Genomic instability score (GIS) and real-world outcomes in patients (pts) with advanced ovarian cancer (aOC) using a U.S. health database.

Abstract

e17565 Background: Identification of homologous recombination deficiency biomarkers are needed to determine which pts with aOC will most likely derive benefit from poly(ADP-ribose) polymerase inhibitor (PARPi) maintenance treatment (tx). Studies of various GIS cutoffs have shown an association with improved outcomes (current standard cutoff, ≥42). Real-world data is needed to explore associations between GIS cutoffs and real-world outcomes. We evaluated the association of GIS with real-world progression-free survival (rwPFS) and time to next tx (TTNT) in pts with aOC treated with PARPi as first-line maintenance (1LM). Methods: The Flatiron Health database was used to retrospectively evaluate eligible adults with aOC who received 1L platinum-based chemotherapy followed by 1LM PARPi between 01Jan2017 and 31Mar2025. Pts were followed from index (start of 1LM PARPi) to earliest of death, loss to follow-up, or study end. Associations between GIS cutoffs (≥33/<33, ≥42/<42, ≥60/<60, GIS-low/medium/high, and unadjusted thresholds for every cutoff from 25 to 70) and rwPFS (time from index to disease progression or death) or TTNT (time from index to start of any second-line tx or death) were estimated per Kaplan Meier and Cox regression (unadjusted and adjusted for demographic and clinical characteristics) analyses. Results: The analysis included 121 pts; most had serous histology (84%), had BRCA wild-type aOC (83%), and received care in a community setting (82%). For each GIS cutoff, pts with a higher GIS had a longer median rwPFS and TTNT (Table). In unadjusted threshold analyses of cutoffs from 25 to 70, every cutoff from 26 to 63 was associated with significant clinical benefit for both rwPFS and TTNT, with the strongest magnitude of association at GIS 41/42 for rwPFS and 42 for TTNT. As only 44% of pts had progression data, adjusted Cox regression for rwPFS was not performed. In adjusted Cox models for TTNT, pts with higher vs lower GIS for each cutoff had a longer TTNT (Table); the strongest association was at GIS cutoff 42. Conclusions: Higher GIS at any cutoff was associated with improved rwPFS and TTNT in pts treated with a 1LM PARPi. GIS cutoff ≥33 showed clinical benefit, with GIS cutoff ≥42 showing the greatest magnitude of rwPFS and TTNT benefit across analyses. GIS cutoff: <33 a ≥33 <42 a ≥42 <60 a ≥60 rwPFS n 23 30 26 27 36 17 Median (95% CI), mo 9.4 (4.2–11.3) 26.1 (11.6–NE) 10.3 (5.6–11.5) 30.4 (13.2–NE) 11.3 (9.4–13.8) NE (10.0–NE) Unadjusted HR (95% CI) 0.25 (0.12–0.53) b 0.21 (0.10–0.47) c 0.36 (0.15–0.88) d TTNT n 64 57 76 45 93 28 Median (95% CI), mo 11.0 (8.2–14.2) 26.2 (12.3–NE) 10.7 (8.2–12.9) NE (21.9–NE) 12.2 (9.9–14.7) NE (17.7–NE) Unadjusted HR (95% CI) 0.43 (0.26–0.71) e 0.26 (0.14–0.47) c 0.36 (0.18–0.72) e Adjusted HR (95% CI) 0.46 (0.26–0.79) e 0.22 (0.11–0.43) c 0.32 (0.15–0.70) e a Ref for HR comparison. b P ≤ 0.001; c P ≤ 0.0001; d P ≤ 0.05; e P ≤ 0.01. HR, hazard ratio; NE, not estimable.