Adding Marrow Adiposity and Cortical Porosity to Femoral Neck Areal Bone Mineral Density Improves the Discrimination of Women With Nonvertebral Fractures From Controls

Roger Zebaze, Marit Osima, Minh Bui, Marko Lukic, Xiaofang Wang, Ali Ghasem-Zadeh, Erik F. Eriksen, Angela Vais, Catherine Shore-Lorenti, Peter R. Ebeling, Ego Seeman, Åshild Bjørnerem

Research output: Contribution to journalArticleResearchpeer-review

13 Citations (Scopus)


Advancing age is accompanied by a reduction in bone formation and remodeling imbalance, which produces microstructural deterioration. This may be partly caused by a diversion of mesenchymal cells towards adipocytes rather than osteoblast lineage cells. We hypothesized that microstructural deterioration would be associated with an increased marrow adiposity, and each of these traits would be independently associated with nonvertebral fractures and improve discrimination of women with fractures from controls over that achieved by femoral neck (FN) areal bone mineral density (aBMD) alone. The marrow adiposity and bone microstructure were quantified from HR-pQCT images of the distal tibia and distal radius in 77 women aged 40 to 70 years with a recent nonvertebral fracture and 226 controls in Melbourne, Australia. Marrow fat measurement from HR-pQCT images was validated using direct histologic measurement as the gold standard, at the distal radius of 15 sheep, with an agreement (R2 = 0.86, p < 0.0001). Each SD higher distal tibia marrow adiposity was associated with 0.33 SD higher cortical porosity, and 0.60 SD fewer, 0.24 SD thinner, and 0.72 SD more-separated trabeculae (all p < 0.05). Adjusted for age and FN aBMD, odds ratios (ORs) (95% CI) for fracture per SD higher marrow adiposity and cortical porosity were OR, 3.39 (95% CI, 2.14 to 5.38) and OR, 1.79 (95% CI, 1.14 to 2.80), respectively. Discrimination of women with fracture from controls improved when cortical porosity was added to FN aBMD and age (area under the receiver-operating characteristic curve [AUC] 0.778 versus 0.751, p = 0.006) or marrow adiposity was added to FN aBMD and age (AUC 0.825 versus 0.751, p = 0.002). The model including FN aBMD, age, cortical porosity, trabecular thickness, and marrow adiposity had an AUC = 0.888. Results were similar for the distal radius. Whether marrow adiposity and cortical porosity indices improve the identification of women at risk for fractures requires validation in prospective studies.

Original languageEnglish
Pages (from-to)1451-1460
Number of pages10
JournalJournal of Bone and Mineral Research
Issue number8
Publication statusPublished - Aug 2019



Cite this