A proof-of-concept phase 2 study of the CHK1 inhibitor prexasertib with translational study outcomes in platinum-resistant recurrent, BRCA wild-type high-grade serous ovarian carcinoma

This proof-of-concept phase 2 trial (NCT02203513) evaluated the clinical activity of the CHK1 inhibitor (CHK1i), prexasertib, in platinum-resistant (PR) recurrent high-grade serous ovarian carcinoma (HGSOC) patients without BRCA mutation. The primary endpoint was objective response rate (ORR). Secondary outcomes were safety and progression-free survival (PFS). Translational research was an exploratory endpoint. Potential biomarkers were investigated using pre-treatment fresh biopsies and serial blood samples. 49 heavily-pretreated patients were enrolled. Among the 39 RECISTv1.1-evaluable patients, ORR was 30.7%, the median PFS was 5.8 months. Toxicity was manageable. Transcriptomic analysis revealed high levels of DNA replication-related genes (POLA1, POLE, GINS3, MCM7) associated with lack of clinical benefit. Subsequent preclinical experiments demonstrated synthetic lethality of POLA1 inhibition in combination with CHK1i in PR-HGSOC cell line models. Therefore, POLA1 expression may be predictive for CHK1i resistance, and the concurrent POLA1 inhibition may improve the efficacy of CHK1i monotherapy in this hard-to-treat population, deserving further investigation. RNAseq was performed using a HiSeq4000 sequencing system (Illumina) at the CCR Sequencing Facility/NCI on pre-treatment biopsy samples from 17 patients.. NovaSeq S1 using Illumina TruSeq Stranded Total RNA Library Prep and paired-end sequencing was used to perform RNAseq using a TruSeq stranded library. Raw RNAseq counts were used as input. STAR v2.7.0f was used to align raw reads in 2-pass mode. RNAseq by expectation-maximization (RSEM) v1.3.1 was used for quantification. Principal component analysis was constructed from the batch-normalized data after centering. The first two PCs were plotted across all data with labels for batch and tumor/normal status. Two samples did not match with ovarian cancer tissue profiles obtained from GTEx database, since they were more likely normal or necrotic tissues. 15 biopsies were selected for further in silico analysis. Quartile normalization and log-transformation prior to analysis were performed on datasets.