Correlation between lumbar dysfunction and fat infiltration in lumbar multifidus muscles in patients with low back pain

Forty-two patients with non-specific LBP and a referral for physiotherapy, who had a recent magnetic resonance imaging (MRI) of their lower spine, were consecutively recruited for this study from different healthcare centers. Patients were not allowed to be familiar with the measures used in this study and they should not have received manual therapy or lumbar stabilization programs prior to this study. Patient data for age, gender, body weight and duration of LBP (acute pain < 12 weeks, chronic pain > 12 weeks) were collected and the body mass index (BMI) for each patient was documented. Patients who were obese (BMI > 35) or not able to perform active lumbar flexion and extension due to pain inhibition were excluded from this study. Those who had prior back surgery, sacroiliac arthritis, acute lumbar trauma, neurological deficits, active malignancy, infectional diseases, and were under the age of 20 or over the age of 75 were excluded as well.

The patients’ MR images were generated prior to this study for medical diagnostics and not for the purpose of this study. MR images were obtained with 1.5 T systems (GE Medical Systems, USA; Siemens Healthcare, Erlangen, Germany) and patients were positioned supine in the MRI device. Each MRI sample contained standard T1- and T2-weighted axial images of the lumbar spine. All images were stored as DICOM format for processing. Due to the clinical setting of this study, we had to accommodate the fact that MR images were produced in different radiology centers and MRI parameters were not standardized. Analyze software (OsiriX, Pixmeo SARL, Switzerland, Version 5.6) was used for image analysis. To determine fat infiltration of the lumbar multifidus muscles, all axial T1-weighted MR images were included in the analysis as such sequences provide excellent anatomical detail.

While many quantitative MRI-based methods like Dixon/IDEAL [24, 25] or proton-density fat-fraction [26] are more accurate to measure fat separation, we had to take the variable MRI parameters into account. Semiquantitative assessment of fat infiltration of lumbar muscles have been reported to be valid and reproducible, and findings correlated with MR spectroscopic measurements [15]. There is also evidence that visual grading of fat infiltration in LMM, using MR images, is reliable [27, 28]. For that reason, a visual evaluation method was used for this study.

Images were analyzed slice per slice within the determined range between L3 and L5 (Fig. 1). In order to optimize image quality, grey scaling was used during analysis. The area demonstrating the highest quantity of fat infiltration in LMM (left and right side combined) was used for grading. To determine fat infiltration of LMM three criteria were used (Fig. 2): Grad 0 (0–10% fat), Grade 1 (10–50% fat) and Grade 2 (>50% fat) [16].

Grading for this study was performed by a doctor of chiropractic who is a clinician and instructor of radiological diagnostics at the national chiropractic academy with 25 years of experience. He was blinded to the patients functional assessments.

Lumbar dysfunction was specified as reduced spinal flexibility, impairment of movement and posture control, attenuation of body awareness and self-assessed functional disability. In order to assess the patients’ extent of LBP and the functional abilities of the low back, a set of different tests were used. All tests were performed by the same investigator, a physical therapist with 20 years of experience in manual therapy. To minimize testing bias, procedures were standardized and trained prior to the study. Patients wore only underwear to allow the observation of the entire body. The investigator was not aware of the patients’ MRI findings.

For each response, we fit a linear model to the data using fat, gender, age, duration of LBP and BMI as covariates. Our linear model for the j-th subject from group i was Y _ij   = μ + α _i  + covariates _ij   + ε _ij , with μ as the intercept, α _i as the fat-effect of group i and the ε _ij as independent and normally distributed errors. We were interested in the covariate-adjusted effect of fat on the outcomes. Pairwise contrasts between the fat-groups were estimated from the estimated model. All simultaneous inference procedures controlled the family-wise error rate of α = 0.05. Residual analysis was performed to check model assumptions, that is, independent and normally distributed errors. Breusch-Pagan tests were performed to test for homogeneous variances and Shapiro-Wilk tests for the normality assumption. For the association between fat content of LMM and age, status, gender and BMI, the Pearson correlation coefficient was calculated. A P value of less than .05 was considered to demonstrate a statistically significant difference.

The statistical analysis was performed with R, version 3.1.0 [34].

Results showed that increased severity of fat infiltration in the lumbar multifidus muscles correlated with decreased range of motion of lumbar flexion (p = 0.032). Pairwise contrasts between the fat-groups indicated a significant difference (p = 0.039) between fat Grade 1 and fat Grade 2 (12.42°, 95% CI 0.513, 24.3) (Table 3).

However, none of the effect moderators (age, gender, duration of LBP, BMI) correlated with lumbar flexion (Table 4).

No significant correlation could be demonstrated between the severity of fat infiltration and movement control and there was no significant correlation found between fat infiltration and posture control (Fig. 4). But patients’ posture control was affected by the duration of LBP. Patients with acute LBP demonstrated a significant (p = 0.003) greater lumbar evasive movement in extension than patients with chronic LBP. The ODI scores and the TPD values did not correlate with the severity of fat infiltration in lumbar multifidus muscles.

The relationship between spinal range of motion and disability in patients with low back pain has already been examined by various authors. Some found no significant correlation between lumbar flexion and reported disability [37, 38], while others reported decreased hip flexion and reduced spinal range of motion in all directions. The latter was found only in patients with LBP and limited straight leg raise [39]. The results of our study demonstrated that fat grade correlated with decrease of lumbar flexion only, whereas the amount of hip flexion increased, albeit not significantly. Therefore the increase of hip flexion might reflect a compensation for the loss of lumbar flexion. To explain the observed reduction in lumbar flexion, possible factors influencing spinal stability have to be highlighted. Shin et al. [40] demonstrated in their work that healthy subjects with the greatest lumbar flexibility had the highest activity levels of the LMM. Panjabi [41] described a model for spinal stability consisting of three subsystems (spinal column, spinal muscles and neural control unit) which together create optimal spinal flexibility and dynamic stability. As these systems are interdependent, one system could compensate for the deficits of another. If LBP occurs, inhibition of neural control is the consequence. Inhibition impedes alpha motor neuron activity in the anterior horn of the spinal cord and inhibits activity of LMM [12]. Thus, lumbar muscles cannot administrate their function anymore, which most likely debilitates postural control. Ongoing pain inhibition leads to alterations in neuro-muscular control, even after remission of LBP [10]. This mechanism becomes chronic and may result in atrophy of LMM. As important stabilizers, LMM act to maintain optimal joint forces, not only in the neutral zone of the spine, but also during prolonged flexion [42]. Therefore, atrophy in the muscular subsystem can lead to instability that must be compensated.

Superficial paraspinal and trunk muscles may have the ability to compensate for the deficit of LMM. Cholewicki et al. [43] investigated the stabilizing function of trunk flexor and extensor muscles in a neutral spine position. They demonstrated that active spinal stability was provided by flexor and extensor coactivation, but participants used different muscle recruitment strategies to achieve lumbar stability. The coactivation of local stabilizers such as LMM and M. tranversus abdominis increase intervertebral stiffness and allow superficial muscles to perform spinal movement. It has been hypothesized that recruitment strategies change in patients with LBP and global muscles try to compensate by global coactivation [44]. Although global coactivation increases stability, it also restricts spinal motion and function. Chan et al. [45] identified changed patterns of elasticity and cross-sectional area in LMM in relation to posture. In upright 25 and 45° forward stooping positions, the multifidus stiffness was higher in LBP patients than that in asymptomatic controls. There is also evidence that altered muscle activation strategies increase trunk stiffness in resting upright postures in recurrent LBP patients [6].

However defined, it cannot be determined whether reduced flexion range of motion of the lumbar spine is a cause or a result of fat infiltration of LMM. Referring to the results of our study, we hypothesize that once muscle activation strategies have changed, fatty infiltration of LMM proceeds. As a consequence, soft and ligamentous tissues that determine lumbar flexion are stiffening to compensate for the loss of dynamic stability. According to O’Sullivan [46], patients with LBP can be subgrouped into different movement dysfunction patterns. Patients with flexion patterns are probably the ones who have a dysfunction of their dorsal stabilizers. Potentially, fatty infiltrations of LMM is only present in the subgroup of patients with flexion patterns. The sample size of our study was too small to subgroup, but for future studies it would be worth trying to determine if there is a trend when subgrouping the patients. Lumbar pathology seen on MRI can play an important role in recurrence of LBP [47]. Whether decreased lumbar flexion and fat infiltration of LMM are risk factors for a recurrence of LBP remain unclear and has to be investigated.

In contrast to our hypothesis we found no significant correlation between the grade of fat infiltration and impaired movement control or impaired postural control.

Likewise, there was no significant correlation found between LBP and impaired movement control and body awareness. This contradicts the findings of other authors [3, 48] but can be explained by the rather small sample size and the fact that patients in this study were not subgrouped according to their LBP specificity. Interestingly, fat infiltration in LMM did not affect postural control, but patients with acute LBP demonstrated significantly higher impairment of posture control than patients with chronic LBP. These findings can be explained by using current evidence of the role of LMM in spinal stability and control. Multifidus muscles are predominantly occupied by muscle fiber type one [49] characterized by an extremly high cross-sectional area with very short muscle fibers that produce large forces over a narrow range of length [42]. The part that contributes the most to spinal stabilization is the deepest and also has a greater percentage of type 1 muscle fibers than the superficial part [50]. Ongoing loss of neural influence and mechanical loading leads to the atrophy of muscle fiber type I [51] whereas age-dependent atrophy of skeletal muscles affects predominately type two fibers [35]. The appearance of acute LBP changes corticomotor excitability [52] and first inhibits the deepest part of LMM which accordingly debilitates postural control more than movement control. This might explain our finding that patients with acute LBP demonstrated significantly higher impairment of posture control than patients with chronic LBP. Nevertheless, a correlation between duration of acute or chronic LBP and decreased lumbar flexion could not be demonstrated. In summary, our results revealed that fat infiltration of LMM has little impact on the measured functions of the lower back muscles.

The results presented in this study should be considered cautiously because of the small sample size and the possible methodological bias of a single-center study with only one tester and one reviewer. Unfortunately, we had to accommodate the fact that patients for this study were referred from different healthcare centers and therefore patients’ MRI were generated in different radiology centers. Differing MRI parameters and visual analysis are limiting factors for grading the amount of fat infiltration in LMM and the authors are aware of the possible inaccuracy of the MRI methodology. Standardized MRI protocols and quantitative MRI-based methods should be used in future studies. Likewise, the selection of patients should be refined in order to harmonize the physical characteristics of the cohort. For financial reasons, a control group with no fat infiltration of LMM and without LBP could not be included. The duration of LBP was measured, but the reoccurrence-rate of LBP was not evaluated. Although lumbar flexion was measured with a reliable instrument, other limiting factors for lumbar flexion (fear avoidance, straight leg raise etc.) were not evaluated in this study and should be considered in continuing investigations.