Gender- and Age-Associated Differences in Bone Marrow Adipose Tissue and Bone Marrow Fat Unsaturation Throughout the Skeleton, Quantified Using Chemical Shift Encoding-Based Water-Fat MRI

Abstract

Bone marrow adipose tissue (BMAT) is a dynamic tissue which is associated with osteoporosis, bone metastasis, and primary bone tumors. The aim of this study is to determine region-specific variations and age- and gender-specific differences in BMAT and BMAT composition in healthy subjects. In this cross-sectional study, we included 40 healthy subjects (26 male: mean age 49 years, range 22-75 years; 14 female: mean age 50 years, range 29-71) and determined the bone marrow signal fat fraction and bone marrow unsaturation in the spine (C3-L5), pelvis, femora, and tibiae using chemical shift encoding-based water-fat imaging (WFI) with multiple gradient echoes (mGRE). Regions of interest covered the individual vertebral bodies, pelvis and proximal epimetaphysis, diaphysis, and distal epimetaphysis of the femur and tibia. The spinal fat fraction increased from cervical to lumbar vertebral bodies (mean fat fraction ( ± SD or (IQR): cervical spine 0.37 ± 0.1; thoracic spine 0.41 ± 0.08. lumbar spine 0.46 ± 0.01; p < 0.001). The femoral fat fraction increased from proximal to distal (proximal 0.78 ± 0.09; diaphysis 0.86 (0.15); distal 0.93 ± 0.02; p < 0.001), while within the tibia the fat fraction decreased from proximal to distal (proximal 0.92 ± 0.01; diaphysis 0.91 (0.02); distal 0.90 ± 0.01; p < 0.001). In female subjects, age was associated with fat fraction in the spine, pelvis, and proximal femur (ρ = 0.88 p < 0.001; ρ = 0.87 p < 0.001; ρ = 0.63 p = 0.02; ρ = 0.74 p = 0.002, respectively), while in male subjects age was only associated with spinal fat fraction (ρ = 0.40 p = 0.04). Fat fraction and unsaturation were negatively associated within the spine (r = -0.40 p = 0.01), while in the extremities fat fraction and unsaturation were positively associated (distal femur: r = 0.42 p = 0.01; proximal tibia: r = 0.47, p = 0.002; distal tibia: r = 0.35 p = 0.03), both independent of age and gender. In conclusion, we confirm the distinct, age- and gender-dependent, distribution of BMAT throughout the human skeleton and we show that, contradicting previous animal studies, bone marrow unsaturation in human subjects is highest within the axial skeleton compared to the appendicular skeleton. Furthermore, we show that BMAT unsaturation was negatively correlated with BMAT within the spine, while in the appendicular skeleton, BMAT and BMAT unsaturation were positively associated.

Keywords: bone marrow adipose tissue; bone marrow adipose tissue distribution; bone marrow fat unsaturation; healthy subjects; water–fat MR imaging.

Figure 1

Dataset and corresponding fit of vertebra Th11 of volunteer #20. The fit is performed on the complex valued data: the magnitude (arbitrary units) in (A), the phase (radians) in (B). The amplitude in the total signal magnitude is larger than in the fat component only, because of the interference of the fat and water signals.

Figure 2

(A) Bone marrow fat fraction increased from cranial to caudal (linear mixed model (LMM): p < 0.001). (B) Unsaturation (number of double bonds, ndb) was highest within the pelvis, followed by the spine, and lowest within the tibia (LMM: p < 0.001).

Figure 3

(A) Bone marrow fat fraction increased from cranial to caudal within the spine (linear mixed model (LMM): p < 0.001). (B) Fat fraction increased in the femur from proximal to distal (LMM: p < 0.001). (C) Fat fraction decreased in the tibia from proximal to distal (LMM: p < 0.001). (D) Unsaturation (number of double bonds, ndb) was similar when comparing the cervical spine, thoracic spine, and the lumbar spine (LMM: p = 0.09; ns). (E) Unsaturation was highest within the diaphysis of the femur (p < 0.001) and (F) within the diaphysis of the tibia (LMM: p < 0.001), compared to the proximal and distal sites. Images on the right side show representative images of ROI placement in the vertebral bodies of the spine, in the proximal epimetaphysis, the diaphysis and the distal epimetaphysis of the femora and tibiae. ns, non-significant.

Figure 4

Correlations (Spearman’s rank correlation coefficient) between age and bone marrow fat fraction (S_FF ) in male (blue) and female (red) subjects. (A) Spinal fat fraction and age were positively correlated in both male and female subjects. (B–D) Within the pelvis, femur, and proximal femur fat fraction and age were positively correlated in female subjects but not in male subjects.

Figure 5

Correlations (Pearson correlation coefficient) between bone marrow signal fat fractions (S_FF ) and unsaturation (number of double bonds ndb) in (A) the spine, (B) the distal femur, (C) the proximal tibia and (D) the distal tibia.