Comparison of neoadjuvant and adjuvant therapy for resectable pancreatic cancer using Markov decision modeling.

Abstract

448 Background: Meta-analysis of smaller studies have shown that neoadjuvant chemotherapy is more beneficial for patients with resectable pancreatic cancer than upfront surgery by comparing life expectancy (LE) and quality-adjusted life expectancy (QALE) computed from Markov models. The study results utilized literature data using several small clinical trials but no individual patient data was used and only gemzar based therapy was studied. Methods: Markov model was used to calculate the LE and QALE for adjuvant and neoadjuvant chemotherapy and individual patient parameters was used in the model to refine certain clinical outcome datapoints. We used 278 patients pancreatic cancer data from 2008 to 2017 from Stony Brook University and used the literature data from randomized clinical trials studying gemzar (GEM), gemzar and capecitabine (GEM+CAP) and modified FOLFIRINOX (mFOL). The median OS for each model was obtained by computer simulation. Results: Intensive adjuvant chemotherapy using mFOL had best simulation outcome with median OS (52.5 months), LE (81.5 months), and QALE (65.0 quality-adjusted life months) compared to using GEM (40.5, 66.5, and 52.9 months for median OS, LE, and QALE), GEM+CAP (16.5, 28.0, and 21.9 months for median OS, LE, and QALE), and 5-FU (16.5, 26.9, and 21.1 months for median OS, LE, and QALE). The neoadjuvant chemotherapy approach improved LE and QALE but not in median OS when compared to adjuvant therapy. Conclusions: Mathematical modeling confirms the improved clinical outcome for modified FOLFIRINOX in resectable pancreatic cancer. The benefit of neoadjuvant chemotherapy approach suggest further clinical trials are needed to determine the better treatment strategy for pancreatic cancer patients.