Background: Alterations in DNA damage repair (DDR) genes are common in metastatic castration-resistant prostate cancer (mCRPC), and are implicated in responses to carboplatin [carbo], PARP inhibitors and immunotherapeutics. Inhibitors of the ATR kinase, which is involved in the DDR response, have been demonstrated to have synergistic activity with platinum compounds in preclinical models. We therefore conducted a phase 2 study of the ATR inhibitor berzosertib [berzo]+carbo vs. docetaxel [doce]+carbo in mCRPC. Methods: Patients (pts) previously treated with at least one secondary hormonal therapy and taxane underwent mandatory pre-treatment biopsy and were randomized 1:1 to receive Arm A (doce 60 mg/m2 day 1 + carbo AUC 4 day 1) or Arm B (berzo 90 mg/m2 days 2,9 + carbo AUC 5 day 1) every 21 days. Pts randomized to Arm A who were not candidates for doce received carbo AUC 5 monotherapy. Stratification factors were 1) prior PARP inhibitor (yes vs. no) and 2) evaluable disease by RECIST 1.1 (yes vs. no). Pts on Arm A crossed over to Arm B (berzo+carbo) at the earlier of PSA or radiographic progression. The primary endpoint was overall response rate (ORR; PSA reduction by ≥ 50% or radiographic response by RECIST 1.1). Secondary endpoints included time to PSA progression, radiographic PFS (rPFS), PFS by PCWG3 criteria, and adverse events (AEs) in each arm. Planned enrollment was 136 pts (for 130 to be treated), with interim analysis for futility after 65 pts were treated. Results: 73 pts were randomized between 6/2019 and 7/2020; 34 pts were treated on Arm A (26 carbo+doce; 8 carbo alone) and 31 on Arm B. Median number of prior systemic therapies (excluding ADT, 5α-reductase inhibitors, 1st generation antiandrogens) was 4 (range 2-8). Median treatment duration was 3 cycles, and 4 pts in each arm discontinued for AEs. Grade 3 or higher treatment-related AEs (TrAE) were seen in 13(38%) pts in Arm A and 21(68%) in Arm B. Pts in Arm B had greater frequency of grade 3-4 thrombocytopenia (8[26%] vs. 3[9%]). 1 pt in Arm B had grade 5 sepsis attributed to study treatment. ORR was 15% in Arm A (5/34; 5/26[19%] in pts who received carbo+doce) and 0% in Arm B (0/31). 14 pts in Arm A crossed over, with no subsequent responses seen. Median rPFS was 2.1(95% CI:2.0,3.2) mo in Arm A and 2.4(1.9,4.2) mo in Arm B. At planned interim analysis, trial enrollment and crossover to Arm B were halted due to futility. Conclusions: Carbo+berzo led to fewer overall responses and a higher rate of grade 3 or higher TrAEs compared to carbo+doce. All responses seen were in pts who received carbo+doce despite requirement for prior progression on taxane, suggesting that this combination is favored over carbo+berzo or carbo monotherapy in a heavily pre-treated biomarker-unselected population. Extensive genetic and molecular studies for DDR assessment from tissue and cfDNA are in progress. Clinical trial information: NCT03517969