The inclusion criteria were revised to include taxane and anthracycline as previous (neo-)adjuvant cytotoxic chemotherapy.

Central testing of gBRCAmut was limited to Myriad only, & all participants were required to have blood samples taken for gBRCAmut testing to determine eligibility. Inclusion criteria were revised to clarify inclusion of participants who were previously treated with platinum; & direct bilirubin was no longer an inclusion criterion. Maximum period of 90 days was permitted from registration to randomization. Further, dose escalation was not allowed. “Treatment related” nonhematologic AEs was added as criteria for treatment interruption. Following dose interruption, time to restart of treatment was extended to 28 days. Thrombocytopenia was specified as one of the key parameters for monitoring hematologic toxicity. For the management of hematologic toxicities, guidelines based on platelet, neutrophil and hemoglobin counts were provided, administration of granulocyte colony-stimulating factor (GCSF) was allowed, and secondary prophylaxis was only allowed for participants in the physician’s choice treatment arm. Guidelines for hematology testing during treatment cycles were provided, and time points for prothrombin time and serum chemistry testing were provided. For the measurement of tumors, chest X-ray was no longer used as a method for measuring tumor lesions. Further, cytology and positron emission tomography were not necessary methods for assessment of residual lesions and, therefore, were removed in Appendix H, the list of drugs known to inhibit or induce cytochrome P450 1A2 (CYP1A2) enzyme was updated.

The study population was revised to allow participants with a deleterious or suspected deleterious germline BRCA1 or BRCA2 mutation to be enrolled into the study. Any participant that did not have a gBRCAmut per central laboratory results, was allowed to continue based on his/her physician discretion and preference. The ITT population was defined as all randomized participants with a central confirmation of gBRCAmut. Few changes were made to the inclusion criteria as follows: Palliative radiotherapy within 3 weeks prior to the start of study was added. The inclusion criterion related to the definition of participants with platinum resistance was revised. Exclusion of participant with platinum resistance was made. For participants with a prior history of ovarian cancer who had peritoneal disease and elevated serum CA-125, a biopsy of the peritoneal disease was required to be enrolled. Exclusion of participants with contraindication to all of the comparator treatments or who received a platelet transfusion within 4 weeks of the first dose of study treatment was added. Niraparib dose modification was revised to be interrupted and/or reduced at any time as considered intolerable by the participant. Missed doses of niraparib were not allowed to be taken at a later date. Information for prophylactic platelet transfusion per platelet count & anticoagulation or antiplatelet concomitant treatment was provided. The primary analysis population for efficacy was modified to constitute all randomized participant who have a gBRCAmut per central laboratory testing. Only 1 interim analysis for futility was revised and planned after 93 (40%) PFS events instead of 2 analyses (35 [15%] and 96 [40%] events).

Two new secondary objectives were added (the first was for the relationship between cytogenetic abnormalities and safety parameters; the second was for descriptive summary statistics of post-treatment data). Inclusion criteria related to persistent Grade >=3 hematologic toxicity or fatigue from prior cancer therapy were provided. In addition, cytogenetic and mutational analysis for myeloid-related genes analysis at Screening and at the end of treatment was required. Exclusion criterion related to history of myelodysplastic syndrome (MDS) was provided. If treatment was resumed for participants with platelets or red blood cells transfusion or hematopoietic growth factor support, dose reduction was required. Involvement of a hematologist was required for participants who had more than 2 transfusions in the absence of non–treatment-related causes or if the treatment-related hematologic toxicities have not recovered to allow retreatment with niraparib after 4 weeks. In addition, if a diagnosis of MDS/ acute myeloid leukemia (AML) was confirmed by a hematologist, permanent discontinuation of niraparib treatment was required. In the physician’s choice treatment arm, assessment of the tumors was revised. Assessment during follow-up visits was updated with collection of information related to new malignancy. In addition, mutational profile testing (mutations of selected myeloid-associated genes) for any highly clinically suspected MDS/AML cases during survival follow-up visits was added to the list of measurement. Evaluation of safety was updated for collection of SAEs/deaths (restricted to at least “likely related” SAEs and “related” death events) after 30 days of the last dose or the initiation of new anticancer therapy (whichever was earlier). Definition of SAEs was updated, and inclusion of any new malignancy at any time for the duration of the study was considered SAEs.

The relationship between cytogenetic abnormalities and safety parameters was no longer one of the secondary objectives. The inclusion criterion related to definition of hormone receptor–positive participants who had hormone-resistant disease was provided; & serum creatinine was revised to >=50 mL/min. Mutational profile testing (for selected myeloid-related genes) was restricted for participants who developed MDS or AML during the study or the follow-up period. Detailed plans for niraparib-related dose modification based on hematologic and nonhematologic toxicity parameters were provided. In the physician’s choice treatment arm, treatment interruption period was extended from 21 days to 28 days. Evaluation of response was revised, and restriction to participants with measurable disease at baseline was removed. In the statistical design, the original final progression-free survival (PFS) analysis of 232 PFS events was revised to occur at 137 PFS events or end of recruitment, whichever occurred later. The power was reduced to 80.0% (with a 1-sided alpha of 0.025) to detect a hazard ratio of 0.6 (equivalent to 3 to 5 months in median PFS [it was 3 to 6 months in the previous version]). A gatekeeping strategy was planned to test PFS and OS. The primary analysis of PFS & overall survival (OS) was determined to use stratified log-rank test, stratifying by the randomization strata. A non-stratified log-rank test was indicated to assess the robustness of the primary results. The PFS events for futility interim analysis was corrected to 68% (from 40%). A gamma family beta spending function with a nonbinding gamma (γ=-5) stopping boundary was indicated for the futility analysis.