Individualized transcriptomic changes in malignant and benign prostate identify markers of disease recurrence

More accurate methods for early prediction of cancer aggressiveness are essential for enhanced prostate cancer (PC) treatment. In this study, the transcriptome profiles from malignant and adjacent benign prostate tissue samples were compared in 41 patients who underwent radical prostatectomy. Hierarchical clustering clearly separated benign and malignant tissues and included expected transcriptional changes associated with carcinogenesis. Using reproducibility-optimized statistical testing (ROTS), we identified 45 genes whose expression differed between paired malignant and benign prostate tissue from the same patient (FDR < 0.05) in patients with and without biochemical recurrence (BCR). Within this individualized approach, two interconnected regulatory pathways underlying the pathogenesis were identified: one centered on FOS, NR4A, and EGR family transcription factors, with strong involvement of TNFα signaling, and another, a less dense network, centered on IL6. Expression of the 45 identified transcripts was also higher in the more aggressive Luminal B-type PC than in Luminal A- and Basal-type PC, further underscoring their association with PC aggressiveness. A machine learning method (SIVS) was then used to further define five transcripts, BPIFB2, C11orf96, DUSP5, NR4A2, and NR4A3, whose individualized malignant to benign expression differences were most strongly associated with an increased risk of BCR, and thus, providing novel information on their potential involvement in processes increasing the PC aggressiveness.