Does olaparib plus tremelimumab work in recurrent ovarian cancer with BRCA mutations?

A single-arm test of a preclinical synergy

The rationale came from mouse work: in Brca1-deficient high-grade serous ovarian cancer models, PARP inhibition combined with CTLA4 checkpoint blockade produced long-term survival, and the effect depended on sustained interferon gamma in the tumor microenvironment. Adding PD1 or PDL1 blockade instead did not raise interferon gamma or extend survival in those models. The trial therefore paired olaparib, an oral PARP inhibitor, with tremelimumab, a fully humanized monoclonal IgG2 antibody against CTLA4, in women whose tumors carried germline BRCA1 or BRCA2 mutations.

This was a multi-site, single-arm phase I/II study (NCT02571725) enrolling 50 women with recurrent epithelial ovarian, fallopian tube, or primary peritoneal cancer between March 2016 and March 2020. All had measurable disease by modified WHO criteria and confirmed pathogenic germline mutations: 72% BRCA1 and 28% BRCA2. The cohort was heavily pretreated — 62% had received three or more prior regimens (range 1–11) and 60% had prior PARP inhibitor exposure. Because there is no randomized comparator, every response rate here describes this regimen in this population, not its advantage over olaparib alone.

Toxicity forced a mid-study schedule change

Treatment was 300 mg olaparib twice daily by mouth plus 10 mg/kg intravenous tremelimumab every 4 weeks for six doses, then every 12 weeks up to 24 months. In phase I, the first three subjects showed no regimen-limiting toxicities and no new toxicity types beyond those already known for either drug, so the starting dose carried into phase II.

Safety then reshaped the protocol. Delayed grade 3 immune toxicities appeared during cycles 5 and 6 in four of the first 14 patients, prompting a 2018 amendment that cut monthly tremelimumab to four doses before moving to every-12-week infusions. In the combined safety cohort of 49 treated subjects, 57% had grade 3 or 4 non-hematologic toxicity, 34.7% (17 subjects) discontinued tremelimumab for immune-related adverse events, and 28.6% had grade 3 or higher hematologic toxicity attributed mainly to olaparib. Gastrointestinal disorders were the most common adverse events at 95.9%, with 32.7% grade 3 or worse. No grade 5 events and no deaths directly attributed to study therapy occurred. Eight of the 17 who stopped tremelimumab continued olaparib alone.

Why 32.7% and 36.4% describe different groups

The primary phase II endpoint was overall response rate, and the protocol counted all 49 patients who received any treatment as the denominator. Forty-four had at least one post-treatment imaging study; the other five — four who withdrew before cycle 3 and one who died without imaging — were counted as non-responders. That yields the reported ORR of 32.7% (95% CI 19.5–45.8%).

Among the 44 imaging-evaluable patients only, the ORR was 36.4% by modified WHO criteria, with a mean response duration of 10.5 months (95% CI 6.44–14.53) and confirmation on a second scan in 10 subjects. The disease control rate — complete response, partial response, or stable disease — was 47.7% (95% CI 33.0–62.5%). The two figures answer different questions: 32.7% is the intention-to-treat-style estimate the endpoint required, while 36.4% excludes patients who never reached imaging. The best-response table shows 4 complete and 12 partial responses among the 50 enrolled.

Median PFS of 3 months hides a long tail

With a mean follow-up of 25 months, median progression-free survival was 3.0 months (95% CI 2.7–5.9) and median overall survival was 24.3 months (95% CI 14.9–30.4, n = 49). A median of 3 months means half the cohort progressed quickly, which is the main reason the efficacy bar was missed.

The distribution matters more than the median for this question. Six subjects (12.2%) had progression-free survival of at least 12 months, and two (4.1%) had no evidence of progression at 24 months. Two patients completed 24 cycles of combined treatment, and two more finished 24 cycles of olaparib alone after stopping tremelimumab for toxicity. Four subjects continued treatment after imaging showed progression because their physicians judged ongoing benefit, for 108 to 337 additional days. The paper's own reading is that this combination may help selected patients beyond olaparib monotherapy, but that claim rests on a single-arm comparison with no control group.

Prior PARP inhibitor exposure separates outcomes

Sixty percent of the cohort had prior PARP inhibitor treatment, and most of those had progressed in a way consistent with acquired resistance. In an exploratory analysis, previously exposed patients had shorter progression-free survival than PARP inhibitor-naive patients (2.6 vs 8.0 months, p < 0.001) and shorter overall survival (20.1 vs 27.5 months, p = 0.053). This is a comparison between non-randomized subgroups, so it describes prognosis in this cohort rather than proving that prior exposure caused the difference.

Two countervailing observations frame the interpretation. First, 24% of patients with prior PARP inhibitor treatment still showed disease control on the combination, which would not be expected from olaparib alone in a resistant tumor. Second, 21.1% of PARP inhibitor-naive patients progressed despite the sensitivity of BRCA-mutated tumors to PARP inhibition. The paper uses both to argue that BRCA mutation status alone does not identify who benefits, and that additional predictive markers are needed.

Low VSTM5 or IFIT1B marks better PFS

Exploratory post-hoc analyses used archived formalin-fixed paraffin-embedded tumor tissue collected at initial cytoreductive surgery, 15 to 154 months before enrollment, from 30 patients with RNA sequencing data. Clinical benefit was defined as disease control lasting at least 12 weeks, and patients not assessable for response were classed as having no benefit.

Low expression of VSTM5 or IFIT1B was associated with treatment benefit and improved progression-free survival (p < 0.001). Established immunotherapy markers did not perform this way: PDL1 expression, tumor mutational burden, and neoantigen load were not associated with response to this regimen. Spatial transcriptomic comparison of VSTM5-high and VSTM5-low tumors (n = 8) showed incoming FAS ligand signaling in epithelial tumor subsets only in the low group, which the authors interpret as greater vulnerability to T cell-mediated control. These are hypothesis-generating associations from a small archival subset, not validated selection tests, and the paper explicitly calls for validation before clinical use.

A biomarker question, not a regimen verdict

The trial's own conclusion is that the CTLA4 checkpoint may be a critical regulator of tumor immunity in BRCA-mutated ovarian cancer and that combining it with PARP inhibition may extend olaparib's benefit in select cases. The evidence for that is the 12.2% with at least 12 months of progression-free survival and the disease control seen in some PARP inhibitor-resistant patients, set against a 3.0-month median PFS and a 34.7% tremelimumab discontinuation rate.

The practical priority the data point to is patient selection rather than wider use of this doublet. The most decision-relevant split in the paper is prior PARP inhibitor exposure — 2.6 versus 8.0 months median PFS — combined with the candidate markers VSTM5 and IFIT1B, which were measured on archival tissue available in routine practice. The limitations sit with those analyses: 30 patients in the biomarker subset, tissue collected up to 154 months before treatment, and no randomized control. A future trial would need to test whether selecting on these markers improves outcomes, and the authors note that strategies to limit toxicity, such as selective PARP inhibitors, will be critical for that work.

Sources used in this answer