MATERIALS AND METHODS
Data of patients who underwent RP as the initial treatment of PCa were reviewed retrospectively in this study. The data source was nationwide PCa database of Turkish Urooncological Association. A total of 3646 patients were found to be treated with RP for localized disease in the database and excluding patients with missing data, study population was downsized to 2454 patients. Patients were divided into two groups according to their waiting period between diagnosis and RP. The waiting period in respective groups was; Group 1: less than or equal to 3 months, Group 2: more than 3 months.
Based on D’amico classification system patients were stratified into low, intermediate and high-risk groups. The date of prostate biopsy was regarded as the diagnosis date and time to treatment is calculated as the number of days between the date of RP and the diagnosis date. Patients who received treatment for PCa (radiotherapy or androgen deprivation therapy etc.) prior to RP or patients who were first enrolled on AS protocol were excluded from the study.
All patients were diagnosed with either standard transrectal ultrasound guided biopsy or magnetic resonance (MR) guided fusion biopsy. All RPs were included in the study irrespective of the surgical approach (robot assisted, laparoscopic or open). Patients were operated by a senior urology staff in each participating center. Both biopsy and RP specimens were evaluated by a dedicated uro-pathologist in each center.
Biochemical recurrence which was defined as a prostate specific antigen (PSA) value measured greater than 0.2 ng/ml during the follow up after RP, is designated as the primary endpoint for this study. Secondary endpoints of the study were surgical parameters, pathological upgrading, metastasis on follow up and the need for additional treatments. For time-based analysis and comparison of oncological outcomes (biochemical recurrence free survival, need for adjuvant treatment or metastasis free survival), only patients with a follow up duration of more than 1 month was included in the statistical tests.
The study data were collected by REDCap data collection software developed by Vanderbilt University and licensed to Turkish Urooncology Association. 5, 6 All data are kept in a secure server and all personal information of patients were anonymized.
For statistical analysis Python Programming Language (Open source v3.7) was used with the help of pandas, matplotlib, numpy, scipy and lifelines7 libraries. JupyterLab (Open source v1.2.6) was used as the coding interface. The scalar variables were analyzed using visual (histograms, probability plots) and analytical methods (Kolmogorov-Simirnov/Shapiro-Wilk’s tests) to determine whether or not they were normally distributed. Descriptive analyses were given as means and standard deviations when the variable was normally distributed. Medians and interquartile ranges were used if the variable was not normally distributed. For the comparison of scalar variables between two groups, t-test or Mann-Whitney U test were used for normally and non-normally distributed variables respectively.
Categorical variables were compared with Chi-square test between groups. If the assumptions of Chi-square do not hold due to low expected cell counts Fisher’s exact test were used for the comparison of categorical variables. For biochemical recurrence free survival variable, Kaplan-Meier survival estimates were calculated. A separate log rank test was used to estimate independent effect of waiting duration group on time to biochemical recurrence. The possible factors identified in univariate analysis further evaluated with Cox regression. The proportional hazard assumption was assessed by means of residual analysis. For all statistical tests, p value below 0.05 was considered statistically significant.