Cortical bone quality as a biomarker for diabetes risk in post-menopausal Chinese-Singaporean women: a preliminary study
Francesca A. A. Leek1, Anna Therese Sjoholm1, Christiani Jeyakumar Henry2, Xiaodi Su3, Marlena C. Kruger4, and John J. Totman1

1A*STAR-NUS Clinical Imaging Research Centre, Singapore, Singapore, 2A*STAR Clinical Nutrition Research Centre, Singapore, Singapore, 3A*STAR Institute of Materials Research and Engineering, Singapore, Singapore, 4School of Food and Nutrition, College of Health, Massey University, Palmerston North, New Zealand

Synopsis

The feasibility of utilising proximal femur cortical bone quality as a biomarker for diabetes risk in post-menopausal Chinese-Singaporean women was investigated. Non-dominant proximal femurs were imaged with quantitative CT (QCT) and MR for the assessment of volumetric bone mineral density (vBMD) and cortical bone porosity. A significant (p<0.01; n=8) positive correlation between MRI vBMD and QCT vBMD for the region of maximum cortical thickness was shown. Whether MRI vBMD is associated with fracture risk and if it is sensitive to changes due to dietary or drug intervention needs to be investigated to fully assess the clinical potential of this method.

Purpose

Bone strength is characterised by bone mineral density (BMD) and bone quality. While cohort studies have found an increased risk of bone fractures in patients with type 2 diabetes, type 2 diabetics do not appear to have lower BMD than non-diabetics.1 The current gold standard for measuring areal BMD is Dual-energy X-ray Absorptiometry (DXA); being a projection method, cortical and trabecular bone are not differentiated nor is bone quality assessed. Three-dimensional imaging methods are needed in order to determine whether volumetric BMD (vBMD) or bone quality may be a potential biomarker for diabetes risk. The purpose of this work is to assess the feasibility of utilising proximal femur cortical bone quality obtained from quantitative imaging as a biomarker for diabetes risk in post-menopausal Chinese-Singaporean (C-S) women.

Methods

100 post-menopausal C-S women will be recruited for this ongoing study. The non-dominant proximal femur of each subject is assessed for vBMD with quantitative CT (QCT, Siemens Biograph mCT) and T1 weighted in-phase gradient echo (T1-VIBE-DIXON) MRI sequence (Siemens Prisma, 18-channel body array coil), and cortical bone porosity with an MRI ultrashort echo-time (UTE) sequence; see Table 1 for imaging parameters.

QCT images are converted to dipotassium phosphate (K2HPO4) equivalent densities using a commercial calibration phantom (Mindways QCT Pro) and the region of maximum cortical thickness in the field of view, found distally from the junction of the lesser trochanter and femoral shaft, are determined from images segmented for cortical bone (threshold > 350 mg/cc). The total effective dose is kept below 2 mSv for each subject.

MRI vBMD is quantified using a T1-VIBE-DIXON MRI sequence with two external calibration references (oil; 400 mg/cc calcium hydroxyapatite (CaHA) in H2O). CaHA equivalent BMDs are estimated from the signal relationship and known concentration of CaHA in the calibration references 2, for the volume of interest (VOI) determined on QCT.

Bone water concentration (BWC, shown to be a surrogate to cortical bone porosity 1) is quantified using a coronal UTE sequence (10,000 radial spokes) and an external reference (10% H2O in D2O doped with MAGNEVIST®). The transverse relaxation time (T2*) is determined by fitting a mono-exponential curve to images with TE<1.0 ms for all voxels within the VOI determined on QCT. BWC is calculated using the method previously described 3; J-modulation is included in the fitting model 4.

Results

For the eight subjects analysed to date (age range 50-67 years; mean age 58.6 years), the mean T2* of cortical bone was found to be 583 ± 22 µs and the mean BWC 16.9 ± 3.6 % at the region of greatest cortical thickness. As would be expected, a negative correlation between BWC calculated from MRI and vBMD from QCT is shown in Table 2; this correlation is not significant (p>0.1).

The absolute vBMDs for each imaging modality are not comparable as they are equivalent to materials with different inherent densities. However, a significant (p<0.01) positive correlation between vBMD derived from MRI and QCT for the region of maximum cortical thickness is shown in Table 2.

Discussion

The T2* of cortical bone was consistent with that found in literature, however, the resultant BWC matched that expected for pre-menopausal women 2; the correlation with QCT vBMD was not significant (p>0.1). Radiofrequency inhomogeneity across the imaging volume and depth-dependent receive-coil sensitivity could be causing signal variations between the external reference and the deep-lying cortical bone resulting in the discrepancies seen. The use of an internal reference standard to calibrate to an external reference of known composition should be investigated.

MRI vBMD significantly positively correlates with QCT vBMD (p<0.01) in the region of maximum cortical thickness in the proximal femur. This correlation suggests that the previously reported signal model 2, used for calculating MRI vBMD, is appropriate.

Conclusion

While initial results indicate that MRI derived cortical bone vBMD has the potential to be used in the assessment of bone quality, more extensive evaluation to determine the method’s clinical potential in the early identification of diabetes risk is needed. Other questions of particular clinical interest would be whether MRI vBMD is associated with fracture risk, whether it is sensitive to changes due to dietary or drug intervention and whether there is the potential to use this low ionising radiation method for greater population screening.

Acknowledgements

This project is part funded by Singapore-New Zealand Foods for Health Grant (BMRC grant 14/1/16/24/008).

References

1. Burghardt AJ, Issever AS, Schwartz AV et al. High-Resolution Peripheral Quantitative Computed Tomographic Imaging of Cortical and Trabecular Bone Microarchitecture in Patients with Type 2 Diabetes Mellitus. J Clin Endocrinol Metab 2010; 95(11): 5045-5055.

2. Ho K-Y, Hu HH, Keyak JH et al. Measuring Bone Mineral Density With Fat-Water MRI: Comparison with Computed Tomography. J Mag Reson Imaging 2013; 37: 237-242.

3. Techawiboonwong A, Song HK, Leonard MB et al. Cortical Bone Water: In Vivo Quantification with Ultrashort Echo-Time MR Imaging. Radiology 2008; 248(3): 824-833.

4. Gajdošík M, Chmelík M, Just-Kukurová I et al. In Vivo Relaxation Behaviour of Liver Compounds at 7 Tesla, Measured by Single-Voxel Proton MR Spectroscopy. J Mag Reson Imaging 2014; 40: 1365-1374.

Figures

Table 1 - imaging parameters.

Table 2 - correlation coefficients (R) of BWC and vBMD from MRI and vBMD from QCT.

Figure 1 - UTE dataset of left proximal femur with external reference; TE=0.04, 0.1, 0.3, 0.5, 1.0 1.23, 2.0, 2.46 ms.

Figure 2 – example plot for T2* of proximal femur cortical bone calculated from J-modulation corrected images with TE = 0.04-1.0 ms.

Figure 3 – T1 weighted in-phase gradient echo MR image of the left proximal femur and two calibration references (oil; 400 mg/cc CaHA in H2O [uppermost calibration reference used for UTE sequence]), converted to CaHA equivalent BMDs.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2258