Whole body fat-free mass was measured with bioelectrical impedance analysis (BIA) and concurrent bioelectrical impedance vector analysis (BIVA) with a single-frequency, phase-sensitive 50 kilohertz bioelectrical impedance analyzer (BIA-101, RJL/Akern Systems, Florence, Italy). This method is highly valid for measuring changes in body composition (Savastano et al. 2010).

Waist, hip, mid-upper arm, forearm, mid-thigh and calf circumferences were measured with a non-stretching tape using standard techniques.

Determined as the highest oxygen consumption (V̇O2) attained during the cardiopulmonary exercise test

Estimated using the modified V-slope method and confirmed by evaluating ventilatory equivalents and end-tidal pressures. Two analysts independently determined the ventilatory anaerobic threshold, with discrepancies of ≥ 7.5% resolved through discussion and consultation with a third analyst if necessary.

Ratio of minute ventilation to O2 consumption and CO2 output at the ventilatory anaerobic threshold

Ratio of oxygen consumption to heart rate at peak exercise during the cardiopulmonary exercise test

Determined as the linear relationship between oxygen uptake and the logarithmic transformation of minute ventilation during the cardiopulmonary exercise test

Concentrations of total cholesterol, low-density lipoprotein (LDL-C) and high-density lipoprotein (HDL-C) were measured from venous blood.

The concentration of triglycerides was measured from venous blood.

Concentration of total testosterone was measured from venous blood.

Concentration of sex hormone binding globulin was measured from venous blood.

The percent risk of a cardiovascular disease event occurring in the next 10 years was calculated with QRISK®2-2017 Web Calculator (https://qrisk.org/2017). The total score ranges from 0 to 100 %, with higher scores indicating a greater risk of cardiovascular disease.

Gleason scores were measured from prostate biopsies. A score of 1 to 5 is assigned to the cells that make up the largest area of the tumor and another score of 1 to 5 is assigned to the cells that comprise the next largest area. A total Gleason score is then calculated by adding both of these scores together. Total scores ranged from 2 to 10, with 2 representing the most well-differentiated tumors and 10 the least-differentiated tumors.

Concentration of prostate specific antigen was measured from venous blood.

Skinfold thickness was measured at suprailiac and subscapular sites with a skinfold caliper (Harpenden Skinfold Caliper, Baty International, Sussex, UK).

Hand grip strength was measured with an analogue dynamometer (Takei Scientific Instruments Ltd., Tokyo, Japan).

Disease-specific health-related quality of life was assessed with the FACT-P 39-item questionnaire (version 4). Each item has a 5-point Likert-like response scale ranging from 0 ("Not at all") to 4 ("Very much"). A total score is calculated as the sum of each score from items 1 to 39, which can range from 0 to 156. Negatively worded items are reverse scored (subtracted from 4) prior to summing so that higher total scores indicate better health-related quality of life.

Fatigue was assessed with the 13-item FACIT-Fatigue scale (version 4). Each item has a 5-point Likert-like response scale ranging from 0 ("Not at all") to 4 ("Very much"). A total score is calculated as the sum of each score from items 1 to 13, which can range from 0 to 52. Negatively worded items are reverse scored (subtracted from 4) prior to summing so that higher total scores indicate less experience of fatigue.

The Godin Leisure-Time Exercise questionnaire was used to characterise levels of physical activity. The questionnaire contains three items and asks the patient to recall the number of bouts of strenuous, moderate, or mild physical activity that exceed 15 min in duration in which they have engaged in over a typical week. A total score is calculated by multiplying the number of strenuous, moderate, and mild physical activity bouts by weights of 9, 5, and 3, respectively, and summing those values in an overall score that ranges from 0 to 119 in arbitrary units. Higher scores indicate participation in higher levels of physical activity.

Measured as the number of eligible participants who were eligible and consented to participate in the trial. This will be reported in a Consolidated Standards of Reporting Trials (CONSORT) participant flowchart.

The number of adverse events were recorded to determine the feasibility of the exercise intervention. An adverse event was defined as the occurrence of any untoward medical occurrence in a participant, which does not necessarily have a causal relationship with the exercise intervention. The type of adverse events was also noted.

Established as the number of patients in the experimental group who discontinued the exercise intervention

In the experimental group, patients lost to follow-up were characterised as those who completed the exercise intervention but did not complete endpoint testing. In the control group, patients lost to follow-up were the number of patients who did not complete endpoint testing.

The number of exercise sessions completed by each participant in the intervention group was recorded. These values were than averaged to provide a measure of adherence. The maximum number of exercise sessions that participants could complete was 24, so adherence ranged from 0 to 24 sessions, with higher scores indicating greater adherence.