Abstract
PURPOSE: NCI-MATCH (EAY131) is a precision medicine trial using genomic testing to allocate patients with advanced malignancies to targeted treatments. Arm Y evaluated capivasertib, a pan AKT inhibitor, in patients with an AKT1 E17K –mutated tumor. Here, we report on the translational objectives of the study, a molecular and genomic analysis of specimens to identify potential biomarkers of response or resistance to capivasertib. METHODS: Eligible patients had AKT1 E17K –mutated metastatic tumors that progressed with standard treatment and received capivasertib 480 mg orally twice daily for 4 days on and 3 days off weekly in 28-day cycles. The primary end point was objective response rate (ORR). We performed whole-exome sequencing, RNA sequencing, and gene set and pathway enrichment analysis on 25 pretreatment tissue samples and evaluated findings in responders (complete response [CR], n = 0, and partial response, n = 9) and nonresponders (stable disease, n = 13, and progressive disease, n = 3). RESULTS: The ORR was 28.6% (10 of 35) in the reported primary trial and 36% (9 of 25) in this translational cohort. Mutations in the TP53 gene were more frequent in responders, whereas the PI3K/AKT/mTOR pathway genes TYRO3 , SYNJ1 , and CDIPT were significantly altered in nonresponders. DNA repair, p53, E2F, and Wnt-beta catenin pathways were enriched in the responder group. Unsupervised clustering of gene expression identified five genes, ANKRD30A , SUSD4 , TTC6 , POTEJ , and POTEI , that were significantly higher in responders and lower in nonresponders. In addition, EGFR expression was significantly increased in nonresponders. CONCLUSION: In patients with AKT1 E17K –mutated tumors, capivasertib achieved a clinically significant ORR. TP53 mutations appear to be associated with response, whereas certain additional PI3K/AKT/mTOR pathway mutations and EGFR overexpression appear to be associated with nonresponse to capivasertib. Further investigation of predictive biomarkers is warranted.

This publication has 61 references indexed in Scilit: