Characteristics of paraspinal muscle degeneration in patients with adult degenerative scoliosis

Adult degenerative scoliosis (ADS) is a 3D deformity that greatly affects the quality of life of patients and is closely related to the quality of paraspinal muscles (PSMs), but the specific degenerative characteristics have not been described. This study included ADS patients who were first diagnosed in our hospital from 2018 to 2022. Muscle volume (MV) and fat infiltration (FI) of PSM were measured by 3D reconstruction, and spinal parameters were assessed by X-ray. The values of convex side (CV) and concave side (CC) were compared. Fifty patients were enrolled with a mean age of 64.1 ± 5.8 years old. There were significant differences in MV, FI, and Cobb angle between male and female groups. The MV of MF and PS on the CC was significantly larger than that on the CV. In the apex and the segments above the apex, the FI of the MF on the CC is greater than the CV, and in the CV of the segment below the apex, the FI of the MF is greater than the CC. Besides, there was a significant positive correlation between the FI and Cobb angle in the MF of the CC-CV. There were significant differences in the MV and FI of PSM on both sides of the spine in ADS patients. It was determined that the PSM of ADS showed different degrees of degeneration in different levels of the lumbar spine and were positively correlated with Cobb angle.


Introduction
Adult degenerative scoliosis (ADS) is a condition in which the spine curves to one side as a result of age-related degeneration of the intervertebral discs, intervertebral joints, paraspinal muscle (PSM), and other soft tissues, primarily in the lumbar region [1].ADS frequently has a significant negative impact on the patient's quality of life and health.The psoas (PS), multifidus (MF), and erector spinae (ES) are among the PSM, a group of muscles that keep the spine stable.Recent research has demonstrated that significant degeneration and atrophy of the PSM are closely associated with ADS, although the pathogenesis of the condition is not entirely known [2][3][4].Muscle atrophy is characterized by a loss of muscle quality and an increased fat infiltration (FI), typically protein breakdown levels increased intermuscular adipose tissue (IMAT), including the number of fat cells stored between muscle fibers and muscle groups [5].IMAT is frequently displayed as low HU in CT and high signal in MRI to distinguish it from muscle tissue.
Despite the existing difficulties and dearth of modalities to measure muscle quality, some studies have shown a correlation between ADS and imaging characteristics of the PSM, such as cross-sectional area (CSA), size, shape, and density.These imaging characteristics, however, are often quantified in two-dimensional axial pictures and have several drawbacks.1. ADS is a holistically degenerative, threedimensional rotational deformity with a complex process; individual vertebral bodies or two-dimensional scans do not accurately depict the alterations in the spine as a whole.2. A variety of factors can affect the strength or rigidity of an individual muscle such as the number of muscles present and fiber-type characteristics.These characteristics are often 1 3 not clearly described by 2D sections, thus leading different researchers to different conclusions.When compared to CSA, muscle volume (MV) and the total degree of FI are better reflections of muscles [6].
In this work, we measured the MV and FI of the PSM in ADS patients to examine the characteristics of the alterations in the PSM of ADS.

Patient selection
The patients with ADS who entered our hospital between 2018 and 2022 were included in this retrospective cohort analysis.Inclusion criteria: age between 50 and 75 years old at the time of presentation, boasting complete anteroposterior, lateral spine radiographs, and CT imaging of the lumbar spine.The Cobb angle (CA) of the curved coronal plane of the lumbar spine in the anteroposterior and posterior position is greater than 10°.Exclusion criteria included a history of scoliosis in children or adolescents, a history of spinal surgery, a local infection, peri-spinal inflammation, a history of severe spinal trauma, a history of spinal tumors, a history of severe PSM injury, giant osteophytes in the zygapophysis, and the presence of other systemic diseases that may affect muscle quality (sarcopenia, ankylosing spondylitis, muscular dystrophy, Parkinson's, etc.).
This study was conducted following guidelines approved by the Ethics Committee and the Declaration of Helsinki.

Measurement of muscle volume
Axial, coronal, and sagittal reconstructions were acquired from the upper endplate edge of T12 to the lower end plate edge of S2 using an Optima CT660 multilayer CT scanner.The CT scan parameters were 120 kV, 250-750 mAS, collimation 0.625 mm, and layer thickness 0.625 mm.According to our institutional review board's guidelines, the patients' original clinical and radiological pictures were gathered.
The DICOM files were imported into the imaging histology software 3D Slicer (available from http:// mipav.cit.nih.gov/ free download), and the region of interest (ROI) was manually outlined with chemical shift coding plugin [7], morphological contour interpolator, and logical operators to 3D reconstruct the target muscle group (Fig. 1), and calculate the MV and effective muscle volume (EMV).Muscle tissue pixels were taken in the range -29 to 150 HU (Fig. 2) [8,9], such software can demonstrate the included PSM in a fairly accurate and rapid analysis [10][11][12].
Calculate the overall muscle FI:

Measurement of spinal parameters
Standing full spine X-rays were used to evaluate the patient's spine parameters.Patients were also reviewed with a postoperative full spine X-ray to measure X-ray parameters on radiographs, including coronal and sagittal parameters (Fig. 3) [13].All radiological parameters were measured three times.Two spine surgeons with extensive experience performed the measurements, none of whom were aware of the patient's condition before measurement.The final measurements were averaged after each sample was measured three times.

Statistical analysis
All results were examined for normality using the Shapiro-Wilk test.PSM of the concave side (CC) and the convex side (CV) were compared using the paired T-test.Comparison between the two groups using the student T-test.Correlations between muscle measurements and other parameters were analyzed using the Pearson correlation test.Statistical significance was set at P < 0.05, and all statistics were performed using SPSS software (version 27.0.1;SPSS, Chicago, IL).

Patient statistics
This study included 50 patients with ADS (21 men and 29 women) with a mean age of 64.1 ± 5.8 years and a mean BMI of 25.6 ± 3.5 (Table 1).The scoliosis main curve usually has an apex at L3 (L3:27, L4:10, L2:13).Among them, there were 14 cases of CC on the left side and 36 cases on the right side (Table 2).LL: 25.4 ± 12.9°, SVA: 6.7 ± 4.1, CBD: 2.9 ± 1.9; these parameters were not significantly correlated with patient sex and CC position.CA (18.4 ± 6.1), MV and FI were significantly different between men and women, but not related to the position of CC (Fig. 4). 3  and 4)

Comparison of MV/FI on the CC and CV (Tables
The MV and FI of PSM from Apex-2 to Apex + 2 levels are shown in Tables 3 and 4. Comparing CC and CV, the FI of MF: there was no significant difference in total (P = 0.09) and showed a significant difference in each stage (P < 0.001).The FI of ES had different in Apex-1 (P < 0.001) and Apex + 2 (P = 0.025); there was no significant difference in total and the rest of the stages.The FI of MW was significantly different in Apex-1 (P < 0.001), Apex (P = 0.001), and Apex + 2 (P = 0.025), and there was no significant difference in total and the rest of the stages.The FI of PS: the FI of total and Apex + 2 was not significantly different, and the rest of the stages were different (Apex-2: P = 0.002, Apex-1: P < 0.001, Apex: P < 0.001, Apex + 1: P = 0.001).The MV of MF: Apex + 1 showed no difference, the total and other segments had different (total: P = 0.02, Apex-2: P = 0.016, Apex-1: P < 0.001, Apex: P < 0.001, Apex + 2: P = 0.02).The MV of ES: Apex (P = 0.011) and Apex + 2 (P = 0.015) had difference, the total and other segments had no difference.MV of MW had difference in Apex-1 (P = 0.001), Apex (P < 0.001), Apex + 2 (P < 0.001), the total and other segments had no significant difference.The MV of PS: Apex-2 had no difference, the total and other segments had different (total: P = 0.02; Apex-1: P < 0.001, Apex: P < 0.001, Apex + 1: P = 0.001, Apex + 2: P = 0.003).5 and 6)

Discussion
Our research showed that the PSM on the CC and CV of the spine in patients with ADS were significantly different,  as shown by varying degrees of MV and FI change, and that the degeneration of PSM was closely linked to ADS.ADS is a complex, three-dimensional degenerative condition marked by variable degrees of asymmetric degeneration of the soft tissues around the lumbar vertebral body.
According to the vicious circle hypothesis put forth by AEBI [14], PSM hypofunction will affect the initial biomechanical relationships, causing overloading of the PSM and increased or uneven loading of numerous structures in the passive subsystem, which in turn causes low back pain and disc degeneration.Conversely, the original biomechanical linkages will change as a result of changes in the passive subsystem, leading to imbalances in muscle loads and progressive imbalances.
We measured the MV and FI of the PSM at all levels of the lumbar spine and after comparing them, we found that the PSM of ADS patients showed different degrees of degeneration at different levels.Unlike previous studies, our statistical results showed that the volume of MF on the CC was larger than that on the CV.Due to the different lengths of the two sides of the spine caused by the curvature of the spine, the muscles on the CV may be elongated and thinner, while the muscles on the CC will shorten and become thicker and thicker [3].And the larger the CA, the higher the degree of degeneration of the MF and ES.Among MF, ES, and PS, the degeneration of the MF was more significant and had certain regularity.Unlike previous studies, our statistical results showed that in the apex and the segments above the apex, the FI of the MF on the CC is greater than the CV.In the segment below the apex, the IF of the MF on the CV is greater than the CC, which we believed was the result of the body maintaining its balance (Fig. 5).Sun suggested that significant asymmetric atrophy of the MF may be the main cause of ADS and may also be used to predict the progression of ADS [15].In the setting of ADS, the level below the apex was close to the lumbosacral junction, and the compensatory curve below the main curve was called the fractional curve, which is opposite to the main curve, and may be related to the changing rules of MF [16].
The asymmetric degeneration of PSM on both sides of the spine in patients with ADS was caused by many factors, not only unilateral radiculopathy due to deformity but also related to the deformity of the spine itself.Consistent with previous studies, the rate of FI in the PSM was positively correlated with CA, but not with CBD, SVA, and LL.A previous study showed that SVA and LL were affected by pelvic reclination and gluteal medial muscle quality [17].The FI of PSM in patients with coronary imbalance was higher than that in the noncoronary imbalance group, and the degree of PSM fat infiltration  on the CC was positively correlated with CBD.We did not measure the corresponding results.It may be that this study did not group and compare coronal disbalance typing.According to the characteristics of PSM degeneration, it is meaningful to provide different rehabilitation guidance and formulate suitable surgical treatment plans for ADS patients, and muscle quality has an impact on surgical outcomes, complications, and disease recurrence rates after spinal surgery, there is a close correlation.For example, Hyun et al. found that increased FI of the ES greatly increases the incidence of postoperative junctional kyphosis [18].
In patients with ADS as compared to those with LSS, Yagy's analysis of the CSA of PSM at the L5-S1 level revealed that the CSA of MF and PS was considerably lower in the ADS group [2].Kim compared the CSA difference indexes of the PS and MF at the level of apex in ADS patients.The CSA of PSM in CV was significantly greater than the CC, and the FI of each muscle did not significantly differ from one another [3], however, Najibullaha found that the FI of paravertebral musculature in CV was significantly greater than the CC [19].As can be seen, CSA and FI are frequently used to evaluate the quality of muscles, especially for muscles like the MF and ES with a specific shape and position.However, this might not be trustworthy enough and could be affected by several variables, such as measuring edges, levels, etc., which could result in inaccurate results.To better understand the onset and progression of PSM degeneration, new methods are needed, including precise techniques for measuring MV, so 3D measurement of the PSM is a key step in analyzing muscle condition [20].Researchers can manually cut the region of interest for each muscle layer throughout, but this method lacks precision and requires a significant investment of time and effort, which is undesirable, especially in studies with multiple sample data.Wei et al. proposed a method for segmenting PSM from 2D axial CT images based on the 2D mapping [21].Kumei et al. worked on 3D scans based on deep convolutional neural networks (CNNs) to automatically segment the PSM at the level of the twelfth thoracic vertebra in CT images [22].Fully automated segmentation techniques can easily cut out the PSM with a high degree of accuracy [20].However, due to the close anatomical relationship between the MF and ES, achieving accurate segmentation of the MF and ES by automated segmentation techniques has some difficulties.For volume measurements and longitudinal investigations of individual muscles semiautomated, single-atlas-based methods must be preferred [23].In applications where fully automated segmentation algorithms are not available or cannot be performed with the desired accuracy, semiautomated segmentation can be a time-saving alternative to manual segmentation, allowing human experts to guide segmentation while minimizing expert effort spent on repetitive tasks that can be automated.

Limitations and prospects
There are some limitations in this study, including: 1. a relatively small sample size, which may explain the lack of significant difference or correlation on some results.The CT-based 3D reconstruction of paraspinal muscles is highly specialized and still requires a significant amount of time.In order to ensure the accuracy of the measurements, researchers need to examine and optimize ROI layer by layer.However, it is crucial in the analysis of paraspinal muscle quality in patients with ADS.In the future, our work will increase the sample size and combine paraspinal muscle volume and FI with electromyography to better assess paraspinal muscle function and degradation.2. This study cohort is a cross-sectional study that does not reflect the development of muscle degeneration in the PSM of ADS and cannot determine the causal relationship between passive and active sublineage degeneration.This requires us to focus on and collect patients at different stages of disease progression.The difficulty is that most patients usually choose to seek medical attention when the disease is more severe, so the prevention of scoliosis in the middle-aged and elderly needs equal attention.3. The method of measuring MV still requires a more accurate automatic segmentation model needs to be trained.4. HU used in this experiment is -29-150; however, different muscle groups have different environments, so the optimal threshold is different for each group, and using the optimal threshold can greatly increase the accuracy of the measurement, but no data were found to support this. 5. Spinal degenerative diseases are closely related to the strength of the core muscle group.Due to technical limitations, this experiment only measured the MF, ES, and PS and did not observe other core muscle groups such as quadratus lumbar, rectus abdominis, and transverse abdominis.We believe that these muscles also play a key role in the development of ADS.
The future direction of our research is to explore the correlation between preoperative paraspinal muscle quality and postoperative functional outcomes in ADS patients.Additionally, further research is needed to investigate the effect of professional rehabilitation therapy on paraspinal muscle quality in ADS patients.

Conclusion
In general, in this experiment, there were significant differences in the MV and FI of PSM on both sides of the spine in ADS patients.It was determined that the PSM of ADS showed different degrees of degeneration in different levels of the lumbar spine and were positively correlated with CA.

Fig. 2 Fig. 3
Fig. 2 Use 3D Slicer to outline the muscle ROI (A), select the muscle gray release value between -29HU and + 150HU (B), and calculate the muscle volume and the effective muscle volume of the target mus-

Fig. 4
Fig. 4 Subgroup analysis of paravertebral muscle volume and fat infiltration.A, B: Comparison of various muscle volumes and fat infiltration in male and female groups.C, D: Comparison of various muscle volumes and fat infiltration in groups with CC position on the MF multifidus, ES erector spinae, PS psoas, WM whole muscles, CA Cobb's angle, SVA sagittal vertical axis, CBD coronal balance distance, LL lumbar lordosis, R Pearson correlation coefficient.*.At the 0.05 level, the correlation was significant, **.At the 0.01 level, the correlation was significant 1 3

Table 1
Patient demographics SVA sagittal vertical axis, CBD coronal balance distance, LL lumbar lordosis

Table 2
Subgroup analysis of spinal parameters CV: P = 0.017) on the CC-CV were positively correlated with CA; the MV of MF (CC: P = 0.035; CV: P = 0.029) on the CC-CV and PS (CC: P = 0.021) on the CC was positively correlated with LL.

Table 3
Comparison on CC or CV fat infiltration at the upper or lower intervertebral level of the apical vertebra MF multifidus, ES erector spinae, PS psoas, WM whole muscles, FI fat infiltration, MV muscle volume

Table 4
Comparison on CC or CV muscle volume(cm 3 ) at the upper or lower intervertebral level of the apical vertebra MF multifidus, ES erector spinae, PS psoas, WM whole muscles MF multifidus, ES erector spinae, PS psoas, WM whole muscles, CA Cobb's angle, SVA sagittal vertical axis, CBD coronal balance distance, LL lumbar lordosis, R Pearson correlation coefficient.*.At the

Table 6
Correlation analysis between muscle volume of paraspinal muscle and spinal parameters