par Vernerey, Déwi;André, Thierry;Mineur, Laurent;Chibaudel, Benoist;Benetkiewicz, Magdalena;Louvet, Christophe;Hammel, Pascal;Bonnetain, Franck;Huguet, Florence;Vienot, Angelique;Goldstein, David;Paget-Bailly, Sophie;Van Laethem, Jean-Luc ;Glimelius, Bengt;Artru, Pascal;Moore, Malcolm J
Référence British Journal of Cancer, 115, 3, page (281-289)
Publication Publié, 2016-07
Référence British Journal of Cancer, 115, 3, page (281-289)
Publication Publié, 2016-07
Article révisé par les pairs
Résumé : | Background: The management of locally advanced pancreatic cancer (LAPC) patients remains controversial. Better discrimination for overall survival (OS) at diagnosis is needed. We address this issue by developing and validating a prognostic nomogram and a score for OS in LAPC (PROLAP). Methods: Analyses were derived from 442 LAPC patients enrolled in the LAP07 trial. The prognostic ability of 30 baseline parameters was evaluated using univariate and multivariate Cox regression analyses. Performance assessment and internal validation of the final model were done with Harrell's C-index, calibration plot and bootstrap sample procedures. On the basis of the final model, a prognostic nomogram and a score were developed, and externally validated in 106 consecutive LAPC patients treated in Besançon Hospital, France. Results: Age, pain, tumour size, albumin and CA 19-9 were independent prognostic factors for OS. The final model had good calibration, acceptable discrimination (C-index=0.60) and robust internal validity. The PROLAP score has the potential to delineate three different prognosis groups with median OS of 15.4, 11.7 and 8.5 months (log-rank P<0.0001). The score ability to discriminate OS was externally confirmed in 63 (59%) patients with complete clinical data derived from a data set of 106 consecutive LAPC patients; median OS of 18.3, 14.1 and 7.6 months for the three groups (log-rank P<0.0001). Conclusions: The PROLAP nomogram and score can accurately predict OS before initiation of induction chemotherapy in LAPC-untreated patients. They may help to optimise clinical trials design and might offer the opportunity to define risk-Adapted strategies for LAPC management in the future. |