Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Urology. Author manuscript; available in PMC 2017 April 20.
Published in final edited form as:
PMCID: PMC5398897

Impact of Prostate Specific Antigen on a Baseline Prostate Cancer Risk Assessment Including Genetic Risk



To determine to what extent prostate cancer risk prediction is improved by adding PSA to a baseline model including genetic risk.

Materials and Methods

Peripheral blood DNA was obtained from Caucasian men undergoing prostate biopsy at the University of Toronto (9/1/2008 – 1/31/2010). Thirty-Three PCa risk-associated single neucleotide polymorphisms (SNPs) were genotyped to generate the PGS-33. Primary outcome is PCa on study prostate biopsy. Logistic regression, area under the receiver-operating characteristic curves (AUC) and net reclassification improvement (NRI) were used to compare models.


Among 670 subjects, 323 (48.2%) were diagnosed with PCa. The PGS-33 was highly associated with biopsy detectable PCa (OR 1.66, p=5.86E-05, AUC 0.59) compared to PSA (OR 1.33, p=0.01, AUC 0.55). PSA did not improve risk prediction when added to a baseline model (age, family history, DRE, and PGS-33) for overall risk (AUC 0.66 vs. 0.66; p=0.86) or Gleason ≥ 7 PCa (AUC 0.71 vs. 0.73; p=0.15). NRI analyses demonstrated no appropriate reclassifications with the addition of PSA to the baseline model for overall PCa but did show some benefit for reclassification of men thought to be at higher baseline risk in the high-grade PCa analysis.


In a baseline model of PCa risk including the PGS-33, PSA does not add to risk prediction for overall PCa for men presenting for “for cause” biopsy. These findings suggest that PSA screening may be minimized in men at low baseline risk.

Keywords: Prostate Cancer, Screening, Genetics


Prostate cancer (PCa) is the most common malignancy affecting American men with an estimated 241,740 new cases and 28,170 deaths expected in 2013.(1) Unfortunately, prostate specific antigen (PSA) based PCa screening has not lead to a definitive improvement in mortality in North American studies and little benefit in European analyses.(2, 3) Due to the large number needed to screen to save one life and the concern of side effects from overtreatment, the U.S. Preventative Task Force has advocated against PSA-based PCa screening.(4) Professional societies such as the American College of Physicians and the American Urological Association have, as a result downplayed the importance of PSA based screening for men at “average” risk.

However, controversy regarding PSA screening continues with proponents of screening pointing to several studies demonstrating improved overall and progression free survival with early PCa treatment.(58) One strategy to reduce the morbidity of screening while maintaining the benefits of early detection is by limiting aggressive screening to those men at highest risk. An easily assessable baseline risk factor, which does not change through out a man’s life, is his germ-line DNA.

Single nucleotide polymorphisms (SNP’s), germ-line DNA markers, identified from previous genome wide association studies (GWAS) studies have demonstrated improved PCa risk prediction in various settings.(911) More recently, the prostate cancer genetic score (PGS-33) has been developed which incorporates 33 PCa associated SNP’s. The PGS-33 improved PCa risk prediction in a randomized controlled trial where men underwent non for-cause prostate biopsy.(10)

The current clinical paradigm for PCa screening employs age, family history, digital rectal examination (DRE) and PSA. Herein we describe the use of the PGS-33 in context with these baseline characteristics and investigate the benefit of adding or excluding PSA in predicting PCa in a cohort of men presenting for prostate biopsy.


Study Population

Following institutional review board approval and informed consent, a total of 670 consecutive men undergoing prostate biopsy for the diagnosis of PCa at the University Health Network in Toronto, Canada, were enrolled from September 1st 2008 until January 31, 2010. Patients were biopsied for elevated PSA, elevated PSA velocity, a nodule on digital rectal exam (DRE), or otherwise at the discretion of the urologists in discussion with patients. Blood specimens were obtained by venipuncture prior to biopsy. Clinical data was collected at the time of phlebotomy, including date of birth, most recent PSA (prior to biopsy), family history of prostate cancer (limited to first degree relatives) and race. A single radiologist performed all DRE assessments, and prostate biopsies as standard protocol at University of Toronto. Caucasian men presenting for prostate biopsy were enrolled sequentially as the PGS-33 score has not been fully investigated in other ethnicities at this point. All prostate biopsies had 10–12 cores sampled and pathology was read by one of 3 pathologists that specialize in urologic oncology.

PCa Associated Single Nucleotide Polymorphisms and Genetic Score

Generation of the PGS-33 was performed as previously described.(10) In brief, SNPs from all PCa GWASs reported before December 2009, which exceeded genome-wide significance levels in initial reports (P < 10−7), and independently replicated were selected for inclusion in the PGS-33.(10) The SNPs were genotyped using the Sequenom MassARRAY platform. One duplicated CEPH (Centre d’Etude du Polymorphisme Humain) sample and two water samples (negative controls) that were blinded to technicians were included in each 96-well plate. The concordance rate between the two genotype cells of the duplicated CEPH sample was 100% for all SNPs.

Statistical analyses

Derivation of the PGS-33 has been previously described.(10) In brief, the genetic score is determined by weighted odds ratios developed based on external meta-analyses for each SNP.(12, 13)

Differences in binary variables (family history, DRE) and continuous variables (age, PSA measurements, prostate volume, number of cores at pre-study entry biopsy, and genetic score) between men with and without positive prostate biopsy were tested using a Chi-square and t-test, respectively. Total PSA levels and genetic score were log transformed to approach a normal distribution.

Area under the receiver operating characteristic curves (AUC) was used to assess the ability of each of the clinical variables, PSA, and the genetic score to predict for PCa at biopsy. The proposed best clinical model was based on clinical information easily obtained without invasive testing. Logistic regression was used to determine the predictive values from combination models. Differences in AUC were assessed using Delong’s test.(14)

Net reclassification improvement (NRI) was used to measure the degree to which PCa risk was appropriately reclassified using a baseline model including age, family history, DRE, and PGS-33 with or without PSA.(15) Each risk result was classified into a low (1st quartile), intermediate (2nd and 3rd quartiles), or high (4th quartile) risk categories for the detection of a positive prostate biopsy as previously described.(10) In addition, an assessment of the performance of prediction models at discriminating high-grade PCa (Gleason score ≥7) was performed.


Among 670 subjects, 347 (51.8%) had a negative biopsy and 323 (48.2%) had PCa found on prostate biopsy. Of patients with a positive biopsy, 202 (62%) had clinical stage T1c, 119 (37%) had clinical stage T2, and only 2 (1%) patients had clinical stage T3 or higher. Clinical variables between the two groups are displayed in Table 1. Patients diagnosed with prostate cancer were older, had a higher PSA, abnormal DRE, and a higher PGS-33. Family history did not differ between the two groups.

Table 1
Clinical variables and genetic score of the subjects in the study

In univariate analysis, the PGS-33 was highly associated with biopsy detectable PCa (OR 1.66, p=<0.001), exceeding the association seen with PSA (OR 1.33, p=0.01) (Table 2). PSA exceeded the PGS-33 in predicting high-grade cancer (Gleason ≥ 7) with odds ratios of 2.1 (p<0.001) and 1.70 (p<0.001), respectively. AUC values for overall PCa risk for clinical variables including age, family history, and DRE were 0.56, 0.52, and 0.61, respectively. The AUC for the PGS-33 was 0.59 exceeding that of PSA at 0.55.

Table 2
Univariate model of demographics and risk of overall and high-grade prostate cancer.

In multivariate analysis, the predictive value of a baseline four variable model (age, family history, DRE, and PGS-33) had an AUC of 0.66 for overall and 0.71 for high-grade PCa (Gleason ≥ 7). PSA did not improve risk prediction when added to this 4 variable model with an AUC of 0.66 (p=0.86) for overall and 0.73 for high-grade PCa (p=0.15).

Reclassification analysis was performed to determine if using the clinical model with genetic score would be reclassified based on the addition of PSA. There was no net improved reclassification for overall PCa detection (all p values >0.05). (Table 3) In regards to high-grade prostate cancer, there was a trend towards improved reclassification (p=0.06) on the total analysis. On subset analyses, PSA did appropriately reclassify 16% of intermediate (2nd and 3rd quartile, p=0.06) risk men and 22.3% of men in the highest quartile of risk (4th quartile, p<0.001), but no improvement in reclassification of risk was seen for men in the lowest risk quartile (p=0.25). (Table 4)

Table 3
Number of men classified as low, middle, or high risks from predictive models with or without genetic score
Table 4
Number of men classified as low, middle, or high risks for high-grade PCa from predictive models with or without genetic score


The PGS-33, calculated from inherited PCa risk associated SNP’s, is an independent predictor of prostate cancer risk in this population of Caucasian men undergoing “for-cause” prostate biopsy. The genetic score performed well as compared to known clinical variables for elevated risk of prostate cancer with an overall odds ratio of 1.66 (1.29–2.12; p<0.001) (Table 2). PSA did not add to overall PCa risk prediction but may have some value for high-grade PCa risk prediction in men at elevated baseline risk.

These results are similar to previous studies showing strong independent associations of genetic markers with PCa risk.(9, 10, 1618) However, despite demonstrating comparable performance of a genetic test using a 12 SNP model,(AUC 0.57) Klein et. al suggested superior risk prediction with PSA (AUC 0.79). It is important to note that this was a nested case-control study performed on men in the Malmö Diet and Cancer Swedish undergoing PSA testing early in life.(19) In contrast, our study showed the AUC for the genetic score (AUC 0.59) exceeded that of any other PCa predictor including PSA (AUC 0.54). The discrepancy in PSA performance is likely due to its analysis in an unscreened population as compared to our for-cause biopsy cohort of men regularly screened for PCa. Our findings are supported by a study by Aly and colleagues who employed a 35 SNP genetic test to show improved prediction of biopsy outcomes in more than 5,000 Swedish men undergoing for-cause prostate biopsy.(11) The investigators showed that PSA had the lowest AUC at 0.55 in a multivariate model including age, family history and genetic score.

Despite several studies demonstrating improved PCa risk prediction with the addition of germ line genetic markers to current predictors, no study to date has examined the impact of adding PSA to a baseline model including genetic markers. In an attempt to better risk stratify men considering or being considered for PSA based screening, the impact of adding PSA to a baseline model including easily obtainable clinical variables (age, family history and DRE) and the PGS-33 (an inexpensive, saliva based test) was assessed in this study. The baseline, four variable model had an AUC of 0.66 for overall and 0.71 for high-grade PCa (Gleason ≥ 7), which was not improved by PSA. To further confirm the results we performed a NRI analysis to determine if PSA would reclassify patients previously stratified by the baseline model.

The NRI analysis did not show significant benefit of adding PSA to the risk prediction model in terms of overall PCa risk assessment. However, in the complete analysis, there was a trend towards improved reclassification for high-grade prostate cancer risk assessment (p=0.06) and upon subset analysis, intermediate (p=0.06) and high-risk groups (p<0.001) had a significantly improved reclassification with the addition of PSA. The biggest reclassification was noted for men in the highest (4th quartile) risk category where 29% (26/89) of individuals were appropriately reclassified into lower risk categories.

While PSA has prognostic value, prostate cancer can occur at any PSA value and, due to its elevation in non-malignant conditions it suffers from poor sensitivity and specificity.(2022) To examine the impact of PSA based PCa screening and the early detection of PCa, two large randomized controlled screening trials were conducted.(2, 3) The American Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial did not demonstrate a life saving benefit and the European Randomized Study of Screening for Prostate Cancer suggested that over 1000 men needed to be screened in order to save 1 life. It is difficult to reconcile this with the fact that PCa is the second leading cause of cancer related mortality. It is clear that a better screening algorithm is needed to focus early detection on men at greatest risk. One possible approach could be to use PSA selectively after the PGS-33 to more accurately stratify men at elevated baseline risk with respect to their risk of high-grade PCa. This combined approach may maintain the benefits of early detection while reducing the harms associated with broad-based PSA screening.

Limitations of this study include the limited sample size in a PSA screened study population. Furthermore, test performance can be altered due to changes in “pre-test probability” given that the population studied are presenting for “for cause” prostate biopsy this may play a factor in this study. While there were over 600 participants, evaluation of higher-risk and higher-grade patients was limited due to loss of power. Prostate biopsies and DREs in this analysis were performed by a single radiologist, which may have improved the accuracy of these tests. This may have impacted the AUC such that the addition of PSA was imperceptible. Moreover, the limitation of transrectal ultrasound guided prostate and possibility of a false negative result could have had an impact on our findings. Genetic guided risk based screening may need to be tested in a prospective screening cohort.


PSA does not add to risk prediction for overall PCa but may add to risk prediction for high-grade disease in men at higher baseline risk. These findings suggest that PSA screening may be minimized particularly for men at low risk based on a clinical model including the PGS-33.


Funding: The study is partially supported by a National Cancer Institute RC2 grant (CA148463) to Dr. Xu.


Disclosures: A patent application has been filed by the Wake Forest University School of Medicine, to preserve patent rights for the technology and results related to the 33 SNPs used in this study. Neil E. Fleshner has received research grants and honorarium from GlaxoSmithKline, consultant/advisor for Merck and GlaxoSmithKline.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin. 2012;62(1):10–29. [PubMed]
2. Andriole GL, Crawford ED, Grubb RL, 3rd, et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med. 2009;360(13):1310–9. [PMC free article] [PubMed]
3. Schroder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009;360(13):1320–8. [PubMed]
4. Moyer VA. Force USPST. Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2012;157(2):120–34. [PubMed]
5. Hugosson J, Carlsson S, Aus G, et al. Mortality results from the Goteborg randomised population-based prostate-cancer screening trial. Lancet Oncol. 2010;11(8):725–32. [PMC free article] [PubMed]
6. Roobol MJ, Steyerberg EW, Kranse R, et al. A risk-based strategy improves prostate-specific antigen-driven detection of prostate cancer. Eur Urol. 2010;57(1):79–85. [PubMed]
7. Etzioni R, Tsodikov A, Mariotto A, et al. Quantifying the role of PSA screening in the US prostate cancer mortality decline. Cancer Causes Control. 2008;19(2):175–81. [PMC free article] [PubMed]
8. Wilt TJ, Brawer MK, Jones KM, et al. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med. 2012;367(3):203–13. [PMC free article] [PubMed]
9. Zheng SL, Sun J, Wiklund F, et al. Cumulative association of five genetic variants with prostate cancer. N Engl J Med. 2008;358(9):910–9. [PubMed]
10. Kader AK, Sun J, Reck BH, et al. Potential Impact of Adding Genetic Markers to Clinical Parameters in Predicting Prostate Biopsy Outcomes in Men Following an Initial Negative Biopsy: Findings from the REDUCE Trial. Eur Urol. 2012;62(6):953–61. [PMC free article] [PubMed]
11. Aly M, Wiklund F, Xu J, et al. Polygenic risk score improves prostate cancer risk prediction: results from the Stockholm-1 cohort study. Eur Urol. 2011;60(1):21–8. [PMC free article] [PubMed]
12. Kim ST, Cheng Y, Hsu FC, et al. Prostate cancer risk-associated variants reported from genome-wide association studies: meta-analysis and their contribution to genetic Variation. Prostate. 2010;70(16):1729–38. [PMC free article] [PubMed]
13. Pharoah PD, Antoniou AC, Easton DF, Ponder BA. Polygenes, risk prediction, and targeted prevention of breast cancer. N Engl J Med. 2008;358(26):2796–803. [PubMed]
14. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–45. [PubMed]
15. Pencina MJ, D’Agostino RB, Sr, D’Agostino RB, Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008;27(2):157–72. discussion 207–12. [PubMed]
16. Fitzgerald LM, Kwon EM, Koopmeiners JS, et al. Analysis of recently identified prostate cancer susceptibility loci in a population-based study: associations with family history and clinical features. Clin Cancer Res. 2009;15(9):3231–7. [PMC free article] [PubMed]
17. Xu J, Sun J, Kader AK, et al. Estimation of absolute risk for prostate cancer using genetic markers and family history. Prostate. 2009;69(14):1565–72. [PMC free article] [PubMed]
18. Salinas CA, Koopmeiners JS, Kwon EM, et al. Clinical utility of five genetic variants for predicting prostate cancer risk and mortality. Prostate. 2009;69(4):363–72. [PMC free article] [PubMed]
19. Klein RJ, Hallden C, Gupta A, et al. Evaluation of multiple risk-associated single nucleotide polymorphisms versus prostate-specific antigen at baseline to predict prostate cancer in unscreened men. Eur Urol. 2012;61(3):471–7. [PMC free article] [PubMed]
20. Kilpelainen TP, Tammela TL, Roobol M, et al. False-positive screening results in the European randomized study of screening for prostate cancer. Eur J Cancer. 2011;47(18):2698–705. [PubMed]
21. Thompson IM, Pauler DK, Goodman PJ, et al. Prevalence of prostate cancer among men with a prostate-specific antigen level < or =4.0 ng per milliliter. N Engl J Med. 2004;350(22):2239–46. [PubMed]
22. Lilja H, Cronin AM, Dahlin A, et al. Prediction of significant prostate cancer diagnosed 20 to 30 years later with a single measure of prostate-specific antigen at or before age 50. Cancer. 2011;117(6):1210–9. [PMC free article] [PubMed]