4621 Background: The POLO study (NCT02184195) showed that mPaC patients (pts) with a deleterious or suspected deleterious gBRCAm, and whose disease had not progressed during ≥16 weeks of first-line platinum-based chemotherapy, had significantly longer progression-free survival (PFS, primary endpoint) with maintenance olaparib vs placebo: median 7.4 vs 3.8 months, hazard ratio (HR) 0.53; P=0.004. PFS benefit was observed in pts with g BRCA1m (HR 0.40) and g BRCA2m (HR 0.63). The POLO study represents the largest BRCAm prevalence study in pancreatic cancer. We report additional exploratory analysis to further characterize patient gBRCAm profiles, including the relationship with efficacy. Methods: Pts were enrolled based on either a previously identified gBRCAm status from a local test result and subsequently confirmed by central testing, or a prospectively identified gBRCAm. Pts received maintenance olaparib 300 mg twice daily (tablet) or placebo. PFS was assessed by blinded independent central review (modified RECIST v1.1). Results: Of 3194 prospectively screened pts, a valid BRCA test result was obtained for 3175 (99%) from 12 countries; gBRCAm prevalence was 6.2% in pts not previously known to harbor a gBRCAm (196/3175; 1.6% g BRCA1m, 4.5% g BRCA2m). In countries (n=8) with >100 pts prospectively tested, highest gBRCAm prevalence was 9.2% (USA) and lowest 4.0% (Spain). Prevalence by race (>100 pts); 6.4% Caucasian, 4.6% Asian. In total, 154 pts with a gBRCAm satisfied all eligibility criteria and were randomized (106 prospectively tested and 48 by local test [44/48 subsequently confirmed by Myriad testing]). 37/154 (24%) randomized pts carried a common Ashkenazi Jewish founder mutation, the majority being from Israel (21 pts). From a total of 151 variants, frameshift mutations were most frequent (g BRCA1m 69.6%, g BRCA2m 71.4%) followed by nonsense mutations (g BRCA1m 6.5%, g BRCA2m 17.1%). The efficacy (PFS) of olaparib vs placebo in the different subgroups are shown in the table. Conclusions: In pts with mPaC enrolled in POLO, g BRCA2m were more prevalent than g BRCA1m and mutation type was predominantly frameshift. PFS benefit was consistent across a heterogenous spectrum of gBRCAm and with the previously reported full analysis set. Clinical trial information: NCT02184195 . [Table: see text]