Background: Programmed death-ligand 1 (PD-L1) has been recently found in tumor-derived exosomes and associated with progressive disease (PD) (Poggio, Cell 2019). In localized prostate cancer, PD-L1 tumor expression has been shown to be a biomarker for node-positive patients (pts) (Petitprez, Eur Urol Focus 2019) but its role in mCRPC is largely unexplored. We aimed to determine the potential of PD-L1 exosomal expression to predict lymph nodal PD in mCRPC receiving abi or enza. Methods: In a biomarker prospective study (REC 2192/2013), exosomes were isolated from plasma samples collected before first-line therapy with abi/enza between December 2014 and October 2018. RNA was extracted for analysis of PD-L1 by digital droplet PCR (Del Re, Prostate Cancer Prostatic Dis 2021). Results: In our cohort of 55 node-positive mCRPC pts (median age: 76 years, range 47-89), the amount of exosomal PD-L1 expression (pts dichotomized in “high” and “low” as below or above median PD-L1/ACTB ratio = 0.0052) was associated with radiographic response after the first 3 months abi/enza therapy [high PD-L1 vs low PD-L1: 8 (25.8%) vs 0 (0%) PD, 23 (74.2%) vs 18 (75%) stable disease, and 0 vs 6 (25%) partial/complete response] (p = 0.0009). We observed an overall median progression-free survival (PFS) of 14.3 months [95% confidence interval (CI) 8.6-not reached] shorter in high PD-L1 compared to low PD-L1 10.6 (6.5-23.4) vs not reached [hazard ratio (HR) 3.1 (95% CI 1.1-8.5), p = 0.023]. In multivariable analysis including exosomal PD-L1 and androgen receptor (AR) variant 7, AR copy number gain detected in plasma, neutrophil lymphocyte ratio (NLR), lactate dehydrogenase, exosomal PD-L1; AR-V7 and AR gain were independent predictors of nodal PFS (HR 5.8, 95% CI 1.4-24.2, p = 0.01, HR 3.2, 95% CI 1.1-9.2, p = 0.03 and HR 12.9, 95% CI 2.3-73.2, p = 0.004), whereas AR-V7, AR gain, NLR > 3 were independent predictors of OS (HR 4.9, 95% CI 1.8-13.1, p = 0.001, HR 4.4, 95% CI 1.1-17, p = 0.02, and HR 3.2 95% CI 1.1-9.4, p = 0.03, respectively). Moreover, a trend of a worse OS was reported in high PD-L1 compared to low PD-L1 pts (HR 2.6, 95% CI 0.9-7.5, p = 0.07). Conclusions: In mCRPC pts treated with ARSI, exosomal PD-L1 expression may be considered as a predictive biomarker of lymph nodal response, playing a potential role in individualized disease monitoring and treatment combining between ARSI and immunotherapy. Larger evaluation of treatment decisions based on exosomal PD-L1 expression is now required.