Main Prostate Cancer Risk Calculators Adjust to Different Populations

The calibrations for the European Randomized Study on Prostate Cancer (ERSPC) and Prostate Cancer Prevention Trial (PCPT) risk calculators can be successfully adapted to different populations than those for which the calculators were originally designed, according to the results of a systematic review published in JMIR Cancer.

Risk calculators that relied on multiple individual factors could accurately predict cases of prostate cancer and their treatment, and as a result, would be able to decrease the number of intrusive and unnecessary biopsies and their complications, according to the review results.

The combination of serum prostate specific antigen (PSA) and predictive models, such as the Prostate Health Index (PHI), complement each other and increase the detection of prostate cancer; PSA alone has decreased prostate cancer mortality by 53%, according to the review results.

The review, designed to evaluate and analyze publications about risk calculators, found that the most effective ones were those that had been calibrated, created, or optimized with a specific population in mind. It was found that these risk calculators had higher predictions but could be adjusted for other populations.

Although advanced imaging and biomarkers are benefits of prostate cancer detection, these are not widely available options and cannot be applicable at the primary level.

However, despite this, the ERSPC and PCPT risk calculators are the most common predictors being used, and the review found that when applied to different populations, similar results can be yielded.

When the ERSPC risk calculator was used in a Chinese population, the results showed that the area above the curve (AUC) was 0.74 for any prostate cancer or that of a higher grade. The range of AUCs for the ERSPC risk calculator fell between 0.68 and 0.86, while the Chinese Prostate Cancer Consortium risk calculator had an AUC of 0.77 for any or higher-grade prostate cancer.

In a South African population, the AUC was 0.833 for high-grade prostate cancer, according to the review results.

Likewise, the PCPT, when used in a Mexican population, yielded a higher AUC at 0.785 for high-grade prostate cancer, when the range was from 0.562 to 0.813, according to the review results.

The ERSPC and PCPT risk calculators were shown to have similar predictability rates.

The review showed that of the 36 studies include, fewer than 10 focused on non-European and non-North American populations: 1 each were in Mexico and South America, while the rest were in Asia.

Risk calculators that were targeted at new populations are also a valid alternative to help calibrate existing ones and yield beneficial results. In a calculator for an Indonesian population, the AUC was 0.938 and used just 5 factors.

Another one in South Korean populations had an AUC of 0.887 and highlighted the use of epidemiological factors instead of serum markers, according to the review results.

The ERSPC risk calculator originally included “[magnetic resonance imaging] information, PSA levels, results of a prior biopsy, results of [digital rectum results (DRE)], and prostate volume measured by transrectal ultrasound,” according to the review results.

The PCPT calculator originally included “age, race, PSA levels, family history of prostate cancer, results of a DRE, results of a prior biopsy, and when available, free PSA, prostate cancer antigen 3, and T2:ERG,” according to the review.

The PHI uses a more mathematical approach including “PSA, free PSA, and prostate specific antigen isoform p2,” according to the review.

The review noted that some risk calculators that were developed use factors from both the ERSPC and PCPT.

The review had 18 articles that evaluated the PCRT risk calculator, 14 that evaluated the ERSPC calculator, and 9 that evaluated new calculators.

The new risk calculators that were created from scratch had higher prediction rates on their respective target populations, according to the review results.
None of the studies included allowed for a meta-analysis of the individual risk factors or overall predictive capabilities.

Reference

Bandala-Jacques A, Esquivel K D C, Pérez-Hurtado F, Hernández-Silva C, Reynoso-Noverón, N. Prostate cancer risk calculators for healthy populations: systematic review. JMIR Cancer. 2021;7(3):e30430. doi:10.2196/30430