Correlation between the complex PSA/total PSA ratio and the free PSA/total PSA ratio, sensitivity and specificity of both markers for the diagnosis of prostate cancer.

Abstract Objective To compare the behaviour of the PSAcomplex/PSAtotal percentage (PSAc%) against the PSA free/PSA total (PSAl%) and analyse both markers for their usefulness in diagnosing prostate cancer. Material and methods We measured total PSA (PSAt), free PSA (PSAl), complex PSA (PSAc), PSAl% and PSAc% levels in 158 patients. Of these, 98 (62%) were biopsied for presenting PSAt ≥3 ng/dl and PSAl%   10, suspicious rectal examination or suspicious ultrasound node. We performed linear regression and Passing–Bablok regression analyses. The ROC curves were calculated to study the sensitivity and specificity of PSAl% and PSAc% and were compared to each other. The prostate cancer diagnoses were analysed by PSAl% and PSAc% by applying the χ 2 test. Results The correlation coefficient ( r ) was good (0.7447, p r 2 ) was 0.5. The result of the Passing–Bablok analysis was a slope of 1.658 (1.452–1.897) and an intersection of 2.044 (−0.936 to 5.393). The optimal cutoff for PSAl% (≤14.7854) showed a sensitivity of 89.29% [95% CI, 0.642–0.823] and a specificity of 54.29% (95% CI, 0.642–0.823). The optimal cutoff for PSAc% (>89.7796) had a sensitivity of 71.43% (95% CI, 0.616–0.802) and a specificity of 71.43% (95% CI, 0.616–0.802). There were no significant differences when comparing the areas under the curve of both markers ( p  = .59). The PPV of PSAl% was less than that of PSAc% (45.7% vs. 71%). Conclusion There was a good correlation between PSAl% and PSAc%. PSAc% has demonstrated greater specificity and efficacy than PSAl% in the diagnosis of prostate cancer.