Detection of Prostate Cancer by Ultrasonographically-guided
Transrectal Prostate Biopsies at the Department of Urology of the First Faculty of Medicine, Charles
University and General Teaching Hospital in Prague, Analysis of 1464 Cases
1Šafařík, L., 1Novák, K., 1Stolz, J., M., 2Zlatohlávek, J., 1Pešl M., 1Babjuk, M., 1Pavlík, I., 1Hanuš, T., 1Dvořáček J.
1Urologická klinika 1. LF UK a VFN, Praha 2Matematicko-fyzikální fakulta UK, Praha
Background. Early diagnostics of prostate cancer is still the most important factor in tumor-specific
survival of patients harbouring this malignant disease. Without better understanding of the etiology and
without relevant markers of the disease progression, only the early diagnostics of organ-confined disease
can save the patient’s life.
Methods and Results. Throughout 7 consecutive years, 1464 transrectal prostate biopsies in 1302
patients were performed. In all cases, the age, DRE (positive or negative), prostatic volume (ccm), total
PSA in peripheral blood (ng/ml) were assigned, as well as free/total PSA ratio (%) in most of them.
Apart from previously used linear regression, we applied logistic regression, since only age grows linear
and DRE is determined only as positive or negative. The surrogate endpoint was ROC, which determines
the area applying to the relations of sensitivity and specificity of any marker. Multifactorial logistic
regression then reached best results at values over 0.8 in all tested age categories with maximal deviation
of 8%, which had not been achieved before.
Conclusions. Despite hundreds of papers published on this topic, the question of when and how the
patient is indicated to the biopsy of the prostate has not been solved. A computer driven model based on
1464 documented examinations on the relatively homogeneous population is presented. On the basis of
the combination of biological and statistical methods, the model can give correct predictions in 9 out of
prostate cancer, early diagnostics, computer model.