We examine the hypothesis that internal dual-energy X-ray absorptiometry (DXA) structural variables measured at the femoral neck and trochanter cross sections will improve prediction of fractures at these sites in addition to areal BMD (aBMD). We present an analysis of the internal distribution of bone at the site of these fractures using baseline hip DXA measurement of 1151 participants, mean age (SD) 75 (3) years, in the 14.5-year Perth Longitudinal Study of Aging in Women. Structural differences at the femoral neck (FN) and trochanter (TR) cross sections were compared in 69 participants who went on to sustain femoral neck fracture (FNF) and 59 participants who went on to sustain trochanter fracture (TRF), on average 10 years later, to those who did not fracture. The new structural variables, in addition to aBMD and bone width (W), were Sigma (distribution of bone within scanned area) and Delta (distance between center-of-mineral mass and geometric center), which can be assessed by current DXA programs. At baseline, compared with nonfracture cases, FNF cases had a FNaBMD 7% lower, a FNWidth 3% higher and a FNDelta 29% higher, associated with reduced bone in the superior segment. In FNF Cox proportional hazard analysis, age (hazard ratio [HR 1.39]), total hip aBMD (THaBMD; 0.79), and Delta (1.70) were significant. Addition of FN Delta to the base case of THaBMD and age improved the C statistic from 0.62 to 0.69, p = 0.01. At baseline, TRF cases had a 15% lower TRaBMD with minor differences in Sigma (1%) and Width (2%). In TRF Cox analysis, neither addition of Sigma nor Width improved the model that included age and THaBMD as predictors and did not alter the C-statistic of 0.81. In conclusion, adding measurement of FN Delta, an assessment of superior sector bone loss, to hip aBMD and age substantially improved clinical FNF prediction in older women.
Keywords:
STRUCTURAL GEOMETRY; FEMORAL NECK FRACTURES; TROCHANTERIC FRACTURES; SIGMA; DELTA