Quantification of mineralized bone response to prostate cancer by noninvasive in vivo μCT and non-destructive ex vivo μCT and DXA in a mouse model

Background: To compare nondestructive in vivo and ex vivo micro-computed tomography (μCT) and ex vivo dual-energy- X-ray-absorptiometry (DXA) in characterizing mineralized cortical and trabecular bone response to prostate cancer involving the skeleton in a mouse model. Methodology/Principal Findings: In vivo μCT was performed before and 10 weeks after implantation of human prostate cancer cells (MDA-PCa-2b) or vehicle into SCID mouse femora. After resection, femora were imaged by nondestructive ex vivo specimen μCT at three voxel sizes (31 μ, 16 μ, 8 μ) and DXA, and then sectioned for histomorphometric analysis of mineralized bone. Bone mineral density (BMD), trabecular parameters (number, TbN; separation, TbSp; thickness, TbTh) and mineralized bone volume/total bone volume (BV/TV) were compared and correlated among imaging methods and histomorphometry. Statistical tests were considered significant if P<0.05. Ten weeks post inoculation, diaphyseal BMD increased in the femur with tumor compared to the opposite femur by all modalities (p<0.005, n = 11). Diaphyseal BMD by in vivo μCT correlated with ex vivo 31 and 16 mm μCT and histomorphometry BV/TV (r = 0.91-0.94, P<0.001, n = 11). DXA BMD correlated less with bone histomorphometry (r = 0.73, P<0.001, n = 11) and DXA did not distinguish trabeculae from cortex. By in vivo and ex vivo μCT, trabecular BMD decreased (P<0.05, n = 11) as opposed to the cortex. Unlike BMD, trabecular morphologic parameters were threshold-dependent and when using "fixed-optimal-thresholds," all except TbTh demonstrated trabecular loss with tumor and correlated with histomorphometry (r = 0.73-0.90, P<0.05, n = 11). Conclusions/Significance: Prostate cancer involving the skeleton can elicit a host bone response that differentially affects the cortex compared to trabeculae and that can be quantified noninvasively in vivo and nondestructively ex vivo.