The Immune Landscape of Prostate Cancer and Nomination of PD-L2 as a Potential Therapeutic Target
Shuang G Zhao,Jonathan Lehrer,S Laura Chang,Rajdeep Das,Nicholas Erho,Yang Liu,Martin Sjöström,Robert B Den,Stephen J Freedland,Eric A Klein,R Jeffrey Karnes,Edward M Schaeffer,Melody Xu,Corey Speers,Paul L Nguyen,Ashley E Ross,June M Chan,Matthew R Cooperberg,Peter R Carroll,Elai Davicioni,Lawrence Fong,Daniel E Spratt,Felix Y Feng
DOI: https://doi.org/10.1093/jnci/djy141
2018-10-13
Abstract:BACKGROUND: Immunotherapy has been less successful in treating prostate cancer than other solid tumors. We sought to better understand the immune landscape in prostate cancer and identify immune-related biomarkers and potential therapeutic targets.METHODS: We analyzed gene expression data from 7826 prospectively collected prostatectomy samples (2013-2016), and 1567 retrospective samples with long-term clinical outcomes, for a total of 9393 samples, all profiled on the same commercial clinical platform in a CLIA-certified lab. The primary outcome was distant metastasis-free survival (DMFS). Secondary outcomes included biochemical recurrence-free survival (bRFS), prostate cancer-specific survival (PCSS), and overall survival (OS). All statistical tests were two-sided.RESULTS: Unsupervised hierarchical clustering of hallmark pathways demonstrated an immune-related tumor cluster. Increased estimated immune content scores based on immune-specific genes from the literature were associated with worse bRFS (hazard ratio [HR] = 1.26 [95% confidence interval [CI] = 1.12 to 1.42]; P < .001), DMFS (HR = 1.34 [95% CI = 1.13 to 1.58]; P < .001), PCSS (HR = 1.53 [95% CI = 1.21 to 1.92]; P < .001), and OS (HR = 1.27 [95% CI = 1.07 to 1.50]; P = .006). Deconvolution using Cibersort revealed that mast cells, natural killer cells, and dendritic cells conferred improved DMFS, whereas macrophages and T-cells conferred worse DMFS. Interestingly, while PD-L1 was not prognostic, consistent with its low expression in prostate cancer, PD-L2 was expressed at statistically significantly higher levels (P < .001) and was associated with worse bRFS (HR = 1.17 [95% CI = 1.03 to 1.33]; P = .01), DMFS (HR = 1.25 [95% CI = 1.05 to 1.49]; P = .01), and PCSS (HR = 1.45 [95% CI = 1.13 to 1.86]; P = .003). PD-L2 was strongly associated with immune-related pathways on gene set enrichment analysis suggesting that it is playing an important role in immune modulation in clinical prostate cancer samples. Furthermore, PD-L2 was correlated with radiation response pathways, and also predicted response to postoperative radiation therapy (PORT) on multivariable interaction analysis (P = .03).CONCLUSION: In the largest study of its kind to date, these results illustrate the complex relationship between the tumor-immune interaction, prognosis, and response to radiotherapy, and nominate PD-L2 as a potential novel therapeutic target in prostate cancer, potentially in combination with radiotherapy.