Presentation Authors: Alp Tuna Beksac*, Ishan Paranjpe, Ugo Falagario, Alberto Martini, Shivaram Cumarasamy, Sara Lewis, Bachir Taouli, Art Rastinehad, Ash Tewari, New York, NY
Introduction: Multiparametric magnetic resonance imaging (mpMRI) has revolutionized the diagnosis of prostate cancer. However, there is a group of prostate cancer patients for whom mpMRI has not been able to demonstrate a cancerous lesion. We sought to analyze the transcriptomic and histologic differences visible and non-visible mpMRI lesions.
Methods: From a single surgeon series, we have performed a retrospective analysis on 283 patients who underwent radical prostatectomy (RP). All patients had preoperative mpMRI and postoperative genomic classifier (GC) test performed. We identified the location of the tissue block submitted for GC score analysis and retrospectively analyzed whether cancer tissue corresponded to an mpMRI lesion or not. PI-RADS v2 was used for characterizing lesions. Microarray based gene expression analysis was performed to gather RNA expression data. Weighted gene correlation network analysis (WGCNA) was applied to identify transcriptomic signatures associated with mpMRI visibility.
Results: Tumors were visible in 195 (68.9%) of patients. Among baseline cahracteristics, there were no significant differences with regards to age (p=0.682), race (p=0.687), body mass index (p=0.449), PSA (p=0.603), Gleason score(p=0.535), pT stage (p=0.425), or Decipher risk group (p=0.211) between the two groups. WGCNA identified networks of correlated genes related to mpMRI visibility. After adjusting for PSA, race, pT stage, and Gleason score, WGCNA revealed 1 gene network (424 genes) whose eigengene was significantly downregulated in patients with mpMRI visible tumors (p = 0.077). These downregulated genes showed significant enrichment (FDR < 0.01) of genes related to extracellular matrix organization, antibody-mediated complement activation, smooth muscle contraction, and integrin signaling. Using a previously published gene signature, we also found that mpMRI-visible tumors displayed greater expression of an androgen receptor signaling gene signature compared to mpMRI-invisible tumors.
Conclusions: Network based transcriptomic analysis revealed a 424 gene network enriched for genes related to extracellular matrix organization and complement activation that offers a novel insight into molecular mechanisms of mpMRI visibility in prostate cancer.