Comprehensive Evaluation of Machine Learning Models and Gene Expression Signatures for Prostate Cancer Prognosis Using Large Population Cohorts

Overtreatment remains a pervasive problem in prostate cancer management due to the highly variable and often indolent course of disease. Molecular signatures derived from gene expression profiling have played critical roles in guiding prostate cancer treatment decisions. Many gene expression signatures have been developed to improve the risk stratification of prostate cancer and some of them have already been applied to clinical practice. However, no comprehensive evaluation has been performed to compare the performance of these signatures. In this study, we conducted a systematic and unbiased evaluation of 15 machine learning (ML) algorithms and 30 published prostate cancer gene expression-based prognostic signatures leveraging 10 transcriptomics datasets with 1,558 primary patients with prostate cancer from public data repositories. This analysis revealed that survival analysis models outperformed binary classification models for risk assessment, and the performance of the survival analysis methods-Cox model regularized with ridge penalty (Cox-Ridge) and partial least squares (PLS) regression for Cox model (Cox-PLS)-were generally more robust than the other methods. Based on the Cox-Ridge algorithm, several top prognostic signatures displayed comparable or even better performance than commercial panels. These findings will facilitate the identification of existing prognostic signatures that are promising for further validation in prospective studies and promote the development of robust prognostic models to guide clinical decisionmaking. Moreover, this study provides a valuable data resource from large primary prostate cancer cohorts, which can be used to develop, validate, and evaluate novel statistical methodologies and molecular signatures to improve prostate cancer management. Significance: This systematic evaluation of 15 machine learning algorithms and 30 published gene expression signatures for the prognosis of prostate cancer will assist clinical decision-making.

Automated summary

This study systematically evaluated the performance of machine learning algorithms and gene expression signatures for prostate cancer prognosis. Survival analysis models outperformed binary classification models, with Cox-Ridge and Cox-PLS methods being the most robust. These findings can assist clinical decision-making.