Background: mPDAC is a deadly disease with dismal survival. Newer systemic therapies including gemcitabine/nab-paclitaxel (GA) and FOLFIRINOX (FF) have resulted in modest increases in median overall survival(mOS). Currently, no good predictors exist to prioritize the use of one regimen over the other. We investigated imaging patterns in patients receiving these two regimens. Methods: A single institution retrospective analysis of pts with mPDAC receiving either GA or FF as the front-line therapy between Jan 2012 - Dec 2014 was conducted. Data included demographics and systemic therapy.Baseline CT evaluated six imaging features; lung metastases, liver metastases, peripancreatic adenopathy, retroperitoneal adenopathy, mesentery/omentum infiltration and ascites. Statistical analysis included Kaplan Meier survival analysis with Chi-square test to compare imaging features that predict mOS or median response duration (mRD). Results: N = 27 pts; median age:GA-66 years( range: 52-76), FF-55 years (range: 40-67); 13 pts (48 %) were women and 24 pts had an ECOG performance status of ≤ 1. 18 pts received GA and 9 received FF. mOS with GA = 6.1 months(m) ( range: 1.9-15.7) and FF = 9.9 m (range: 2.5-24.1). 5/18 (28%) of the GA group and 7/9(78%) of the FF group received subsequent therapies. The presence of peripancreatic adenopathy in GA pts correlated with prolonged mRD (2.2 vs. 0.6 m; p < 0.001) without statistical difference in mOS (8.5 vs. 2.9 m; p = 0.4). In pts receiving FF the absence of hepatic metastasis resulted in a significant mRD (2.3 vs. 0.9 m; p = 0.01) and mOS (7.9 vs. 1.9 m; p = 0.04). The remainder of the radiological features did not show any statistically significant difference. Conclusions: In patients with mPDAC receiving GA or FF in front-line therapy; we observed that metastasis to the liver predicts worse outcome with FF than GA. Interestingly, peripancreatic lymphadenopathy was predictive of longer mRD for GA pts, without difference in mOS. Our pilot data demonstrates the potential of using CT imaging features to predict likelihood of response to different chemotherapy regimens. A larger study is needed to confirm these findings.