Effect of exercise and androgen deprivation therapy on prostate cancer: a meta-analysis

We searched the PubMed, EMBASE, Web of Science, EBSCO, and Cochrane Library databases for reports that were published until October 22, 2021. The search terms generally focused on ‘exercise’, ‘training’, ‘physical activity’, ‘prostate cancer’, and ‘androgen deprivation therapy’.

The inclusion criteria were based on the PICOS framework.

(1) Population: all participants were adult men diagnosed with prostate cancer, and were currently receiving ADT during the intervention, regardless of whether they had received chemotherapy, radiotherapy or other therapy.

(2) Intervention: the interventions included aerobic exercise, resistance exercise, and/or impact exercise.

(3) Outcomes: the report should describe at least one relevant outcome (lean body mass, body fat mass, body fat rate, regional and whole-body bone mineral density).

(4) Study: reports of RCTs and studies were considered appropriate if they were published in English.

The exclusion criteria were:

(1) studies that did not provide ADT to all patients (unless the results were stratified according to ADT use); and (2) studies that involved interventions that combined exercise and diet, nutrition or other lifestyle.

Data were independently extracted by two reviewers (Shao WJ and Zhang HY) using a standardised data extraction sheet, and any disputes were discussed and settled by a third person (Zhang YM). Relevant data included study-related information (first author name, publication year, study location, participant characteristics, and details of the intervention), quality assessment information, sample size and outcomes data.

The outcomes of intra-group differences in the exercise and control groups were selected as priority data. Secondly, the outcomes at baseline and post-intervention were selected in the exercise and control groups, and then calculated the intra-group differences using formulas. The data were presented by mean and standard deviation. If the standard deviation was not originally reported, it was calculated based on related data (e.g., quartiles and 95% confidence intervals using the relevant formulas.

The Cochrane risk of bias tool was used for the quality assessment, which assigns high, low, or unclear risks for selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias.

The statistical analyses were performed using RevMan software (version 5.3) and Stata software (version 12.0). Absolute net differences between the intervention and control groups were used to estimate merger effects. Outcomes were expressed as the weighted mean difference and its 95% CI. Random effect models were used given the heterogeneity of interventions.Subgroup analyses were performed according to exercise type, exercise intensity, exercise duration, and ADT duration.The Egger’s and Begg’s tests were used to judge publication bias. Differences were considered statistically significant at P -values of <0.05.