Skip to main content

15-01-2012 | Surgery | Article

PCPT risk calculator varies in predictive value between cohorts


Free abstract

MedWire News: The validity of the Prostate Cancer Prevention Trial (PCPT) Risk Calculator (PCPTRC) tool shows considerable variation depending on the biopsy cohort it is applied to, report US researchers.

This is most likely due to different criteria and work-up for biopsy between such cohorts, they say.

Since its publication in 2006, reports of the successful or failed validation of the PCPTRC have led to confusion over whether the tool should be recommended in clinical practice.

The PCPTRC was developed from the placebo arm of the PCPT in which participants underwent end-of-study biopsy regardless of prostate-specific antigen (PSA) or digital rectal exam (DRE) indication.

The tool is therefore not subject to the ascertainment bias that results when only participants with clinical indications for biopsy are actually biopsied, explain Donna Ankerst (University of Texas, San Antonio) and colleagues.

As reported in the World Journal of Urology, the researchers tested the external validity of the PCPTRC in 10 European and US contemporary biopsy cohorts, some with more than six-core biopsies already performed and most with significant numbers of PSA values higher than those in the PCPT cohort.

The team analyzed data from five cohorts from the European Randomized Study of Screening for Prostate Cancer (ERSPC) as well as three additional screening cohorts, namely the San Antonio Biomarkers Of Risk of prostate cancer study (SABOR), Texas, USA, the Prostate testing for cancer and Treatment (ProtecT), UK, and Tyrol, Austria, cohorts. Data from two US clinical cohorts from Cleveland, Ohio, and Durham, North Carolina, were also included.

The researchers report that the areas under the receiver operating characteristic curve (AUC) for PCPTRC ranged from a low of 56.2% to a high of 72.0% across the study populations.

In addition, while the AUC for PCPTRC exceeded that of PSA in all cohorts, it was only significantly greater in four of the eight screening cohorts.

Calibration plots indicated that PCPTRC overestimated the risk for prostate cancer for men of low, medium, and high risk for all of the ERSPC cohorts except for one where there was overlap between observed risk and predicted PCPTRC risk.

In all five ERSPC cohorts and in the Cleveland and ProtecT cohorts, there was limited to no clinical benefit in using the PCPTRC for biopsy referral compared with biopsying all of those meeting cohort-specific criteria for biopsy. For the remaining three cohorts (SABOR, Tyrol, and Durham) there was clinical net benefit observed at reasonable risk threshold ranges of 15-45%, 18-41%, and 25-100%, respectively.

The authors say the variation in validity of the tool is not unique to the PCPTRC, but would rather extend to validation studies of all risk-prediction tools.

They suggest that future validation models explore the properties of the model in different cohorts, and investigate the aspects of a cohort that affect model performance.

"In addition, clinicians should be cautious in using a model unless it has been shown to provide added value, such as benefit, in a very similar population to the one in which it is being used clinically," conclude Ankerst and team.

By Sally Robertson

Related topics