Gene expression in normal-appearing tissue adjacent to prostate cancers are predictive of clinical outcome: evidence for a biologically meaningful field effect

// Cristina Magi-Galluzzi 1,2 , Tara Maddala 3 , Sara Moscovita Falzarano 1 , Diana B. Cherbavaz 3 , Nan Zhang 3 , Dejan Knezevic 3 , Phillip G. Febbo 3 , Mark Lee 3 , Hugh Jeffrey Lawrence 3 and Eric A. Klein 2 1 Pathology and Laboratory Medicine Institute, Cleveland Clinic, Cleveland, Ohio, USA 2 Glickman Urological and Kidney Institute, Cleveland Clinic, Cleveland, Ohio, USA 3 Genomic Health, Inc., Redwood City, California, USA Correspondence to: Hugh Jeffrey Lawrence, email: // Keywords : prostate cancer, gene expression profiling, molecular diagnostics, prognosis, risk assessment Received : April 07, 2016 Accepted : April 13, 2016 Published : April 22, 2016 Abstract Purpose: We evaluated gene expression in histologically normal-appearing tissue (NT) adjacent to prostate tumor in radical prostatectomy specimens, assessing for biological significance based on prediction of clinical recurrence (cR - metastatic disease or local recurrence). Results: A total of 410 evaluable patients had paired tumor and NT. Forty-six genes, representing diverse biological pathways (androgen signaling, stromal response, stress response, cellular organization, proliferation, cell adhesion, and chromatin remodeling) were associated with cR in NT (FDR < 20%), of which 39 concordantly predicted cR in tumor (FDR < 20%). Overall GPS and its stromal response and androgen-signaling gene group components also significantly predicted time to cR in NT (RM-corrected HR/20 units = 1.25; 95% CI: 1.01−1.56; P = 0.024). Experimental Design: Expression of 732 genes was measured by quantitative reverse transcriptase polymerase chain reaction (RT-PCR) separately in tumor and adjacent NT specimens from 127 patients with and 374 without cR following radical prostatectomy for T1/T2 prostate cancer. A 17-gene expression signature (Genomic Prostate Score [GPS]), previously validated to predict aggressive prostate cancer when measured in tumor tissue, was also assessed using pre-specified genes and algorithms. Analysis used Cox proportional hazards models, Storey’s false discovery rate (FDR) control, and regression to the mean (RM) correction. Conclusions: Gene expression profiles, including GPS, from NT adjacent to tumor can predict prostate cancer outcome. These findings suggest that there is a biologically significant field effect in primary prostate cancer that is a marker for aggressive disease.