Presentation Authors: Taylor Y. Sadun*, Los Angeles, CA, Kathleen E. Houlahan, Toronto, Canada, Amirali Salmasi, San Diego, CA, Aydin Pooli, Ely R. Felker, Steven S. Raman, Preeti Ahuja, Anthony E. Sisk, Paul C. Boutros, Robert E. Reiter, Los Angeles, CA
Introduction: Multiparametric MRI (mpMRI) has transformed prostate cancer (PCa) management by improving identification of clinically significant disease. However, ~20% of primary prostate tumors are invisible to mpMRI. We hypothesize that differences in functional mpMRI visibility reflect fundamental molecular properties of a tumor.
Methods: We profiled the transcriptomic and copy number profile of 40 Gleason Grade Group 2 tumors treated by prostatectomy. Twenty tumors were mpMRI invisible (PI-RADSv2: 1-2), while 20 tumors were visible (PI-RADsv2: 5).
Results: Copy number aberrations (CNAs) and mRNA abundance were analyzed. Univariate analysis identified 102 transcripts differentially abundant between visible vs invisible tumors. Unexpectedly, non-coding transcripts comprised the majority of differentially abundant RNAs (57/102 transcripts). In particular, snoRNAs were significantly more likely to have elevated abundance in visible tumors (OR=4.4; FDR=1.6x10-3). Perhaps most provocatively, SCHLAP1, a lncRNA linked to PCa progression, was more abundant in visible tumors (log2FC=3.2, FDR=0.028; Figure 1A). Additionally, visible tumors harbored significantly more unstable genomes, quantified as the percentage of the genome altered via CNAs (PGA; P=0.036; log2FC=2.3; Figure 1B). Concordantly, intraductal carcinoma (IDC) and cribriform architecture (CA) were enriched in PI-RADSv2 5 tumors (OR = 7.0; P=0.031; Figure 1C). Finally, we quantified a synergy between hallmarks and found the odds of visibility to be 10-fold higher with co-occurrence of â‰¥2 hallmarks (OR=10; P=5.7x10-3; Figure 1D). Nimbosus hallmarks synergized with snoRNA levels to predict visibility with 87% accuracy, superior to the 60% accuracy of the clinical signature, suggesting elevated snoRNA abundance may be a novel hallmark of nimbotic tumors (AUC=0.87, 95% CI: 0.75-0.99; Figure 1E).
Conclusions: This work points to a novel model for the origin of mpMRI visibility involving the co-occurrence of multiple aggressive hallmarks, reminiscent of nimbosus. These hallmarks include IDC/CA pathology, increased PGA and overexpression of key non-coding transcripts, such as SCHLAP1 and snoRNAs. This co-occurrence results in an aggressive tumor phenotype, with poor patient outcome.
Source of Funding: UCLA SPORE In Prostate Cancer, NIH/NCI grant number P50 CA092131