Abstract
Prostate tumors with TP53 gene mutations are molecularly heterogenous, and the presence of TP53 gene mutations has been linked to inferior outcomes. We developed an RNA-based gene signature that detects underlying TP53 gene mutations and identifies wild-type prostate tumors that are analogous to TP53-mutant tumors.
Using genomic expression profiles from The Cancer Genome Atlas, we developed a mutation signature score to predict prostatic tumors with a molecular fingerprint similar to tumors with TP53 mutations. Area under the receiver operating characteristic curve assessed model accuracy in predicting TP53 mutations, and Cox regression models measured association between the signature and progression-free survival and metastasis-free survival (MFS).
The TP53 signature score achieved an area under the receiver operating characteristic curve of 0.84 in the training and 0.82 in the validation cohorts for predicting an underlying mutation. In three retrospective cohorts, a high score was prognostic for poor 5-year MFS: 46% versus 81% (hazard ratio [HR], 3.05; P < .0001; Johns Hopkins University cohort), 64% versus 83% (HR, 2.77; P < .0001; Mayo Clinic cohort), and 71% versus 97% (HR, 6.8; P = .0001; Brigham and Women’s Hospital cohort). The signature also identified TP53 wild-type tumors molecularly analogous to TP53 mutant tumors, wherein high signature score correlated with worse 5-year MFS (50% vs. 82%; HR, 3.05; P < .0001).
This novel mutational signature predicted tumors with TP53 mutations, identified TP53 wild-type tumors analogous to mutant tumors, and was independently associated with poor MFS. This signature can therefore be used to strengthen existing clinical risk-stratification tools.
TP53 is the most frequently mutated gene in cancer. We developed a molecular signature detecting underlying TP53 alterations. This signature identified aggressive tumors, and after adjusting for known predictors of worse outcomes, the signature identified patients with prostate cancer likely to develop metastases. These findings may help doctors identify patients with prostate cancer with aggressive tumors and hence, choose the optimal treatment.
