Responsive signatures established by pharmaco-transcriptomic correlation analysis identifies subsets for PARP-targeted therapy and reveals potential synergistic interactors

Abstract

(1) Background: Poly (ADP-ribose) polymerases (PARPs) have pleiotropic roles including canonical DNA-damage response (DDR) pathways. PARP inhibition is initially proposed as a synthetic lethal interactor with cancer harboring homologous recombination deficiency (HRD), thus becoming a key therapeutic option for genetically-defined subsets of patients. Recently, there has been increasing evidence supporting the expansion of PARP-targeted therapy beyond HRD. Besides, synthetic lethality pathways for PARP-targeted therapy are being studied extensively due to the rapidly developed resistance to PARP inhibitors (PARPi). (2) Methods: We perform integrative pharmaco-transcriptomic analyses by correlating the drug response profiles of clinically-approved PARPi olaparib with the transcriptomes of solid cancer cell lines (n=659) to establish PARPis responsive gene signatures, which are then evaluated for their reliability using independent drug response datasets, and applied to identify tumor subsets primed for PARPi and potential targets synergistically interacting with PARPi. (3) Results: Based on the pharmaco-transcriptomic correlation analysis, we delineate gene signatures to predict the sensitivity and resistance to olaparib in pan-solid cancer cells, which is confirmed by independent drug response datasets. In further exploring the PARPi sensitivity signature, we identify IDH1/2 (isocitrate dehydrogenase 1/2)-mutated low-grade glioma (LGG) and NEUROD1-driven small cell lung cancer (SCLC) as potential subsets for prioritized PARPi, highlighting relaunching PARPi as a promising and innovative strategy to target these malignancy subtypes. Interestingly, the PARPi responsive signatures display a high degree of heterogeneity in the correlation with the curated HRD signatures across TCGA pan-solid cancer cohort, suggesting that these signatures predictive of PARPi responsiveness are HRD-independent. With the PARPi resistance signature, we identified several potentially synthetic lethal interactors with PARPi, e.g. dasatinib, EGFR, or MEK inhibitors. (4) Conclusions: The established PARPi responsive (sensitive/resistant) signatures in solid tumors exhibit robustness in identifying cancer subtypes that are highly primed for PARP-targeted therapy, and combined targets that synergistically augment the efficacy of PARPi.