Résumé : Background: Positive findings from the phase III MPACT trial led to the regulatory approval of nab-paclitaxel plus gemcitabine as a treatment option for patients with metastatic pancreatic cancer. This report is an update of overall survival (OS) based on longer follow-up. Methods: Patients (n = 861) with metastatic pancreatic cancer and a Karnofsky performance status of 70 or greater were randomly assigned one to one to receive nab-paclitaxel + gemcitabine or gemcitabine alone. Efficacy data for this post hoc analysis were collected through May 9, 2013. Exploratory analyses of carbohydrate antigen 19-9 (CA19-9) and neutrophil-to-lymphocyte ratio (NLR) were conducted. The primary efficacy endpoint was OS, which was analyzed for all randomly assigned patients by the Kaplan-Meier method. All statistical tests were two-sided. Results: The median OS was statistically significantly longer for nab-paclitaxel plus gemcitabine vs gemcitabine alone (8.7 vs 6.6 months, hazard ratio [HR] = 0.72, 95% confidence interval [CI] = 0.62 to 0.83, P <. 001). Long-term (>three-year) survivors were identified in the nab-paclitaxel plus gemcitabine arm only (4%). In pooled treatment arm analyses, higher CA19-9 level and NLR at baseline were statistically significantly associated with worse OS. There appeared to be a treatment effect for OS favoring nab-paclitaxel plus gemcitabine over gemcitabine alone in poor-prognosis subgroups defined by these factors (HR = 0.612, P <. 001 for CA19-9 level ≥ median and HR = 0.81, P =. 079 for NLR > 5). Conclusions: These data confirm and extend the primary report of OS, supporting the superior efficacy of nab-paclitaxel plus gemcitabine over gemcitabine alone. Subgroup analyses support the relevance of CA 19-9 and NLR as prognostic markers in metastatic pancreatic cancer.