There is controversy over the usefulness of prostate-specific antigen (PSA) as a prostate cancer (PCa) biomarker. This controversy arises when there are differences in the results of PSA assay modalities. In this study, which aimed to evaluate a proper validation between the two PSA assay modalities, the agreement between the results of the two modalities was analyzed. PSA examinations were conducted using two PSA assay modalities in 4810 patients. The intra-class correlation coefficient (ICC) and weighted kappa analysis were used to evaluate the agreement between the two assay modalities. A linear regression was performed to evaluate the association between the two assay modalities. According to ICC values (ICC: 0.999, p < 0.001) and weighted kappa analysis values (kappa: 0.951, alpha's standard error (ASE): 0.001, p < 0.0001), the agreement between the assay modalities was rated as excellent. However, the strength of agreement was poor in the following PSA sub-groups: 0.05-0.1 ng/mL (ICC: 0.281, p = 0.0860); 0.15-0.2 ng/mL (ICC: 0.288, p = 0.0036); 1.5-2.0 ng/mL (ICC: 0.360, p = 0.0860); and 2.0-2.5 ng/mL (ICC: 0.303, p = 0.0868). In linear regression analysis, when modality B PSA yielded a value of 0.2 ng/mL, the expected value for modality A was 0.258 ng/mL (95% CI: 0.255-0.260), and when modality B PSA yielded a value of 4 ng/mL, the expected value for modality A was 3.192 ng/mL (95% CI: 3.150-3.235). The difference in the PSA values between the two PSA assay modalities is confirmed, and this difference may be clinically meaningful.
Keywords: diagnosis; prostate-specific antigen; statistics.