Transcriptomic Analysis of ATM Expression in Localized Prostate Cancer – Implications for Radiotherapy Treatment

Abstract

<h3>Purpose/Objective(s)</h3> The prognostic impact and therapeutic implication of ATM (ataxia telangiectasia mutated) somatic and germline alterations have been previously characterized in prostate cancer. However, ATM expression, particularly in the context of radiotherapy (RT) for localized disease, has not been well documented. Herein, genomic expression analyses were performed to evaluate associations between ATM expression, biological pathways, clinical outcomes and RT treatment response. <h3>Materials/Methods</h3> Whole-transcriptome assays were used to interrogate 5,780 radical prostatectomy (RP) samples for ATM expression (5,000 from a prospective cohort was used to evaluate biological signatures and 780 samples from a retrospective cohort with long term follow-up data to evaluate ATM prognostic impact). Additionally, we assessed 245 patients (subset of the 780) who were treated with adjuvant or salvage RT after RP. A 1:1 matched cohort [n=98 treated with RT, n=98 with no RT] was generated to assess ATM expression in relation to RT response. We next investigated the relationship using Pearson's correlation tests between ATM expression and more than 200 genomic signatures, biological pathways, and a Genomic Classifier (GC) (molecular subtypes). ATM high, medium, and low were defined as the top 25%, middle 50%, and bottom 25% of ATM expression, respectively. The prognostic and predictive ability of ATM expression was examined for the metastasis-free survival (MFS) endpoint with hazard ratios (HRs) and 95% confidence intervals (CIs) from Cox Proportional Hazard model. <h3>Results</h3> In the prospective cohort, ATM was significantly overexpressed in high GC patients (molecular surrogate for poor outcome) (p<0.0001). ATM high (top 25%) was associated with worse MFS (HR = 1.16, 95% CI: 1.51-1.96, p=0.002). ATM expression was highly correlated to signatures related to DNA repair, angiogenesis, immune cell infiltration and genomic instability and highly correlated to CDK2 and PARP1 genes (R>0.5, p<0.0001 for all). In patients treated with RT, ATM high was associated with worse MFS compared to lower ATM (10-year MFS 30% vs 61%; p=0.009), whereas in the patients treated without RT, ATM high was not significantly associated with worse MFS (10-year MFS 52% vs 66%; p=0.26) <h3>Conclusion</h3> Prostate cancer patients harboring high ATM expression by transcriptomic analysis had poorer outcomes with a significantly higher risk of metastasis. Given the worse outcomes, particularly in patients treated with RT, a high ATM expression may be a marker of resistance to DNA damaging agents. Additional studies are needed to elucidate further the predictive ability of ATM overexpression for RT personalization.