JU INSIGHT Does Perioperative Testosterone Predict Post-Prostatectomy Genomic Risk Score?

By: Mohammed Shahait, MBBS; Joseph G. Cheaib, MD; Elai Davicioni, PhD; Yang Liu, PhD; Ibrahim Abu Ghaida, MD; Ryan W. Dobbs, MD; Mohamed Alshalalfa, PhD; Priti Lal, MD; Daniel J. Lee, MD; David I. Lee, MD | Posted on: 01 Sep 2022

Shahait M, Cheaib JG, Davicioni E, et al. Does perioperative testosterone predict post-prostatectomy genomic risk score? J Urol. 2022;(6)1214-1221.

Study Needs and Importance

The role of endogenous testosterone in de novo prostate cancer pathogenesis in humans remains unclear. The Decipher^® genomic classifier (GC) post-radical prostatectomy has been shown to be highly correlated with Gleason grade and predictive of the risk of metastasis and prostate cancer-specific mortality. If serum testosterone levels truly affect high-risk prostate cancer pathogenesis, this should be reflected in the tumor transcriptome. As such, a low testosterone level should be a predictor of the presence of a high GC score. The effect of testosterone on the tumor genome is not explored.

“The Decipher^® genomic classifier (GC) post-radical prostatectomy has been shown to be highly correlated with Gleason grade and predictive of the risk of metastasis and prostate cancer-specific mortality.”

Figure. Association of serum testosterone levels with previously described prostate cancer gene-expression signatures. The log p value of the association between signature scores and serum testosterone levels are plotted by the odds ratio effect size, after adjusting for PSA, BMI, Gleason score, and tumor pathological stage. NS, not significant. Beta, beta value (regression coefficient). Adv Path, adverse pathology.

What We Found

We explored the correlation between perioperative testosterone level on genomic risk score in 339 men who underwent radical prostatectomy and had adverse pathological features in their final surgical specimens (positive margin and/or pT3a or higher) at final pathology. The median genomic risk score was lower in the low testosterone group compared to the intermediate and normal testosterone groups (0.38 vs 0.52 vs 0.53, respectively; p=0.049). There was no difference in biochemical recurrence-free survival between the 3 testosterone groups (p=0.9). Patients with low testosterone levels had higher odds of receiving secondary treatment (OR: 2.27; 95% CI: 1.14–4.50; p=0.02) than those with normal levels. A total of 43 (of 188) gene expression signatures were associated with testosterone level (p <0.05). In total, 33 signatures were positively associated with serum testosterone levels, including 12 signatures involved in DNA repair pathways (see figure).

“A total of 43 (of 188) gene expression signatures were associated with testosterone level (p <0.05).”

Limitations

The generalizability of our results is limited by this being a single-center, single-surgeon study with a short followup. Also, in this study a commercial prostate cancer diagnostic kit was used that examines only an index tumor which is associated with significant pathological parameters, such as the highest Gleason score, the largest tumor volume and extraprostatic extension, and clinical events such as metastasis; however, this approach can miss identification of some aggressive tumors due to intratumoral and intertumoral heterogeneity, and given the multifocal nature of localized prostate cancer.

Interpretation for Patient Care

This is the first study to assess the correlation of preoperative testosterone levels on the tumor transcriptome and showed no significant clinical correlation between pre-defined GC score risk groups and the testosterone group. This study adds to the notion that endogenous testosterone plays a limited role in the development of de novo high-risk localized prostate cancer.