Age, PSA levels, and PSAD were higher in the prostate cancer diagnosis group than in the non-prostate cancer diagnosis group (P<0.05). Hb, albumin, prostate volume, and prostate transitional zone volume were significantly lower in the prostate cancer diagnosis group than in the non-prostate cancer diagnosis group (P<0.05). The presence of diabetes mellitus and hypertension, as well as serum ALP levels, were not significantly different between the two groups (Table 1).
1) Clinical and imaging characteristics after propensity-score nearest matching
We performed 1:1 propensity score matching using R to evaluate the significance of newly reported parameters such as RETzPSA, EPHPSA, PSAD, PZPSAD, and ETzD when there were no differences in age, underlying diseases (diabetes mellitus and hypertension), and serologic test (Hb, albumin, serum ALP levels and PSA levels) between the two groups. RETzPSA and EPHPSA were 4.78±1.33 ng/mL and 1.00±0.10 ng/mL in the prostate cancer diagnosis group, respectively, which were higher than those of the non-prostate cancer diagnosis group (3.99±1.16 ng/mL and 0.95±0.12 ng/mL, respectively) (p≤0.001). In contrast, prostate volume and prostate transitional zone volume were smaller in the prostate cancer diagnosis group (p<0.001). PSAD, PZPSAD, and ETzD in the prostate cancer diagnosis group were 0.24±0.10 ng/mL/cc, 0.35±0.14 ng/mL/cc, and 0.049±0.01 ng/mL/cc, respectively, which were statistically higher than those of the non-prostate cancer diagnosis group (0.17±0.07 ng/mL/cc, 0.27±0.10 ng/mL/cc, and 0.038±0.01 ng/mL/cc, respectively) (p<0.001). In DRE, hardness finding showed an odds ratio (OR) of 1.99 in the prostate cancer diagnosis group (95% CI: 1.21-3.27). In TRUS findings, the hypoechoic lesion (OR=2.86, 95% CI: 1.65-4.94) in the prostate cancer diagnosis group and calcification lesion (OR=0.54, 95% CI: 0.34-0.86) in the non-prostate cancer diagnosis group were statistically significant (p<0.05) (Table 2).
2) ROC curves analysis of several factors compared to PSA values
Analysis of ROC curves showed that the AUC for PSA value was 0.553 (95% CI: 0.495-0.610). Compared with the volume-adjusted PSA parameters, PSA density showed the highest value with 0.745 (95% CI: 0.693-0.793), followed by ETzD with 0.731 (95% CI: 0.677-0.780) and PZPSAD with 0.677 (95% CI: 0.611-0.719) (Fig. 1A). On the other hand, compared with adjusted PSA, RETzPSA showed the highest value with 0.673 (95% CI: 0.617-0.725), and then EPHPSA was 0.611 (95% CI: 0.554-0.666) (Fig. 1B).
Cut-off values of parameters were obtained using the ROC curve, and sensitivity, specificity, and positive predictive rate were analyzed according to each cut-off value. PSAD had a sensitivity of 67.11%, a specificity of 71.71%, and a predictive rate of 70.34% at 0.18 ng/mL/cc. ETzD showed a sensitivity of 69.08%, a specificity of 64.47%, and a predictive rate of 66.04% at 0.04 ng/mL/cc. Sensitivity, specificity, and predictive rates for RETzPSA at 3.8 ng/mL were 77.63%, 50.66%, and 61.14%, respectively, and EPHPSA at 1.0 ng/mL had a sensitivity of 56.58%, a specificity of 64.47%, and a predictive rate of 61.43%.
3) Logistic regression analysis according to cut-off values of various parameters
Sections were divided by the cut-off values obtained through ROC curve analysis. When the cut-off value of PSA in the gray zone was divided by 8.0 ng/mL, the OR was 1.731 (95% CI, 1.030-2.907), and at 0.15 ng/mL3, which was the cut-off value of conventional PSAD, the OR was 3.432 (95% CI, 2.095-5.623). When the cut-off value of PSAD was increased to 0.18 ng/mL3, the best results were obtained with an OR of 5.171 (95% CI, 3.171-8.432), followed by ETzd with 4.054 (95% CI, 2.513-6.540). The volume-adjusted PSA value showed higher OR than the adjusted PSA value (Table 3).