Introduction: Prostate cancer (PCa) is the second leading cause of cancer death in Australian men. Despite molecular heterogeneity, care still centres on androgen-deprivation therapy, from which many tumours develop treatment resistance, progressing to metastatic castration-resistant PCa. Although personalised medicine has improved patient clinical outcomes and quality of life in other cancer types, its adoption in PCa still lags. Synthetic lethality (SL), where dual perturbation is lethal but either alone is tolerated, offers a route to genotype-guided therapy. SL can be inferred from mutual exclusivity of mutations or deletions in tumours; however, PCa has few recurrent drivers (e.g., SPOP, RB1, TP53) and relatively low mutation/CNA burden, limiting discovery from patient data. We therefore developed Prostate Synthetic Lethal Miner (ProSLeM), which integrates mutation and copy-number data with transcriptomics, adding a loss-of-function dimension via decreased gene expression to prioritise PCa-specific SL pairs.
Objective: To generate a PCa–specific, genotype-anchored map of synthetic-lethal (SL) interactions with functional validation
Methods: We built ProSLeM, integrating copy-number and transcriptomic mutual-exclusivity signals from primary (TCGA-PRAD) and metastatic (SU2C) prostate cancer datasets to prioritise mutation-specific SL partners, then filtered candidates by pathway convergence and druggability.
Results: ProSLeM identified hundreds of predicted SL pairs across common PCa drivers.As a proof of concept, we discovered a SL interaction between SPOP mutation and glutathione metabolism genes (e.g. GSS, CBS, SLC3A2) and validated this SL interaction in vitro and in vivo. Because GSH constrains reactive oxygen species and detoxifies lipid peroxides from polyunsaturated fatty acids (PUFAs), its impairment is predicted to enable iron-dependent lipid peroxidation and ferroptosis. Consistent with this, SPOP-mutant LNCaP cells exhibited heightened sensitivity to GPX4 inhibitors (ML210, RSL3). Multi-omic analyses revealed induction of ferroptosis-associated transcriptional programs and PUFA accumulation in SPOP-mutant cells. Importantly, sorafenib, an FDA-approved multikinase inhibitor with ferroptosis-inducing activity, robustly triggered ferroptosis in vitro and in patient-derived explants (PDEs); SPOP-mutant models showed greater sensitivity, lower IC50 values, and prolonged xenograft survival relative to wild-type controls.
Conclusions: ProSLeM reveals a genotype-linked SL landscape in PCa and pinpoints a glutathione–ferroptosis vulnerability in SPOP-mutant disease, nominating sorafenib as a precision therapeutic.