BACKGROUND: To investigate whether using a new concept of relative probability for prostate cancer (RPpca) can increase sensitivity and specificity in detecting prostate cancer. METHODS: For 217 patients with total prostate-specific antigen (PSA) level of 4-20 ng/mL, prostate volume and free to total PSA (F/T) ratio were measured. From the fitted curves between positive predictive values for prostate cancer and age, prostate volume, total PSA or the F/T ratio, each function predicting prostate cancer was determined. RPpca defined by the combined functions of age, prostate volume, total PSA and F/T ratio was calculated for each individual patient. The probability of prostate cancer was also calculated, using logistic regression analysis (LRPpca). Receiver-operating characteristic analysis was performed to elucidate the areas under the curve (AUC), sensitivities and specificities and cutoff values of the conventional and new parameters. Finally, we investigated the applicability of the above parameters in the other patient group using a different PSA assay kit (AxSYM). RESULTS: Although RPpca had the largest AUC in the total PSA range of 4-20 ng/mL, it did not reach statistical significance between RPpca and F/T ratio or LRPpca. The cutoff values of F/T ratio, LRPpca and RPpca were determined as 0.15, 0.12 and 0.20, respectively. Using these cutoff values in AxSYM data, RPpca had the highest sensitivity (91%) and specificity (57%). CONCLUSIONS: RPpca can provide more precise information to avoid unnecessary biopsy than LRPpca or F/T ratio. RPpca could be valuable to decide whether to perform prostate biopsy when using various PSA kits.
BACKGROUND: To investigate whether using a new concept of relative probability for prostate cancer (RPpca) can increase sensitivity and specificity in detecting prostate cancer. METHODS: For 217 patients with total prostate-specific antigen (PSA) level of 4-20 ng/mL, prostate volume and free to total PSA (F/T) ratio were measured. From the fitted curves between positive predictive values for prostate cancer and age, prostate volume, total PSA or the F/T ratio, each function predicting prostate cancer was determined. RPpca defined by the combined functions of age, prostate volume, total PSA and F/T ratio was calculated for each individual patient. The probability of prostate cancer was also calculated, using logistic regression analysis (LRPpca). Receiver-operating characteristic analysis was performed to elucidate the areas under the curve (AUC), sensitivities and specificities and cutoff values of the conventional and new parameters. Finally, we investigated the applicability of the above parameters in the other patient group using a different PSA assay kit (AxSYM). RESULTS: Although RPpca had the largest AUC in the total PSA range of 4-20 ng/mL, it did not reach statistical significance between RPpca and F/T ratio or LRPpca. The cutoff values of F/T ratio, LRPpca and RPpca were determined as 0.15, 0.12 and 0.20, respectively. Using these cutoff values in AxSYM data, RPpca had the highest sensitivity (91%) and specificity (57%). CONCLUSIONS: RPpca can provide more precise information to avoid unnecessary biopsy than LRPpca or F/T ratio. RPpca could be valuable to decide whether to perform prostate biopsy when using various PSA kits.