This retrospective study is the first to report real-life data from patients with platinum-sensitive recurrent non-gBRCA mutated ovarian cancer receiving niraparib as maintenance treatment. Compared with the NOVA trial, the time to first subsequent treatment was comparable, whereas the progression-free survival was somewhat shorter in our study5. We confirmed a clinically meaningful prolongation of the time to next treatment, as patients who had started chemotherapy at the end of follow-up had remained off chemotherapy for almost a year. Patients with normalized CA125 before niraparib treatment had an improved time to first subsequent treatment and progression-free survival of 5–6 months compared with patients without normalized CA125. Overall, we found a lower incidence of adverse events compared with the NOVA trial, especially with regard to hematologic toxicity.

The introduction of PARP inhibitors has been the most important development in the treatment of ovarian cancer in the last decade. The efficacy in our study is comparable with the NOVA trial, including a similar time to first subsequent treatment of 10.7 months compared with 11.8 months in NOVA. However, the comparison of time to first subsequent treatment is limited by the fact that 38% had not yet commenced chemotherapy at last follow-up. The observed median progression-free survival of 6.9 months is numerically shorter than the 9.3 months reported in NOVA. Thus far, only five reports on real-life data of PARP inhibitors have been published,22–26 but none of these included the present efficacy outcomes; only time to discontinuation,23 24 number of cycles administered,22 24 and progression-free survival in BRCA mutated and non-mutated platinum-sensitive disease (n=48)26 and platinum-sensitive and platinum-resistant recurrent disease (n=51)25 were reported.

Our study cohort is in many respects similar to the non-gBRCA group in the NOVA study.5 The age distribution is comparable, as well as the number of previous treatment lines. However, in the NOVA trial, 50% of the patients had complete response to chemotherapy before entering the study, while in our study only 11.7% had investigator-assessed complete response prior to niraparib treatment. This may explain the shorter progression-free survival observed in this study.

The striking difference in progression-free survival and time to first subsequent treatment depending on CA125 normalization before niraparib treatment may be useful when estimating the potential benefit for the patients, and to our knowledge has not been reported before. Patients without normalized CA125 should be followed closely for early progression to avoid excess toxicity. Studies to date have focused on potential differences dependent on complete versus partial response to platinum-based chemotherapy, but no clear pattern has emerged.27 Any response to platinum-based chemotherapy is a surrogate for PARP inhibitor response, and has been adapted as part of the indication for PARP inhibitors.

A shorter progression-free survival after chemotherapy preceding PARP inhibitor treatment has recently been observed in patients progressing on PARP inhibitors compared with placebo,28 with mechanisms of cross-resistance between PARP inhibitors and platinum-based chemotherapy.29 In this study, only 49% responded to subsequent platinum-based chemotherapy, indicating a need to develop successful maintenance therapy beyond re-exposure to PARP inhibitors.30 31

Compared with the NOVA study, our study shows an overall lower incidence of adverse events, with fewer grade 3–4 adverse events (43% vs 74% in the NOVA study). Among these, anemia (16%), thrombocytopenia (13%), and neutropenia (7%) were most frequent, compared with 25%, 34%, and 20%, respectively in NOVA. However, we found a slightly higher rate of hypertension, both for any grade (23% vs 19%) and grade 3–4 (11% vs 8%). In this study, the number of dose interruptions (39) and dose reductions (46) of the patients had dose- interruptions and reductions, respectively, lower than those reported in NOVA, but higher than in other real-word studies.22 Still, the number of patients discontinuing niraparib owing to an adverse event was similar: 13.8% versus 14.7% in the NOVA study.

Aiming for minimization of treatment-related toxicity, individualized starting dose is essential. Patients with individualized starting dose had significantly fewer dose reductions than patients with non-individualized starting dose, in line with the post hoc analysis of the NOVA trial. At enrollment in this patient access program, the general recommendation for individualized starting dose based on weight was <77 kg. However, the initial Summary of Product Characteristics used 58 kg as the weight limit for dose reduction to 200 mg/day, instead of 77 kg, which may have contributed to deviations from individualized starting dose recommendations in this study.

Real-life data are important for evaluating whether the benefit of new anti-neoplastic agents observed in phase III trials can be translated into clinical practice. Clinical trials are restrictive in their eligibility criteria, and real-life data may therefore be helpful to estimate the benefit in an unselected population and in counseling the individual patient. The higher prevalence of co-morbidity in our study is in line with other real-life reports, and allow conclusions regarding tolerability in a less strictly selected population compared with a clinical trial. Another strength is the large number of patients included, as well as the complete follow-up of all patients.

A limitation of this study is the lack of data on sBRCA and HRD status, positively influencing the endpoints.32 In line with the procedures of the NOVA trial,5 only gBRCA testing was performed at our institutions at the time of enrollment. Since then, recommendations for BRCA and HRD testing have been updated.33 Furthermore, the study had no control group and lacked retrospective radiologic response re-evaluation.