Background

Stabilizing therapy through muscle training is one of the main physiotherapeutic interventions for low back pain [1,2,3]. A uniform theoretical background for this treatment is lacking [4], however, and more recent studies report contradictory results following this treatment [5,6,7]. There are no explanations for how stabilizing therapy could have such opposing effects on patients with low back pain. Given that the lumbar multifidus (LM) is regarded as the major stabilizing muscle of the spine [5, 8,9,10,11,12,13,14], the anatomy and topography of this muscle might offer at least some explanation for the opposing effects of stabilizing therapy.

The morphology of the LM is complex, and several anatomical descriptions have appeared in the literature [15,16,17,18,19]. Anatomical studies have concluded that the LM has the largest cross-sectional area (CSA) of paraspinal muscles with short levers located at the most medial part of the spine between approximately L4 and S1 [16, 20, 21]. Important factors in spinal stabilization include CSA, deformation or stress in ligaments, and muscle type, activity pattern, force, mass, and length [8,9,10, 19, 22]. According to other studies, however, LM muscle mass is too small to play a substantial stabilizing role, and the primary stabilizing role should be attributed to the erector spinae (ES) [4]. Furthermore, an ongoing debate concerns the topography of the LM to the ES (i.e., deep vs superficial) [19, 23,24,25]. This discussion has led to many different approaches to investigating the morphology and functional characteristics of the LM. For example, LM muscle morphology has been quantified through ultrasound imaging (USI), MRI scanning, CT scanning, surgery, biopsy, and cadaver research. The outcomes of these methods have led to varying conclusions about CSA, muscle thickness, percentage of fat infiltration, fiber-bundle angle, and fiber length [19, 26,27,28,29]. Although each method has its own strengths and limitations [30], the results also depend on other variables, including population, spine-level measurement, and methodological quality.

At present, there is no clear overview of the similarities and differences between anatomy atlases and LM topography studies with regard to LM topography in humans. Such an overview is essential to improving understanding concerning the theoretical background of stabilizing therapies in the treatment of low back pain, as well as with regard to its role in basic anatomy training. The present study is therefore intended to review the literature on LM morphology.

Materials and Methods

This study was conducted according to the guidelines formulated by Arksey and O’Malley 2005 and by Grudniewicz et al. 2016 [31, 32], using the following five-step framework: (1) identification of the research question, (2) identification of relevant literature, (3) study selection, (4) data extraction, and (5) collation, summary, and reporting of results. The identification of the research question (1) is explained in the Background section. Steps 2, 3, and 4 are explained in this Materials & Methods section, and Step 5 is discussed in the Results section.

(2) Identification of relevant literature

Search strategy

To identify relevant studies on LM morphology, two databases—PubMed (Medline) and EMBASE—were searched, as well as gray literature (anatomy atlases) until June 2019. Search strategies were built, consisting of a combination of database-specific MeSH terms, title/abstract, free text, “wild cards” (words truncated by using “*”), and Boolean operators (“AND”, “OR”). The search string is provided in Additional file 1. The snowball method was used to identify additional papers from the reference lists of studies that were included.

Eligibility

All of the studies included were reviewed in terms of population, method, and outcome. To be included, studies had to be published in English and be based on studies of adult humans or human cadavers. A supplemental search of the Dutch literature did not reveal any relevant studies. Letters to the editor, abstract-only articles, and review papers were excluded. The initial search identified an extensive number of studies and gray literature. To minimize the inclusion of low-quality studies, we limited inclusion to peer-reviewed studies. All of the studies included were screened for the methods used to measure LM morphology: USI, MRI, CT scanning, modeling (biomechanical model of muscles), and cadaver studies. Furthermore, the parameters by which LM morphology was defined were described for each study (i.e., images, photos, drawings, models, trajectory descriptions, thickness or CSA, spine levels, and location of the LM).

(3) Study selection

The selection procedure started with the identification of studies in the databases and the elimination of duplicates using the duplicate function in Endnote X9. Further, studies were screened according to title, abstract, and full-text, and additional papers were identified from reference lists of the included studies. Two authors (AH and RS) independently selected and assessed studies for quality and subsequently discussed them to reach consensus. When no consensus was achieved, a third reviewer was consulted (GJG).

Anatomy atlases were included as well, given their importance as basic anatomical introductions to LM topography. The anatomy atlases were selected through a university library system, followed by a snowball procedure, as they were not included in medical databases. Major anatomy atlases available in English, Dutch, and German were retrieved from the university libraries (including specialized medical libraries) of the University Medical Center Groningen and Saxion University of Applied Sciences, as well as from online resources. The results of anatomical studies and atlases are presented separately.

(4) Data extraction

All studies and atlases were extracted according to the following LM parameters: location (deep/superficial), imaging methods, spine levels, muscle trajectory (origin and insertion), muscle diameter (anteroposterior diameter), and CSA. The risk of bias assessment was not determined, as it primarily has to do with the methodology of studies [33]. Instead, a quality-assessment tool was developed to rate the quality of the descriptions of LM morphology. The tool consists of five items, each worth one point, with a maximum score of five (Table 1). The reliability of LM morphology descriptions was assessed by checking for the presence of an image and determining whether this image was an original photograph (as opposed to a model or drawing) [34]. Furthermore, the validity of the images was assessed by checking for the labelling of the LM, depiction of spinal levels, and description of planes. Descriptions scoring 5/5 were regarded as being of high quality, with scores of (3-4)/5 representing moderate quality and scores (≤ 2)/5 representing low quality. The inter-rater reliability (% agreement) of the two reviewers was calculated using a kappa value. In cases where the LM location was not described explicitly despite the presence of adequate imaging, the LM location was determined in consensus by the authors (AH, RS and GG).

Table 1 Quality assessment tool

Results

Study selection

The search yielded 2450 original studies, 299 of which were ultimately included, along with 4 additional studies. In addition, 19 anatomy atlases were identified that described parameters of LM morphology (Additional file 2) [25, 55,56,57,58,59,60,61,62,63,64,65,66,67]. The study-selection procedure is depicted in Fig. 1.

Fig. 1
figure 1

Flowchart of the selection process for the literature review

LM parameters in studies

The characteristics of the studies included are presented in Additional file 3 [17,18,19,20, 22, 26, 27, 29, 34,35,36,37, 39,40,41,42,43,44,45,46,47,48,49,50,51,52,53, 71,72,73,74,75,76,77,78,79,80, 82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255,256,257,258,259,260,261,262,263,264,265,266,267,268,269,270,271,272,273,274,275,276,277,278,279,280,281,282,283,284,285,286,287,288,289,290,291,292,293,294,295,296,297,298,299,300,301,302,303,304,305,306,307,308,309,310,311,312,313,314,315,316,317,318,319,320,321,322,323,324,325,326,327,328,329,330,331,332,333,334,335,336,337,338,339,340,341,342,343]. In descriptions of moderate to high quality, the most frequently applied methods for measuring or visualizing LM morphology were MRI (51%), USI (36%), and drawings (8%).

Location

In 153 of the 303 studies, LM was presented only as a superficial muscle at one or more levels between L4–S1. In 72 studies, it was presented only as a deep muscle and, in 35 studies, it was presented as both a superficial and a deep muscle. We were unable to identify the precise location of LM in 43 studies (Additional file 3).

Origin and insertion

The origin of the LM is described at the spinous process of L4 and L5 [35]. In some studies, however, LM origin was also described at the caudal and dorsal surface of each lamina (L1–L5) (Table 2) [23]. Whereas some studies described LM insertion as being at the lateral or medial side of the dorsal foramen of the sacrum [23, 35], others stated that the superficial LM muscle fibers are inserted at the posterior superior iliac spine (PSIS) [19, 23].

Table 2 Studies describing the fiber trajectory of LM

Muscle thickness and cross-sectional area (CSA)

We identified a variety of methods of measuring the CSA of LM. These methods include USI, CT scanning, and/or MRI at various levels of the lumbar spine (Additional file 3) between L1 and S1 (Fig. 2). This focus on L4 and L5 measurements was found in nearly all studies on different locations (Fig. 2).

Fig. 2
figure 2

Overview of spine levels at which LM is measured in all studies (Total), in studies of moderate to high quality referring to deep LM (Deep), in studies referring to superficial LM (Superficial), and in studies referring to deep and superficial LM (Deep & Superficial). Deep muscles lie closer to bone or internal organs, and superficial muscles are close to the surface of the skin

The CSA of LM has been measured in a variety of populations, resulting in an extensive range of LM CSA outcomes [27, 47, 48]. The total range in square millimeters varied between 9.08 and 2500 mm2, possibly due to the incorrect description of corresponding units of value. Variations in LM thickness were found with regard to the level of measurement (L3/L4, L4/L5, or L5/S1) and LM activation conditions (rest vs (sub)maximal voluntary contraction), as well as in terms of body position (e.g., prone vs standing position) (Additional file 4) [38, 47,48,49,50,51,52,53, 51,52,53, 74, 75, 98, 108, 113, 115, 127, 131, 135, 136, 144, 151, 159, 162, 163, 168, 175, 187, 203, 211,212,213, 233, 235, 236, 239, 253, 259, 263, 286, 301, 305, 308, 309, 311, 317,318,319,320,321,322, 325, 326, 335, 336, 340] [48,49,50,51,52,53]. The total range in LM thickness in millimeters varied between 2.4 and 41.1 mm [53, 54].

LM parameters in anatomy atlases

Within the anatomy atlases, we observed variations in the description and presentation of LM (Table 5), although the majority of atlases showed the same configuration of the LM. In 16 of the 19 atlases reviewed, the LM was depicted as a deep back muscle [24, 25, 55,56,57,58,59,60,61,62, 64,65,66,67,68,69,70,71], either covered by the thoracolumbar fascia and/or as being covered by the ES. Moreover, LM imaging varied in terms of the presence of spine levels (cervical-sacrum), imaging planes (transversal, dorsal, sagittal), and of whether it was with or without other low back muscles in a single figure.

Location and muscle diameter

Variations were found with regard to the location, diameter, and topography of the LM. In one anatomy atlas (Gray’s Anatomy) [61], the superficial part of the LM extended from T11 cranially to S3 caudally as a wide (large anterior-posterior diameter) muscle next to the median sacral crest. In a Radiology Anatomy Atlas Viewer [63], the LM was depicted in the axial spinal cross-sections, albeit with inconsistent labelling of the LM.

Variations were found in the diameter of the LM between the various lumbar levels. The location of the widest part of the LM varied between the level of PSIS [55, 58, 60, 61] and L5–S1 [25, 56, 57, 69]. In some atlases, however, the widest part of the LM was undefined, due to the overlying low back muscles [24, 67].

Various origins and insertions of the LM were identified in the anatomy atlases (Table 3), with the (lumbar) multifidus extending between the dorsal part of the sacrum [69] and the transverse processes of T1 [24, 25, 55,56,57,58, 60, 65, 67, 70, 71], and as attaching to the iliac [68] or ischium [55, 60] part of the pelvis.

Table 3 Atlases describing the fiber trajectory of lumbar multifidus

Overall, deep LM trajectories were consistently described between L1 and S5 [25, 56,57,58, 60, 61, 67, 70], although some superficial LM fibers were illustrated as originating from the spinous process of T10 [69] or T12 [61, 70]. Furthermore, some atlases did not illustrate the origin and insertion of the LM, as other muscles were more superficially presented and/or because these features were not described [24, 59, 62, 64, 66, 68].

Quality assessment

Quality scores were determined for each description and anatomical image of the LM in the literature. The per-item quality scores for descriptions and anatomical images are presented in Table 4. The agreement between the reviewers of the quality assessment had a kappa value of 0.67, and all differences were resolved in a consensus meeting. There were no major differences between the descriptions and anatomical images of the LM in the literature with regard to the presence of images. The most difficult item to score was clear labeling.

Table 4 Total scores for quality items, in numbers and percentages

The total quality scores of the studies varied between 1 and 5 (out of 5). The highest score [5/5] was found in 43% of the studies (129/303), with moderate scores [[2,3,4]/5] found in 39% of the studies (117/303) and low scores [(≤2)/5] found in 19% of the studies (57/303). In the studies with quality scores of 5/5 and 4/5, MRI and USI were the most commonly used methods for visualizing the LM muscle (Fig. 3). More detailed data are presented in Additional file 3. The majority of atlases were rated as being of moderate quality [(3–4)/5] (79%, 15/19) or low quality [(≤2)/5] (16%, 3/19). Only one atlas was found to be of high quality [5/5] (Table 5) [55].

Fig. 3
figure 3

Overview of the percentage of total scores on the quality-assessment tool (inside ring) and the associated percentage of techniques used (outside ring). Scores per study are presented in Additional file 3

Table 5 Data extraction of the atlases included (n=19), sorted by quality score

Discussion

Substantial contradictory results were found across a large number of anatomy studies included in the review, and there appears to be no general consensus concerning the trajectory and muscle description of the LM [19, 23, 25, 29, 35]. Particularly with regard to the descriptions of “fiber trajectory” and “location”, major differences were found between the studies by Macintosh and Bogduk (1986), Rosatelli et al. (2008), and Moseley et al. (2002), and those by Kim et al. (2015), Bojadsen et al. (2000), and De Foa et al. (1989) [17,18,19, 29, 36, 40]. Discrepancies were also identified with regard to LM diameter, especially the distance between the spinous process and the lateral margin of the LM at levels L4–S1 [46, 72] and its location relative to the ES [46, 72, 73].

Each method that is used in literature to measure LM characteristics has its own strengths and limitations [30]. The architecture and function of the LM has been studied predominantly according to morphological and imaging methods. One disadvantage of cadaver studies [15, 17, 22, 29, 36, 40, 46, 74] is that the studies do not clearly identify the type and amount of structures (skin, fat, fascia, and muscle) that were removed from the cadaver. In these studies, it could be difficult to describe the exact location of the LM relative to other lumbar muscles and structures. The MRI and USI methods offer the advantage of being able to present undisturbed anatomy. This finding could have positive implications for clinical practice, given that USI is a user-friendly and affordable way to measure LM morphology in physiotherapy practice.

In anatomy atlases, the LM was depicted primarily as a relatively small deep muscle, in contrast to some research studies that refer to its large size and the presence of superficial slips at L4–S2 levels. Differences in LM images were identified even within anatomy atlases [25, 60, 61, 68]. In Wolf-Heidegger’s Atlas of Human Anatomy, LM insertion is depicted at the ventral side of the sacrum, in contrast to Gray’s Anatomy, in which it is depicted at the dorsal side of the sacrum [61, 68]. Furthermore, in some atlases, the diameter and location of the LM is undefined, due to the overlying low back muscles [24, 67]. Overall, anatomy atlases reflect no consensus about the fiber trajectory of the LM, thus making it difficult for therapists, clinicians, and students to know and learn what is correct about LM morphology.

The inconsistencies in the descriptions and imaging of LM morphology in studies and atlases could be due to the differences in many parameters. Some of these variables are related to methodology (Fig. 2), spine levels, and/or type of population [26, 36, 75, 76]. Of particular note are the variations we found between LM images in anatomy atlases and those in studies. In some large studies [15, 26, 77, 78], the location and presentation of the LM differ from those in the most recent anatomy atlases [24, 25, 55]. One consistent finding in the anatomy atlases was that they all depict the LM as a deep lumbar muscle, whereas most studies presented it as a superficial lumbar muscle at the levels of L4–S1. We nevertheless identified some consistency in studies based on a USI LM protocol that had been developed in an earlier study (e.g., by Belavy et al.) [79,80,81]. The same protocol, which referred to similar images, has been used in different studies based on different research questions to present new knowledge about LM morphology.

Limitations

One possible limitation of this study could be that it might have overlooked some anatomy atlases, due to the lack of a database of anatomy atlases. Another limitation could be related to the reliability and validity of the quality-assessment tool that was developed and used by three authors of the current study. This quality-assessment tool was developed for lack of an existing “risk-of-bias assessment tool” with which to assess the quality of descriptions of LM morphology. It would be advisable to improve this assessment tool by conducting a validation and/or reliability study, as well as by expanding the tool beyond the current five items [34]. Any validation study regarding this quality-assessment tool will nevertheless be hampered by the current lack of a gold standard.

Variations in the images and measurement of LM morphology could be influenced by a number of potentially confounding factors, including research methods, level, side of measurement, population, intra-individual differences, intervention, research objectives, measurement technique (e.g., with or without contraction), and the relative experience of the assessors and/or practitioners creating the images. The variation that we observed in LM morphology emphasizes the importance of correct reference to morphology, although no gold standard for LM morphology has been developed to date. To reduce some of the existing variation, the authors call for improvement in the standardization of research protocols (e.g., in studies using EMG, USI, or MRI). The proper measurement of LM function could allow measurement of the contribution of the LM to spine movements in patients with non-specific LBP or other conditions. This knowledge could help clinicians and therapists to improve their diagnosis of patients.

Clinical implications

Remarkable differences in the reporting of LM morphology were found within anatomical studies, as well as between anatomical studies and anatomy atlases, especially with regard to trajectories of the musculature and its location relative to the ES. Such differences in the reporting of LM could have implications for clinical practice, given that knowledge of morphology provides the foundation for the diagnosis and treatment of patients by physiotherapists. For example, if the topography of a low back muscle (in terms of origin, insertion, deep/superficial) is clear, it should provide a clearer indication of the function of this low back muscle. This could make it easier to identify a cause or diagnosis of low back function or muscle impairment. For therapists or clinicians, these inconsistencies make it difficult to conclude which results are correct. Once therapists or clinicians know the correct LM morphology, this will clarify the function of the LM. Such knowledge could enhance understanding concerning the role of the LM in patients with LBP. It could also enhance the quality and consistency of decision-making by specialists concerning treatments for patients with LBP. Although a recent review on the effects of stabilizing therapy compared to usual care identified significant benefits of stabilizing therapy on pain and disability, these differences were not interpreted as clinically important [82]. Improved diagnosis may allow better sub-grouping, possibly enhancing the therapeutic effects for patients with LBP.

Conclusion

We identified a lack of standardization in the depiction and description of LM morphology, which may affect the precise understanding of its role in the background and therapy for patients with LBP. Standardization of research methodology with regard to LM morphology is recommended. Anatomy atlases should be updated on LM morphology.