Introduction/Background We sought to identify predictors of Dose-Limiting Adverse Events (DLAE) (adverse events (AE) leading to dose reduction or discontinuation) in patients who received standard of care PARP inhibitors (PARPi) maintenance therapy and the impact of toxicities on overall survival (OS) in advanced ovarian cancer (aOC).

Methodology Retrospective data collection was performed for patients (newly diagnosed or recurrent) who received at least one dose of maintenance Olaparib or Niraparib between April/2015-November/2021, at the Royal Marsden, UK. Pearson’s Chi-square and Log rank Kaplan Meier tests were used for categorical and continuous variables, respectively. Logistic regression was used to predict DLAE; Cox regression for OS. STATA SE 17.0 was used for statistical analysis.

Results 160 patients (median age 62.5 years, 41% (66/160) first-line, 49% (79/160) BRCA-mutated; median follow up on PARPi of 18.7 months; 68/160 were deceased at data cut-off) were included. DLAE were reported in 46.2% (74/160). Grade (G) 2 and G3 AE led to DLAE in 52.7% (39/74) and 32.4% (24/74) of cases, respectively. 78.2% (140/179) ≥G2 AEs occurred during the first 3 months. Hypertension (OR 2.6, p=0.03), upfront surgery (OR 2.7, p=0.01), previous G2 AE on chemotherapy (OR 1.8, p=0.01), residual disease (OR 2.4, p=0.04), and creatinine clearance<60 ml/min (OR 3.5, p=0.01) predicted higher risk of DLAE. HRD (OR 0.4, p=0.04), and Niraparib at 200 mg (OR 0.4, p<0.001) predicted lower risk of DLAE. G3/G4 hematological AE predicted better PFS at 24 months (OR 0.4, p=0.047). ≥G2 AEs in the first 3 months predicted better 5-year OS from diagnosis (OR 0.4, p=0.005) for the overall population. Dose reductions did not impact on OS (p=0.65).

Conclusion This is the first real-world data analysis suggesting that the development of early PARPi toxicities predicts improved 5-year OS in aOC. This model warrants further validation in prospective cohorts.