Hormuzd Katki

NCI

Investigator

PLCO (Learn more about this study)

PLCO-127

Feb 3, 2015

Assessing the value of a SNP-based score to predict outcomes from prostate cancer screening

To improve the efficiency of PSA-based prostate cancer screening, many researchers have suggested basing screening guidelines on individual risk calculations. We propose to examine how risk of each important outcome (total prostate cancer, aggressive prostate cancer, and prostate cancer death) stratifies based on PSA and elicitable information (age, race, family history, etc.), and most notably, the SNPs that have been recently associated with prostate cancer. We plan to use the most currently-published SNPs (currently about 100 SNPs from http://www.ncbi.nlm.nih.gov/pubmed/25217961), and use a SNP-score that is the sum of the number of SNPs at that locus multiplied by their published log-odds ratios. We will examine if the score improves risk stratification of each of the 3 outcomes beyond PSA and elicitable information alone. We will calculate risk by weighting up the case-control sample in the PLCO screening arm with SNP information (about 5000 prostate cancer cases and 3000 controls) to represent the full PLCO screening arm. We will use a weighted Cox survival model to incorporate all risk factors to estimate risk. We will develop the sampling weights by using the known selection criteria for SNP genotyping. Within ranges of PSA results, we will examine if the SNP-score substantially changes the risk of each of the 3 outcomes. We also want to consider risk of a false-positive biopsy (i.e. The man was referred for biopsy based on PSA, but no cancer was found), develop a logistic regression risk model for false-positive biopsy, and see if this risk also stratifies by the SNP-score (if so, this suggests that SNP-score might be able to triage men with positive PSA test results to biopsy or watchful-waiting). Although the current SNPs are not known to differentiate between lethal and non-lethal prostate cancer, combining PSA results and SNP-score may identify men at very low risk of prostate cancer who could either be safely excluded from further PSA screening, or at least safely return for their next screen only after an extended period of time.