br ADT treated men were
ADT-treated men were recruited between April 2014 and November 2017 via clinician referral from Alfred Health (Melbourne, Australia), Peter MacCallum Cancer Centre (Victoria, Australia) and six private urology practices (Victoria, Australia), as well as from 32 PCa support groups (Victoria, Australia) and advertisements in state/local news-papers. PCa and healthy controls were recruited between October 2014
and February 2016 from PCa support groups and advertisements in state/local newspapers. The study was conducted in accordance with Declaration of Helsinki and was approved by the human research ethics committees at Deakin University (HREC 2013-184), Alfred Health (Project No: 455/15) and Peter MacCallum Cancer Centre (Project No: 17/118). All participants gave their informed written consent prior to participation.
Lumbar spine (L1-L4) and proximal JQ1 (femoral neck and total hip) aBMD (g/cm2) were assessed using DXA (Lunar iDXA, GE Lunar Corp., Madison, WI, USA) and analysed using enCORE software (ver-sion 12.30.008), by a researcher blinded to group allocation. All scans were reviewed for artefacts or local structural abnormalities known to influence BMD and where necessary, artefacts were marked and ex-cluded from analysis. For the lumbar spine, T-scores from adjacent vertebrae were reviewed to determine any large differences in BMD from one vertebrae to the next. Individual vertebrae affected by structural abnormalities causing artificial elevation of BMD (T-score ≥ 2 SDs different to adjacent vertebrae) were excluded from the analysis such that the overall lumbar spine BMD results included only the remaining vertebrae (N = 11 participants with verterbrae ex-cluded). Additionally, two participants (PCa control, N = 1; healthy control, N = 1) with hip arthroplasty were excluded from the analysis of proximal femur sites. One ADT-treated participant was excluded from analysis of total hip aBMD due an abnormally high value (T-score of 7.3). The short-term coefficient of variation (CV) for aBMD measures range from 0.6% to 1.0% within our laboratory. The lowest T-score of the lumbar spine, femoral neck and/or total hip was used to classify participants with osteoporosis (T-score ≤ −2.5) or osteopenia (T-score between −2.5 and −1), consistent with World Health Organization criteria .
2.2.2. Volumetric BMD, bone structure and strength
Proximal (66%) and distal (4%) sites of the non-dominant radius and dominant tibia were scanned using pQCT (XCT 3000, Stratec Medizintechnik GmbH, Pforzheim, Germany). Volumetric BMD (total and cortical vBMD [mg/cm3]), structure (total, cortical and medullary area [mm2]) and strength (density-weighted polar cross-sectional mo-
ment of inertia [Ipolar, mg/cm]) at proximal sites as well as vBMD (total and trabecular vBMD [mg/cm3]), bone size (total area [mm2]) and
strength (bone strength index [BSI, mg2/mm4]) at distal sites were as-sessed. The slice thickness was 1 mm and voxel size was 0.5 mm at a scanning speed of 20 mm/s. pQCT images were analysed in the Fiji image analysis platform  using the BoneJ plugin  as previously reported . Distal radius and tibia (4%) total bone area were ana-lysed based on thresholding at 169 mg/cm3. Trabecular density was determined by peeling single layers of pixels until 55% of the total bone area was peeled away. The remaining inner 45% of the total bone area was considered to be exclusively trabecular bone. BSI was calculated as follows: BSI = total area multiplied by the square of total vBMD . For the 66% proximal radius and tibia, the periosteal surface was de-termined based on a threshold of 280 mg/cm3, and cortical bone a threshold of 550 mg/cm3. Medullary area was calculated by subtracting cortical area from total area. Ipolar was determined using the bone threshold of 480 mg/cm3 . Three participants (ADT-treated, N = 1; PCa controls, N = 1; healthy control, N = 1) did not complete pQCT scans due to maintenance of the scanner at the time of testing. Scans at each site were reviewed for errors and excluded according to the visual inspection rating scale of participant movement [36,37]. Distal tibia scans of 10 participants (ADT-treated men, N = 3; PCa controls, N = 3; healthy controls, N = 4), proximal tibia scans of three participants (ADT-treated men, N = 3), the distal radius scan of one ADT-treated participant and the proximal radius scans of nine participants (ADT-
Participant characteristics of men with prostate cancer (PCa) treated with androgen deprivation therapy (ADT), PCa controls and healthy controls.
Healthy controls PCa controls ADT-treated men P-value
Time since PCa diagnosis (months)