Study design and setting
This was a retrospective cohort study. This study was conducted at the Uganda Cancer Institute (UCI). The UCI is a comprehensive national cancer referral center, and the East Africa Center of Excellence in Oncology receives approximately 4,000 new cancer patients annually and approximately 20 new prostate cancer patients per month. The Cancer Institute has a records department where patient charts are stored according to histological diagnosis, and each patient is assigned a unique identifier.
Inclusion criteria and sampling procedure
Patients who were admitted to the UCI from January 2015 to December 2019 with a documented histological diagnosis of prostate cancer, radiological evidence of metastasis and receiving a combination of taxane-based chemotherapy and ADT were included in the study.
Purposive sampling: All patients who fulfilled the inclusion criteria were purposively included in the study.
Sample size estimation
The sample size was estimated using Stata/IC 16.0, power analysis for the Cox proportional hazards model, a one-sided test, adjusting for censoring; power=80%, significance level=0.05, hazard ratio=1.2, standard deviation=0.842, N=263
The estimated sample size for Cox PH regression was 263 participants.
Study variables
For each chart, we extracted data on demographics, comorbidities, baseline laboratory parameters, treatments received and vital status (alive, dead and lost to follow-up). The demographic variables included age at diagnosis with cancer and region of residence in Uganda. Clinical characteristics included patient comorbidities, ECOG status, symptoms at presentation and Gleason score grouping. The laboratory characteristics included histology grade, baseline PSA levels, white blood cell (WBC) counts, and hemoglobin levels.
Data collection
The patient register at the UCI records office was used to identify all patients with prostate cancer during the study period. The following data were extracted using a pretested data abstraction form: patient demographics, clinical symptoms at presentation, comorbidities, physical exam findings, diagnostic tests, treatments received and overall outcomes. PSA levels (ng/ml) at diagnosis were grouped into <100 and ≥100. The Gleason score was grouped into low grade, intermediate grade and high grade. Low/intermediate scores were categorized as a total score of 6, 7 or 8. A high Gleason score was categorized as a total score of 9 or 10. We determined the vital status of patients by follow-up phone calls and by recording the date of death for those in whom death was documented in their charts or through death certificates. The date of last contact (e.g., hospital visit or follow-up phone calls) for patients who were alive during the follow-up period (i.e., censoring) was recorded. Patients who could not be reached by phone calls or who did not have records of vital status in their charts were defined as lost to follow-up.
Data Analysis
The data were analyzed using Stata/IC 16.0. Numerical data were summarized using frequencies, percentages, means (± standard deviations) for normally distributed continuous data or medians (interquartile ranges) for continuous but skewed variables. The biochemical response (TPSA) of patients with mPCa treated with ADT and docetaxel was calculated as the proportion of patients with a PSA concentration ≤4.0 ng/mL according to the 95% confidence interval. The Kaplan‒Meier method was used to estimate overall survival with 95% confidence intervals (95% CIs) at 1, 3 and 5 years. Survival time was defined as the time from the diagnosis of mPCa to death. Cox proportional hazards regression was applied to estimate mortality hazard ratios (HRs) with 95% confidence intervals (CIs). Variables with P <0.2 in univariable models were selected for multivariable models. All tests were two-sided. The multivariate Cox regression model was tested for proportionality using the Schoenfeld residuals. For models that violated the proportionality assumption, we used stratified Cox model analysis. The threshold for statistical significance was set to a two-tailed α < 0.05.
Ethical considerations
This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The Uganda Cancer Institute Research and Ethics Committee approved the study protocol and provided waiver of consent (approval number UCI-2021-15).