Background: Metastatic castration resistant prostate cancer (mCRPC) is a heterogeneous disease in which molecular stratification is needed to improve clinical outcomes. Targeting programmed cell death protein 1 (PD-1) elicits durable antitumor responses in multiple cancer types. Putative biomarkers predictive of response to anti-PD-1 therapies include programmed death-ligand-1 (PD-L1), tumour mutational burden (TMB), T cell-inflamed 18-gene expression profile (GEP), and DNA repair defects. Objective: To study potential biomarkers of response in mCRPC to anti-PD-1 therapy (PD-L1, GEP, mismatch repair (MMR) protein), other biomarkers of interest including BRCA2, PALB2, CDK12, PTEN, TP53, SOX2, and to determine association with clinical outcomes. Methods: The study population included 100 men with mCRPC treated at the Royal Marsden Hospital with available fresh mCRPC biopsy tissue. All men had received at least one line of therapy with a next generation hormonal agent and one line of taxane-based chemotherapy. Clinical characteristics and outcomes were extracted from medical records. mCRPC biopsies were assayed by whole exome sequencing (WES), targeted next generation sequencing (NGS), RNA sequencing (RNAseq), GEP score (Nanostring), PD-L1 immunohistochemistry (IHC; DAKO 22C3 assay), SOX2 IHC, ATM IHC. Correlations among these biomarkers and clinical outcomes were assessed. Results: The median age of patients was 68.0 years; 46/84 (54.8%) had de novo metastatic disease at diagnosis and 24/100 (24.0%) patients had visceral disease. Median follow-up from mCRPC biopsy was 56.2 months. The prevalence of loss of protein expression by IHC and/or pathogenic mutation by NGS of MMR was 7/100 (7%). Loss of PTEN and ATM by IHC was 29/100 (29%) and 13/100 (13%) respectively. SOX2 expression (defined as expression in >5% of cells) was 27/100 (27%). The prevalence of TP53 mutation was 25/100 (25%); deleterious alterations of BRCA2 was 9/100 (9%), CDK12 3/100 (3%) and PALB2 1/100 (1%). PDL1 and GEP results were available for 70 and 93 samples respectively. PD-L1 was expressed (combined positive score ≥1) in 24 (33%) mCRPC biopsies; 23 (26%) had high GEP scores (> -0.318). PD-L1 and GEP expression were positively correlated (Phi 0.63). No other biomarkers showed strong correlations. Of 5 samples with MMR loss for which PD-L1 was available, 1 (20%) had PD-L1 ≥1; one of the CDK12 samples had PD-L1 ≥1 (33%). 0/6 BRCA2 mutated biopsies expressed PD-L1 (0%). Based on univariate analysis, PD-L1 expression [HR: 1.75 (1.00;3.06), p=0.045], high GEP score [HR: 2.00 (1.18;3.39), p=0.0083] and SOX2 expression [HR: 1.81 (1.12;2.94), p=0.015] were associated with worse overall survival (OS). No other biomarkers associated with OS. Conclusions: PD-L1 IHC expression was detected in 33% of mCRPC patients and associated with high GEP score. Higher PD-L1, GEP, and SOX2 expression were associated with poor prognosis.