Introduction: Ultrasound-guided transrectal prostate biopsy is currently an indispensable test for diagnosing prostate cancer. Many variables have been related to the presence of cancer in the biopsy (e.g. digital rectal examination [DRE], serum levels of prostate-specific antigen [PSA], free PSA fraction [PSAI/PSAt]). Multivariate mathematical models integrating these variables (nomograms, artificial network models) and improving the capacity to predict tests results are currently available.
Objective: To develop a nomogram for predicting the probability of a positive prostate biopsy in patients in whom this test is requested, and to use such nomogram in subsequent patients to assess its predictive ability.
Material and methods: A total of 410 consecutive patients undergoing biopsy due to a suspicious digital rectal examination or two serum PSA values higher than 4 ng/mL were enrolled into the study. Ten cores were taken in the prostate biopsy. Patients with both PSA levels >20 ng/ml and prior biopsies were excluded. The following variables were recorded in each patient: age, total PSA, free PSA fraction, prostate volume, transition zone volume, PSA density, PSA density adjusted by transition zone volume, digital rectal examination, and findings suggesting cancer during transrectal ultrasound (hypoechogenic nodules). Prospective external validation was performed with 185 patients who met the same inclusion criteria. Statistical analysis consisted of four phases: a univariate study, a multivariate logistic regression study which was used to develop the nomogram, internal validation, and prospective external validation. S-Plus#r Programme Design and SPSS 12.0#r software was used for the procedure.
Results: Variables found to be independently and significantly associated to the presence of cancer included age, digital rectal examination, trnsition zone volume, PSA density, and the presence of hypoechogenic nodules during transrectal ultrasound. Such variables were therefore used to develop the nomogram. The goodness-of-fit of the nomogram was 84%. Validation with an external sample showed a 73% concordance index.
Conclusion: A nomogram having a satisfactory predictive ability and fit that allows for predicting the prostate biopsy result with a high accuracy rate was developed.