124 (PB114) - Defining activity and patient selection of a novel CDK7 inhibitor, GTAEXS-617, through AI-supported primary cancer tissue profiling

Abstract

Background: Dual inhibition of CDK4/6 has become a mainstay treatment for several cancers, including HR-positive/HER2-negative breast cancer. CDK7 represents another potentially attractive CDK target that could combine the cell cycle inhibition found in CDK4/6 inhibitors with inhibition of transcription. We have generated a highly potent, selective and bioavailable inhibitor of CDK7, GTAEXS-617 (‘617). CDK7 is expressed in both normal and diseased tissue, thus understanding expression, sensitivity profiles and effects of CDK7 inhibition on cancer and non-transformed cells (e.g. stromal or immune) is crucial to elucidate MOA and determine the therapeutic window. In order to identify those patients more likely to benefit from CDK7 inhibition and ‘617 therapy in particular, we have used machine learning and other approaches in part to analyse multiparameter response in heterogeneous patient tissue samples. Materials and Methods: To achieve full understanding of the potency and activity of ‘617, including versus other CDK7 and CDK4/6 inhibitors, we deployed disease relevant primary sample model systems from, for instance, ovarian, breast, lung and MCL patients that represent the cancer microenvironment. Single cell ex vivo functional screening combined with transcriptomics after CDK7 perturbation in disease relevant primary human cancer samples helps reveal cancer-specific effects and patient selection methods. A translatable high content imaging platform amenable to primary human material supported by deep learning driven image analysis is used for all functional profiling. Results: Primary cancer samples from various indications had a broad differential sensitivity to ‘617, with specific sensitivity in high grade ovarian cancer versus low grade. Ex vivo cancer cell sensitivity differences in various ovarian cancer grades, as identified by correlation of phenotypic response to differential single cell transcriptomics after ‘617 perturbation, indicates a potential stratification point that could be used for patient enrichment in clinical trials. Further, ‘617 displayed less cytotoxic effects on the immune compartment than other CDK7 and CDK4/6 inhibitors. Conclusions: We used single cell functional profiling and transcriptomics to begin to maximise our understanding of ‘617 effect with the goal of enriching likely responding patient groups. We’ve described approaches which enable the rapid assessment of new targeted therapies in primary human disease samples containing host immune cells, together with functional data. Conflict of interest: Ownership: MS, ED, JB, MS, OB, JlC, BE, RP, TWP, RS, CB, GIV are shareholders and employees of Exscientia. FA and JJ are shareholders and employees of GT Aperion.