Prostate cancer is one of the most common solid malignancies worldwide. Most patients with non-metastatic disease receive either prostatectomy or radiotherapy. Radiotherapy of prostate cancer may be associated with significant acute urinary toxicity such as cystitis, particularly if the volume of the urinary bladder is small. Three studies showed that bladder volumes \<200 ml or \<180 ml, respectively, were associated with increased acute or late urinary toxicity. Therefore, it appears important to achieve bladder volumes \>200 ml at as many radiation fractions as possible. Several studies investigated the role of drinking protocols. However, drinking a certain amount of water at a specific point in time may require considerable discipline from the mainly elderly or very elderly patients. These considerations led to the idea of developing a mobile application (app) that reminds the patients prior to each radiation session to drink a certain amount of water. The present study investigates the number of radiation fractions with bladder volumes \<200 ml in a prospective cohort of patients with high-risk prostate cancer using a such a reminder app. In addition, the study evaluates whether the use of the reminder app leads to a significant reduction of the proportion of fractions with bladder volumes \<200 ml when compared to a historical control group not supported by an app. The app reminds the patients to drink 300 ml of water 45 minutes prior to each radiotherapy session.
Endpoints: The primary endpoint is the number of radiation fractions with bladder volumes \<200 ml after 35 fractions of radiotherapy. In addition, the following endpoints will be evaluated: Number of radiation fractions with bladder volumes \<200 ml at the end of radiotherapy, patient satisfaction with the reminder app: Evaluation at the end of radiotherapy, and impact of the reminder app on the use of health technology: Evaluation at the end of radiotherapy.
General trial design: This is a single-arm prospective study, which will investigate the effect of a reminder app on the number of radiation fractions with bladder volumes \<200 ml during a course of radiotherapy for definitive treatment of high-risk prostate cancer in comparison to a historical control group of 23 patients. The historical control group is considered appropriate for the comparison to the cohort of the present study, since all patients of the control group were treated very recently (in 2022 or 2023) with external beam radiotherapy alone in three of the five centers participating in the present study and received a Cone Beam Computed Tomography prior to each radiation fraction, which allowed assessment of the bladder volume as precisely as in the present study. To ensure comparability of total number of fractions administered in the prospective study with those in the historical control group, the primary endpoint is restricted to the first 35 fractions administered.
The recruitment of all 15 patients is planned to be completed within 18 months. The follow-up period will end directly after the radiotherapy course, which is scheduled to take 7 to 8 weeks. This equals a total running time for the study of 20 months. In accordance with the previous study assessing the number of radiation fractions with bladder volumes \<200 ml during a course of radiotherapy for definitive treatment of high-risk prostate cancer, the following characteristics will be recorded to allow adequate comparison with the historical control group: Initial (pre-radiotherapy) bladder volume, body-mass index (BMI), age, prostate volume prior to radiotherapy, Karnofsky performance score, prostate-specific antigen (PSA), Gleason score, tumor stage, and antihormonal therapy.
Sample size calculation: The primary goal of this study is to assess the impact of an app that reminds patients irradiated for high-risk prostate cancer to drink water prior to each radiotherapy session on the number of fractions with bladder volumes \<200 ml during the radiotherapy course and to demonstrate that this number is lower than without using an app (historical control group). To allow for a skewed distribution of the primary endpoint, the Wilcoxon-Mann-Whitney test will be applied for confirmatory statistical analysis. In the external historical control group consisting of 23 patients, the mean number of radiation fractions with bladder volumes \<200 ml was 29.2 (SD 6.6), and the median number was 32 fractions (IQR 24-35). A decrease of this mean value by roughly 30% (to 20.4 fractions) is considered clinically relevant. For illustrative purposes, translating this decrease into a non-parametric effect size framework (assuming for simplicity normal distribution) leads to a probability of roughly 0.25 that the number of fractions \<200 ml with the reminder app is larger than without the app. Based on this effect size, a sample size of 13 patients in the prospective trial is required for the comparison with the historical control group to ensure 90% power to reach statistical significance with a two-sided Wilcoxon Mann-Whitney-U-test and a 5% significance level. Assuming that roughly 10% of the enrolled patients will not be eligible for the primary analysis, since they have received less than 35 fractions, a total number of 15 patients should be enrolled in this study.
Statistical analyses: All data recorded in the case report forms describing the study population (demographic and clinical characteristics recorded at baseline), efficacy, safety and quality of life will be analyzed descriptively. Categorical data will be presented in contingency tables with frequencies and percentages and their 95% confidence intervals. Continuous data will be summarized with at least the following: frequency (n), median, quartiles, mean, standard deviation (standard error), minimum and maximum. Number of subjects with protocol deviations during the study and listings describing the deviations will be provided. In general, chi-square tests will be used to compare percentages in a two-by-two contingency table, replaced by Fisher´s exact test if the expected frequency in at least one cell of the associated table is less than 5. Stratified two-by-two contingency tables will be analyzed using Cochran-Mantel-Haenszel tests. Logistic regression models serve as multivariable methods for binary endpoint data. Comparison of ordinal variables between treatment arms will be performed using the asymptotic Wilcoxon-Mann-Whitney test, replaced by its exact version in case of ordinal categories with small number of categories and/or sparse data within categories. Any shift in location of quantitative variables between study groups will be performed with the Wilcoxon-Mann-Whitney tests as well. All patients who have started the radiotherapy and provide data on the primary endpoint will be analyzed (Full Analysis Set). The data analysis will be performed according to the statistical analysis plan (SAP), and which will be finalized prior to database lock and prior to any statistical analysis.
Primary Endpoint: The primary study endpoint is defined as the number of radiation fractions with bladder volumes \<200 ml after 35 fractions of radiotherapy. Descriptive measures of location and dispersion will be used to describe the results of the prospective study. The impact of patient characteristics on the primary endpoint will be assessed by means of Wilcoxon-two sample tests. These factors include age (\<75 vs. ≥75 years), Karnofsky performance score (90-100 vs. ≤80), body-mass index (\<30 vs. ≥30 kg/m2), prostate volume prior to radiotherapy (\<60 vs. ≥60 ml), PSA level (\<10 vs. ≥10 ng/ml), Gleason score (≤8 vs. 9), and primary tumor stage (T1 or T2 vs. T3 or T4). For confirmatory analysis, the prospective study will be compared with the historical control group by means of a two-sided Wilcoxon-Mann-Whitney two sample test and significance level of 5%. A high degree of comparability of patient population is expected between the prospective trial data and the retrospective patient data set. However, potential heterogeneity of study populations will be identified by comparing the patient characteristics listed above with Wilcoxon-Mann-Whitney tests. Homogeneity is assumed if all resulting p-values are above 20%. Any factor indicating a tendency towards heterogeneity (i.e. p\<0.20) will be included in a multivariable count data Poisson regression model with the number of radiation fractions with bladder volumes \< 200 ml as dependent variable and including the respective factors and the binary factor (prospective study vs historical control) as independent variables. In case of evidence of overdispersion the Poisson model will be replaced by a negative binomial model.
Secondary Endpoints: In addition to the primary study endpoint, secondary endpoints will be subjected to statistical analysis. Since no comparison with historical data is possible, these analyses focus on descriptive statistical analysis only. The results of patient satisfaction will be used to decide whether the app needs modification: In case of a dissatisfaction rate \>20%, the reminder app needs modifications. In case of a dissatisfaction rate \>40%, it will be considered not useful.
