Focal osteoporosis defects play a key role in hip fracture

Abstract

Background: Hip fractures are mainly caused by accidental falls and trips, which magnify forces in well-defined areas of the proximal femur. Unfortunately, the same areas are at risk of rapid bone loss with ageing, since they are relatively stress-shielded during walking and sitting. Focal osteoporosis in those areas may contribute to fracture, and targeted 3D measurements might enhance hip fracture prediction. In the FEMCO case-control clinical study, Cortical Bone Mapping (CBM) was applied to clinical computed tomography (CT) scans to define 3D cortical and trabecular bone defects in patients with acute hip fracture compared to controls. Direct measurements of trabecular bone volume were then made in biopsies of target regions removed at operation.

Methods: The sample consisted of CT scans from 313 female and 40 male volunteers (158 with proximal femoral fracture, 145 age-matched controls and 50 fallers without hip fracture). Detailed Cortical Bone Maps (c.5580 measurement points on the unfractured hip) were created before registering each hip to an average femur shape to facilitate statistical parametric mapping (SPM). Areas where cortical and trabecular bone differed from controls were visualised in 3D for location, magnitude and statistical significance. Measures from the novel regions created by the SPM process were then tested for their ability to classify fracture versus control by comparison with traditional CT measures of areal Bone Mineral Density (aBMD). In women we used the surgical classification of fracture location ('femoral neck' or 'trochanteric') to discover whether focal osteoporosis was specific to fracture type. To explore whether the focal areas were osteoporotic by histological criteria, we used micro CT to measure trabecular bone parameters in targeted biopsies taken from the femoral heads of 14 cases.

Results: Hip fracture patients had distinct patterns of focal osteoporosis that determined fracture type, and CBM measures classified fracture type better than aBMD parameters. CBM measures however improved only minimally on aBMD for predicting any hip fracture and depended on the inclusion of trabecular bone measures alongside cortical regions. Focal osteoporosis was confirmed on biopsy as reduced sub-cortical trabecular bone volume.

Conclusion: Using 3D imaging methods and targeted bone biopsy, we discovered focal osteoporosis affecting trabecular and cortical bone of the proximal femur, among men and women with hip fracture.

Keywords: Fracture prediction; Hip fracture; Osteoporosis; Pathogenesis.

Graphical abstract

Fig. 1

a) Boundaries for classifying the two main types of hip fracture; femoral neck (FN) and trochanteric (TR), shown on a 3D Bone Map of cortical thickness from a clinical CT scan b) Standard clinical two-dimensional areal bone mineral density (aBMD) measurements of i) the total hip region and ii) the femoral neck region. Approximate boundaries for aBMD measurement used in this study (QCTpro CTXA version 5.1.3, Mindways Software Inc., Austin, Texas).

Fig. 2

Biopsy regions. The cartoon shows the locations for the biopsies, and a resulting XTEK high resolution scan image through the femoral head of a FEMCO study participant who had donated their femoral head at operation. Also shown are the five 100-slice segments.

Fig. 3

Study 1.1. Bone Mapping (CBM) ROIs. Statistically significant differences in cortical (a) and trabecular (b) bone between hip fracture (n = 70, 50 female, 20 male) and 70 healthy controls shown as a colour map on the canonical femur model. CMSD Cortical Mass Surface Density, ECTD Endocortical Trabecular Density Study 1.2. Cortical (c) and trabecular (d) differences between female femoral neck patients (n = 86) and controls (n = 125). Cortical (e) and trabecular (f) differences between female trochanteric fracture patients (n = 52) and controls (n = 125). The black arrow highlights the biopsy site for Study 2.1. Study 1.3. Cortical (g) differences between frail female patients with at least one injurious fall (n = 50) versus 50 healthy controls (no difference in trabecular bone (h)).

Fig. 4

Results from biopsy study. The graph shows the statistically significantly lower BV/TV (with standard error of the mean) in the trabecular area highlighted by the Bone Mapping technique (black arrow in Fig. 3d). p = 0.046 for the paired difference between 1 and 100 slices.

Fig. 5

a) Area Under the Curve (AUC) from Receiver Operating Characteristic analysis for the ability of age, height, Femoral neck (Fn) and Total Hip (Th) DXA-like areal Bone Mineral Density (aBMD) from CT to correctly categorise hip fracture types (grey lines- dotted for TR or dashed for FN) as well as all hip fractures (black solid lines-ALL HIP FX). ROC analysis for different combinations of the novel 3D Cortical Bone Mapping (CBM) measures to correctly discriminate hip fractures (5b–d). An average single measure of 3D Cortical Mass Surface Density or Trabecular Density (ECTD) was taken for each patient from the bone mapping ROIs (shown as patches in Fig. 3c–f). The ability of age, height and an average 3D measure of either CMSD (5b), ECTD (5c) or both CMSD and ECTD (5d) to correctly discriminate fractures, as well as the corresponding AUC values and 95% confidence intervals for discriminating all fractures (ALL FX), Trochanteric fractures (TR) and Femoral neck (FN) are shown.