Clinical utility of transcriptomic signatures to identify androgen receptor and neuroendocrine signaling in prostate cancer


Yu-Wei Chen, Joanne Xiu, Kelsey Anne Poorman, Charles J. Ryan, Brady Gilg, Matthew James Oberley, George W. Sledge Jr., Himisha Beltran, and Rana R. McKay


Background: Neuroendocrine prostate cancer (NEPC) is a relatively androgen receptor (AR)-independent and aggressive variant of PC that can arise from adenocarcinoma with varied degrees of adenocarcinoma and high-grade neuroendocrine histologic features seen during its progression. This study evaluated the clinical and molecular features of histologic subtypes of prostate tumors (adenocarcinoma, NEPC, mixed adeno-NE) based on previously validated AR signaling and NEPC transcriptomic signatures of PC.
Methods: Whole exome (WES) and whole transcriptome sequencing (WTS) were performed on PC tumors. Molecular subsets were defined by AR signaling signature [AR score positive (+) vs negative (-)] and NEPC gene signature [NE score positive (+) vs negative (-)]. The prevalence of key molecular alterations (RB1, TP53, PTEN, AR, FANC, SLFN11 expression) were investigated. Genes encoding cell surface antigens with potential therapeutic implications were described across signature groups. Overall survival (OS) was obtained from claims data and analyzed with Kaplan-Meier estimator.
Results: A total of 4,476 prostate tumors were analyzed in this study including: 3,623 prostate adenocarcinoma, 56 NEPC, and 25 tumors with mixed adeno-NE. Tissue from 2,641 (67%) tumors were collected from prostate and 1,326 (33%) were metastatic tumors [lymph node (14%), bone (9%), liver (7%), lung (2%), CNS (2%)]. The four AR/NEPC signature-defined subtypes were AR+/NE+ (N=649, 14%), AR+/NE- (N=1614, 36%), AR-/NE+ (N=875, 20%), and AR-/NE- (N=1358, 30%). AR+/NE- was most common in prostate (39%), LN (42%) and lung metastasis (35%); AR-/NE+ was most common in bone (34%) and liver metastasis (43%); AR-/NE- was most common in CNS metastasis. The most common gene alterations in each molecular subtype are defined in Table. AR-/NE+ tumors had the lowest mRNA expression of FOLH1(PSMA), STEAP1, TACSTD2 (Trop-2), CD276 (B7-H3), and PSCA among the four groups. DLL3 expression was higher among NE+ subtypes. FOLH1 and PSCA expression were higher among NE- subtypes. The median OS (months) was longer in NE- than NE+ subtypes [AR+/NE- (78.3), AR-/NE- (78.7) vs AR+/NE+ (70.7), AR-/NE+ (69.1)]. In the platinum-treated subgroup (N=146), the median OS was numerically longer among tumors with SLFN11 overexpression (17.6 vs 12.0) although this was not statistically significant (p-value: 0.99).
Conclusions: Molecularly distinct subtypes of PC can be further characterized by AR signaling signature, NEPC gene expression signature, and co-occurring molecular alterations. The phase II biomarker-driven PREDICT trial will integrate the above signatures/molecular alterations to allocate treatment regimens in metastatic castration-resistant PC.
External Link