Between-method differences in prostate-specific antigen assays affect prostate cancer risk prediction by nomograms.

Area of Science:

• Urology
• Oncology
• Medical Diagnostics

Background:

• Prostate cancer (PCa) risk prediction nomograms currently do not account for variability between different prostate-specific antigen (PSA) assay methods.
• This oversight may impact the reliability of PCa risk assessments.

Purpose of the Study:

• To evaluate the impact of inter-assay variability in prostate-specific antigen (PSA) measurements on the accuracy of established PCa risk prediction nomograms.
• To assess how different PSA assays affect the calibration and discrimination capabilities of PCa nomograms.

Main Methods:

• Measured total PSA (tPSA) and free PSA in 780 biopsy-referred men using five distinct assays.
• Applied these measurements and other clinical data to five established PCa detection nomograms.
• Evaluated nomogram accuracy using discrimination and calibration criteria.

Main Results:

• Significant variations in nomogram calibration and discrimination were observed across different PSA assays.
• Median PCa probabilities varied from 0.59 to 0.76 within the same nomogram when different PSA assays were used.
• While areas under the ROC curve were comparable, sensitivities and specificities differed considerably at various probability cutoffs.

Conclusions:

• The agreement between different PSA assays limits the accuracy of PCa risk probabilities generated by nomograms.
• Inter-assay variability can lead to unacceptable variations in PCa risk prediction.
• A more cautious approach to applying PCa nomograms is recommended due to assay-dependent discrepancies.