). What's new in pancreatic cancer treatment? J Hepatobiliary Pancreat Surg 9, 61-75.
52. Oettle, H., Arnold, D., Hempel, C., and Riess, H. (2000). The role of gemcitabine alone and in combination in the treatment of pancreatic cancer. Anticancer Drugs 11, 771-86.
53. Heinemann, V. (2002). Gemcitabine-based combination treatment of pancreatic cancer. Semin Oncol 29, 25-35.
54. Heinemann, V. (2002). Gemcitabine in the treatment of advanced pancreatic cancer: a comparative analysis of randomized trials. Semin Oncol 29, 9-16.
55. Kozuch, P., Petryk, M., and Bruckner, H. W. (2002). A comprehensive update on the use of chemotherapy for metastatic pancreatic adenocarcinoma. Hematol Oncol Clin North Am 16, 123-38.
56. Haller, D. G. (2003). Chemotherapy for advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 56, 16-23.
57. Louvet, C., Andre, T., and Artru, P. (2002). [Gemcitabine-based combinations in inoperable pancreatic cancers]. Bull Cancer 89 Spec No, S96-101.
58. Claude, L., and Mornex, F. (2003). [Chemoradiation in pancreatic carcinoma]. Cancer Radiother 7, 254-65.
59. Karnofsky, D. A., and Burchenal, J. H. (1949). The clinical evaluation of chemotherapeutic agents in cancer. In "Evaluation of Chemotherapeutic Agents" (M. CM, Ed.), Columbia University Press, New York.
60. Zubrod, C. G., Schneiderman, M. A., and Frei, A. (1960). Appraisal of methods for the study of chemotherapy in man: comparative therapeutics trial of nitrogen mustard and triethylene thiophosphoramide. J Chron Dis 11, 7-33.
61. Roila, F., Lupattelli, M., Sassi, M., Basurto, C., Bracarda, S., Picciafuoco, M., Boschetti, E., Milella, G., Ballatori, E., Tonato, M., and et al. (1991). Intra and interobserver variability in cancer patients' performance status assessed according to Karnofsky and ECOG scales. Ann Oncol 2, 437-9.
62. Buccheri, G., Ferrigno, D., and Tamburini, M. (1996). Karnofsky and ECOG performance status scoring in lung cancer: a prospective, longitudinal study of 536 patients from a single institution. Eur J Cancer 32A, 1135-41.
63. Storniolo, A. M., Enas, N. H., Brown, C. A., Voi, M., Rothenberg, M. L., and Schilsky, R. (1999). An investigational new drug treatment program for patients with gemcitabine: results for over 3000 patients with pancreatic carcinoma. Cancer 85, 1261-8.
64. Rothenberg, M. L., Abbruzzese, J. L., Moore, M., Portenoy, R. K., Robertson, J. M., and Wanebo, H. J. (1996). A rationale for expanding the endpoints for clinical trials in advanced pancreatic carcinoma. Cancer 78, 627-32.
- 72 -
65. Wiersema, M. J. (2001). Accuracy of endoscopic ultrasound in diagnosing and staging pancreatic carcinoma. Pancreatology 1, 625-32.
66. Howard, T. J., Chin, A. C., Streib, E. W., Kopecky, K. K., and Wiebke, E. A. (1997). Value of helical computed tomography, angiography, and endoscopic ultrasound in determining resectability of periampullary carcinoma. Am J Surg 174, 237-41.
67. Fritscher-Ravens, A., Knoefel, W. T., Krause, C., Swain, C. P., Brandt, L., and Patel, K. (2005). Three-dimensional linear endoscopic ultrasound-feasibility of a novel technique applied for the detection of vessel involvement of pancreatic masses. Am J Gastroenterol 100, 1296-302.
68. Miller, A. B., Hoogstraten, B., Staquet, M., and Winkler, A. (1981). Reporting results of cancer treatment. Cancer 47, 207-14.
69. Burris, H., and Storniolo, A. M. (1997). Assessing clinical benefit in the treatment of pancreas cancer: gemcitabine compared to 5-fluorouracil. Eur J Cancer 33 Suppl 1, S18-22.
70. Gattani, A. M., Mandeli, J., and Bruckner, H. W. (1996). Tumor markers in patients with pancreatic carcinoma. Cancer 78, 57-62.
71. Stemmler, J., Stieber, P., Szymala, A. M., Schalhorn, A., Schermuly, M. M., Wilkowski, R., Helmberger, T., Lamerz, R., Stoffregen, C., Niebler, K., Garbrecht, M., and Heinemann, V. (2003). Are Serial CA 19-9 Kinetics Helpful in Predicting Survival in Patients with Advanced or Metastatic Pancreatic Cancer Treated with Gemcitabine and Cisplatin? Onkologie 26, 462-7.
72. Lutz, M. P., Koniger, M., Muche, R., Ellenrieder, V., Steinkamp, M., Adler, G., and Gress, T. M. (1999). A phase II study of weekly 24-h infusion of high-dose 5-fluorouracil in advanced pancreatic cancer. Z Gastroenterol 37, 993-7.
73. Cartwright, T. H., Cohn, A., Varkey, J. A., Chen, Y. M., Szatrowski, T. P., Cox, J. V., and Schulz, J. J. (2002). Phase II study of oral capecitabine in patients with advanced or metastatic pancreatic cancer. J Clin Oncol 20, 160-4.
74. Wagener, D. J., Verdonk, H. E., Dirix, L. Y., Catimel, G., Siegenthaler, P., Buitenhuis, M., Mathieu-Boue, A., and Verweij, J. (1995). Phase II trial of CPT-11 in patients with advanced pancreatic cancer, an EORTC early clinical trials group study. Ann Oncol 6, 129-32.
75. Stehlin, J. S., Giovanella, B. C., Natelson, E. A., De Ipolyi, P. D., Coil, D., Davis, B., Wolk, D., Wallace, P., and Trojacek, A. (1999). A study of 9-nitrocamptothecin (RFS-2000) in patients with advanced pancreatic cancer. Int J Oncol 14, 821-31.
76. Okada, S., Sakata, Y., Matsuno, S., Kurihara, M., Sasaki, Y., Ohashi, Y., and Taguchi, T. (1999). Phase II study of docetaxel in patients with metastatic pancreatic cancer: a Japanese cooperative study. Cooperative Group of Docetaxel for Pancreatic Cancer in Japan. Br J Cancer 80, 438-43.
- 73 -
77. Androulakis, N., Kourousis, C., Dimopoulos, M. A., Samelis, G., Kakolyris, S., Tsavaris, N., Genatas, K., Aravantinos, G., Papadimitriou, C., Karabekios, S., Stathopoulos, G. P., and Georgoulias, V. (1999). Treatment of pancreatic cancer with docetaxel and granulocyte colony-stimulating factor: a multicenter phase II study. J Clin Oncol 17, 1779-85.
78. Whitehead, R. P., Jacobson, J., Brown, T. D., Taylor, S. A., Weiss, G. R., and Macdonald, J. S. (1997). Phase II trial of paclitaxel and granulocyte colony-stimulating factor in patients with pancreatic carcinoma: a Southwest Oncology Group study. J Clin Oncol 15, 2414-9.
79. Rougier, P., Adenis, A., Ducreux, M., de Forni, M., Bonneterre, J., Dembak, M., Clouet, P., Lebecq, A., Baille, P., Lefresne-Soulas, F., Blanc, C., and Armand, J. P. (2000). A phase II study: docetaxel as first-line chemotherapy for advanced pancreatic adenocarcinoma. Eur J Cancer 36, 1016-25.
80. Wils, J. A., Kok, T., Wagener, D. J., Selleslags, J., and Duez, N. (1993). Activity of cisplatin in adenocarcinoma of the pancreas. Eur J Cancer 29A, 203-4.
81. Rothenberg, M. L., Moore, M. J., Cripps, M. C., Andersen, J. S., Portenoy, R. K., Burris, H. A., 3rd, Green, M. R., Tarassoff, P. G., Brown, T. D., Casper, E. S., Storniolo, A. M., and Von Hoff, D. D. (1996). A phase II trial of gemcitabine in patients with 5-FU-refractory pancreas cancer. Ann Oncol 7, 347-53.
82. Ulrich-Pur, H., Kornek, G. V., Raderer, M., Haider, K., Kwasny, W., Depisch, D., Greul, R., Schneeweiss, B., Krauss, G., Funovics, J., and Scheithauer, W. (2000). A phase II trial of biweekly high dose gemcitabine for patients with metastatic pancreatic adenocarcinoma. Cancer 88, 2505-11.
83. Tempero, M., Plunkett, W., Ruiz Van Haperen, V., Hainsworth, J., Hochster, H., Lenzi, R., and Abbruzzese, J. (2003). Randomized phase II comparison of dose-intense gemcitabine: thirty-minute infusion and fixed dose rate infusion in patients with pancreatic adenocarcinoma. J Clin Oncol 21, 3402-8.
84. Burris, H. A., 3rd, Moore, M. J., Andersen, J., Green, M. R., Rothenberg, M. L., Modiano, M. R., Cripps, M. C., Portenoy, R. K., Storniolo, A. M., Tarassoff, P., Nelson, R., Dorr, F. A., Stephens, C. D., and Von Hoff, D. D. (1997). Improvements in survival and clinical benefit with gemcitabine as first-line therapy for patients with advanced pancreas cancer: a randomized trial. J Clin Oncol 15, 2403-13.
85. Bramhall, S. R., Rosemurgy, A., Brown, P. D., Bowry, C., and Buckels, J. A. (2001). Marimastat as first-line therapy for patients with unresectable pancreatic cancer: a randomized trial. J Clin Oncol 19, 3447-55.
86. Moore, M. J., Hamm, J., Dancey, J., Eisenberg, P. D., Dagenais, M., Fields, A., Hagan, K., Greenberg, B., Colwell, B., Zee, B., Tu, D., Ottaway, J., Humphrey, R., and Seymour, L. (2003). Comparison of gemcitabine versus the matrix metalloproteinase inhibitor BAY 12-9566 in patients with advanced or metastatic adenocarcinoma of the pancreas: a phase III trial of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 21, 3296-302.
- 74 -
87. Smith, D., and Gallagher, N. (2003). A phase II/III study comparing intravenous ZD9331 with gemcitabine in patients with pancreatic cancer. Eur J Cancer 39, 1377-83.
88. Van Custem, E., Karasek, P., and Oettle, H. (2002). Phase III trial comparing gemcitabine+R115777 (Zarnestra) versus gemcitabine+placebo in advanced pancreatic cancer (abstract). Proc Am Soc Clin Oncol 21, 517.
89. Cheverton, P., Friess, H., Salek, Geddes, C., Bodoky, G., Valle, J., and Humblet, Y. (2004). Phase III results of exatecan (DX-8951f) versus gemcitabine (Gem) in chemotherapy of patients with advanced pancreatic cancer (abstract). In "ASCO Fortieth Annual Meeting, New Orleans LA" (G. SM, Ed.), American Society of Clincal Oncology 2004.
90. Schulz, L., Schalhorn, A., and Wilmanns, W. (1998). Synergistic interaction of gemcitabine and 5-fluorouracil in colon cancer cells (abstract). Proc Am Soc Clin Oncol 17, A965.
91. Castellano, D., Paz-Ares, L., and Pronk, L. (2000). A phase I/II clinical and pharmacologic study of dose-escalating and dose-sequencing administration of gemcitabine and folinic acid/fluorouracil in advanced unresectable pancreatic cancer (abstract). Ann Oncol 11, 61.
92. Correale, P., Cerretani, D., Marsili, S., Pozzessere, D., Petrioli, R., Messinese, S., Sabatino, M., Roviello, F., Pinto, E., Francini, G., and Giorgi, G. (2003). Gemcitabine increases systemic 5-fluorouracil exposure in advanced cancer patients. Eur J Cancer 39, 1547-51.
93. Oettle, H., and Riess, H. (2002). Gemcitabine in combination with 5-fluorouracil with or without folinic acid in the treatment of pancreatic cancer. Cancer 95, 912-22.
94. Di Costanzo, F., Sdrobolini, A., Carlini, P., Massidda, B., Mattioli, R., Lop, A., Barletta, E., Moscetti, L., Recchia, F., Tralongo, P., and Ospedaliera, A. (2001). Gemcitabine alone or in combination with 5-FU continuous infusion in the treatment of advanced pancreatic cancer: a GOIRC reandomized phase II trial (abstract). Proc Am Soc Clin Oncol 20, 612.
95. Berlin, J. D., Catalano, P., Thomas, J. P., Kugler, J. W., Haller, D. G., and Benson, A. B., 3rd (2002). Phase III study of gemcitabine in combination with fluorouracil versus gemcitabine alone in patients with advanced pancreatic carcinoma: Eastern Cooperative Oncology Group Trial E2297. J Clin Oncol 20, 3270-5.
96. Peters, G. J., Ruiz van Haperen, V. W., Bergman, A. M., Veerman, G., Smitskamp-Wilms, E., van Moorsel, C. J., Kuiper, C. M., and Braakhuis, B. J. (1996). Preclinical combination therapy with gemcitabine and mechanisms of resistance. Semin Oncol 23, 16-24.
97. Philip, P. A., Zalupski, M. M., Vaitkevicius, V. K., Arlauskas, P., Chaplen, R., Heilbrun, L. K., Adsay, V., Weaver, D., and Shields, A. F. (2001). Phase II study of gemcitabine and cisplatin in the treatment of patients with advanced pancreatic carcinoma. Cancer 92, 569-77.
98. Heinemann, V., Wilke, H., Mergenthaler, H. G., Clemens, M., Konig, H., Illiger, H. J., Arning, M., Schalhorn, A., Possinger, K., and Fink, U. (2000). Gemcitabine and cisplatin in the treatment of advanced or metastatic pancreatic cancer. Ann Oncol 11, 1399-403.
- 75 -
99. Colucci, G., Giuliani, F., Gebbia, V., Biglietto, M., Rabitti, P., Uomo, G., Cigolari, S., Testa, A., Maiello, E., and Lopez, M. (2002). Gemcitabine alone or with cisplatin for the treatment of patients with locally advanced and/or metastatic pancreatic carcinoma: a prospective, randomized phase III study of the Gruppo Oncologia dell'Italia Meridionale. Cancer 94, 902-10.
100. Faivre, S., Raymond, E., Woynarowski, J. M., and Cvitkovic, E. (1999). Supraadditive effect of 2',2'-difluorodeoxycytidine (gemcitabine) in combination with oxaliplatin in human cancer cell lines. Cancer Chemother Pharmacol 44, 117-23.
101. Louvet, C., Andre, T., Lledo, G., Hammel, P., Bleiberg, H., Bouleuc, C., Gamelin, E., Flesch, M., Cvitkovic, E., and de Gramont, A. (2002). Gemcitabine combined with oxaliplatin in advanced pancreatic adenocarcinoma: final results of a GERCOR multicenter phase II study. J Clin Oncol 20, 1512-8.
102. Alberts, S. R., Townley, P. M., Goldberg, R. M., Cha, S. S., Sargent, D. J., Moore, D. F., Krook, J. E., Pitot, H. C., Fitch, T. R., Wiesenfeld, M., and Mailliard, J. A. (2003). Gemcitabine and oxaliplatin for metastatic pancreatic adenocarcinoma: a North Central Cancer Treatment Group phase II study. Ann Oncol 14, 580-5.
103. Louvet, C., Labianca, R., Hammel, P., Lledo, G., de Braud, F., Andre, T., Cantore, M., Ducreux, M., Zaniboni, A., and de Gramont, A. (2004). GemOx (Gemcitabine + Oxaliplatin) versus Gem (Gemcitabine) in non resectable pancreatic adenocarcinoma: final resutls of the GERCOR/GISCAD Intergroup Phase III. In "ASCO Fortieth Annual Meeting, New Orleans LA" (G. SM, Ed.), American Society of Clincal Oncology 2004.
104. Rocha Lima, C. M., Savarese, D., Bruckner, H., Dudek, A., Eckardt, J., Hainsworth, J., Yunus, F., Lester, E., Miller, W., Saville, W., Elfring, G. L., Locker, P. K., Compton, L. D., Miller, L. L., and Green, M. R. (2002). Irinotecan plus gemcitabine induces both radiographic and CA 19-9 tumor marker responses in patients with previously untreated advanced pancreatic cancer. J Clin Oncol 20, 1182-91.
105. Stathopoulos, G. P., Rigatos, S. K., Dimopoulos, M. A., Giannakakis, T., Foutzilas, G., Kouroussis, C., Janninis, D., Aravantinos, G., Androulakis, N., Agelaki, S., Stathopoulos, J. G., and Georgoulias, V. (2003). Treatment of pancreatic cancer with a combination of irinotecan (CPT-11) and gemcitabine: a multicenter phase II study by the Greek Cooperative Group for Pancreatic Cancer. Ann Oncol 14, 388-94.
106. Jacobs, A. D., Otero, H., and Picozzi, V. J. (2000). A phase I/II study of gemcitabine and docetaxel in patients with unresectable pancreatic cancer (abstract). Proc Am Soc Clin Oncol 19, 265.
107. Cascinu, S., Gasparini, G., Catalano, V., Silva, R. R., Pancera, G., Morabito, A., Giordani, P., Gattuso, D., and Catalano, G. (1999). A phase I-II study of gemcitabine and docetaxel in advanced pancreatic cancer: a report from the Italian Group for the Study of Digestive Tract Cancer (GISCAD). Ann Oncol 10, 1377-9.
108. Lueck, A., Ridwelski, K., and Kettner, E. (2000). Final results of a phase I study of weekly gemcitabine and docetaxel in pancreatic carcinoma and preliminary results of a phase II study (abstract). Proc Am Soc Clin Oncol 19, 318.
- 76 -
109. Stathopoulos, G. P., Mavroudis, D., Tsavaris, N., Kouroussis, C., Aravantinos, G., Agelaki, S., Kakolyris, S., Rigatos, S. K., Karabekios, S., and Georgoulias, V. (2001). Treatment of pancreatic cancer with a combination of docetaxel, gemcitabine and granulocyte colony-stimulating factor: a phase II study of the Greek Cooperative Group for Pancreatic Cancer. Ann Oncol 12, 101-3.
110. Ryan, D. P., Kulke, M. H., Fuchs, C. S., Grossbard, M. L., Grossman, S. R., Morgan, J. A., Earle, C. C., Shivdasani, R., Kim, H., Mayer, R. J., and Clark, J. W. (2002). A Phase II study of gemcitabine and docetaxel in patients with metastatic pancreatic carcinoma. Cancer 94, 97-103.
111. Scheithauer, W., Kornek, G. V., Raderer, M., Hejna, M., Valencak, J., Miholic, J., Kovats, E., Lang, F., Funovics, J., Bareck, E., and Depisch, D. (1999). Phase II trial of gemcitabine, epirubicin and granulocyte colony-stimulating factor in patients with advanced pancreatic adenocarcinoma. Br J Cancer 80, 1797-802.
112. Neri, B., Cini, G., Doni, L., Fulignati, C., Turrini, M., Pantalone, D., Mini, E., De Luca Cardillo, C., Fioretto, L. M., Ribecco, A. S., Moretti, R., Scatizzi, M., Zocchi, G., and Quattrone, A. (2002). Weekly gemcitabine plus Epirubicin as effective chemotherapy for advanced pancreatic cancer: a multicenter phase II study. Br J Cancer 87, 497-501.
113. O'Reilly, E. M., Abou-Alfa, G. K., Letourneau, R., Harker, W. G., Modiano, M. R., Hurwitz, H., Tchekmedyian, N. S., Ackerman, J., De Jager, R. L., and Eckhardt, S. G. (2004). A randomized phase III trial of DX-8951f (Exatecan Mesylate; DX) and Gemcitabine (GEM) vs. Gemcitabine alone in advanced pancreatic cancer (abstract). In "ASCO Fortieth Annual Meeting, New Orleans LA" (G. SM, Ed.), American Society of Clincal Oncology 2004.
114. Richards, D. A., Kindler, H. L., Oettle, H., Ramanathan, R. K., Van Laethem, J. L., Peeters, M., Fuchs, M., John, W., Arning, M., and Von Hoff, D. (2004). A randomized phase III study comparing gemcitabine + pemetrexed versus gemcitabine in patients with locally advanced and metastatic pancreas cancer (abstract). In "ASCO Fortieth Annual Meeting, New Orleans LA" (G. SM, Ed.), American Society of Clincal Oncology 2004.
115. Doyle, T. H., Mornex, F., and McKenna, W. G. (2001). The clinical implications of gemcitabine radiosensitization. Clin Cancer Res 7, 226-8.
116. Lawrence, T. S., Eisbruch, A., McGinn, C. J., Fields, M. T., and Shewach, D. S. (1999). Radiosensitization by gemcitabine. Oncology (Huntingt) 13, 55-60.
117. McGinn, C. J., Lawrence, T. S., and Zalupski, M. M. (2002). On the development of gemcitabine-based chemoradiotherapy regimens in pancreatic cancer. Cancer 95, 933-40.
118. Ostruszka, L. J., and Shewach, D. S. (2000). The role of cell cycle progression in radiosensitization by 2',2'-difluoro-2'-deoxycytidine. Cancer Res 60, 6080-8.
119. Talamonti, M. S., Catalano, P. J., Vaughn, D. J., Whittington, R., Beauchamp, R. D., Berlin, J., and Benson, A. B., 3rd (2000). Eastern Cooperative Oncology Group Phase I trial of protracted venous infusion fluorouracil plus weekly gemcitabine with concurrent
- 77 -
radiation therapy in patients with locally advanced pancreas cancer: a regimen with unexpected early toxicity. J Clin Oncol 18, 3384-9.
120. Osteen, R. T., Zinner, M. J., and Fuchs, C. S. (1999). Phase I trial of concurrent gemcitabine, infusional 5 fluorouracil and radiation therapy in patients with localized, unresectable pancreatic adenocarcinoma (abstract). Proc Am Soc Clin Oncol 18, 284.
121. Martenson, J. A., Vigliotti, A. P., Pitot, H. C., Geeraerts, L. H., Sargent, D. J., Haddock, M. G., Ghosh, C., Keppen, M. D., Fitch, T. R., and Goldberg, R. M. (2003). A phase I study of radiation therapy and twice-weekly gemcitabine and cisplatin in patients with locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 55, 1305-10.
122. Safran, H., Dipetrillo, T., Iannitti, D., Quirk, D., Akerman, P., Cruff, D., Cioffi, W., Shah, S., Ramdin, N., and Rich, T. (2002). Gemcitabine, paclitaxel, and radiation for locally advanced pancreatic cancer: a Phase I trial. Int J Radiat Oncol Biol Phys 54, 137-41.
123. Karasek, P., Skacel, T., Kocakova, I., Bednarik, O., Petruzelka, L., Melichar, B., Bustova, I., Spurny, V., and Trason, T. (2003). Gemcitabine monotherapy in patients with locally advanced or metastatic pancreatic cancer: a prospective observational study. Expert Opin Pharmacother 4, 581-6.
124. Crino, L., Mosconi, A. M., Calandri, C., Corgna, E., Porrozzi, S., Chiara, S., Nobili, M. T., and Tonato, M. (2001). Gemcitabine in advanced pancreatic cancer: a phase II trial. Am J Clin Oncol 24, 296-8.
125. Meyer, F., Hirbaschek, A., Kuhn, R., Eichelmann, K., Luck, A., Lippert, H., and Ridwelski, K. (2001). Antineoplastic effect and quality of life in patients with pancreatic carcinoma who received gemcitabine monotherapy. Viszeralchirurgie 36, 85-90.
126. Okada, S., Ueno, H., Okusaka, T., Ikeda, M., Furuse, J., and Maru, Y. (2001). Phase I trial of gemcitabine in patients with advanced pancreatic cancer. Jpn J Clin Oncol 31, 7-12.
127. Aykan, F., Argon, A., Alici, S., and Bulutlar, G. (2000). A phase II trial of gemcitabine in patients with advanced pancreatic cancer (APC) (abstract). Proc Am Soc Clin Oncol 19, 1241.
128. Kurtz, J. E., Trillet-Lenoir, V., Bugat, R., Negrier, S., Adenis, A., Kayitalire, L., Ripoche, V., and Dufour, P. (1999). Compassionate use of gemcitabine in advanced pancreatic cancer: a French multicentric study. Bull Cancer 86, 202-206.
129. Manzano, H., Esquerdo, G., Rita, J., Agutar, D., Arrivi, A., Pazo, R., Martin, J., and Terrasa, J. (1998). Clinical benefit with gemcitabine as first-line therapy for patients with advanced pancreatic cancer (abstract). Ann Oncol 9, 256.
130. Petrovic, Z., Jovic, J., Perisic, N., and Brocic-Pecelj, T. (1998). Monotherapy with gemcitabine in patients with advanced pancreatic carcinoma (abstract). Ann Oncol 9, 641.
131. Carmichael, J., Fink, U., Russell, R. C., Spittle, M. F., Harris, A. L., Spiessi, G., and Blatter, J. (1996). Phase II study of gemcitabine in patients with advanced pancreatic cancer. Br J Cancer 73, 101-5.
- 78 -
132. Casper, E. S., Green, M. R., Kelsen, D. P., Heelan, R. T., Brown, T. D., Flombaum, C. D., Trochanowski, B., and Tarassoff, P. G. (1994). Phase II trial of gemcitabine (2,2'-difluorodeoxycytidine) in patients with adenocarcinoma of the pancreas. Invest New Drugs 12, 29-34.
133. Heinemann, V., Quietzsch, D., Gieseler, F., Schonekas, H., Rost, A., Neuhaus, H., Clemens, M., Szymala, M., Hiddemann, W., Schalhorn, A., and Grosshadern, K. (2001). Gemcitabine plus cisplatin versus gemcitabine in advanced pancreatic cancer : preliminary results of a randomized phase III trial (abstract). Proc Am Soc Clin Oncol 20, 625.
134. Kaplan, E. L., and Meier, P. (1958). Nonparametric estimation of incomplete observations. J Am Stat Assoc 53, 457-481.
135. (1976). "ICD-O International classification of diseases for oncology," 1st. ed., Geneva.
136. (1990). "ICD-O International classification of diseases for oncology," 2nd. ed., Geneva.
137. Feinstein, A. R., Sosin, D. M., and Wells, C. K. (1985). The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 312, 1604-8.
138. Andre, T., Louvet, C., and de Gramont, A. (2004). Colon cancer: what is new in 2004? Bull Cancer 91, 75-80.
139. Schrag, D. (2004). The price tag on progress--chemotherapy for colorectal cancer. N Engl J Med 351, 317-9.
