Spatial patterns of fat within the deep multifidus as a biomarker for chronic low back pain

Abstract

Background context: Chronic low back pain (cLBP) patients often have elevated fat infiltration (FI) in the multifidus (MF), but it is unclear how this relates to pain and degenerative spine features. Most prior work assess MF degeneration as the average whole-muscle fat content even though deep and superficial fascicles of the MF have different structural and functional characteristics. Assessing the spatial distribution of MF FI may provide regional context for causal mechanisms which may have distinct regional presentations within the muscle.

Purpose: This study assesses spatial patterns of MF FI at each lumbar level to identify regional differences associated with cLBP symptoms and degenerative spine features.

Study design: This is an observational cross-sectional study.

Patient sample: Our study sample consisted of 230 cLBP patients from the BACPAC comeback cohort who reported low back pain that has persisted for the past 3 months.

Outcome measures: Measures included MF fat-maps-a FI curve showing the spatial distribution of fat moving radially through the MF-created at each lumbar level (L1L2-L5S1). Other measures included average fat fraction in the deepest 15% of the MF (deep15 FI%), average whole-muscle fat fraction (overall FI%) and the Pain, Enjoyment of Life, and General Activity (PEG) survey score.

Methods: We collected 3T MRI and used advanced sequences (IDEAL) to map the spatial distribution of MF FI at each lumbar level. We used statistical parametric mapping to identify spatial patterns of fat in the MF associated with age, sex, and BMI. Then, we tested for differences in spatial patterns of MF FI associated with pain and adjacent disc degeneration. Next, we calculated the fat fraction in a region of interest in the deepest 15% of the MF (deep15 FI%) and used linear mixed effects modeling to compare ho w age, sex, BMI, pain, and degenerative spine features associate with the deep15 FI% and the overall FI% separately. Lastly, we used linear regression models of PEG to compare FI measures as predictors for pain.

Results: Elevated FI associated with PEG and adjacent disc degeneration only within the deepest 10% (p<.05) and deepest 25% (p<.01) of the MF respectively at the lower lumbar levels. Associations between demographic factors and FI were not specific to the deep MF. Older age and female sex associated with elevated FI throughout the muscle (p<.001) while higher BMI associated with elevated FI in only the superficial 60% of the MF (p<.001). Lastly, higher mean deep15 FI% but not mean overall FI% at the lower lumbar levels associated with higher PEG (p=.023).

Conclusions: FI in deep regions of the MF at L4L5 and L5S1 is more strongly associated with pain and adjacent disc degeneration and less associated with age, sex, and BMI than overall FI%. We identify regional differences in MF FI related to pain which improves our understanding of cLBP-related MF degeneration and provides additional context for possible causal mechanisms of FI. Further, the "deep15" FI% is a novel MF FI measure, and possible biomarker, more strongly associated with cLBP than summary measures.

Keywords: Biomarker; Chronic low back pain; Fat infiltration; Multifidus; Muscle quality; Spatial distribution.

Fig. 1.

Workflow for multifidus fat-map creation and SPM analysis A) Circular regions of interest, centered at the vertebral center of rotation, are defined in the muscle segmentations at increments of 2 pixels. B) For each circular ROI, the average fat infiltration across all pixels is in the ROI is averaged. C) The fat infiltration at each ROI is plotted and distance normalized to 0% to 100% of the muscle length, resulting in a fat map curve, allowing for comparisons. ex. Fat infiltration curves for males and females at the L4L5 vertebral level D) SPM statistical tests are used to identify regions of significance (clusters) in the distribution. A region is considered significant if the SPM test statistic (black line) crosses the critical threshold (dotted line) Ex. Females have higher FI than males at L4L5 from 1.1%−88.8% of the distribution.

Fig. 2.

Model descriptions for linear mixed effects modeling. For both deep15 FI% and overall FI%, we created 3 nested linear mixed effects models, resulting in a total of 6 models.

Fig. 3.

Multifidus fat distribution patterns differ across lumbar levels with lower lumbar levels having elevated fat infiltration in the deep region of the muscle. SPM t-tests showing regional differences in FI between upper lumbar levels and lower lumbar levels. Lower levels have higher FI in the deep MF and lower FI in the superficial MF compared to the upper lumbar levels.

Fig. 4.

SPM Canonical correlation analysis shows that older age associates with elevated FI across the entire distribution of the multifidus. Posthoc testing confirms this relationship at every lumbar level.

Fig. 5.

SPM Canonical correlation analysis shows that elevated BMI associates with higher levels of FI in different regions of the multifidus at the upper and lower levels. Posthoc testing reveals that this relationship is level specific. Interestingly, FI in the deep MF at L4L5 and L5S1 is not associated with BMI, but at upper lumbar levels, elevated BMI associates with elevated FI across the entire MF.

Fig. 6.

SPM Hotellings analysis shows that females have higher levels of FI across the entire distribution of the multifidus. Posthoc testing confirms this relationship at every lumbar level, except in the superficial MF at L4L5 and L5S1.

Fig. 7.

Multivariable regression accounting for age, sex, and BMI, identifies spatial patterns of FI associated with elevated adjacent disc degeneration. At lower levels, these disc degeneration related FI patterns are in the deep MF. At upper levels, an association is observed only at L1L2 in the superficial MF.

Fig. 8.

Multivariable regression accounting for age, sex, and BMI, shows that higher PEG score associates with distinct patterns of FI accumulation specifically in the deep MF at the lower lumbar levels. At upper levels, we did not find an association between PEG and FI distribution.

Fig. 9.

Elevated fat infiltration in the deepest 15% of the multifidus at L4L5 and L5S1 is uniquely associated with pain and disc degeneration but not with BMI.