733 Integrative molecular profiling of high-grade primary prostate cancer identifies patients with a biomarker profile that favors the combination of standard of care (SOC) therapy with immunotherapy

BackgroundProstate cancer (PCa) is primarily driven by androgen receptor (AR) signaling and has a highly immunosuppressive microenvironment. Although genomic and histopathological differences between low- and high-grade primary PCa (lgPCa and hgPCa) have been reported, an integrative assessment of multiple molecular features in the context of disease grade and metastatic outcome is lacking. We propose that a subset of hgPCa patients who relapse under SOC may benefit from adjuvant immune-checkpoint blockade (ICB) added to SOC to overcome immune suppression.MethodsWe analyzed treatment naive prostatectomy tissue from a cohort of 124 primary PCa patients (n= 58, Gleason score ≤6; n= 66, Gleason score ≥ 8). We performed RNAseq expression profiling, whole-exome sequencing (WES) and immunohistochemistry. We employed digital spatial analysis in tumor vs. stromal regions to characterize differences in CD8+ T-cell topology between lgPCa and hgPCaResults1.Comparisons in lg vs. hgPCA: Digital spatial analysis assessing the proximity of CD8+ T-cells to tumor cells revealed a T-cell exclusion phenotype that is more prominent in hgPCa, whereas evaluation of overall CD8+ T-cell density in tumor and stromal regions did not differentiate disease grades. HgPCa had a higher frequency of at least one functional mutation in either TP53, RHPN2, or KMT2D genes compared to lgPCa. Assessment of MHC-I deficiency by IHC and mRNA revealed that hgPCa has significantly lower MHC-I protein expression compared to lgPCa. Interestingly, MHC-I loss in hgPCa associated with a T-cell exclusion phenotype. Moreover, RNAseq gene expression signatures revealed higher expression of tumor-associated macrophage (TAMs), T-regs, Cancer-Associated Fibroblasts (CAFs), DNA damage repair (DDR) genes and lower Interferon-γ (IFN- γ) expression in hgPCa compared to lgPCa. Overall, hgPCa is characterized by a combined phenotype of ‘MHCIloss/IFN- γ low/CAFhigh/TAMhigh/T-reghigh/DDRhigh’. 2.Comparisons within hgPCA that develop metastasis: Unsupervised analysis of molecular features in hgPCa patients that developed metastases identified a subset of patients that exhibit a less immunosuppressive phenotype with lower tumor AR expression, retained tumor MHC-I expression, moderate CD8+ T-cell infiltration and a high IFN-γ RNA signature (figure 1), suggesting potential benefit from ICB therapyAbstract 733 Figure 1Unsupervised analysis of primary hgPCa that develop metastasis. Features used for unsupervised classification into Cluster 1 and Cluster 2 are: IHC CD8+ cell proximity to tumor cells (IHC_Infiltrated/Desert/Excluded), CD8+ overall cell density in tumor area, MHCI IHC H-score in tumor (% of tumor cells that are positive X stain intensity), RNAseq signatures for AR pathway in tumor, T-cell exhaustion, Interferon-y, Macrophage M1, Neuroendocrine phenotypes and DNA repair pathway. Patient cohort annotated for at least one mutations in driver genes (TP53, RHPN2, KMT2D), percent tumor expression of MHCI IHC (>25% high, <25% low), CD8 infiltration type in relation to tumor and cancer subtypes as defined by Mortensen mRNA profiling (Mortensen at al, Science Reports 2015).ConclusionsOur analysis suggests that hgPCa is characterized by low antigenicity as assessed by loss of MHC-I protein expression and an immunosuppressive microenvironment rich in CAFs, macrophages, T-regs and T-cell exclusion phenotypes. Unlike lgPCa, hgPCA can have a poor prognosis (within 5 years relapse). However, a subset of hgPCa patients that metastasized while on SOC exhibited a biomarker profile that might benefit from combination of SOC with ICBEthics ApprovalThis study was approved by BMS Cambridge Massachusetts Institutional Biosafety Committee, approval number CAM_2020_12050_6Consent‘Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.’