Machine Learning on a Genome-Wide Association Study to Predict Late Genitourinary Toxicity Following Prostate Radiotherapy

Publication date: Available online 31 January 2018
Source:International Journal of Radiation Oncology*Biology*Physics
Author(s): Sangkyu Lee, Sarah Kerns, Harry Ostrer, Barry Rosenstein, Joseph O. Deasy, Jung Hun Oh
PurposeLate genitourinary (GU) toxicity after radiotherapy limits the quality of life of prostate cancer survivors, but efforts to explain GU toxicity using patient/dose information remain unsuccessful. We aimed to identify patients with higher congenital GU toxicity risk by identifying and integrating patterns in genome-wide single nucleotide polymorphisms (SNPs).Materials and MethodsWe applied a pre-conditioned random forest regression (PRFR) method for predicting risk from the genome-wide data to combine the effects of multiple SNPs and overcome statistical power limitations of single-SNP analysis. We studied a cohort of 324 prostate cancer patients who were self-assessed for four urinary symptoms at 2 years after radiotherapy using the International Prostate Symptom Score.ResultsThe predictive accuracy of the methodology varied across symptoms. Only for the weak stream endpoint, it achieved a significant area under the curve of 0.70 (95% confidence interval: 0.54 – 0.86, p = 0.01) on hold-out validation data that outperformed competing methods. Gene ontology analysis highlighted key biological processes, such as neurogenesis and ion transport, from the genes known to be important for urinary tract functions.ConclusionWe applied machine learning methods and bioinformatics tools to genome-wide data to predict and explain GU toxicity. Our approach enabled designing a more powerful predictive model and finding plausible biomarkers and biological processes associated with GU toxicity.



http://ift.tt/2EszzwN