Center for Global Research Data

Celgene: Using generalized pairwise comparisons to assess the benefit/risk balance of new treatments for patients with metastatic pancreatic cancer

Julien Péron

Hospices Civils de Lyon

None

None

6/17/2016

Background: We sought to assess the benefit-risk balance of nab-paclitaxel combined with gemcitabine versus gemcitabine alone in patients with metastatic pancreatic adenocarcinoma.Methods We will use generalized pairwise comparisons. Generalized pairwise comparisons are useful to perform a quantitative assessment of the benefit-risk balance of new treatments. It provides a clinically intuitive way of comparing patients with respect to all important efficacy and toxicity outcomes. This statistical method permits the simultaneous analysis of several prioritized outcome measures. Here, the first priority outcome will be survival time (OS). Differences in OS that exceeding two months will be considered clinically relevant. The second priority outcome will be toxicity. The overall treatment effect will be quantified using the chance of a better outcome with nab-paclitaxel, which can be interpreted as the net probability for a random patient treated in the nab-paclitaxel group to have a better overall outcome than a random patient in the gemcitabine alone group.

Background: Traditional statistical tests can be used to compare two groups of subjects in terms of simple outcomes, such as response to treatment, duration of survival, adverse events rate etc. The method proposed here fundamentally differs from such traditional tests in several important ways. In particular, it makes it possible to compare two samples in terms of several outcome measures simultaneously, as long as these outcome measures can be prioritized (Buyse 2010, Péron 2015). The method might be used to perform an assessment of the benefit-risk ratio of interventions in randomized trials. Of note, generalized pairwise comparisons are equivalent to traditional tests for simple outcome measures, such as Fisher’s exact test for binary outcomes, the Wilcoxon’s test for continuous outcomes, and Gehan’s generalized Wilcoxon test for time to event outcomes. Purpose of project: The purpose of the present project is to use generalized pairwise comparisons to analyse data from randomized clinical trials evaluating innovative chemotherapy combinations for metastatic pancreatic cancer. One trial compared nab-paclitaxel plus gemcitabine with gemcitabine alone as first-line therapy in patients with metastatic pancreatic cancer (Von Hoff, 2013). A second trial compared a combination chemotherapy regimen consisting of oxaliplatin, irinotecan, fluorouracil, and leucovorin (FOLFIRINOX) with gemcitabine alone in the same patient population (Conroy, 2011). The two innovative chemotherapy regimens significantly improved patient outcomes. The larger benefit was shown with the FOLFIRINOX combination, but this regimen also had significant chemotherapy-related toxicities.Hence the simultaneous analysis of Overall Survival (OS) and adverse events rate should allow a standardized assessment of the benefit/risk balance of the two innovative regimens. For the main analysis the first priority outcome will be OS. Only pairs of patients with differences in OS exceeding two months will be considered informative, because smaller differences in OS will not be considered clinically meaningful. The second priority outcome will be treatment-related AEs, with patients experiencing the lower grade related AE considered to have had a more favorable outcome. The impact of the choice of outcomes, thresholds, and priority on the results will be assessed in sensitivity analyses.Methods: All pairwise comparisons will be performed to estimate the proportion in favor of the innovative treatment as compared with gemcitabine with respect to the following prioritized outcomes:1. Death 2. Treatment-related adverse eventsA randomization test will be used to estimate the confidence interval of this proportion, and its statistical significance. Communication plan:The benefit-risk assessment of nab-paclitaxel in combination with gemcitabine will be published in a medical oncology journal. The targeted journal is Annals of Oncology.

All the 861 patients randomly assigned to nab-paclitaxel plus gemcitabine or gemcitabine alone will be included in this post-hoc study.Data required: The following data items should be available for all individual patients randomized in the NCT00844649 trial (codes are provided for the sake of convenience; any other codes may be used. Moreover, the full database of the trial can be provided if this proves to be less time consuming): BASELINE1. Institution ID2. Patient ID3. Date of Randomization4. Stratification factors used in the randomization (if any)5. Treatment groupSURVIVAL1. Date of last observation (dd/mm/yyyy)2. Overall survival status at the date of last observation (0=alive; 1=death; 2=lost to follow-up)3. Cause of death (0=alive; 1=non disease, non toxicity; 2=disease related, non toxicity; 3= toxicity related; 9=unknown)RESPONSE AND PROGRESSION1. Date of tumor response (dd/mm/yyyy)2. Tumor response, confirmed (0=CR; 1=PR; 2=SD; 4=PD; 9=unknown)3. Date of progression (or date of last observation if patient did not progress, dd/mm/yyyy)4. Progression indicator (0=no progression; 1=progression)TOXICITIES1. Treatment-related adverse event higher grade2. Nature of the higher grade treatment-related adverse event3. Adverse event higher grade4. Nature of the higher grade adverse event

The benefit-risk assessment of nab-paclitaxel in combination with gemcitabine will be published in a medical oncology journal. The targeted journal is Annals of Oncology. The expected date of publication is december 2016.

Marc Buyse, IDDI consulting

None

Primary endpoint: proportion in favor of the nab-paclitaxel plus gemcitabine treatment with respect to the following prioritized outcomes: 1. Death 2. Treatment-related adverse eventsDifferences in OS that exceeding two months will be considered clinically relevant.A randomization test will be used to estimate the confidence interval of this proportion, and its statistical significance. Secondary endpoints : Several threshold values for clinically relevant differences in OS will be evaluated as a sensitivity analysis. Progression-free survival will be included among prioritized outcomes as another sensitivity analysis.

The publication citation will be added after the research is published.