Marrow Adipose Tissue Quantification of the Lumbar Spine by Using Dual-Energy CT and Single-Voxel (1)H MR Spectroscopy: A Feasibility Study

Abstract

Purpose: To test the performance of dual-energy computed tomography (CT) in the assessment of marrow adipose tissue (MAT) content of the lumbar spine by using proton (hydrogen 1 [(1)H]) magnetic resonance (MR) spectroscopy as a reference standard and to determine the influence of MAT on the assessment of bone mineral density (BMD).

Materials and methods: This study was institutional review board approved and complied with HIPAA guidelines. Written informed consent was obtained. Twelve obese osteopenic but otherwise healthy subjects (mean age ± standard deviation, 43 years ± 13) underwent 3-T (1)H MR spectroscopy of the L2 vertebra by using a point-resolved spatially localized spectroscopy sequence without water suppression. The L2 vertebra was scanned with dual-energy CT (80 and 140 kV) by using a dual-source multi-detector row CT scanner with a calibration phantom. Mean basis material composition relative to the phantom was estimated in the L2 vertebra. Volumetric BMD was measured with and without correction for MAT. Bland-Altman 95% limits of agreement and Pearson correlation coefficients were calculated.

Results: There was excellent agreement between (1)H MR spectroscopy and dual-energy CT, with a mean difference in fat fraction of -0.02 between the techniques, with a 95% confidence interval of -0.24, 0.20. There was a strong correlation between marrow fat fraction obtained with (1)H MR spectroscopy and that obtained with dual-energy CT (r = 0.91, P < .001). The presence of MAT led to underestimation of BMD, and this bias increased with increasing MAT content (P < .001).

Conclusion: Dual-energy CT can be used to assess MAT content and BMD of the lumbar spine in a single examination and provides data that closely agree and correlate with (1)H MR spectroscopy data.

Figure 1:

Bland-Altman plot of MAT content with ¹H MR spectroscopy (MRS) and dual-energy CT (DECT). All values are within limits of agreement (dotted lines), corresponding to ±1.96 standard deviation (SD) from the mean.

Figure 2:

Plot of correlation analysis of MAT content assessed with ¹H MR spectroscopy (MRS) and dual-energy CT (DECT) . There is a strong correlation between the two techniques (r = 0.91, P < .001).

Figure 3a:

Plot of correlation analysis between the differences in corrected and uncorrected BMD and MAT content. (a) There is a strong correlation between the difference in corrected and uncorrected BMD and MAT content with dual-energy CT (DECT) (r = 0.99, P < .001). (b) There is a strong correlation between the difference in corrected and uncorrected BMD and MAT content with ¹H MR spectroscopy (MRS) (r = 0.87, P < .001).

Figure 3b:

Plot of correlation analysis between the differences in corrected and uncorrected BMD and MAT content. (a) There is a strong correlation between the difference in corrected and uncorrected BMD and MAT content with dual-energy CT (DECT) (r = 0.99, P < .001). (b) There is a strong correlation between the difference in corrected and uncorrected BMD and MAT content with ¹H MR spectroscopy (MRS) (r = 0.87, P < .001).