Deep state.

Top Line: You already know that luminal and basal subtypes of breast cancer have different clinical behavior and different treatment approaches. Prostate cancer can also be classified as luminal or basal. 

The Study: In this retrospective molecular analysis of over 600 patients with metastatic castration-resistant prostate cancer, 45% were classified as luminal and 55% as basal. As you might expect, luminal prostate cancers were more likely to overexpress androgen receptor pathway genes than basal tumors (which included 90% of small cell neuroendocrine cancers). Patients with basal tumors had worse survival than those with luminal tumors and did not derive as much benefit from androgen signaling inhibitors. Such classifications could better identify which patients may benefit from androgen inhibition and which may have greater benefit from other forms of therapy. The molecular diversity of prostate cancer is further emphasized by another study that developed a biologically informed deep learning model (P-NET) to characterize treatment resistant prostate cancer. Obviously, P-NET stands for pathway-aware multi-layered hierarchical network. Among > 1000 tumors (67% primaries, 33% castration resistant metastatic cancers), P-NET accurately categorized 73% of primary tumors and 80% of metastatic tumors. More informative, though, is visualization of the inner workings of the P-NET model, which shows the complex interactions among genetic alterations and biological pathways that drive treatment resistance. In fact, the 27% of primary tumors “mis-classified” as castration resistant were significantly more likely to develop biochemical recurrence after definitive treatment than their correctly classified counterparts.

TBL: Luminal and basal subtypes of prostate cancer have different prognoses and respond differently to androgen signaling inhibition. | Aggarwal, JAMA Oncol 2021 and Elmarakeby, Nature 2021


Popular Posts