Abstract A110: Identifying sensitive patient populations for CDK7 inhibitors using cell panel screens and bioinformatic approaches

• Published in: Molecular cancer therapeutics 2023-12, Vol.22 (12_Supplement), p.A110-A110
• Main Authors: Hearn, Keisha, Ahn, Maria, Bevan, Luke, Brothwood, Jessica, East, Charlotte, Esteve-Arenys, Anna, Fazal, Lynsey, Hindley, Christopher, James, Sabrina, Kucia-Tran, Justyna, Lyons, John, Munck, Joanne, Saini, Harpreet, Unbekandt, Mathieu, Varshney, Dhaval, Wagner, Steve, Woodhead, Andrew, Wallis, Nicola
• Format: Article
• Language: English
• ISSN: 1538-8514; 1538-8514
• DOI: 10.1158/1535-7163.TARG-23-A110

Background Dysregulation of cell cycle and transcriptional processes promote tumorigenesis and tumor growth.  Due to its dual role in regulating both cellular processes, CDK7 is an attractive therapeutic target, with several CDK7 inhibitors in clinical trials. Preclinical data suggests that such agents could have utility across a range of tumor types particularly those driven by defects in the regulation of cell cycle and transcriptional processes.  Consequently, potential patient populations are broad and often lack well-defined patient stratification. To identify specific cancers with greater dependence on CDK7 activity, we employed the use of a cell line panel screen coupled with bioinformatic analysis of association of drug sensitivity to molecular features. Materials and Methods To explore tumor sensitivities to CDK7 inhibition we profiled two CDK7 inhibitors (THZ1 and LY3405105; Eli Lilly) and the non-selective kinase inhibitor staurosporine as a control across a panel of 468 human cancer cell lines with varied genetic backgrounds.  Activity area was used to define drug responses whilst multi-omics features of cell lines obtained from DepMap were used to define potential molecular features associated with drug response. Differential gene expression between sensitive and resistant cell lines was used to identify expression signatures associated with sensitivity to CDK7 inhibition.  Results were validated in vitro by characterisation of cell line models in viability assays, western blots and RT-qPCR.   Our findings were further validated in vivo in a NSCLC mouse xenograft model.  Results Cell panel data showed a wide range of sensitives to selective CDK7 inhibition (LY3405105) compared to the profile of the less selective THZ1 which had a similar distribution of sensitivity as the control staurosporine.  Bioinformatics analysis identified a global c-MYC signature with a significant correlation between CDK7 expression and c-MYC expression across the entire cell panel which confers sensitivity to CDK7 inhibition and was highly correlated in lung cancer. Further analysis of CDK7 inhibitor sensitivity in lung cancer as a function of c-MYC expression found that SCLC and NSCLC had the highest correlation.   Using an independent set of cell lines with varying c-MYC levels, we confirmed these findings and the results were further validated in vivo where we observed partial tumor regressions.  Conclusions Cell panel data coupled to bioinformatic analysis was us