Muscle impairment is an important component of spinal disorders. In fact, muscle dysfunction can be considered both a cause and a consequence of low back pain (LBP) (1)—for instance, several studies have shown an association between fatty infiltration and the reduced cross-sectional area (CSA) of paraspinal muscles with chronic LBP (2–7). Moreover, fatty degeneration may lead to functional limitations such as deterioration of balance and alignment of the spine (8, 9). However, to date, the contribution of compromised muscles to pain in spinal pathologies, such as facet arthropathy, disc degeneration, spinal stenosis, or deformity, is unclear. Paraspinal muscle fatty degeneration and atrophy have been reported in patients with LBP and disc herniation (10–13) or degenerative lumbar flat back (14). In patients with disc herniation, muscle infiltration may be asymmetric and more pronounced on one side than the other (15). Moreover, greater fatty infiltration in muscles such as multifidus, longissimus, and psoas correlates with poorer functional outcomes with an even stronger relation in persons with a history of LBP (16). Similarly, patients with lumbar spinal stenosis (LSS) or symptomatic LSS (sLSS) show increased muscle atrophy and fatty infiltration (17). LSS is one of the most frequent spinal disorders and the most common reason for spine surgery in the population aged above 65 years (18). While there are several studies investigating the correlation of muscle composition and morphology in patients with LBP, to date, evidence for this relationship in patients with LSS is lacking.

Fatty degeneration of the multifidus and reduced CSA of the psoas muscles have been associated with a lower functional performance in terms of higher scores on the Oswestry Disability Index (ODI) (19–21). In addition, a reduced CSA of the multifidus may predispose patients to worse outcome after surgery for sLSS (22). Moreover, the relation between CSA of paraspinal muscles and anthropometric parameters has been explored. The results of previous studies are inconclusive, where some studies did not observe a significant association between paraspinal muscle CSA and body height or body mass (23–25), while others reported a greater paraspinal CSA in taller and heavier persons (26).

Magnetic resonance imaging (MRI) facilitates the detailed investigation of the lumbar spine and the estimation of the morphology and composition of paraspinal muscles, including multifidus and erector spinae (longissimus and iliocostalis). These assessments can be performed in a qualitative or in a quantitative way. The Goutallier classification is a visual grading system to qualitatively assess fatty infiltration (27–30). Initially proposed for grading fatty degeneration of the rotator cuff muscles on computer tomography, the Goutallier classification has been expanded to MRI and to the evaluation of other muscles, including back muscles (6, 31, 32). Moreover, there is evidence of positive correlations between Goutallier grades and the severity of disc degeneration as well as age (33). Previous studies have shown a substantial to excellent intraobserver and a good interobserver reliability (32, 33) as well as a significant positive correlation of Goutallier grades, with the percentage of fat in the multifidus muscle measured in a quantitative way (32). Although the Goutallier classification is simple to apply and a useful tool for clinicians when evaluating their patients, it has the disadvantages of qualitative measurements and ordinal scales. This classification depends on the experience of the assessor and is reported on an ordinal scale with five discrete levels. In contrast, the quantitative MRI measurements of paraspinal muscles overcome such limitations by being objective and continuous measures. However, these quantitative measures are more time consuming and hence more suitable for research rather than clinical settings. To date, data on comparisons between qualitative and quantitative assessments of paraspinal muscle fatty infiltration and their association with patient characteristics in patients with sLSS are lacking. The aims of this study were to compare paraspinal muscle fatty infiltration as assessed using the Goutallier classification vs. quantitative MRI measurements, assess asymmetry in muscle degeneration, and investigate the association between anthropometric parameters and paraspinal muscle morphology and fatty infiltration in patients with sLSS.

This single center cross-sectional study was approved by the regional ethics committee and conducted in accordance with the Declaration of Helsinki. All enrolled patients provided written informed consent.

The aims of this study were to compare paraspinal muscle fatty infiltration as assessed using the Goutallier classification vs. quantitative MRI measurements and to investigate the association between anthropometric parameters and paraspinal muscle morphology and fatty infiltration in patients with sLSS. We observed a moderate to strong correlation between the Goutallier classification system of muscle fatty infiltration and the quantitative assessment of the LeanCSA and lean fraction of the paraspinal muscle CSA, named as LeanCSA/CSA. Moreover, paraspinal muscle morphology and fatty infiltration correlated strongly with body height. These results support the value of assessing fatty infiltration at all levels in the lumbar region and the importance of considering the stature of a patient when interpreting fatty infiltration in the context of LSS.

Patients with higher Goutallier grades had lower LeanCSA and LeanCSA/CSA, representing smaller absolute and relative lean muscle CSA. Similarly, Battaglia et al. reported a strong to very strong correlation for MR images of 25 randomly selected subjects between Goutallier grades and mean percent fat value measured with the same method as in our study (32). Though, to date, the role of fatty infiltration and atrophy of paraspinal muscles is not clear, there is evidence of their association with LBP—for instance, Kjaer et al. found that fatty infiltration of the multifidus muscle was strongly associated with LBP (5). These results are consistent with those of Goubert et al., who found a positive correlation of fatty infiltration with chronic LBP (4). Other authors investigated the association of muscle morphology and composition with spinal disorders. Yanik et al. reported a significant increase in fatty infiltration in individuals affected by disc herniation (6), and this was corroborated by other studies (15, 41). Yarjanian et al. found a reduced functional CSA of paraspinal muscles in patients with sLSS compared to asymptomatic individuals; however, the difference was not significant when compared to subjects with chronic LBP without evidence of LSS (17). Overall, the evidence for the relation of back muscle to spinal disorders emphasizes the importance for clinicians to regularly evaluate the morphology of paraspinal muscles on the MR images of patients. However, quantitatively measuring paraspinal muscle CSA and lean fraction is time consuming and hence not feasible in the clinical setting. Our results showed that fatty atrophy, assessed using the Goutallier classification system, strongly correlates with quantitative measures and thus—despite being a qualitative measure—is a good and, most importantly, efficient method for evaluating the degeneration of paraspinal muscles in sLSS.

The quantitative measures of paraspinal muscle morphology and fatty infiltration CSA, LeanCSA, and LeanCSA/CSA correlated strongly between the left and right sides across all levels of the lumbar spine. We interpret this as a sign of symmetry in muscle morphology and degeneration. However, because we did not record data on the laterality of symptoms nor of spinal stenosis, we cannot exclude that, in case of asymmetric symptoms or spinal stenosis, an asymmetric muscle degeneration may be observed. In fact, in a study on patients affected by disc herniation, Battié et al. reported greater fatty infiltration in the multifidus muscle on the side of the radicular compression at the level below the herniation, although the total paraspinal muscle CSA was greater on the affected side (15). In another study on patients with symptomatic posterolateral disc herniation at L4 and L5, Fortin et al. did not observe an asymmetry of the multifidus but of the erector spinae muscle, with smaller CSA and greater fatty infiltration on the affected side both of the multifidus muscle and the erector spinae muscle (41). The mechanism of these findings seems to be related to abnormal muscle activation as a result of the altered neural signal (42). Conversely, we did not observe any significant asymmetry of the paraspinal muscles. While we did not assess a possible laterality of symptoms and laterality, we would have only been able to detect a systematic laterality of all patients, which was not the scope of this study. Another possible explanation for this discrepancy is that LSS and its symptoms are more commonly bilateral compared to disc herniation. Nevertheless, further investigation is needed to support this statement.

Fatty infiltration generally increased from cranial to caudal, with the highest value at L5. This result confirms previous studies by Kjaer et al. (5) and Lee et al. (14), who showed that L4 and L5 are the segments most affected by fatty infiltration. In our study, both LeanCSA and LeanCSA/CSA differed significantly between L5 and all other levels. One possible explanation is that most degenerative spinal processes involve the segments L4/L5 and L5/S1 (43), but there is a lack of evidence regarding a causal link between fatty atrophy of paraspinal muscles and spinal disorders. Furthermore, Lee et al. (14) observed an increasing fatty infiltration toward the more caudal levels in 10 healthy volunteers. Kjaer et al. (5) found that the lowest lumbar levels had the greatest signs of fatty infiltration regardless of age in adults and adolescents with a history of LBP. These results suggest that the most caudal lumbar levels are physiologically more prone to fatty infiltration even in young and healthy individuals. Spinal pathologies known to be more frequent at these levels and to be associated with an increase in fatty infiltration of the back muscles may further exaggerate the difference in muscle composition between the cranial and caudal portions of the paraspinal muscles.

Our secondary aim was to determine the association between quantitative measurements of paraspinal muscle morphology and fatty infiltration and anthropometric parameters. We found that CSA and LeanCSA had a strong and LeanCSA/CSA had a weak correlation with body height. The results of previous studies on this association are inconclusive—for instance, our results confirm those reported by Gibbons et al., who observed a positive correlation between the CSA of paraspinal muscles and body height, body mass, and BMI in a large sample of male monozygotic twins (130 subjects) (26). In contrast, two smaller studies by Wood et al. (25) and McGill et al. (23) (13 and 26, respectively) did not find any significant association of CSA with body height, body mass, or BMI in persons with various suspected injuries and diseases. Moreover, Reid et al. reported that greater body height was not a predictor of larger CSA (24). Biomechanically, an association of CSA with body height is intuitive because the greater a person is, the greater are the moment arms of the forces generated by the weight of the upper torso, arms, and head relative to the lumbar spine area during daily activities. Frequently, the paraspinal total and lean CSA are normalized to the vertebral body or intervertebral disc CSA (19, 44–47). It is interesting to note that, in our study, both overall CSA and LeanCSA correlated more strongly with body height than with vertebral CSA (strong vs. moderate correlation). Although we consider the vertebral body or the disc as reasonable choice of reference for reducing the bias of differences in anthropometric measures between persons, our results suggest that body height greatly influences paraspinal muscle CSA and should hence be used for normalization.

In our study, only CSA, but not LeanCSA or LeanCSA/CSA, correlated with BMI. This result is consistent with the study by Kjaer et al. (5) who did not find BMI to have an influence on the amount of fatty infiltration in the paraspinal muscles. Moreover, Gibbons et al. (26) found that a greater BMI was a predictor of greater CSA, and Kalichman et al. (48) reported a negative correlation between muscle density on CT scan and BMI. Our results show that people with higher BMI are not prone to greater fatty infiltration of paraspinal muscles, possibly because stronger muscles are required to account for the larger body weight in obese persons when stabilizing the spine. These observations suggest that the higher levels of fatty infiltration of paraspinal muscles are likely not related to obesity but rather to the degenerative process of the muscle itself or to the spinal disorder affecting the patient.

The main strength of our study is the measurement of qualitative and quantitative parameters of paraspinal muscles at each level of the lumbar spine regardless of the level of stenosis. Despite the evidence that fatty infiltration is greater between L4 and S1 and given the anatomy and function of the erector spinae and multifidus muscles, we believe that these entire muscles should be considered as a single unit. Hence, we introduced global qualitative and quantitative parameters for assessing fatty atrophy of the entire paraspinal muscle unit which aGoutallier and aLeanCSA defined as the average of the respective parameters across all levels. The main limitation of our study is the small sample size, and thus we could not correct for multiple testing. However, while this study can be considered a proof-of-principle investigation, the strong correlations and clear differences observed in our study suggest that these results should hold true when investigated in larger samples. Moreover, we included only individuals affected by sLSS and scheduled for surgery. Similar analyses should be conducted in patients with less severe LSS or who were treated conservatively and in age-matched healthy controls to elucidate the role of muscle atrophy and fatty infiltration in the etiology of the disease and the effects of different treatments on these parameters. The MR images were either obtained at our clinic or transferred by external providers. While using the same make and model for all measurements may have provided even better results, we intended to conduct this study in a real-world environment. In other words, if the same make and model had been used for all patients, the generalizability of the data to other MRI systems would have been unknown/limited. In contrast, our data showed strong correlations despite the different make and models of the MRI systems used, suggesting that the agreement between the assessments is robust regarding the specific MRI system, making our result even more relevant in a clinical context. Finally, while the quantitative method employed here has shown good interobserver and excellent intraobserver reliability, the accuracy of the thresholding procedure has yet to be confirmed with MRI sequences based on chemical shift (Dixon) or spectroscopy. Nonetheless, the sequence used here is a standard clinical sequence, and extracting information on fatty infiltration in standard clinical sequences is relevant especially in cases where Dixon sequences are not available.

The correlation of the Goutallier classification with the quantitative assessment of fatty infiltration of paraspinal muscles suggests that clinicians should consider this classification as an efficient tool for evaluating paraspinal muscle fatty infiltration. Nonetheless, this qualitative measure does not consider muscle morphology, which may add insight into the role that paraspinal muscle status plays in the etiology of LSS. Because paraspinal muscle CSA correlates with body height, we suggest taking body height as a reference for normalization in future studies assessing paraspinal muscle atrophy and fatty infiltration.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethikkommission Nordwest- und Zentralschweiz (EKNZ). The patients/participants provided their written informed consent to participate in this study.

FM: conceptual study design, data collection and processing, and manuscript writing. CNü: conceptual study design, patient recruitment, data collection and processing, statistical analysis, and manuscript review. YZ: data compilation and manuscript review. FH: statistical analysis and manuscript review. SS: conceptual study design and manuscript review. AM: conceptual study design, data analysis, statistical analysis, and manuscript writing and review. CNe: conceptual study design, patient recruitment, and manuscript review. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.