673P - Development and validation of a predictive model for disease control (DC) at first disease assessment, among patients (pts) with metastatic squamou...

Date 29 September 2014
Event ESMO 2014
Session Poster Display session
Topics Anti-Cancer Agents & Biologic Therapy
Oesophageal Cancer
Presenter Nuria Kotecki
Citation Annals of Oncology (2014) 25 (suppl_4): iv210-iv253. 10.1093/annonc/mdu334
Authors N. Kotecki1, E. Tresch2, N. Penel3, S. Hiret4, P.L. Etienne5, E. Francois6, M. Galais7, A. Ben Abdelghani8, P. Michel9, L. Dahan10, F. Ghiringhelli11, L. Bedenne12, E. Samalin13, G. Piessen14, J. Bennouna15, C. Peugniez3, F. El Hajbi1, A. Kramar16, C. Mariette17, A. Adenis1
  • 1Gastrointestinal Oncology Department, Centre Oscar Lambret, 59020 - Lille/FR
  • 2Biostatistics Unit, Centre Oscar Lambret, 59000 - Lille/FR
  • 3Medical Oncology, Centre Oscar Lambret, 59000 - Lille/FR
  • 4Oncologie, ICO site René Gauducheau, ST HERBLAIN/FR
  • 5Oncologie Medicale, Clinique Armoricaine de Radiologie, St Brieuc/FR
  • 6Medical Oncology, Antoine Lacassagne Cancer Institute, Nice/FR
  • 7Cancerologie Digestive, CRLCC FRANCOIS BACLESSE, Caen/FR
  • 8Oncology, Centre Paul Strauss, Strasbourg/FR
  • 9Gastroenterology And Digestive Oncology, CHU de Rouen, 76000 - Rouen/FR
  • 10Digestive Oncology, APHM, Marseille/FR
  • 11Oncology, CLCC GF Leclerc, Dijon/FR
  • 12Hepatogastroenterology Department, University Hospital, 21079 - DIJON Cedex/FR
  • 13Gastroenterologie, Val d'Aurelle CRLC Centre Régional de Lutte contre le Cancer, Montpellier/FR
  • 14Department Of Digestive And Oncological Surgery, Claude Huriez University, Lille/FR
  • 15Department Of Medical Oncology, Institut de Cancérologie de l'Ouest, 44805 - Saint-Herblain cedex/FR
  • 16Biostatistics, Centre Oscar Lambret, 59000 - Lille/FR
  • 17Department Of Digestive And Oncological Surgery, Claude Huriez University, 59037 - Lille/FR

Abstract

Aim

There is little evidence that CT has any impact on outcome of pts with MSCC. This analysis was aimed to identify predictors for DC at first disease assessment, in MSCC pts receiving first-line CT.

Methods

In the development cohort (68 pts from the multicentric and ongoing E-Dis trial [NCT01248299]), we have developed predictors for DC using a logistic regression model. Then, we applied the predictive model to an independent retrospective cohort of 60 consecutive pts from one center (Lille), and assessed Sensitivity (Se), Specificity (Sp), Positive Predictive Value (PPV), Negative Predictive Value (NPV), Accuracy (A), Discriminative slope (DS) and the calibration with the Brier score (BS).

Results

In the development cohort, gender (women), ECOG 0-1, BMI ≥ 18.5, no bone metastasis, albumin ≥ 35g/l and hemoglobin ≥ LLN were associated with DC. Using logistic regression we identified the following 3 prognosticators: BMI ≥ 18.5 (OR 4.5; 95% CI: 0.91-22.5), no bone metastasis (OR 4.6; 95% CI: 0.91-23.2), albumin ≥ 35 g/l (OR 3.5; 95% CI: 1-12.1). Based on the presence or absence of these 3 independent prognosticators, we have built a predictive model using a score from 0 to 3. 64/68 pts were assessable. The DC rate was 64% (41/64), 14.3% (2/14), and 78% (39/50), in the overall population, when score = 0-1, and when score = 2-3, respectively. In the validation cohort, 51/60 patients were assessable. The DC rate was 39.2% (20/51), 12.5% (1/8), and 44.2% (19/43) in the overall population, when score = 0-1, and when score = 2-3, respectively. Se, Sp, PPV, NPV, A, and DS were 95%, 22%, 44%, 88%, 51% and 31.5%, respectively. Model calibration was good since the BS was very low (0.203). Our score was also predictive for death within 90 days (data not shown).

Conclusions

We have developed and validated a predictive score for DC at first disease assessment, which includes BMI assessment, bone metastasis status, and albumin level for pts with MSCC receiving first-line CT.

Disclosure

E. Francois: Participation to boards : Roche, Merck, Sanofi, Novartis, Celgène; A. Adenis: Bayer (advisory board, research funding). All other authors have declared no conflicts of interest.