For this study, we recruited patients aged 41–79 years with nonspecific CLBP who presented with symptoms persisting for more than 3 months despite conservative treatment such as medication or therapeutic exercise. CLBP was defined as pain, stiffness, and discomfort in the lower back from the 12th rib down toward the lumbar or lumbosacral area, with unclear source of pain. The exclusion criteria were the presence of (1) a systemic inflammatory disease; (2) lower extremity symptoms; (3) prior spinal surgery; (4) neoplasm, infection, or acute trauma; (5) history of spinal fracture; (6) spinal deformities such as spondylolisthesis with or without obvious instability (defined as sagittal translation of >3 mm, segmental motion of >20°, or a posterior opening of >5° on flexion/extension radiographs) or scoliosis (>10°); (6) advanced disc degeneration (Pfirrmann grade V) [18]; and (8) diabetes, hypertriglyceridemia, or other metabolic/endocrine disorders. The Visual Analog Scale (VAS, 0–100) was used to measure the intensity of LBP. Patients with VAS scores >30 mm were selected. Body mass index (BMI) was calculated using self-reported body weight (kg) divided by the square of height (m²) as an indicator of obesity, which is associated with the total body fat.

We assessed full-length spine and pelvis radiographs taken with participants in a standing position as recommended in a previous report [19]. The following sagittal spinal radiological parameters (Fig. 1) were recorded in the sagittal plane of the spine radiographs: lumbar lordosis (LL), measured from the superior endplate of L1 to the superior endplate of S1; thoracic kyphosis (TK), measured from the superior endplate of T4 to the inferior endplate of T12; and sagittal vertical axis (SVA), measured as the horizontal offset from the posterior-superior corner of S1 to the center of the corpus of C7. Moreover, the following sagittal pelvic parameters were recorded in the sagittal plane of pelvic radiographs: sacral slope (SS), the angle between a horizontal reference line and the upper sacral endplate; pelvic tilt (PT), the angle between a vertical reference and a line passing through the midpoint of the sacral plate to the femoral head axis; and pelvic incidence (PI), the angle between a line perpendicular to the upper S1 endplate at its midpoint and a line connecting this point to the femoral head axis. Intra- and interobserver reliability values for the measurement of spinopelvic parameters were assessed using the measurements of the two investigators (I.O. and H.T.). Intra- and interobserver reliability values for the measurement of spinopelvic parameters were also assessed, yielding similar results (Pearson’s correlation coefficient: 0.84–0.89, 0.87–0.92).

This study followed a previously described MRI protocol [20-24]. Briefly, T2-weighted sagittal and transverse magnetic resonance (MR) images were obtained using the Signa HDx 1.5 T MRI system (GE Healthcare, Milwaukee, WI, USA) with a spine coil. Using these images, the proton MRS volume of interest (VOI) was placed at the center of the right psoas major at the L4/5 level (Fig. 2). A single-voxel point-resolved spectroscopy pulse sequence was acquired with the following parameters: repetition time, 2,000 ms; echo time, 35 ms; average number of signals, 64; VOI size, 15×15×15 mm (3.4 mL); and acquisition time 164 seconds.

CSA and FI of the right psoas major were measured using axial T2-weighted MR images at L4/5 (the same level and side as for MRS). CSA was determined by manually tracing the fascial boarder of the muscle (Fig. 3A) as previously described [15]. The regions of interest were analyzed based on areas and histograms of signal intensity using a digital image processing software (ImageJ; National Institutes of Health, Bethesda, MD, USA). The FI percentage was measured using a pseudo-coloring method, in which bright pixels presenting fat tissue were colored red using software tools. Subsequently, the percentage of red pixels in the muscle compartment was calculated (Fig. 3B). Then, the relative CSA (mm²/m²) was calculated as the absolute CSA (mm²) divided by the square of the patient’s height (m²) to eliminate any influences of body constitution and gender on these measured anatomical variables. Intra- and inter-reliability values for the measurement of CSA and FI were 0.93 and 0.91, respectively.

MRS data were analyzed according to a previously described method [20-24]. Spectral data were used to measure EMCL and IMCL contents using the LCModel software (Stephen Provencher Inc., Oakville, ON, Canada). Data were transferred from the scanners to a Linux workstation, and metabolites were quantified using eddy current correction and water scaling. Data for EMCL (1.5 ppm) and IMCL (1.3 ppm) methylene protons were used for statistical analyses. Assessments of EMCL and IMCL contents were scaled to an unsuppressed water peak (4.7 ppm) using the automatic function of the software and were expressed in institutional units. These data are displayed graphically, with the chemical shift on the x-axis and peak intensity on the y-axis, to enable metabolic identification (Fig. 4). Participants with psoas major EMCL or IMCL content >15% above the standard deviation were excluded from the LCModel analysis.

We assessed the possible correlations of LBP VAS scores, age, CSA, and FI with the EMCL and IMCL contents of the psoas major using Pearson’s correlation coefficient test according to gender. Mann-Whitney U-test was used to analyze gender differences. The associations of EMCL and IMCL contents with spinopelvic parameters were also analyzed using Pearson’s correlation coefficient test according to gender. Significance was set at p<0.05. All numerical data are expressed as means±standard error of the mean.