Need Help?
Copied to clipboard!

A Phase 1-2 Study of the Combination of Olaparib and Tremelimumab in BRCA1 and BRCA2 Mutation Carriers with Recurrent Ovarian Cancer (NCT02571725)

INST1419: A phase 1-2 study of the combination of olaparib and tremelimumab in BRCA1 and BRCA2 mutation carriers with recurrent ovarian cancer (NCT02571725) was designed to evaluate the safety and efficacy of combined treatment with the PARP inhibitor olaparib and the CTLA4 immune checkpoint inhibitor tremelimumab in women with recurrent ovarian cancer and a germline BRCA1 or BRCA2 mutation. This was a multi-site single-arm open-label study that enrolled 50 subjects with recurrent epithelial ovarian, fallopian tube, or primary peritoneal cancer between 2016 and 2020. Treatment efficacy was assessed using best overall response by modified WHO (mWHO) criteria, and secondary endpoints included disease control rate, duration of response, and progression-free and overall survival.

The median age was 56 (range 39-78). All patients had measurable disease and pathogenic germline mutations in BRCA1 (72%) or BRCA2 (28%). Subjects received 300mg of olaparib by mouth, twice daily, and tremelimumab at 10mg/kg IV every four weeks for 6 doses, and then every 12 weeks for up to 24 months. Following an interim data safety monitoring review in 2018, the protocol was modified to reduce the frequency of tremelimumab infusions to 10mg/kg IV every four weeks for 4 doses, followed by infusions every 12 weeks.

Exploratory biomarker assessments were performed using archived formalin-fixed paraffin-embedded (FFPE) tumor samples collected at the time of initial cytoreductive surgery from 33 subjects. Clinical characteristics and outcomes for the biomarker cohort were similar to the overall and response cohorts. Candidate biomarkers were identified based on differential gene expression in samples from patients with clinical benefit from treatment, defined as disease control for at least 12 weeks, versus those with progressive disease. For RNAseq, sample integrity was assessed using Agilent Tapestation, to ensure the DV200 was greater than or equal to 30. Sequencing was done on the Illumina NovaSeq instrument by Personalis. Gene expression used in this analysis was the normalized measure defined as transcripts per million (TPM) total RNA molecules. Welch's t-test was used to assess whether there was a difference in gene expression based on clinical benefit. The area under the receiver operating characteristic curve (AUC) was also computed for TPM gene expression based on clinical benefit. The AUC, as determined by the trapezoidal rule, was estimated by the concordance index (SAS 9.4 LOGISTIC procedure). DNA quality was confirmed if the DNA integrity number was greater than or equal to 3. Exome enrichment was performed using the Accuracy and Content Enhanced (ACE) version 4 (ACE ImmunoID NeXT) enrichment platform of Personalis. Briefly, DNA from tumor and cryopreserved PBMC samples were used to construct whole-exome capture libraries, and additional custom probes and optimized chemistry improved uniformity of coverage in sequence regions of high G-C content across the exome using ACE technology. Sequencing was performed at 20 Gigabase (G) sequencing depth for normal samples and 35G sequencing depth for tumor samples. All DNA and RNA reads were aligned to the hs37d5 reference genome built in accordance with the Personalis ACE Cancer Exome Analysis Pipeline, duplicate reads were removed, and the Genome Analysis toolkit was used to correct base quality scores and improve sequence alignment. Mutational burdens were generated using tumor-specific genomic events (SNVs, indels and fusions) of at least 5% allelic fraction that were verified using transcriptomic data. All potential neoepitopes were created for each mutation and tested for presentation, and considered putative neoepitopes if they were predicted to bind to any patient-specific alleles. The Xenium platform was used for spatial transcriptomic analyses using Xenium slides and sample prep reagents (10x Genomics). The immune panel was augmented with custom-ordered genes. Initial preprocessing included quality control, normalization, batch correction, and clustering. Cells with fewer than 20 detected features were excluded as low-quality. High-quality cell-by-gene count matrices were processed using Seurat 5.3.0, with normalization performed using SCTransform. Batch correction was carried out using Harmony integration, following standard preprocessing: individual dataset normalization, merging, PCA (RunPCA), and integration via RunHarmony. Clustering was performed on the Harmony-corrected embeddings using FindNeighbors and FindClusters (resolution = 0.03), and visualized with UMAP (RunUMAP). Cell types were initially assigned using SingleR with the HPCA reference dataset. To resolve cellular heterogeneity, subclustering was performed on select clusters using the FindSubCluster function in Seurat. Canonical marker genes used for cell type identification included CD3E for T cells, MS4A1 for B cells and EPCAM for epithelial and DCN for fibroblasts.

Results from this trial demonstrated grade 3 or 4 non-hematologic toxicity in 57% of subjects. The overall response rate was 32.7% (95% CI 19.5-45.8%), with a disease control rate of 47.7% (95% CI 33.0-62.5%). In exploratory biomarker analyses, low VSTM5 or IFIT1B expression predicted treatment benefit and progression-free survival (p<0.001).

Data available through dbGaP include RNAseq, whole exome sequencing, and Xenium spatial transcriptomic data.