Predicting prostate cancer : on the use of biomarkers in prostate cancer diagnostics

Abstract

Aims: The aims of this thesis work were to answer the following questions. Paper I: How prevalent is testing and retesting with prostate-specific antigen (PSA)?; Paper II: Is a genetic score based on single-nucleotide polymorphisms (SNPs) informative regarding the risk of prostate cancer (PCa) in men with low PSA?; Paper III: Are the commercially available tests Prostate Health Index (PHI) and the four-kallikrein panel comparable in aiding biopsy decisions?; Paper IV: Do commonly used medications affect PSA and the risk of PCa?

Methods: In Paper I and Paper IV, the population-based PSA cohort STHLM0 was used together with registry-based data. Paper I described limited-duration point prevalence of testing and survival analysis describing retesting with PSA. Paper IV determined differences in PCa risk and PSA level among men using aspirin, statin, metformin or no medication. Paper II included 172 men with PSA at 1‒3ng/ml. Participants were invited according to their genetic score and underwent prostate biopsy. Risk of prostate cancer was assessed using logistic regression. Paper III included 531 men who had undergone a first prostate biopsy. Predictive models were compared using receiver-operating characteristics (ROC/AUC) and calculation of biopsies that could be avoided.

Results: Paper I. During a 9-year study period, 46%, 68%, and 77% of men without previous PCa and aged 50–59 years, 60–69 years, and 70–79 years, respectively, had a PSA test. The probability of retesting with PSA was PSA- and age-dependent, with a 26-month cumulative incidence of 0.34 if the first PSA value was < 1 ng/ml. Paper II. PCa was diagnosed in 47 of 172 men with PSA levels of 1‒3ng/ml (27%), with Gleason sum of ≥ 7 in 10 of them (5.8%). There was an increase in the odds ratio of 1.60 with increasing genetic risk score. The absolute difference in risk of positive biopsy was 19 percentage points, comparing the high and low genetic risk groups (37% vs. 18%). Paper III. The four-kallikrein panel showed AUCs of 69.0 when predicting PCa of any grade and 71.8 when predicting high-grade cancer (Gleason score ≥ 7). Similar values were found for PHI (70.4 and 71.1, respectively). Both models had higher AUCs than a base model with PSA value and age. Using a 10% predicted risk of high-grade PCa by the four-kallikrein panel or PHI = 39 as cutoff for biopsy saved 29% of the biopsies performed, at a cost of delayed diagnosis for 10% of the men with highgrade cancer. Paper IV. There were no significant associations between aspirin or any antidiabetic medication and the risk of PCa. Men using any statin had an increased risk of both high-grade PCa and PCa overall (OR = 1.25; OR = 1.16). Compared to men without the medication, the level of the first PSA was lower in men using aspirin, statin, metformin, or insulin.

Conclusions: Although screening for PCa is not recommended in Sweden, PSA testing in Stockholm County was high in men aged over 50 years. A risk score based on SNPs predicts biopsy outcome in previously unbiopsied men with PSA levels of 1–3 ng/ml. Furthermore, we found that two blood tests, the PHI and the four-kallikrein panel, performed similarly in predicting prostate biopsy outcome. Introduction of such risk stratification tools could increase the proportion of men being classified in line with their true risk of PCa. We found no protective effect of aspirin, statins, or antidiabetics in terms of overall risk of prostate cancer or high-grade cancer.