br Localised removed br Advanced br Unknown br ADT
ADT duration (months), median (IQR)
Previous prostatectomy, N (%)
Previous radiotherapy, N (%)
Previous chemotherapy, N (%)
Current active surveillance, N (%)
Data are: mean ± standard deviation unless stated otherwise.
a ADT-treated men compared to PCa controls only. b Comorbidities included asthma/respiratory problems, chronic bronchitis, muscle/ligament problems, back pain, angina/stroke/heart condition, diabetes, hy-pertension and hypercholesterolaemia.
2.2.3. Cortical bone distribution
Cortical bone density distribution was calculated from the pQCT scans of the proximal tibia and radius. As we have previously reported , radial bone density distribution within the bone cross-section was calculated by measuring the average bone density for three concentric rings (endocortical, midcortical and pericortical) from the centre of bone mass to the outer bone edge. The short-term CVs for repeated measures of endocortical, midcortical and pericortical vBMD in samples of healthy men range from 1.6% to 4.3% for the radius  and 1.4% to 2.4% for the tibia .
2.2.4. Anthropometry, physical activity and diet
Height and body mass was assessed using AWD 131-138 portable stadiometer (SECA, Hamburg, Germany) and scales (A&D, Tokyo, Japan) respec-tively, with participants wearing light clothing and no shoes. Body mass index (BMI) was calculated as body mass (kg) divided by height (m) squared (kg/m2). The Community Healthy Activities Model Programme for Seniors (CHAMPS) physical activity questionnaire was used to as-sess to total habitual activity levels (kJ/d) based participation in a comprehensive list of low, moderate and vigorous physical and leisure activities . Diet was assessed using a 24-hour food recall. Dietary analysis was performed using Australia-specific dietary analysis soft-ware (FoodWorks, Xyris software, Highgate Hills, Australia).
2.3. Statistical analysis
All statistical analyses were performed using Stata statistical soft-ware (Version 15, Stata, College Station, TX, USA). Initially, all out-come data were screened for outliers and descriptive statistics were computed to compare the three groups on known confounding variables of the outcomes of interest. Equality of variances and normality of distribution of all data were assessed using Levene's test and Shapiro-Wilk's test, respectively. Outcomes with non-normally distributed re-siduals (lumbar spine aBMD, distal tibia total area and BSI, distal radius total and trabecular vBMD, total area, medullary area, cortical vBMD and Ipolar of the proximal tibia and radius, and endocortical, midcortical and pericortical vBMD of the proximal tibia and radius) were trans-formed using natural logarithms prior to analysis and percentage dif-ferences were calculated on the natural log transformed data for abnormal hemoglobin outcomes . One-way analyses of variance/covariance (ANOVA/ ANCOVA) were used to compare continuous variables between groups. Chi-squared tests, or Fisher's exact test for small expected frequencies, were used to compare categorical variables between groups. BMI was included as a covariate in analyses of all outcomes, and tibia and radius length were included as covariates in analyses of all tibia and radius pQCT outcomes, respectively. Bonferroni corrected pairwise compar-isons were used where appropriate to further investigate any significant main effects between groups. Subgroup analysis was completed to compare ADT-treated men with treatment duration less than and > 12-months. This cut-point was chosen as bone loss with ADT is shown to be greatest within the first 12-months of treatment [13,42]. Spearman rank-order correlation coefficients were used to investigate the re-lationship between ADT duration and bone outcomes. Continuous data were reported as mean ± standard deviation (or mean [95% con-fidence intervals] for adjusted values), whereas categorical variables
were reported as frequency and percentage, unless stated otherwise. A significance level of P < 0.05 was adopted for all statistical tests.
3.1. Participant characteristics
The characteristics of the 192 men included in the three groups are shown in Table 1. For ADT-treated men, the median (IQR) treatment duration was 12 (5–23) months. ADT-treated men had 7.9% higher BMI than PCa controls (P = 0.012), but not healthy controls. ADT-treated men were more likely to have advanced PCa and be previously treated with radiotherapy or chemotherapy compared to PCa controls, while PCa controls were more likely to have had a prostatectomy. There were no other differences between groups for other demographic, diet, physical activity or medical treatment outcomes.