2. UroToday Home
  3. Recent Abstracts
  4. Urologic Oncology
  5. Prostate Cancer

How Well Do Polygenic Risk Scores Identify Men at High Risk for Prostate Cancer? Systematic Review and Meta-Analysis - Beyond the Abstract

Prostate cancer (PCa) has an extremely variable clinical course and especially indolent, low-risk cases have posed a major problem of overdiagnosis and overtreatment for PSA-based screening, even though detection of aggressive cases has resulted in decreased PCa-specific mortality.2 To better balance benefits and harms, strategies to improve PCa screening focus on multistep testing, where PSA is combined with other biomarkers and with magnetic resonance imaging (MRI).3,4 Thus, the goal is to detect potentially harmful PCa, while minimizing overdiagnosis of low-risk disease.

Prostate cancer has high heritability and genome-wide association studies (GWAS) have found over two hundred susceptibility loci for PCa, with large majority exerting only a small impact on the risk of developing PCa.5 Polygenic risk scores (PRS) integrating the effect across single nucleotide polymorphisms (SNPs) have potential for identifying men with high-risk PCa and hence allow the development of a personalized, risk-stratified screening strategy for PCa. It may be possible to reduce the number of men who need further examination for PCa by selecting men for screening based on PRS classification (e.g. men in the highest decile of PRS). Empirical evidence for the application of such a strategy in screening trials is, however, largely lacking. In the BARCODE1 trial pilot, which aimed to evaluate the usability of PRS in the selection of men for PCa screening, 39% of the men in the highest PRS decile were diagnosed with PCa at screening.6 All cases were low-risk PCa. In the IMPACT study, 640 men with mutations in DNA repair genes (MLH1, MSH2, or MSH6) had a higher detection rate in PSA-based screening than the men in the control group.7 In that study, the majority of the detected cases were classified as clinically significant.

Currently, evidence is lacking on what extent using genetic predisposition is able to predict the risk of PCa. To evaluate this, we performed a systematic review and meta-analysis to summarize the evidence on the accuracy of PRS in predicting the risk of PCa.8

We calculated pooled area under the receiver-operating curve (ROC AUC) estimate from 16 individual publications. Overall, the predictive accuracy was mediocre (pooled AUC 0.63 (95% CI 0.62-0.64). As a subgroup analysis, we pooled AUC values from the studies where PRS was combined with clinical variables. Then the pooled estimate for the combined clinical and genetic risk score increased to 0.74, which still remained of limited value. Heterogeneity was high or moderate in all analyses, indicating substantial differences between the original publications. However, we found no indication of publication bias. We also performed sensitivity analyses but found no clear predictors of differences in results between the studies. Also, we attempted to conduct pooled analysis using odd ratios as a measure of effect, however, it ended up impossible due to a highly inconsistent fashion in terms of subject grouping causing high heterogeneity.

This study was conducted in collaboration within the PIONEER consortium. The main goal of this consortium is to use Big Data for improved outcomes, health system efficiency, and the quality of health and social care for PCa patients in Europe.9 To answer unknown questions, one option might conduct an analysis using the Big Data to analyze the usability of PRS in an environment with harmonized data. In this approach data set will be wide with all variables in a common format with a common representation, PRS is formed and calculated equally, and same clinical variables can be fit into the models.

Based on our meta-analysis, the predictive value of PRS is comparable to PSA. Also, based on individual studies, adding PRS to clinical prediction models, e.g. with PSA, increased AUC less than 0.05. Furthermore, standardized procedures for estimating PRS is needed to decrease heterogeneity, e.g. selection of SNPs, methods in calculation of PRS, and what is the cut-off value for following examination (such as other biomarkers, MRI, biopsies, digital rectal examination).

Currently, evidence is lacking to evaluate whether using genetic predisposition as a criterion for targeting screening can increase detection of aggressive versus non-aggressive PCa. Our results show that performance of PRS in predicting PCa is comparable to PSA. More accurate prediction of PCa risk is needed. Improved methods are being developed and new results published continuously, especially future research focusing on prediction of aggressive PCa can improve usability of PRS in PCa prediction.

Written by: Aino Siltari,1 Anders Bjatell,2 Anssi Auvinen3

  1. Faculty of Medicine and Health Technology, Tampere University, Tampere Finland
  2. Department of Translational Medicine, Lund University, Malmö, Sweden
  3. Faculty of Social Sciences, Tampere University, Tampere, Finland

Reference:

  1. Sung H, Ferlay J, Siegel RL et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209-249.
  2. Hugosson J, Roobol MJ, Månsson M et al. A 16-yr Follow-up of the European Randomized study of Screening for Prostate Cancer. Eur Urol 2019;76:43-51.
  3. Auvinen A, Rannikko A, Taari K et al. A randomized trial of early detection of clinically significant prostate cancer (ProScreen): study design and rationale. Eur J Epidemiol. 2017;32:521-527.
  4. Eklund M, Jäderling F, Discacciati A, Bergman M, Annerstedt M, Aly M, Glaessgen A, Carlsson S, Grönberg H, Nordström T; STHLM3 consortium. MRI-Targeted or Standard Biopsy in Prostate Cancer Screening. N Engl J Med. 2021;385:908-920
  5. Conti DV, Darst BF, Moss LC et al. Trans-ancestry genome-wide association meta-analysis of prostate cancer identifies new susceptibility loci and informs genetic risk prediction. Nat Genet 2021;53:65-75.
  6. Benafif S, Ni Raghallaigh H, McGrowder E, et al. The BARCODE1 Pilot: a feasibility study of using germline single nucleotide polymorphisms to target prostate cancer screening. BJU Int 2022;129:325-336.
  7. Bancroft EK, Page EC, Brook MN, et al. A prospective prostate cancer screening programme for men with pathogenic variants in mismatch repair genes (IMPACT): initial results from an international prospective study. Lancet Oncol. 202;22:1618-1631.
  8. Siltari A, Lönnerbro R, Pang K et al. How Well do Polygenic Risk Scores Identify Men at High Risk for Prostate Cancer? Systematic Review and Meta-Analysis. Clin Genitourin Cancer. 2022: S1558-7673(22)00196-3. doi: 10.1016/j.clgc.2022.09.006.
  9. Omar MI, Roobol MJ, Ribal MJ et al. Introducing PIONEER: a project to harness big data in prostate cancer research. Nat Rev Urol. 2020;17:351-362. Erratum in: Nat Rev Urol. 2020;17:482.
Read the Abstract