Abstract

OBJECTIVE:

To assess the effects on the performance characteristics, in a prostate-cancer detection programme using prostate-specific antigen (PSA) levels, of a lower PSA threshold and the incorporation of reflex (free or complexed PSA) tests.

METHODS:

We reviewed publications and extracted data on PSA distributions and performance characteristics of the PSA test and isoform tests from population-based surveys. We estimated the rate of biopsy, cancers detected, and cancers missed that would result from lowering PSA thresholds and including reflex testing.

RESULTS:

Lowering the PSA threshold for biopsy referral to 2 ng/mL would increase the number of referrals from 110 to 230 per 1000 men tested, with most of the extra biopsies being among men with no cancer, i.e. an increase from 74 to 172 per 1000 men tested. However, this increased testing would result in an increase in the cancer-detection rate from 3.6% to 5.8%. Including a reflex test for men with moderately elevated PSA levels has little effect on programme performance, with only a modest (10-15%) reduction in unnecessary biopsies and a small increase in the numbers of missed cancers.

CONCLUSIONS:

Lowering PSA thresholds, with or without the concurrent introduction of reflex tests, would increase both the numbers of cancers detected and the number of patients referred for biopsy procedures, of which most would be unnecessary. As the extra cancers detected are likely to be clinically localized, and with no evidence that their treatment improves the outcome of the disease, such changes place a possibly unjustified additional burden on the healthcare provider.