Assessment of subchondral bone marrow lipids in healthy controls and mild osteoarthritis patients at 3T

Abstract

The compartment-specific lipid changes in femoral-tibial bone of healthy controls and mild osteoarthritis (OA) patients were quantified at 3.0 T. Healthy volunteers [Kellgren-Lawrence (KL) grade = 0; n = 15, 4 females, 11 males, mean age 39 ± 16 years, age range = 24-78 years] and mild OA patients (KL = 1, 2; n = 26, 12 females, 14 males, mean age 61 ± 14 years, age range = 27-80 years) were scanned on a 3 T scanner. Clinical proton density (PD)-weighted fast spin echo (FSE) images in the sagittal (without fat-saturation), axial and coronal (fat-saturation) planes were acquired for cartilage Whole-Organ MR Imaging Score (WORMS) grading. A voxel of 10 × 10 × 10 mm(3) was positioned in the medial and lateral compartments of the tibia [medial tibial (MT) and lateral tibial (LT)] and femur [medial femoral (MF) and lateral femoral (LF)] for MRS measurements using the single voxel-stimulated echo acquisition mode (STEAM) pulse sequence. All MRS data were processed with Java-based Magnetic Resonance User Interface (JMRUI). Wilcoxon's rank sum test and mixed model two-way analysis of variance (ANOVA) were performed to determine significant differences between different compartments as well as examine the effect of OA grade and compartment, and their interactions. Generally, the MF compartment index of unsaturation was increased in healthy subjects compared with OA subjects (whether graded by KL or WORMS score). Differences between MF at KL0 and all other compartments at KL1 except LF approached statistical significance (p < 0.05). Differences in saturated lipids signals could be observed predominantly in the 2.03 p.p.m. frequency shift. Healthy controls in the MF compartment had the lowest saturated lipid signals, and mild OA patients with KL2 and WORMS5-6 in the MF compartment had the highest saturated lipid signals compared with other compartments at 2.03 p.p.m. (p < 0.05).

Fig. 1

First two rows of Fig.1 (a) show the representative images of the voxel positions in the external reference water phantom and the corresponding spectra with the voxel position being top left, top right, bottom left, and bottom right, respectively. The third row of Fig. 1 (a) shows the corresponding coefficients of variation (CVs) of measurement with each voxel position kept the same between four times of repeated MRS measurements. Fig.1 (b) shows the representative images of the voxel position shift of the in-vivo MRS measurement and the corresponding spectra in LF, LT, MF, and MT compartment, respectively. Four knee OA patients were scanned two times with different voxel position shift between two measurements. For subject 1 in LF compartment, Anterior-Posterior (AP)/Left-Right (LR)/Feet-Head (FH) = 81 mm/102 mm/10 mm for measurement 1, 68 mm/102 mm/10 mm for measurement 2, in LT, 75 mm/99 mm/44 mm for measurement 1, 64 mm/102 mm/43 mm for measurement 2, in MF, 76 mm/63 mm/10 mm for measurement 1, 61 mm/63 mm/11 mm for measurement 2, in MT, the lipid peaks were unable to be differentiated, so ignored (also see Table. 2); for subject 2 in LF, 75 mm/115 mm/19 mm for measurement 1, 80 mm/113 mm/20 mm for measurement 2, in LT, 79 mm/111 mm/53 mm for measurement 1, 61 mm/109 mm/53 mm for measurement 2, in MF, 66 mm/73 mm/25 mm for measurement 1, 82 mm/75 mm/ 26 mm for measurement 2, in MT, /61 mm/81 mm/55 mm for measurement 1, 73 mm/83 mm/56 mm for measurement 2; for subject 3 in LF, 80 mm/73 mm/10 mm for measurement 1, 68 mm/71 mm/10 mm for measurement 2, in LT, 64 mm/71 mm/43 mm for measurement 1, 75 mm/73 mm/43 mm for measurement 2, in MF, 85 mm/ 29 mm/13 mm for measurement 1, 73 mm/27 mm/13 mm for measurement 2, in MT, 78 mm/28 mm/46 mm for measurement 1, 65 mm/26 mm/46 mm for measurement 2; for subject 4 in LF, 68 mm/109 mm/25 mm for measurement 1, 68 mm/109 mm/29 mm for measurement 2, in LT, 67 mm/111 mm/62 mm for measurement 1, 67 mm/111 mm/59 mm for measurement 2, in MF, 85 mm/61 mm/42 mm for measurement 1, 61 mm/61 mm/45 mm for measurement 2, and in MT, 79 mm/64 mm/66 mm for measurement 1, 80 mm/62 mm/65 mm for measurement 2, respectively. Fig. 1(c) shows the corresponding CVs of measurement at 0.9 ppm, 1.3 ppm, 2.03 ppm, and 5.31 ppm, respectively in each compartment with different voxel positions between two times of repeated MRS measurements.

Fig. 2

Representative cartilage images from normal to severe degeneration based on WORMS grading. Arrowheads mark the location of cartilage lesions. WORMS = 0 for all regions, a) coronal proton density (PD)-weighted FSE fat-saturated, b) lateral and c) central sagittal PD-weighted without fat saturation; WORMS = 1 for lateral patellar facet, d) axial PD-weighted FSE fat-saturated; WORMS = 2 for medial posterior tibial plateau, e) sagittal PD-weighted without fat saturation; WORMS = 2.5 for medial femoral condyle, f) coronal PD-weighted FSE fat-saturated, g) sagittal PD-weighted without fat saturation; WORMS = 4 for medial femoral central condyle, h) coronal PD-weighted FSE fat-saturated, i) sagittal PD-weighted without fat saturation; WORMS = 5 for lateral patellar facet, j) axial PD-weighted FSE fat-saturated; WORMS = 5 for medial tibial anterior plateau, k) coronal PD-weighted FSE fat-saturated; WORMS = 6 for central tibial plateau, l) coronal PD-weighted FSE fat-saturated, m) sagittal PD-weighted without fat saturation; WORMS = 6 for lateral tibial plateau, n) sagittal PD-weighted without fat saturation.

Fig. 3

Box and whisker plots comparing median bone marrow index of unsaturation among the four compartments (LF, LT, MF, MT) based on KL grade (a) and WORMS (b) in healthy controls and mild OA subjects. The horizontal dashed lines on the boxes show the corresponding mean values. The box and whisker plot shows the five statistics (minimum, first quartiles, median, third quartiles, and maximum). Fig. 3(a), there were significant differences (P < 0.05) in index of unsaturation between MF (KL0) and all other compartments (KL1) except LF (KL1). Fig. 3(b), significant differences (P < 0.05) existed in index of unsaturation between LF (WORMS0–1) and all other compartments (WORMS5–6) except LF (WORMS5–6), MF (WORMS0–1) and all other compartments (WORMS5–6) except LF (WORMS5–6), LT (WORMS5–6) and all other compartments (WORMS0–1) except LT (WORMS0–1).

Fig. 4

Box and whisker plots comparing median saturated lipids at 0.9 ppm (a), 1.3 ppm (b), 2.03 ppm (c), and (d) median unsaturated lipids at 5.31 ppm among the four compartments (LF, LT, MF, MT) based on KL score in healthy controls and mild OA subjects. The horizontal dashed lines on the boxes show the corresponding mean values. Fig. 4(b), there were significant differences (P < 0.05) in saturated lipids at 1.3 ppm between MT (KL0) and all compartments (KL1), LF (KL1) and all other compartments (KL0) except LF (KL0), LT (KL1) and all other compartments (KL0) except LF (KL0), MT (KL0) and all compartments (KL2). Fig. 4(c), there were significant differences (P < 0.05) in saturated lipids at 2.03 ppm between LF (KL0) and all other compartments (KL1) except MF (KL1), MF (KL0) and all compartments (KL1), LT (KL1) and all compartments (KL0), LF (KL0) and all other compartments (KL2) except MT (KL2), MF (KL0) and all compartments (KL2), LT (KL2) and all other compartments (KL0) except LT (KL0). Fig. 4(d), there were significant differences (P < 0.05) in unsaturated lipids at 5.31 ppm between MT (KL0) and all other compartments (KL1) except MT (KL1), LF (KL0) and all other compartments (KL2) except LF (KL2).

Fig. 5

Box and whisker plots comparing median saturated lipids at 0.9 ppm (a), 1.3 ppm (b), 2.03 ppm (c), and median unsaturated lipids at 5.31 ppm (d) among the four compartments (LF, LT, MF, MT) based on WORMS in healthy controls and mild OA subjects. The horizontal dashed lines on the boxes show the corresponding mean values. Fig. 5(b), there were significant differences (P < 0.05) in saturated lipids at 1.3 ppm between MT (WORMS0–1) and all compartments (WORMS5–6), LT (WORMS5–6) and all other compartments (WORMS0–1) except LF (WORMS0–1), MT (WORMS2–4) and all other compartments (WORMS5–6) except MF (WORMS5–6). Fig. 5(c), there were significant differences (P < 0.05) in saturated lipids at 2.03 ppm between LF (WORMS2–4) and all compartments (WORMS0–1), LT (WORMS2–4) and all compartments (WORMS0–1), LF (WORMS0–1) and all compartments (WORMS5–6), LT (WORMS0–1) and all other compartments (WORMS5–6) except MF (WORMS5–6), MF (WORMS0–1) and all compartments (WORMS5–6), LF (WORMS5–6) and all other compartments (WORMS0–1) except MT (WORMS0–1), LT (WORMS5–6) and all compartments (WORMS0–1), MT (WORMS5–6) and all other compartments (WORMS0–1) except MT (WORMS0–1).

Fig. 6

Box and whisker plots comparing median T₂* values at 0.9 ppm (a), 1.3 ppm (b), 2.03 ppm (c), and 5.31 ppm (d) among the four compartments (LF, LT, MF, MT) based on KL score in healthy controls and mild OA subjects. The horizontal dashed lines on the boxes show the corresponding mean values. Fig. 6(a), there were significant differences (P < 0.05) in mean T₂* values at 0.9 ppm between MT (KL1) and all other compartments (KL0) except MT (KL0). Fig. 6(b), there were significant differences (P < 0.05) in mean T₂* values at 1.3 ppm between LF (KL0) and all compartments (KL1), MF (KL0) and all other compartments (KL1) except LF (KL1), LT (KL1) and all other compartments (KL0) except MT (KL0), MT (KL1) and all other compartments (KL0) except MT (KL0), LF (KL0) and all compartments (KL2), MF (KL0) and all other compartments (KL2) except LF (KL2), LT (KL2) and all other compartments (KL0) except MT (KL0), MT (KL2) and all other compartments (KL0) except MT (KL0). Fig. 6(d), there were significant differences (P < 0.05) in median T₂* values at 5.31 ppm between LF (KL0) and all compartments (KL1), MF (KL0) and all other compartments (KL1) except LF (KL1), LT (KL1) and all compartments (KL0), MT (KL1) and all compartments (KL0), LF (KL0) and all other compartments (KL2) except LF (KL2), MF (KL0) and all other compartments (KL2) except LF (KL2), LT (KL2) and all compartments (KL0).

Fig. 7

Box and whisker plots comparing median T₂* values at 0.9 ppm (a), 1.3 ppm (b), 2.03 ppm (c), and 5.31 ppm (d) among the four compartments (LF, LT, MF, MT) based on WORMS in healthy controls and mild OA subjects. The horizontal dashed lines on the boxes show the corresponding mean values. Fig. 7(a), there were significant differences (P < 0.05) in median T₂* values at 0.9 ppm between LT (WORMS5–6) and all compartments (WORMS0–1), MT (WORMS5–6) and all other compartments (WORMS0–1) except MT (WORMS0–1). Fig. 7(b), there were significant differences (P < 0.05) in median T₂* values at 1.3 ppm between LF (WORMS0–1) and all other compartments (WORMS2–4) except LF (WORMS2–4), LF (WORMS0–1) and all compartments (WORMS5–6), MF (WORMS0–1) and all other compartments (WORMS5–6) except LF (WORMS5–6), LT (WORMS5–6) and all other compartments (WORMS0–1) except MT (WORMS0–1), MT (WORMS5–6) and all other compartments (WORMS0–1) except MT (WORMS0–1). Fig. 7(d), there were significant differences (P < 0.05) in median T₂* values at 5.31 ppm between LF (WORMS0–1) and all other compartments (WORMS2–4) except LF (WORMS2–4), LF (WORMS0–1) and all compartments (WORMS5–6), MF (WORMS0–1) and all other compartments (WORMS5–6) except LF (WORMS5–6), LT (WORMS5–6) and all compartments (WORMS0–1), MT (WORMS5–6) and all other compartments (WORMS0–1) except LT (WORMS0–1), LF (WORMS2–4) and all other compartments (WORMS5–6) except LF (WORMS5–6), LT (WORMS5–6) and all other compartments (WORMS2–4) except LT (WORMS2–4).