Women with fracture, unidentified by FRAX, but identified by cortical porosity, have a set of characteristics that contribute to their increased fracture risk beyond high FRAX score and high cortical porosity

Abstract

The Fracture Risk Assessment Tool (FRAX) is widely used to identify individuals at increased risk for fracture. However, cortical porosity is associated with risk for fracture independent of FRAX and is reported to improve the net reclassification of fracture cases. We wanted to test the hypothesis that women with fracture who are unidentified by high FRAX score, but identified by high cortical porosity, have a set of characteristics that contribute to their fracture risk beyond high FRAX score and high cortical porosity. We quantified FRAX score with femoral neck areal bone mineral density (FN aBMD), and femoral subtrochanteric architecture, in 211 postmenopausal women aged 54–94 years with non-vertebral fractures, and 232 fracture-free controls in Tromsø, Norway, using StrAx software. Of 211 fracture cases, FRAX score > 20% identified 53 women (sensitivity 25.1% and specificity 93.5%), while cortical porosity cut-off > 80th percentile identified 61 women (sensitivity 28.9% and specificity 87.9%). The 43 (20.4%) additional fracture cases identified by high cortical porosity alone, had lower FRAX score (12.3 vs. 26.2%) than those identified by FRAX alone, they were younger, had higher FN aBMD (806 vs. 738 mg/cm²), and fewer had a prior fracture (23.3 vs. 62.9%), all p < 0.05. They had higher cortical porosity (48.7 vs. 42.1%), thinner cortices (3.75 vs. 4.12 mm), lower cortical and total volumetric BMD (942 vs. 1053 and 586 vs. 699 mg HA/cm³), larger medullary and total cross-sectional areas (245 vs. 190 and 669 vs. 593 mm²), and higher cross-sectional moment of inertia (2619 vs. 2388 cm⁴) all <i>p</i> < 0.001. When the fracture cases and controls with high cortical porosity were compared, cases had higher cortical porosity, lower cortical vBMD, lower total vBMD, smaller cortical CSA/Total CSA, larger medullary CSA and larger total CSA than controls (all <i>p</i> ≤ 0.05). Thus, fracture cases, unidentified by FRAX, but identified by cortical porosity, had an architecture where the positive impact of larger bone size did not offset the negative effect of thinner cortices with increased porosity. A measurement of cortical porosity may be a marker of other characteristics that capture additional fracture risk components, not captured by FRAX.