Olaparib as Treatment Versus Nonplatinum Chemotherapy in Patients With Platinum-Sensitive Relapsed Ovarian Cancer: Phase III SOLO3 Study Final Overall Survival Results

Abstract

Olaparib treatment significantly improved objective response rate (primary end point) and progression-free survival versus nonplatinum chemotherapy in patients with BRCA-mutated platinum-sensitive relapsed ovarian cancer in the open-label phase III SOLO3 trial (ClinicalTrials.gov identifier: NCT02282020). We report final overall survival (OS; prespecified secondary end point), post hoc OS analysis by number of previous chemotherapy lines, and exploratory BRCA reversion mutation analysis. Two hundred sixty-six patients were randomly assigned 2:1 to olaparib tablets (300 mg twice daily; n = 178) or physician's choice of single-agent nonplatinum chemotherapy (pegylated liposomal doxorubicin, paclitaxel, gemcitabine, or topotecan; n = 88). OS was similar with olaparib versus chemotherapy (hazard ratio [HR], 1.07 [95% CI, 0.76 to 1.49]; P = .71, median 34.9 and 32.9 months, respectively, full analysis set). OS with olaparib was favorable in patients with two previous chemotherapy lines (HR, 0.83 [olaparib v chemotherapy] [95% CI, 0.51 to 1.38]; median 37.9 v 28.8 months); however, a potential detrimental effect was seen in patients with at least three previous chemotherapy lines (HR, 1.33 [95% CI, 0.84 to 2.18]; median 29.9 v 39.4 months). BRCA reversion mutations might have contributed to this finding. No patient randomly assigned to olaparib with a BRCA reversion mutation detected at baseline (6 of 170 [3.5%]) achieved an objective tumor response.

FIG 1.

Kaplan-Meier estimates of OS. The median (range) treatment duration in months was 13.1 (0.1-67.5, olaparib), 6.0 (0.9-15.4, pegylated liposomal doxorubicin), 5.1 (1.8-20.0, paclitaxel), 3.3 (0.7-14.3, gemcitabine), and 6.2 (2.3-9.7, topotecan) (safety analysis set). One hundred and nineteen (66.9%, olaparib) and 54 (61.4%, chemotherapy) patients received subsequent anticancer therapy. The P value presented for OS is for descriptive purposes only since the trial protocol only allowed a formal type 1 controlled test of OS if PFS2 was statistically significant. HR, hazard ratio; OS, overall survival; PFS2, time to second progression or death.

FIG 2.

Summary of post hoc subgroup results. (A) OS, PFS, and ORR in patients with 2, 3, ≥3, and ≥4 previous chemotherapy lines. Post hoc analysis by the number of previous chemotherapy lines was based on the final DCO (April 16, 2021) for OS and the primary DCO (October 10, 2018) for PFS and ORR. The analysis in all patients was performed using a stratified log-rank test with factors as recorded in the Interactive Voice Response System for time to disease progression after the end of last PBC (6-12 v >12 months) in the full analysis set (for OS and PFS) and in patients with measurable disease at baseline (for ORR). The analysis in the previous line of chemotherapy subgroups was performed using a single Cox proportional hazards model containing the treatment term, the subgroup covariate of interest, and the treatment by subgroup interaction for each subgroup. Size of circle is proportional to the number of events. Gray shading represents the 95% CI for the overall (all patients) HR; unconfirmed ORR is based on blinded independent central review in the measurable disease population. For OS, the number of patients with events/total number of patients in population was 116/178 (65.2%) for olaparib and 46/88 (52.3%) for chemotherapy in all patients, 53/88 (60.2%) for olaparib and 23/46 (50.0%) for chemotherapy in the two previous lines of chemotherapy group, 31/42 (73.8%) for olaparib and 15/25 (60.0%) for chemotherapy in the three previous lines of chemotherapy group, 63/90 (70.0%) for olaparib and 23/42 (54.8%) for chemotherapy in the ≥3 previous lines of chemotherapy group, and 32/48 (66.7%) for olaparib and 8/17 (47.1%) for chemotherapy in the ≥4 previous lines of chemotherapy group. For PFS, the number of patients with events/total number of patients in population was 110/178 (61.8%) for olaparib and 49/88 (55.7%) for chemotherapy in all patients, 47/88 (53.4%) for olaparib and 27/46 (58.7%) for chemotherapy in the two previous lines of chemotherapy group, 27/42 (64.3%) for olaparib and 18/25 (72.0%) for chemotherapy in the three previous lines of chemotherapy group, 63/90 (70.0%) for olaparib and 22/42 (52.4%) for chemotherapy in the ≥3 previous lines of chemotherapy group, and 36/48 (75.0%) for olaparib and 4/17 (23.5%) for chemotherapy in the ≥4 previous lines of chemotherapy group. For ORR, the number of patients with response/total number of patients in population was 109/151 (72.2%) for olaparib and 37/72 (51.4%) for chemotherapy in all patients, 65/76 (85.5%) for olaparib and 23/38 (60.5%) for chemotherapy in the two previous lines of chemotherapy group, 25/37 (67.6%) for olaparib and 7/22 (31.8%) for chemotherapy in the three previous lines of chemotherapy group, 44/75 (58.7%) for olaparib and 14/34 (41.2%) for chemotherapy in the ≥3 previous lines of chemotherapy group, and 19/38 (50.0%) for olaparib and 7/12 (58.3%) for chemotherapy in the ≥4 previous lines of chemotherapy group. (B) OS in patients with two previous chemotherapy lines. (C) OS in patients with ≥3 previous chemotherapy lines. (D) Best objective response by blinded independent central review. Although reversions can occur in BRCA1 and BRCA2,^, all baseline reversions in SOLO3 were found in the BRCA2 gene, possibly reflecting a higher prevalence of reversions in BRCA2 as observed in other data sets. These BRCA reversions are likely to have occurred on previous PBC. Patients had single or multiple reversions, present at relatively low allelic frequency (0.2%-8%), suggestive of a clonal event and ctDNA likely capturing tumor heterogeneity. All reversion mutations overcame the original mutation by either a small deletion removing the original mutation and restoring the open reading frame or introducing a missense mutation instead of a stop codon and would be predicted to restore BRCA function partially or fully. Best overall RECIST response was determined by the best response a patient had during their time in the study up until RECIST progression or the last evaluable assessment in the absence of RECIST progression and was derived from target and nontarget lesion responses combined with new lesion information. A negative change in the best change from baseline in target legion size (%) represents a positive outcome. ^aPatients with a BRCA reversion mutation at baseline. ctDNA, circulating tumor DNA; DCO, data cutoff; HR, hazard ratio; NC, not calculated; OS, overall survival; ORR, objective response rate; PBC, platinum-based chemotherapy; PFS, progression-free survival.