[Predicting prostate cancers using a logistic regression model with transrectal ultrasound characteristics, age and serum PSA].

Objective: To determine the value of transrectal ultrasound characteristics in combination with age and prostate specific antigen (PSA) for predicting prostate cancers.

Methods: 579 patients with prostate diseases were examined with color doppler, followed by biopsies of prostates. A logistic regression model was developed to determine the relationship between prostate cancers and transrectal ultrasound characteristics, age and PSA levels.

Results: The biopsy confirmed 194 (33. 5%) prostate cancer cases and 385 (66. 5%) noncancerous patients with prostate diseases. The prostate cancer patients showed hypoechoic nodule, uneven echo, abnormal blood flow, abnormal prostate shape, unclear boundary of internal and external glands, unclear dividing line of prostate and seminal vesicle glands, unclear dividing line of prostate and rectal, and lymph node metastasis. Our prediction model had a 83.59% accuracy, 64.95% sensitivity, 92.99% specificity, 82.35% positive predictive value, and 84.04% negative predictive value.

Conclusion: Transrectal ultrasound characteristics in combination with patients age and serum PSA is a good predictor of prostate cancer.