Multi-Omics Profiling and CRISPR Dependency Screening Identify WEE1 Inhibition and Rational BET Inhibitor Combinations as a Therapeutic Strategy for Triple-Negative Breast Cancer

Triple-negative breast cancer (TNBC) is an aggressive subtype with limited targeted therapies. We integrated multi-omics data (RNA expression, DNA methylation, copy number variation, and somatic mutations) from 186 TNBC patients using Similarity Network Fusion (SNF), identifying 8 molecular clusters. Cluster 1 comprised 85% of patients and was characterized by MTORC1, IGF1R, and MYC pathway activation. Using ElasticNet regression, we predicted sensitivity to 286 drugs across all clusters. The same 5 drugs—IGF1R_3801, Mitoxantrone, OTX015, AZD8055, and I-BET-762—were top hits across all clusters, nominating a candidate convergent vulnerability along the IGF1R→PI3K/mTOR→BET axis. We then tested this hypothesis across three independent, orthogonal layers of evidence. External validation in METABRIC TNBC (n = 320) showed that PIK3CA expression was not prognostic for overall survival (HR = 0.920, 95% CI: 0.783–1.082, p = 0.315), though it was associated with relapse-free survival (HR = 0.869, 95% CI: 0.769–0.981, p = 0.024). CRISPR dependency screening in 15 TNBC cell lines showed that IGF1R was not essential (mean Chronos = 0.010, 0% of lines dependent) and PIK3CA was essential in only 20% of lines (mean Chronos = -0.304), refuting both nodes of the predicted axis as viable targets despite their computational prioritization. In contrast, WEE1 — a gene outside the original signature — emerged as the single most essential gene across all 15 lines (mean Chronos = -2.754, 100% dependent), exceeding BRD4 (-0.972, 100%) and MTOR (-1.214, 100%). Synergy analysis using the Bliss independence model identified Mitoxantrone + I-BET-762 as the most synergistic combination (score = 0.572, p < 0.001), consistent with BET inhibition suppressing transcription of DNA-repair genes and sensitizing cells to DNA damage. Together, these results show that computational drug-sensitivity convergence does not guarantee target validity: the pathway nominated by AI-driven drug prediction (IGF1R, PIK3CA) failed functional and clinical validation, while the strongest functional dependency (WEE1) fell outside that prediction entirely. We propose a revised therapeutic roadmap for TNBC that prioritizes WEE1 inhibition and Mitoxantrone + BET inhibitor combinations, while deprioritizing PIK3CA and IGF1R targeting. These findings are hypothesis-generating and await wet-lab validation.