Simultaneous Quantification of Bone Edema/Adiposity and Structure in Inflamed Bone Using Chemical Shift-Encoded MRI in Spondyloarthritis

Abstract

Purpose: To evaluate proton density fat fraction (PDFF) and R2* as markers of bone marrow composition and structure in inflamed bone in patients with spondyloarthritis.

Methods: Phantoms containing fat, water, and trabecular bone were constructed with proton density fat fraction (PDFF) and bone mineral density (BMD) values matching those expected in healthy bone marrow and disease states, and scanned using chemical shift-encoded MRI (CSE-MRI) at 3T. Measured PDFF and R2* values in phantoms were compared with reference FF and BMD values. Eight spondyloarthritis patients and 10 controls underwent CSE-MRI of the sacroiliac joints. PDFF and R2* in areas of inflamed bone and fat metaplasia in patients were compared with normal bone marrow in controls.

Results: In phantoms, PDFF measurements were accurate over the full range of PDFF and BMD values. R2* measurements were positively associated with BMD but also were influenced by variations in PDFF. In patients, PDFF was reduced in areas of inflammation and increased in fat metaplasia compared to normal marrow. R2* measurements were significantly reduced in areas of fat metaplasia.

Conclusion: PDFF measurements reflect changes in marrow composition in areas of active inflammation and structural damage and could be used for disease monitoring in spondyloarthritis. R2* measurements may provide additional information bone mineral density but also are influenced by fat content. Magn Reson Med 79:1031-1042, 2018. © 2017 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Keywords: bone marrow; chemical shift-encoded MRI; inflammation; spondyloarthritis.

Figure 1

Fat–water (a, b) and fat–water–bone (c, d) phantoms. (a) 0%, 30%, and 70% FF vials are shown lying flat to demonstrate solidity. (b) Microscopic images of 50% FF tube shows the arrangement of fat and water in the emulsion (the fat globules are circular and outlined by dark lines). (c) Photograph of the fat–water–bone phantom, showing fat content varying by row and bone mineral density varying by column. (d) Microscopic images of trabecular bone granules in dilute emulsion (approximately 10% FF); a number of fat droplets are visible coating the surface of the trabeculae. FF, fat fraction.

Figure 2

Delineation of edema and fat metaplasia. Areas of bone marrow edema and fat metaplasia were identified on T2‐weighted STIR (a) and T1‐weighted (d) spin echo images, respectively. Freehand regions of interest were placed on these images (b, e) and directly transferred to the corresponding anatomical location on PDFF maps (c, f) using the sacroiliac joints and sacral foraminae as anatomical landmarks. PDFF, proton density fat fraction; STIR, short tau inversion recovery.

Figure 3

Fat–water and fat–water–bone phantoms. (a) PDFF maps from the fat–water phantom. Labels indicate reference FF values, which vary between 0 and 100%. (b and (c) PDFF and R2* maps from the fat–water–bone phantom. Reference FF values vary by row, whereas bone mineral density values vary by column. FF, fat fraction; PDFF, proton density fat fraction.

Figure 4

Relationship of PDFF and R2* with reference FF and BMD. (a) PDFF measurements agree closely with reference FFs in the absence of bone. (b) Increases in bone density have minimal impact on PDFF measurements over a physiological BMD range. (c) R2* measurements are linearly related to BMD, although the slope varies depending on fat content. (d) Variations in FF also influence the measured R2* for a given BMD. BMD, bone mineral density; FF, fat fraction; PDFF, proton density fat fraction.

Figure 5

PDFF and R2* values in areas of normal bone marrow, bone marrow edema, and fat metaplasia. (a) PDFF values were significantly increased in areas of fat metaplasia, and significantly reduced in areas of bone marrow edema compared to normal subchondral marrow. (b) R2* values were significantly reduced in areas of fat metaplasia but not significantly altered in areas of edema. (c) PDFF and R2* measurements are shown on a scatterplot; areas of fat metaplasia exhibit both increases in PDFF and reductions in R2*. BMD, bone mineral density; FF, fat fraction; PDFF, proton density fat fraction.

Figure 6

PDFF maps (a, b) and R2* maps (c, d) demonstrating bone marrow edema and fat metaplasia in patients with sacroiliitis. Bone marrow edema (a, c; solid red arrow) causes a decrease in PDFF but no change in R2*. Fat metaplasia (b, d; dashed red arrow) causes an increase in PDFF and a reduction in R2*. The reduction in R2* may indicate a loss of bone mineral density. PDFF, proton density fat fraction.

Figure 7

Bland‐Altman plots. The figures demonstrate the repeatability of observations for PDFF (a, b) and R2* (c, d) measurements, using scatterplots (a, c) and Bland‐Altman 95% limits of agreement (b, d). Limits were adjusted to account for the presence of multiple observations for each individual. PDFF, proton density fat fraction.