Impact of prior irinotecan exposure on outcomes of metastatic pancreatic cancer (mPC) patients.

Abstract

667 Background: Published data suggests prior exposure to irinotecan infers a lower likelihood of benefit to liposomal irinotecan. This analysis seeks to expand this hypothesis by evaluating U.S. patterns of care to understand how prior irinotecan therapy impacts outcomes in mPC. Methods: Using the Flatiron Health database, data were extracted and analyzed for treated mPC patients (pts) in the 2L+ setting between Jan 1, 2014 and Jun 30,2019. Therapies of interest included: gemcitabine/ nab-paclitaxel (GnP), FOLFOX, FOLFIRI, FOLFIRINOX (FFX), and liposomal irinotecan/5-FU/LV (nal-IRI). The reference date for each treatment group was the date of treatment initiation. Prior irinotecan was defined as any irinotecan given in a prior regimen in mPC diagnosis. Cox proportional hazard (PH) methods were used to calculate mortality hazard ratios (HRs). HRs were adjusted to account for demographics and relevant covariates. Pts with prior exposure to irinotecan were used as the reference population for the Cox PH model (an HR < 1 represents worse survival for exposed pts relative to the unexposed). Results: N = 1,978 were included in this analysis. The median age at treatment initiation, and the proportion of pts previously treated with irinotecan are reported in table. Crude mortality was: GnP pts, HR 0.93 [95% CI: 0.77 – 1.11, adjusted HR, 0.94, 0.76 – 1.15]; nal-IRI pts, HR 0.81 [0.64 – 1.02, adjusted HR: 0.89, 0.67 – 1.19]; HR for FOLFOX was 0.55 [0.38 – 0.78, adjusted HR: 0.51, 0.33 – 0.79]. HRs are not reported for FFX and FOLFIRI due to the small numbers with prior irinotecan exposure. Conclusions: In mPC, prior irinotecan treatment may not preclude benefit from later treatment with nal-IRI or GnP as can be seen from the adjusted and unadjusted HRs. These findings are hypothesis-generating and need to be considered in the context of wide CI’s, retrospective nature and the limitations of such data. Further study is required to understand the less-favorable signal observed with FOLFOX and prior irinotecan.[Table: see text]