Berzosertib Plus Carboplatin Fails to Improve Responses in Pretreated mCRPC

The ATR inhibitor berzosertib (M6620) plus carboplatin failed to improve responses compared with carboplatin with or without docetaxel in patients with heavily pretreated, biomarker-unselected metastatic castration-resistant prostate cancer (mCRPC), according to findings from a randomized phase 2 trial (NCT03517969) published in Cancer Research Communications.¹

The combination was also associated with more frequent high-grade treatment-related adverse effects (TRAEs), and enrollment was halted for futility at a planned interim analysis.

The overall response rate (ORR) was 0% with berzosertib plus carboplatin (arm B; n = 31) vs 15% in the control arm (arm A; n = 34). All five responses in arm A occurred among patients who received docetaxel plus carboplatin (19%; n = 26); none of the patients who received carboplatin alone (n = 8) responded. Grade 3 or higher TRAEs occurred in 65% of patients in arm B vs 38% of patients in arm A. Enrollment stopped after 65 of a planned 130 patients had been treated.

"Berzosertib with carboplatin led to fewer clinical responses and more frequent serious adverse events compared with a control regimen of docetaxel with carboplatin in a randomized phase II study," lead study author Atish D. Choudhury, MD, PhD, and coauthors wrote in the journal.

Choudhury is affiliated with Dana-Farber Cancer Institute and Harvard Medical School in Boston, Massachusetts.

What was the rationale for combining berzosertib with carboplatin?

Berzosertib inhibits ataxia-telangiectasia and Rad3-related (ATR) kinase, a key DNA damage response protein activated at stressed replication forks. ATR inhibitors have shown synergy with platinum compounds in preclinical models, and the investigators hypothesized that carboplatin-induced replication stress would make prostate cancer cells dependent on ATR, potentially extending benefit to patients both with and without tumor homologous recombination deficiency (HRD).

How was the trial designed?

The open-label trial enrolled adults with mCRPC who had received at least one androgen receptor pathway inhibitor and a taxane, had an ECOG performance status of 0 to 2, and agreed to a mandatory pretreatment tumor biopsy. Prior PARP inhibitor therapy was permitted; prior carboplatin or ATR inhibitor therapy was not.

Patients were randomly assigned 1:1 to arm A, docetaxel at 60 mg/m² plus carboplatin at AUC 4 on day 1 (or carboplatin at AUC 5 alone for patients who were not docetaxel candidates), or arm B, berzosertib at 90 mg/m² on days 2 and 9 plus carboplatin at AUC 5 on day 1. Both regimens were given every 21 days, and patients in arm A could cross over to arm B at progression.¹

The primary end point was ORR, defined as a PSA decline of at least 50% or a confirmed radiographic response per RECIST 1.1 criteria.^1,2 Key secondary end points included radiographic progression-free survival (rPFS), PFS per Prostate Cancer Working Group 3 (PCWG3) criteria, time to PSA progression, safety, and the association between tumor HRD and clinical outcomes. Overall survival (OS) and response after crossover were exploratory end points.

Between June 2019 and July 2020, 73 patients were randomly assigned, and 65 received protocol treatment.¹ The median age was 69 years in both arms. All treated patients had received prior docetaxel and an androgen receptor pathway inhibitor; 41% of patients in arm A and 42% of patients in arm B had also received cabazitaxel, and 9% and 10%, respectively, had received a PARP inhibitor. The median number of prior systemic therapies was 4 in arm A and 3 in arm B.

What did the efficacy and biomarker analyses show?

None of the 14 patients who crossed over from arm A to berzosertib plus carboplatin after progression responded.

Median rPFS was 3.2 months (95% CI, 2.1-5.1) in arm A vs 2.6 months (95% CI, 2.1-4.8) in arm B (HR, 1.14; 95% CI, 0.62-2.1). Median PFS per PCWG3 criteria was 2.5 months (95% CI, 2.1-4.2) vs 2.4 months (95% CI, 2.1-3.5), respectively (HR, 1.05; 95% CI, 0.58-1.90), and median time to PSA progression was 2.1 months (95% CI, 1.4-3.6) vs 1.4 months (95% CI, 1-2.8; HR, 1.65; 95% CI, 0.88-3.09). In the exploratory OS analysis, median OS was 8.4 months (95% CI, 4.5-12.8) vs 5.7 months (95% CI, 4.6-8.1), and 6-month OS rates were 57% (95% CI, 39%-72%) vs 44% (95% CI, 26%-61%), respectively.

HRD status, assessed by whole-exome sequencing and RAD51 foci–based immunohistochemistry (IHC) of tumor tissue and by DirectHRD analysis of circulating cell-free DNA, did not significantly correlate with PFS, and no tissue- or blood-based biomarker identified a subgroup with differential benefit from ATR inhibition. However, an exploratory combined IHC biomarker capturing either low ATM activity or high replication stress, the latter indicated by elevated phosphorylated KAP1 (pKAP1), was associated with significantly longer PFS across both arms (log-rank P = .045). The investigators described the biomarker as prognostic for carboplatin-based therapy and worthy of further study.

What did the safety analysis show?

All 65 treated patients experienced a TRAE of any grade. Grade 3 or 4 thrombocytopenia occurred in 26% of patients in arm B vs 9% of patients in arm A, and grade 3 anemia occurred in 16 of 31 patients vs 5 of 34 patients, respectively. One patient in arm B experienced treatment-related grade 5 sepsis. Among the 14 patients who crossed over to berzosertib plus carboplatin, 29% experienced grade 3 TRAEs.

What do the findings mean for future research?

The investigators called the lack of synergy unexpected based on preclinical data and suggested that the berzosertib dose, limited to 90 mg/m² with carboplatin because of overlapping dose-limiting toxicities, may have been insufficient to inhibit ATR. On-treatment biopsies to confirm target engagement were not performed.

The authors also noted that every response occurred with docetaxel plus carboplatin, despite prior docetaxel progression in all patients and a lower carboplatin dose in that regimen. Although carboplatin monotherapy was reserved for patients who were not docetaxel candidates, they wrote that the findings favor the doublet over carboplatin alone in a biomarker-unselected population.

References

1. Choudhury AD, Zhong C, Xie W, et al. A phase II study of berzosertib in combination with carboplatin compared with docetaxel with carboplatin in metastatic castration-resistant prostate cancer. Cancer Res Commun. 2026;6(9):2206-2219. doi:10.1158/2767-9764.CRC-26-0488
2. A phase 2 study of M6620 (VX-970, berzosertib) in combination with carboplatin compared with docetaxel in combination with carboplatin in metastatic castration-resistant prostate cancer. Updated July 31, 2026. Accessed September 28, 2026. https://clinicaltrials.gov/study/NCT03517969