Two-site evaluation of the relationship between in vivo and carcass dual energy X-ray absorptiometry (DXA) in pigs

A study was conducted to compare swine body composition measurements (fat tissue, lean tissue and bone mineral) made by DXA in vivo and on the right carcass half using two GE Lunar DPX pencil beam instruments at two different locations. Pigs weighing between 60 and 105 kg were first scanned in vivo by DXA and after slaughter the right carcass half of each pig was scanned — 78 pigs at the USDA in Beltsville, MD using a GE Lunar DPX-L instrument and 62 pigs at the LVG in Oberschleissheim, Germany using a GE Lunar DPX-IQ instrument (an updated model of the DPX-L). In addition to a separate analysis, despite confounding effects likely caused by the different study locations and probe preparations (scan modes) the data were pooled. Based on linear regression analysis of the pooled data, the relationship between DXA carcass half (dependent variable) and in vivo measurements resulted in coefficients of determination (R 2 ) of 0.58 for fat percentage (√MSE=3.82%) and of 0.52 for lean percentage (√MSE=4.09%). The R 2 values were expectedly higher when the data sets were analyzed separately for each location: 0.79 (√MSE=2.38%) and 0.85 (√MSE=1.60%) for fat percentage and 0.79 (√MSE=2.41%) and 0.84 (√MSE=1.65%) for lean percentage at USDA and LVG, respectively. Regardless of the relatively close relationship among in vivo and carcass results for the pooled data, caution is advised in comparing measurements among GE Lunar DPX-L and DPX-IQ instruments, since the residuals (√MSE) for the regression analysis of DXA FAT% and DXA LEAN% measurements exceed 2.5%.