Analysis of the association between lumbar paraspinal muscle atrophy, facet joint degeneration, and degenerative lumbar scoliosis

Abstract

Objective

To analyze the correlation between paraspinal muscle atrophy, facet joint degeneration, and degenerative scoliosis (DS).

Methods

A retrospective study included 231 chronic low back pain patients from Zhongda Hospital Affiliated to Southeast University (January 2023-January 2024). Radiographic diagnosis assigned 150 patients to DS group (subclassified into mild [n = 72], moderate [n = 56], severe [n = 22]) and 81 to non-DS control group. Using T2-weighted MRI at L3-S1 levels, ImageJ software measured multifidus (MF) and erector spinae (ES) cross-sectional area (CSA) and functional muscle ratio (LCSA/GCSA). Surgimap software quantified facet joint angle (FJA), facet overhang (FO) length, and facet joint space width (FJSW). Logistic regression analyzed risk factors with ROC curves determining diagnostic thresholds.

Results

The non-DS group demonstrated a significantly higher proportion of males (P = 0.023) and greater bone mineral density (P = 0.043) compared to the DS group. Regarding paraspinal muscle parameters, the non-DS group exhibited significantly larger MF CSA, MF + ES CSA, and LCSA/GCSA at the L3/4, L4/5, and L5/S1 levels, as well as a larger ES CSA at the L3/4 level (all P < 0.05). Conversely, the ES CSA at the L5/S1 level was significantly smaller in the non-DS group. For facet joint parameters, the non-DS group showed significantly smaller FJA, FO Length at the L3/4, L4/5, and L5/S1 levels, and smaller FJSW at the L3/4 and L4/5 levels (all P < 0.05). Within the DS group, significant differences were observed between the convex and concave sides at all L3-S1 levels for LCSA/GCSA, MF CSA, ES CSA, FJA, FO Length, and FJSW (all P < 0.05). With increasing severity of DS, there was a progressive decrease in LCSA/GCSA, MF CSA, and ES CSA, and a progressive increase in FJA and FO Length across the L3-S1 levels (all P < 0.01). Post-hoc analysis revealed significant differences in the majority of muscle parameters between severe DS and mild/moderate DS (P < 0.05). Correlation analysis indicated that, except for FJSW at L5-S1 (P = 0.526), the Cobb angle was negatively correlated with MF CSA, ES CSA, LCSA/GCSA, and FJSW, and positively correlated with FJA and FO Length (all P < 0.001). In both the DS and non-DS groups, most LCSA/GCSA and other CSA measurements demonstrated no significant correlations with FJA, FO length, and FJSW. Among the few statistically significant correlations observed, all were weak (rho < 0.30). Multivariate logistic regression analysis identified the following risk-associated factors for DS: lower BMD (OR = 0.802, P = 0.032), reduced LCSA/GCSA (OR = 0.005, P = 0.003), smaller MF CSA (OR = 0.969, P = 0.027), smaller ES CSA (OR = 0.973, P = 0.014), larger FJA (OR = 1.075, P = 0.016), and greater FO length (OR = 1.067, P = 0.001). ROC analysis yielded AUCs/cut-offs: BMD (0.581/− 0.900 T-score), LCSA/GCSA (0.712/0.805), MF CSA (0.608/635 mm²), ES CSA (0.463/832 mm²), FJA (0.627/57°), FO length (0.651/6.550 mm).

Conclusion

DS patients demonstrate progressive paraspinal muscle atrophy, sagittal-oriented facet joints, and advanced facet degeneration correlating with scoliosis severity. Diagnostic thresholds indicating DS probability are BMD < − 0.900 T-score, LCSA/GCSA < 0.805, MF CSA < 635 mm², ES CSA < 832 mm², FJA > 57°, and FO length > 6.550 mm.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03609-w.

Introduction

Degenerative scoliosis (DS) is a prevalent health concern among the elderly population, attracting increasing attention in the context of global aging. As a degenerative condition, the incidence of DS rises with age. In China, the prevalence of DS is as high as 13.3% [1], and it is even more common in individuals over 65 years old, affecting 32% to 68% of this demographic [2–4]. Functional and structural alterations in the paraspinal muscles are frequently observed in DS patients. Studies by Wang et al. indicated that paraspinal muscle atrophy is a potential risk factor for DS [5, 6]. Concurrently, paraspinal muscle atrophy can lead to facet joint hypertrophy and osteophyte formation, subsequently contributing to facet joint degeneration. Therefore, paraspinal muscle atrophy and facet joint degeneration often coexist.

Within the paraspinal musculature, the erector spinae (ES), located superficially, plays a role in dynamic stabilization during spinal motion. The multifidus muscle (MF), situated deep within the paraspinal muscles, is primarily responsible for spinal stability and maintaining lumbar lordosis. Both muscles act as crucial regulators of postural balance and spinal stability [7, 8]. Through compensatory mechanisms, they counteract spinal imbalance during degenerative processes, often operating under high load for extended periods, leading to pain, fatigue, atrophy, and degeneration. Crawford RJ et al. noted that pathological changes in the paraspinal muscles most commonly occur at the L4–L5 level [9], likely related to spinal biomechanics and load distribution. Furthermore, studies suggest a close association between the severity of scoliosis and the progression of lumbar facet arthropathy (LFA), particularly at the L3–L4, L4–L5, and L5–S1 levels [10]. Based on these findings, this study selected the L3–S1 segments for measurement and evaluation of various parameters related to the lumbar paraspinal muscles and facet joints.

Notably, the functional status of the paraspinal muscles is closely interrelated with the morphology and loading of the facet joints, which together constitute a coordinated “musculoskeletal synergy” essential for spinal stability. Building upon this foundation,

the lumbar facet joints, being the only synovial joints in the spine, play a vital role in load-sharing and maintaining spinal stability [11]. With aging, these joints undergo degeneration, manifested as facet joint hypertrophy, osteophyte formation at the joint margins, and joint space narrowing [12]. Previous research has frequently observed osteophyte formation at vertebral margins and intervertebral disc space narrowing on imaging studies of DS patients [13]. Facet joint orientation (FJO) refers to the angle of the facet joint in the transverse plane relative to the sagittal plane. Dai LY et al. found that a more sagittally oriented FJO increases the risk of developing DS [14].

The paraspinal muscles and facet joints function as a core biomechanical stabilizing unit of the spinal motion segment, exhibiting tight structural and functional coupling. This interdependent relationship demonstrates significant synergy even during the degenerative process. Guven et al. [15] explicitly highlighted a close association between multifidus atrophy and facet joint osteoarthritis, a link potentially mediated by shared mechanisms such as dysfunction of the segmental neural supply (medial branch of the spinal nerve) and aberrant local biomechanical loading on the facets resulting from diminished muscular stability. This finding suggests that rehabilitative training and targeted strengthening of the multifidus may help delay or interrupt the vicious cycle of “muscle atrophy → segmental instability → joint degeneration.” Further supporting this concept, the longitudinal study by Schönnagel et al. [16] provided evidence over a three-year period for the existence of a coordinated “musculoskeletal co-degeneration” mechanism in the spine. Their work pathologically reinforces the theoretical model in which muscular degeneration acts as a driving factor, occurring prior to or concurrently with the degeneration of bony structures.

Although existing studies have preliminarily revealed the coexistence of paraspinal muscle atrophy and facet joint degeneration during the progression of DS, the specific pathways of their interaction and their synergistic mechanisms in DS development remain unclear. Furthermore, differences in paraspinal muscle degeneration and lumbar facet joint degeneration between non-DS and DS patients, as well as the relationship between these two degenerative processes, are still not well defined. The correlation between paraspinal muscle atrophy, facet joint degeneration, and degenerative scoliosis has been scarcely investigated. Most current literature only analyzes the role of individual factors, lacking exploration of the association between paraspinal muscle dysfunction and facet joint degeneration, as well as systematic research on how they collectively exacerbate spinal imbalance. This study aims to bridge this critical gap by integrating various imaging parameters to systematically analyze the association between morphological and functional changes in the paraspinal muscles and the severity of facet joint degeneration, ultimately constructing a model for their role in DS progression. Moreover, moving beyond single-factor analysis, this study adopts an integrated perspective of spinal “musculoskeletal co-degeneration” to elucidate the core pathological mechanisms of DS. The findings are expected to provide new therapeutic targets for early identification and intervention of DS, establish a theoretical basis for developing targeted combined muscle-bone rehabilitation strategies, and hold significant clinical value for preventing DS occurrence, delaying its progression, and improving the quality of life in the elderly population.

Therefore, based on the overarching hypothesis that “paraspinal muscle degeneration and facet joint degeneration are interrelated and collectively contribute to the onset and progression of DS,” this study utilizes CT and MRI imaging data to validate the following specific research aims:

Central Hypothesis 1: The severity of DS correlates with the degree of degeneration in both the paraspinal muscles and facet joints.

Aim 1: To compare differences in degenerative parameters between non-DS and DS patients, confirming their characteristics as DS phenotypic features.

Aim 2: To analyze the correlation between degenerative parameters and DS severity within the DS cohort.

Central Hypothesis 2: Paraspinal muscle degeneration and facet joint degeneration exhibit biomechanical synergy in DS.

Aim 3: To reveal the association between asymmetry in degeneration and biomechanical imbalance by analyzing differences in degenerative parameters between the convex and concave sides of the curve in DS patients.

Aim 4: To investigate the intrinsic relationship between paraspinal muscle and facet joint degeneration through correlation analysis, and explore their combined diagnostic and predictive value for DS.

Materials and methods

Inclusion and exclusion criteria

Inclusion Criteria: (1) Diagnosis of Degenerative Scoliosis (DS) confirmed by radiography or CT (Cobb angle ≥ 10°) [17], with imaging findings consistent with clinical symptoms and signs; (2) Age between 45 and 80 years, regardless of gender; (3) No significant improvement in clinical symptoms after 3 months of conservative treatment; (4) Provision of written informed consent by the patient.

Exclusion Criteria: (1) Dysplastic, pathological, or traumatic scoliosis; (2) History of previous lumbar spine surgery; (3) History of primary or metastatic lumbar tumors, lumbar fractures, ankylosing spondylitis, or rheumatoid arthritis; (4) Inconsistency between imaging findings and clinical symptoms/signs; (5) Unavailability of the essential imaging studies for this research (lumbar MRI, CT, or full-length spine radiographs), thus preventing the measurement and evaluation of the pertinent parameters.

Study participants and grouping

This single-center, retrospective observational study analyzed patients who were hospitalized in the Department of Spinal Surgery at Zhongda Hospital, Southeast University, between January 2023 and January 2024 for symptoms such as chronic low back pain, leg pain, and neurogenic claudication, which could be accompanied by imaging-confirmed structural changes including reduced lumbar lordosis, vertebral tilting, and lateral listhesis [18]. Subsequently, through a review of medical records and imaging reports, patients were further screened according to the predetermined inclusion and exclusion criteria. Following this process, a final cohort of 231 patients was included for analysis.Based on full-spine standing AP radiographs(measuring the Cobb angle), patients were classified as having DS or not non-DS. This yielded 150 patients in the DS group and 81 patients in the non-DS group. The non-DS patients served as the control group. Patients in the DS group (observation group) were further stratified into three subgroups based on curve severity: mild scoliosis (n = 72, Cobb angle 10°–20°), moderate scoliosis (n = 56, Cobb angle 20°–40°), and severe scoliosis (n = 22, Cobb angle > 40°) [18]. Demographic data, including gender, age, body mass index (BMI), and curve direction, were recorded for all groups. All methods were performed in accordance with the relevant guidelines and regulations stipulated by the Clinical Research Ethics Committee of the Zhongda Hospital, Southeast University and the Declaration of Helsinki.This study was approved by the Clinical Research Ethics Committee of Zhongda Hospital Affiliated to Southeast University (Approval No.: 2022ZDSYLL406-P01). The study flowchart is detailed in Fig. 1.

Fig. 1 .

Flowchart illustrating the patient selection process and group allocation. A total of 607 patients hospitalized with chronic low back pain were assessed for eligibility. After applying the inclusion and exclusion criteria, 376 patients were excluded, leaving 231 patients enrolled in this retrospective study. Based on full-spine X-ray measurements, patients were classified into a Control group (Non-degenerative Scoliosis, Non-DS) with a Cobb angle < 10° (n = 81) and an Observation group (DS) with a Cobb angle ≥ 10° (n = 150). The DS group was further subdivided into mild (10°–20°), moderate (20°–40°), and severe (> 40°) scoliosis subgroups. Demographic and clinical data were collected and analyzed for all groups

Baseline characteristics of all patients, including age, gender, and body mass index (BMI), were collected from their initial medical visit records. All imaging data, such as magnetic resonance imaging (MRI), were obtained from routine scans performed for diagnostic clarification prior to any therapeutic intervention, with the acquisition dates falling within the designated study enrollment period.

As a retrospective study, the sample size was determined by all available patient data from our institution that met the predefined inclusion and exclusion criteria during the study period. To ensure the adequacy of the sample size, a post-hoc power analysis was conducted using G*Power software (version 3.1.9.7). Taking the primary group comparison in this study as an example (Non-DS group, n = 81; Mild DS group, n = 72; Moderate DS group, n = 56; Severe DS group, n = 22), an F-test (ANOVA) was selected as the statistical model. The parameters were set as follows: significance level α = 0.05, effect size f = 0.25 (representing a medium effect), number of groups = 4, and total sample size N = 231. The analysis indicated that the achieved statistical power (1-β) for this study was 0.99, which exceeds the conventional threshold of 0.80.

Radiographic measurement parameters

Assessment of curve severity The Cobb angle was used to evaluate the severity of scoliosis [19]. A Cobb angle ≥ 10° defined the presence of DS, while an angle < 10° indicated no DS. Curve severity was categorized as mild (10°–20°), moderate (20°–40°), or severe (> 40°) [16], with higher Cobb angles indicating greater severity. All patients underwent standardized standing full-spine AP and lateral radiographs. The Cobb angle was measured on the AP radiograph by identifying the most tilted vertebrae at the cranial and caudal ends of the major curve. Lines were drawn parallel to the superior endplate of the superior end vertebra and the inferior endplate of the inferior end vertebra; the angle formed by the intersection of these lines (or their perpendiculars) constituted the Cobb angle (See Figs. 2 and 3).

Fig. 2.

Anteroposterior radiograph of one patient with severe DS, demonstrating a Cobb angle of 53°

Fig. 3.

Full-spine standing AP radiographs of DS patients representing different curve severities. A Mild DS (Cobb angle 27°); B Moderate DS (Cobb angle 33°); C Severe DS (Cobb angle 46°)

Assessment of paraspinal muscle atrophy: All enrolled patients (both DS and non-DS groups) underwent conventional lumbar spine MRI scans using a Siemens 3.0 T Vida scanner (Siemens Healthcare GmbH, Erlangen, Germany; Sequence type: T2-weighted images; Scanning levels: At the L3/4, L4/5, and L5/S1 intervertebral disc levels; Slice thickness: 4 mm; Field of view: 200 mm; Pixel matrix: 512*512; Voxel size: 0.65*0.49*4 mm; Contrast agent: Not used). T2-weighted axial images were acquired at the mid-disc level of the L3/4, L4/5, and L5/S1 segments and retrieved using the institutional Picture Archiving and Communication System (PACS). As illustrated in Figs. 4 and 5, the contours of the multifidus (MF) and erector spinae (ES) muscles were manually traced on the axial images, a method validated for good intra- and inter-observer reliability [20, 21]. Using ImageJ software (National Institutes of Health, Bethesda, MD, USA), the gross cross-sectional area (GCSA), encompassing the traced MF and ES muscles, and the lean cross-sectional area (LCSA), representing the fat-free muscle area after subtracting non-muscle tissue (fat, ligaments, bone) within the contours, were measured bilaterally at each level. The sum of the bilateral values for total MF + ES CSA and LCSA was recorded per segment. The degree of muscle atrophy was quantified using the functional cross-sectional area ratio (LCSA/GCSA), calculated as LCSA divided by the corresponding GCSA for each muscle group [22, 23], where a higher ratio indicates less atrophy (better muscle quality) and a lower ratio indicates more severe atrophy. Thus, total CSA reflects muscle quantity, while LCSA/GCSA reflects muscle quality. Three senior spine surgeons received our standardized training and mastered the knowledge of controlling grayscale values, adjusting thresholds, and identifying tissue edges. Additionally, they all had 5 to 10 years of working experience, were blinded to the research objectives, and remained uninformed throughout the study to minimize potential bias. The mean values of the measurements from the three surgeons were taken as the final results.

Fig. 4 .

T2-weighted axial MRI images of one patient with moderate DS. A L3–L4 level. B L4–L5 level. C L5–S1 level. The multifidus muscle region is delineated by the red line; the erector spinae muscle region is delineated by the yellow line

Fig. 5.

Measurement of LCSA/GCSA on axial T2-weighted MRI using ImageJ software. A Total MF and ES cross-sectional area (GCSA) outlined in yellow at L4/5 level in a 65-year-old Female; B Calculation of MF and ES cross-sectional area (LCSA) after threshold adjustment

Assessment of lumbar facet joint degeneration parameters Measurements were performed using Surgimap software (version 2.3.2.1, Nemaris Inc., USA) on axial T2-weighted MRI (T2WI) slices parallel to the superior endplates of L4, L5, and S1 vertebral bodies, corresponding to the L3/4, L4/5, and L5/S1 disc levels. Facet joint angle (FJA) was defined as the angle between the line connecting the anteromedial and posterolateral points of the facet joint surface and the horizontal line (Fig. 5). Facet overhang length (FO Length) was defined as the distance of the osteophyte projecting from the base of the lamina to the posterior margin of the mid-portion of the facet joint (Fig. 6). Facet joint space width (FJSW) was defined as the measurement taken at the widest point of the facet joint space (Fig. 5). Each parameter (FJA, FO Length, FJSW) was measured bilaterally at the L3/4, L4/5, and L5/S1 levels. The average value of the bilateral measurements for each parameter at each level was calculated and used for statistical analysis. A larger FJA indicates a more sagittally oriented facet joint.

Fig. 6 .

Measurement of lumbar facet joint morphological parameters. A Axial T2-weighted MRI (T2WI) at the L4–L5 disc level, acquired parallel to the superior endplate of the L5 vertebral body. Measurements shown: Facet joint angle (FJA), defined as the angle (indicated by red markers) between the line connecting the anteromedial and posterolateral points of the facet joint surface and the horizontal reference line; Facet overhang length (FO Length), defined as the distance (yellow double-headed arrow) of the osteophyte projecting from the base of the lamina to the posterior margin of the mid-portion of the facet joint; Facet joint space width (FJSW), measured as the widest point of the joint space (blue line). B Schematic diagram illustrating the measurement of facet joint parameters

Statistical analysis

Group comparisons were conducted using tests appropriate for the data type and distribution. For comparisons between the DS and non-DS groups, continuous variables were analyzed using the Mann–Whitney U test, while categorical variables were analyzed using the Chi-square (χ²) test or Fisher’s exact test, as appropriate. For comparisons among DS subgroups, continuous variables were analyzed using the Kruskal–Wallis H test. For comparisons between the convex and concave sides in DS patients, paired tests were employed; the appropriate test (paired t-test or Wilcoxon signed-rank test) was automatically selected based on an assessment of the normality of the paired differences.

Spearman’s rank correlation analysis was used to explore the relationship between scoliosis severity (Cobb angle) and various imaging parameters. To investigate factors associated with the occurrence of DS, a multivariate binary logistic regression model was constructed. Variables were included based on results from univariate analyses (P < 0.1) and clinical relevance. Multicollinearity was assessed using the variance inflation factor (VIF), with a threshold of VIF < 5. Model results are reported as odds ratios (OR) with their 95% confidence intervals (95% CI). Model fit was evaluated using the Akaike Information Criterion (AIC).

For the significant predictors identified in the regression model, as well as for a combined multi-parameter model, receiver operating characteristic (ROC) curve analysis was performed. The analysis reported the area under the curve (AUC) with its 95% CI. The optimal cutoff point was determined using the Youden index, and the corresponding sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. All statistical tests were two-sided, with a significance level set at α = 0.05; a P-value < 0.05 was considered statistically significant.

Inter-rater reliability assessment

All measurements were conducted in strict accordance with a predefined, detailed standardized protocol. A complete blinding procedure, including data anonymization, random coding, and concealment of group allocation, was implemented by research assistants to ensure that all assessments were performed by evaluators who were fully blinded to the group assignments. To assess intra-rater reliability, a randomly selected subset of 22% of the cohort (50 patients) underwent repeated measurements after a four-week interval. The inter-rater reliability of measurements performed by the three spine surgeons was evaluated using the intraclass correlation coefficient (ICC). This assessment covered measurements of the Cobb angle, MF CSA, ES CSA, LCSA/GCSA, FJA, FO Length, and FJSW. ICC values were interpreted as follows: < 0.40 indicating poor agreement, 0.40–0.75 indicating fair to good agreement, and > 0.75 indicating excellent agreement.

Results

Baseline data results

In the DS group and the non-DS group, no significant differences were observed between the two groups in terms of BMI, smoking status, alcohol consumption, history of hypertension, or history of diabetes (all P > 0.05). Subjects in the DS group were significantly older than those in the non-DS group (P < 0.001). The distribution of gender also differed significantly between the two groups, with a lower proportion of males in the DS group (32% vs. 48.1%, P = 0.023). Regarding BMD, the DS group exhibited significantly lower values compared to the non-DS group [P = 0.043]. For specific data, please refer to Table 1.

Table 1.

Baseline Characteristics of the Study Population

Variable	Non-DS (n = 81)	DS (n = 150)	P value
Age ᵃ	62.00 (53.00–68.00)	70.00 (62.25–75.75)	< 0.001
BMI ᵃ	24.00 (22.20–26.80)	24.50 (22.09–26.59)	0.996
BMD ᵃ	− 1.30 (− 2.80 to − 1.60)	− 1.90 (− 3.10 to − 0.43)	0.043
Gender (Male, n%) ᵇ	39 (48.1%)	48 (32%)	0.023
Smoking (Yes, n%) ᵇ	15 (18.5%)	29 (19.5%)	1.000
Drinking (Yes, n%) ᵇ	20 (24.7%)	28 (18.7%)	0.364
Hypertension (Yes, n%) ᵇ	34 (42%)	72 (48%)	0.460
Diabetes (Yes, n%) ᵇ	12 (14.8%)	38 (25.3%)	0.092

ᵃData presented as Median (P25–P75). P value calculated using Mann–Whitney U test

ᵇData presented as n (%). P value calculated using Chi-squared test or Fisher’s exact test

Baseline characteristics of the DS patients, stratified by disease severity, are presented in Table 2. The prevalence of diabetes differed significantly across the severity groups (P = 0.027), with the severe group having the highest proportion of diabetic patients (47.6%). No other baseline characteristics, including age, BMI, BMD, gender, smoking, drinking, hypertension, and convexity, showed statistically significant differences among the three groups (all P > 0.05).

Table 2.

Baseline Characteristics of DS Patients by Severity

Variable	Mild (n = 72)	Moderate (n = 57)	Severe (n = 21)	Overall P value
Age ᵃ	69.00 (60.75–74.00)	71.00 (63.00–78.00)	72.00 (64.00–77.00)	0.131
BMI ᵃ	24.20 (22.04–26.63)	24.80 (21.80–26.70)	24.70 (23.10–25.80)	0.750
BMD ᵃ	− 1.45 (− 2.65 to − 0.40)	− 2.10 (− 2.90 to − 0.10)	− 2.50 (− 3.60 to − 1.60)	0.076
Gender (Male) ᵇ	28 (38.9%)	16 (28.1%)	4 (19%)	0.171
Gender (Female) ᵇ	44 (61.1%)	41 (71.9%)	17 (81%)
Smoking (No) ᵇ	56 (77.8%)	46 (80.7%)	18 (85.7%)	0.649
Smoking (Yes) ᵇ	16 (22.2%)	10 (17.5%)	3 (14.3%)
Drinking (No) ᵇ	57 (79.2%)	47 (82.5%)	18 (85.7%)	0.805
Drinking (Yes) ᵇ	15 (20.8%)	10 (17.5%)	3 (14.3%)
Hypertension (No) ᵇ	41 (56.9%)	30 (52.6%)	7 (33.3%)	0.177
Hypertension (Yes) ᵇ	31 (43.1%)	27 (47.4%)	14 (66.7%)
Diabetes (No) ᵇ	58 (80.6%)	43 (75.4%)	11 (52.4%)	0.027
Diabetes (Yes) ᵇ	14 (19.4%)	14 (24.6%)	10 (47.6%)
Convexity (Left) ᵇ	44 (61.1%)	36 (63.2%)	14 (66.7%)	0.884
Convexity (Right) ᵇ	28 (38.9%)	21 (36.8%)	7 (33.3%)

ᵃData presented as Median (P25–P75). P value calculated using Kruskal–Wallis test

ᵇData presented as n (%). P value calculated using Chi-squared test or Fisher’s exact test

Comparisons of muscle and facet joint parameters at various lumbar levels (L3-4, L4-5, L5-S1) between the groups are detailed in Table 3.

Table 3.

Comparison of muscle and facet joint parameters between groups

Parameter	Non-DS (n = 81)	DS (n = 150)	P value
L3-4 Avg LCSA/GCSAᵃ	0.76 (0.70–0.83)	0.71 (0.62–0.77)	< 0.001
L3-4 Total MF CSAᵇ	1193.26 ± 394.77	1078.59 ± 376.33	0.034
L3-4 Total ES CSAᵃ	1844.00 (1520.00–2449.00)	1608.00 (1251.00–2057.50)	< 0.001
L3-4 Avg FJAᵃ	50.00 (45.00–55.00)	55.00 (49.25–60.00)	0.001
L3-4 Avg FO Lengthᵃ	2.69 (1.11–3.61)	3.26 (2.54–4.21)	< 0.001
L3-4 Avg FJSWᵃ	0.00 (0.00–0.68)	0.00 (0.00–0.00)	< 0.001
L4-5 Avg LCSA/GCSAᵃ	0.70 (0.61–0.76)	0.65 (0.58–0.72)	0.020
L4-5 Total MF CSAᵃ	980.00 (756.00–1259.00)	871.50 (671.50–1024.00)	0.003
L4-5 Total ES CSAᵃ	1599.00 (1256.00–1905.00)	1602.00 (1252.00–1886.50)	0.577
L4-5 Avg FJAᵃ	53.00 (45.00–61.00)	59.00 (51.25–64.00)	0.002
L4-5 Avg FO Lengthᵃ	2.21 (0.72–3.71)	4.05 (3.20–5.67)	< 0.001
L4-5 Avg FJSWᵃ	0.70 (0.00–1.33)	0.00 (0.00–0.99)	0.002
L5-S1 Avg LCSA/GCSAᵇ	0.58 ± 0.12	0.54 ± 0.12	0.008
L5-S1 Total MF CSAᵃ	836.00 (652.00–1089.00)	775.50 (601.50–914.50)	0.108
L5-S1 Total ES CSAᵃ	932.00 (664.00–1285.00)	1095.00 (865.00–1387.75)	0.009
L5-S1 Avg FJAᵃ	45.00 (39.00–51.00)	53.00 (47.00–57.00)	< 0.001
L5-S1 Avg FO Lengthᵃ	1.88 (0.00–3.34)	3.51 (2.57–4.30)	< 0.001
L5-S1 Avg FJSWᵃ	0.40 (0.00–1.06)	0.00 (0.00–1.03)	0.166

ᵃData presented as Median (P25–P75). P value calculated using Mann–Whitney U test

ᵇData presented as Mean ± SD. P value calculated using independent t-test

Regarding muscle parameters, the DS group demonstrated generally poorer muscle quality, indicated by significantly lower values of the muscle atrophy indicator Avg LCSA/GCSA at the L3-4, L4-5, and L5-S1 levels compared to the non-DS group (all P < 0.05). Concurrently, the total CSA of the MF and ES muscles were significantly smaller in the DS group at the L3-4 and L4-5 levels (most P < 0.05). Conversely, at the L5-S1 level, the total ES CSA was significantly larger in the DS group (P = 0.009).

Regarding facet joint parameters, the DS group exhibited significant signs of facet joint osteoarthritis across all three levels. This was characterized by significantly larger FJA and greater FO Length compared to the non-DS group (most P < 0.01). Furthermore, the FJSW was significantly wider in the Non-DS group at the L3-4 and L4-5 levels (P < 0.01) (Fig. 7).

Fig. 7 .

Comparative axial MRI images of paraspinal muscles in non-DS and DS patients. A–C Paraspinal muscles at L3-L4, L4-L5, and L5-S1 levels in a representative non-DS patient. D–F Paraspinal muscles at corresponding levels in a representative DS patient

Within the DS group, highly significant differences were observed between the convex and concave sides across all lumbar levels (L3-S1) for both muscle and facet joint parameters (Table 4).

Table 4.

Paired comparison of convex and concave side parameters (DS Group)

Parameter	Convex side	Concave side	P value
L3-4 LCSA/GCSA ᵃ	0.73 (0.65–0.79)	0.69 (0.59–0.76)	< 0.001
L3-4 MF CSA ᵃ	559.00 (439.25–694.75)	497.50 (362.50–628.00)	< 0.001
L3-4 ES CSA ᵃ	809.50 (622.75–1094.00)	803.50 (598.25–997.75)	0.003
L3-4 FJA ᵃ	56.50 (51.00–63.00)	51.00 (46.25–56.00)	< 0.001
L3-4 FO Lengthᵃ	2.92 (2.00–3.72)	3.60 (3.04–4.67)	< 0.001
L3-4 FJSW ᵃ	0.00 (0.00–1.10)	0.00 (0.00–0.00)	< 0.001
L4-5 LCSA/GCSA ᵃ	0.69 (0.60–0.75)	0.63 (0.53–0.71)	< 0.001
L4-5 MF CSA ᵃ	440.50 (326.25–570.25)	403.00 (315.25–526.50)	0.004
L4-5 ES CSA ᵇ	828.11 ± 289.31	795.66 ± 254.98	0.023
L4-5 FJA ᵃ	62.00 (53.00–67.00)	56.00 (49.00–61.00)	< 0.001
L4-5 FO Lengthᵃ	3.52 (2.72–5.06)	4.70 (3.56–6.47)	< 0.001
L4-5 FJSW ᵃ	1.17 (0.00–1.58)	0.00 (0.00–1.11)	< 0.001
L5-S1 LCSA/GCSA ᵇ	0.56 ± 0.13	0.52 ± 0.13	< 0.001
L5-S1 MF CSA ᵃ	402.00 (310.25–483.50)	369.50 (289.00–438.50)	< 0.001
L5-S1 ES CSA ᵃ	561.50 (397.75–720.25)	531.00 (415.00–678.25)	0.036
L5-S1 FJA ᵃ	55.00 (48.00–61.00)	49.00 (44.00–54.00)	< 0.001
L5-S1 FO Lengthᵃ	2.89 (2.14–3.78)	3.92 (2.82–5.23)	< 0.001
L5-S1 FJSW ᵃ	0.00 (0.00–1.28)	0.00 (0.00–0.89)	< 0.001

Data from DS patients only. P values are calculated using paired tests

ᵃData as Median (P25–P75). P value from Wilcoxon Signed-Rank Test (non-parametric paired)

ᵇData as Mean ± SD. P value from Paired t-test (parametric)

Muscle parameters demonstrated a consistent pattern: the convex side exhibited significantly higher LCSA/GCSA values, indicating less muscle atrophy compared to the concave side (all P < 0.05).Facet joint parameters also showed marked asymmetry: the convex side had significantly larger FJA and FJSW, but shorter FO Length than the concave side (all P < 0.001).

These findings indicate a systematic asymmetry between the two sides of the spine in DS patients, with more pronounced changes on the convex side in terms of both muscle characteristics and joint degeneration (Fig. 8).

Fig. 8 .

Axial T2-weighted MRI comparison of paraspinal muscle atrophy at the L4-L5 level in DS patients of varying severities. A Mild DS, LCSA/GCSA = 0.88; B Moderate DS, LCSA/GCSA = 0.67; C Severe DS, LCSA/GCSA = 0.35

Stratification by DS severity revealed significant associations with worsening paraspinal muscle degeneration and facet joint osteoarthritis across all lumbar levels (with overall P-values < 0.01 for most parameters, except for L5-S1 Avg FJSW, P = 0.175). The most pronounced changes were observed at the L4–5 level.

Specifically, worsening muscle degeneration was evidenced by a significant decrease in the muscle quality indicator Avg LCSA/GCSA and a reduction in MF CSA and ES CSA across all severity levels from mild to severe DS. Concurrently, progressive facet joint osteoarthritis was characterized by significant increases in FJA and FO Length with increasing disease severity.

Post-hoc analyses revealed that severe DS was characterized by significant deterioration in both muscle and joint parameters compared to mild and moderate stages. In contrast, the progression from mild to moderate DS was primarily driven by increases in the facet joint angle across all lumbar levels, without widespread significant muscle degeneration (Table 5, Supplementary Table 1).

Table 5.

Comparison of imaging parameters by severity (DS Group)

Parameter	Mild (n = 72)	Moderate (n = 57)	Severe (n = 21)	Overall P value
L3-4 Avg LCSA/GCSAᵃ	0.73 (0.67–0.78)	0.71 (0.63–0.77)	0.60 (0.50–0.67)	< 0.001
L3-4 Total MF CSAᵃ	1145.50 (954.75–1388.25)	1004.00 (730.00–1240.00)	713.00 (602.00–978.00)	< 0.001
L3-4 Total ES CSAᵃ	1797.50 (1306.75–2201.25)	1577.00 (1283.00–2011.00)	1061.00 (884.00–1446.00)	< 0.001
L3-4 Avg FJAᵃ	50.50 (44.00–55.00)	56.00 (53.00–61.00)	61.00 (58.00–63.00)	< 0.001
L3-4 Avg FO Lengthᵃ	3.08 (2.29–3.68)	3.37 (2.76–4.37)	4.07 (3.02–5.52)	0.003
L3-4 Avg FJSWᵃ	0.72 (0.00–1.06)	0.00 (0.00–0.73)	0.00 (0.00–0.00)	< 0.001
L4-5 Avg LCSA/GCSAᵃ	0.68 (0.61–0.75)	0.65 (0.62–0.72)	0.54 (0.46–0.62)	< 0.001
L4-5 Total MF CSAᵃ	925.00 (728.00–1096.00)	858.00 (706.00–1010.00)	637.00 (532.00–858.00)	0.001
L4-5 Total ES CSAᵃ	1693.00 (1374.75–2030.75)	1612.00 (1314.00–1819.00)	1058.00 (879.00–1263.00)	< 0.001
L4-5 Avg FJAᵃ	51.00 (46.00–58.00)	62.00 (59.00–64.00)	68.00 (64.00–72.00)	< 0.001
L4-5 Avg FO Lengthᵃ	3.56 (2.74–4.60)	4.58 (3.75–5.93)	6.57 (4.49–7.73)	< 0.001
L4-5 Avg FJSWᵃ	1.21 (0.55–1.72)	0.88 (0.00–1.32)	0.26 (0.00–1.15)	0.010
L5-S1 Avg LCSA/GCSAᵃ	0.58 (0.50–0.67)	0.55 (0.48–0.60)	0.42 (0.35–0.48)	< 0.001
L5-S1 Total MF CSAᵃ	820.50 (673.75–924.75)	779.00 (609.00–959.00)	519.00 (446.00–725.00)	< 0.001
L5-S1 Total ES CSAᵃ	1130.50 (969.25–1447.75)	1155.00 (919.00–1403.00)	807.00 (489.00–904.00)	< 0.001
L5-S1 Avg FJAᵃ	47.00 (41.00–54.00)	56.00 (52.00–58.00)	59.00 (55.00–62.00)	< 0.001
L5-S1 Avg FO Lengthᵃ	2.99 (2.21–3.98)	3.71 (2.83–4.49)	4.06 (3.71–4.69)	0.002
L5-S1 Avg FJSWᵃ	1.08 (0.00–1.24)	0.00 (0.00–0.91)	0.20 (0.00–1.01)	0.175

ᵃData presented as Median (P25–P75). P value calculated using Kruskal–Wallis H test

Most paraspinal muscle and facet joint parameters showed no significant or only weak correlations (|rho|< 0.30), suggesting that muscle degeneration and facet joint osteoarthritis are relatively independent processes in degenerative spines (Supplementary Table 2). Among DS patients, a higher Cobb angle was consistently associated with poorer muscle status and more severe facet joint degeneration (Table 6).

1. Muscle Parameters: The fat infiltration fraction (Avg LCSA/GCSA) at all levels and nearly all muscle CSA showed significant negative correlations with the Cobb angle (all rho negative, P < 0.05), indicating worse muscle quality and smaller muscle volume were associated with a larger Cobb angle.

2. Facet Joint Parameters: The FJA and FO Length at all levels showed significant positive correlations with the Cobb angle (all rho positive, p < 0.001). The strongest correlation was observed between the L4-5 FJA and the Cobb angle (rho = 0.720).

3. Exception: The FJSW at L5-S1 showed no significant correlation with the Cobb angle (P = 0.526).

Table 6.

Spearman correlation between Cobb angle and imaging parameters (DS Group, n = 150)

Parameter	Spearman’s rho	P value
L3-4 Avg LCSA GCSA	− 0.290	< 0.001
L3-4 Total MF CSA	− 0.330	< 0.001
L3-4 Total ES CSA	− 0.320	< 0.001
L3-4 Avg FJA	0.620	< 0.001
L3-4 Avg FO	0.300	< 0.001
L3-4 Avg FJSW	− 0.300	< 0.001
L4-5 Avg LCSA GCSA	− 0.280	< 0.001
L4-5 Total MF CSA	− 0.250	0.002
L4-5 Total ES CSA	− 0.340	< 0.001
L4-5 Avg FJA	0.720	< 0.001
L4-5 Avg FO	0.460	< 0.001
L4-5 Avg FJSW	− 0.270	< 0.001
L5-S1 Avg LCSA GCSA	− 0.360	< 0.001
L5-S1 Total MF CSA	− 0.230	0.005
L5-S1 Total ES CSA	− 0.260	0.001
L5-S1 Avg FJA	0.620	< 0.001
L5-S1 Avg FO	0.280	< 0.001
L5-S1 Avg FJSW	− 0.052	0.526

Multivariate logistic regression and ROC curve analysis

Higher BMD (OR = 0.802, P = 0.032), greater LCSA/GCSA (OR = 0.005, P = 0.003), larger MF CSA (OR = 0.969, P = 0.027), and larger ES CSA (OR = 0.973, P = 0.014) were protective factors against DS, while larger FJA (OR = 1.075, P = 0.016) and greater FO Length (OR = 1.067, P = 0.001) were independently associated with an increased likelihood of DS. FJSW showed no significant association (P = 0.108). Complete regression outputs are presented in Table 7.

Table 7.

Logistic regression analysis of risk factors for DS

Variable	β	SE	Wald	OR	95% CI	P-value
BMD	− 0.221	0.103	4.605	0.802	0.655–0.981	0.032
L4-5 Avg LCSA/GCSA	− 5.240	1.761	8.851	0.005	0.001–0.167	0.003
L4-5 Avg MF CSA	− 0.031	0.014	4.899	0.969	0.943–0.996	0.027
L4-5 Avg ES CSA	− 0.027	0.011	6.024	0.973	0.953–0.995	0.014
L4-5 Avg FJA	0.072	0.031	5.761	1.075	1.013–1.139	0.016
L4-5 Avg FO Length	0.065	0.014	4.643	1.067	1.038–1.097	0.001
L4-5 Avg FJSW	0.140	0.085	2.713	1.151	0.974–1.359	0.108

Model Fit Index: AIC = 252.58

To address collinearity, parameters from the L4-5 segment were selected as representative for ROC analysis. The combined model demonstrated excellent ability to discriminate between DS and non-DS cases (AUC = 0.832, sensitivity = 80.0%, specificity = 77.8%). Among individual parameters, Avg LCSA/GCSA showed the highest diagnostic value (AUC = 0.712) and high sensitivity (92.0%), while FO Length and FJA had moderate discriminatory ability (AUCs = 0.651 and 0.627). The ES CSA demonstrated no significant discriminatory value (AUC = 0.463), as detailed in Table 8 and Fig. 9.

Table 8.

ROC Curve Analysis of Significant Predictors for DS

Predictor	AUC	95% confidence interval	Optimal Cut-off1	Sensitivity	Specificity	PPV2	NPV3
**Combined Model (All Predictors)**	0.832	0.750–0.875	0.607	0.800	0.778	0.870	0.677
BMD	0.581	0.514–0.647	− 0.900	0.680	0.481	0.708	0.448
L4-5 Avg LCSA/GCSA	0.712	0.640–0.784	0.805	0.920	0.556	0.793	0.789
L4-5 Avg MF CSA	0.608	0.528–0.688	635	0.652	0.548	0.741	0.451
L4-5 Avg ES CSA	0.463	0.383–0.543	832	0.479	0.452	0.608	0.323
L4-5 Avg FJA	0.627	0.547–0.707	57	0.675	0.570	0.749	0.482
L4-5 Avg FO Length	0.651	0.571–0.731	6.55	0.690	0.593	0.772	0.503

¹Optimal cut-off value determined by the Youden’s J statistic (max (sensitivity + specificity − 1))

²PPV: Positive Predictive Value

³NPV: Negative Predictive Value

Fig. 9 .

ROC curves of predictive factors for degenerative scoliosis. A ROC curves for BMD, LCSA/GCSA, MF CSA, ES CSA, FJA and FO Length with their respective area under the curve (AUC) values, generated using SPSS software. B ROC curve of the combined diagnostic model integrating all six parameters (BMD, LCSA/GCSA, MF CSA, ES CSA, FJA, FO Length)

Inter-rater reliability results

Inter-rater reliability assessment for key measurements (Cobb angle, MF CSA, ES CSA, LCSA/GCSA, FJA, FO Length, FJSW) among three spine surgeons demonstrated excellent consistency, with all ICC values exceeding the established reliability threshold of 0.90. These results confirm high measurement agreement and ensure robust data credibility, as detailed in Table 9.

Table 9.

Inter-rater reliability of key parameters

Parameter Category	ICC	95% CI	P-value
Cobb angle	0.98	0.96–0.99	< 0.001
MF CSA	0.95	0.94–0.99	< 0.001
ES CSA	0.93	0.87–0.96	< 0.001
LCSA/GCSA	0.96	0.92–0.99	< 0.001
FJA	0.93	0.88–0.96	< 0.001
FO Length	0.96	0.93–0.98	< 0.001
FJSW	0.91	0.85–0.95	< 0.001

Discussion

The pathogenesis of DS remains controversial, with unclear interrelationships among imaging parameters and undetermined critical thresholds for pathological cascades. This retrospective study analyzes radiographic parameters in DS patients to elucidate their correlations and underlying mechanisms.

As a degenerative condition, DS progresses with age, potentially linked to age-related paraspinal muscle CSA reduction [24]. Previous studies associate paraspinal muscle atrophy with age and sex [25]. Due to physiological differences between sexes, males demonstrate significantly greater paraspinal muscle density and cross-sectional area, along with a lower degree of fat infiltration, compared to females [26]. The lower incidence of DS in males, consistent with the protective role of larger paraspinal muscle cross-sectional area [27], was reflected in our study by a significantly higher proportion of males in the non-DS group. This suggests a protective effect of male sex, likely attributable to their typically greater paraspinal muscle mass and higher bone density. Furthermore, the prevalence of diabetes mellitus increased significantly with DS severity, reaching 47.6% in the severe group. This not only supports previous reports linking diabetes to scoliosis risk [28] but also positions it as a potential aggravating factor in DS, possibly through mechanisms involving microangiopathy, impaired tissue repair, and compromised bone quality [29]. Consequently, stringent glycemic control may be crucial in managing DS progression in diabetic patients.

This study primarily utilized MRI for the assessment of paraspinal muscles, as it is a non-irradiating modality with high sensitivity and excellent soft-tissue contrast for detecting muscular degeneration [30]. It is important to note, however, single-level measurements cannot represent whole-lumbar degeneration despite inter-level correlations [31]. Therefore, we assessed maximal CSA at L3-4, L4-5, and L5-S1 disc midpoints on T2-weighted MRI for comprehensive evaluation [32].

The paraspinal muscles comprise anterior/posterior groups, with MF and ES as key posterior stabilizers. Current research focuses on MF, psoas, and ES due to their anatomical proximity to spinal laminae/processes [33]. In non-pathological populations, ES degenerates earlier and more severely than MF [34]. MF primarily stabilizes the lumbar spine and maintains posture [35], while ES facilitates trunk movement (rotation/flexion/extension) [36]. MF atrophy in degenerative lumbar conditions correlates with functional demands, while ES hypertrophy compensates for instability [37]. This study demonstrated a protective effect of larger paraspinal muscle cross-sectional area in non-DS patients, except for ES CSA at L5-S1 which suggests compensatory hypertrophy in DS patients under mechanical stress. The most severe degeneration of muscles and facet joints occurred at L4-L5, confirming this level as the biomechanical epicenter of DS [6, 38]. These findings provide crucial guidance for surgical planning, including fusion level selection and decompression strategy, highlighting their direct clinical relevance.

From a biomechanical perspective, the loading balance of the paraspinal muscles is disrupted by scoliosis. Chronic compensatory overloading causes pain, fatigue, and atrophy—concave muscles exhibit degenerative atrophy (reduced CSA, higher fat infiltration), while convex muscles hypertrophy [39]. Asymmetric degeneration worsens with Cobb angle progression [40]. At the L3-S1 levels, the severe DS group demonstrated significantly lower CSA and LCSA/GCSA values compared to mild and moderate groups, indicating more advanced muscular atrophy, consistent with existing literature. In contrast, no significant differences were observed between mild and moderate groups. This suggests compensatory mechanisms (e.g., residual muscle fiber hypertrophy, synergistic muscle recruitment) may preserve muscle cross-sectional area during early DS stages, while progression to severe deformity marks a decompensation phase with marked muscle atrophy. Crawford et al. [41] report nonlinear fat infiltration acceleration, supporting that muscle degeneration progresses non-uniformly beyond a critical threshold. These findings collectively indicate that paraspinal muscle degeneration is a complex, non-linear process influenced by multiple factors. The progression appears to follow a critical threshold pattern: beyond a certain point, rapid muscle atrophy occurs, whereas changes remain relatively gradual and metrically subtle prior to reaching this threshold.

Beyond its protective role for the lumbar spine, the paraspinal musculature has also been shown to significantly influence surgical outcomes [42, 43]. Preserving muscle mass is thus crucial. Physical activity improves CSA, particularly at L2-4 [24]. Physiotherapeutic scoliosis-specific exercises (PSSE) balance bilateral muscle strength and are recommended by SOSORT [44], though high-quality evidence is lacking. Targeted training addressing convex/concave asymmetry and superficial/deep muscle layers is needed.

The musculoskeletal system functions interdependently [45]; metabolic imbalances reciprocally drive pathology. Low BMD strongly associates with paraspinal atrophy (5.7-fold risk increase) [46]. In this study, the DS group exhibited lower BMD, suggesting its potential involvement in DS pathogenesis. The lack of significant BMD differences among DS severity subgroups is likely attributable to a generally lower BMD baseline across all DS patients.

Decreased bone mineral density and paraspial muscle atrophy collectively point to a core pathological mechanism—loss of spinal stability. Weakened structural support from bony elements and functional decline of the dynamic stabilizers compromise the spine's ability to maintain normal alignment under load. The resulting instability generates asymmetric stresses, which are ultimately concentrated on the lumbar facet joints—the only true synovial joints in the spine [47]—thereby underscoring their critical role in the progression of DS.

It is well-established that facet orientation transitions from sagittal (L1-2, facilitating flexion/extension) to coronal (L3-5, resisting rotation/shear). With intervertebral discs, they form the “three-joint complex,” bearing 18% axial and 33–50% shear loads [48], crucial for stability [49].

Driven by spinal imbalance, paraspinal muscle atrophy progresses segmentally from L3 to S1, with the most severe deterioration occurring at L5/S1 [50]. Our data consistently showed decreasing CSA and LCSA/GCSA values from L3/L4 to L5/S1 across all DS severity groups. This aligns with Fortin et al.’s findings of particularly severe muscle loss at L5/S1 [22]. Therefore, rehabilitation must specifically target the lumbosacral junction through precise exercises like controlled bridging to activate deep stabilizers at this critical level.

The progressive atrophy of paraspinal muscles—particularly in the heavily loaded lower lumbar segments—serves as a critical substrate for the onset and progression of degenerative spondylolisthesis, while the morphology and structural integrity of the facet joints also play an essential role in spinal stability. Sagittally oriented facets experience higher stress, which accelerates degeneration and the progression of DS [51]. This study found that DS patients showed significantly larger FJA across L3-S1 levels than non-DS controls, with a clear severity gradient (severe > moderate > mild). Although most facet joint parameters differed significantly among DS subgroups at these levels, the average facet joint space width at L5-S1 remained consistent across groups, likely due to its inherent anatomical constraints under high mechanical stress.

Concave facets endure higher loads, worsening scoliosis [52]. We observed concave-side coronarization (reduced FJA) versus convex-side sagittalization (increased FJA), reflecting compressive remodeling. Concave facets develop more osteophytes (FO Length↑) [13]. Convex facets exhibit wider FJSW due to superior articular process sliding [53]. Insignificant FJSW differences in severe DS at L3-4/L5-S1 result from bilateral osteophyte filling, whereas L4-5 differences persist due to slower degeneration [54].

While sagittally-oriented facet joints have been established as a significant risk factor for the development and progression of DS, the biomechanical environment within a scoliotic spine—being a three-dimensional deformity—is inherently heterogeneous. Following the onset of scoliosis, the facet joints are subjected to distinctly asymmetric loading patterns between the concave and convex sides. This mechanical asymmetry, in turn, induces differential adaptive remodeling of the articular morphology. Facet degeneration correlates with osteoarthritis (OA) and degenerative disease [55]. While FJSW is graded in facet OA classifications [56], its reliability as an OA indicator is debated [57]. Thus, we measured FJSW without inferring OA severity.

During spinal motion, paraspinal muscles counteract shear forces—posteriorly in upper lumbar and anteriorly in lower segments [58]. Progressive facet sagittalization reduces shear resistance, activating compensatory muscle overloading that accelerates atrophy. Whether facet degeneration initiates muscle atrophy or vice versa remains undetermined, but their correlation is unequivocal.

Guided by the “muscle-bone unit” concept, this study investigated the correlation between facet joint degeneration and paraspinal muscle atrophy in DS. DS patients demonstrated longer facet osteophytes, narrower joint space, more sagittally-oriented facets, and greater muscle atrophy than controls. Cobb angle correlated negatively with MF CSA, ES CSA, LCSA/GCSA, and FJSW, and positively with FJA and FO length, particularly at L4–S1. Spearman analysis revealed significant negative correlations between sagittal facet orientation/osteophyte severity and muscle parameters in the DS group, unlike controls. This suggests closely linked musculoskeletal degeneration in DS, where paraspinal muscle dysfunction and facet joint degeneration form an integrated pathological complex that progresses with DS severity. Although the retrospective cross-sectional design precludes causal inference, these elements likely interact in a self-reinforcing vicious cycle driving DS progression.

Based on the integrated assessment of paraspinal muscles, facet joints, and bone density, we developed a discriminative model for DS risk stratification. The L4–5 multifidus LCSA/GCSA proved to be a powerful independent screening indicator (sensitivity 92%; cutoff < 80.5%). The combined model showed balanced performance (AUC = 0.832; cutoff > 0.607). While this tool effectively distinguishes current DS status in symptomatic individuals, its prospective predictive value requires validation through longitudinal studies. Patients with low BMD, reduced muscle parameters, or increased facet angles warrant monitoring despite the model's retrospective design.

We evaluated MF/ES CSA (muscle quantity) and LCSA/GCSA (quality), alongside FJA, FO Length, and FJSW for facet degeneration. Cobb angle correlations confirmed significant FJA/FO Length/LCSA/GCSA associations with DS severity. Multivariate logistic regression identified higher BMD, greater LCSA/GCSA, larger MF CSA, and larger ES CSA were protective factors against DS, while larger FJA and greater FO Length were independent risk factors.

Limitations and future directions

This study has several limitations: (1) Its single-center retrospective case–control design cannot establish causality, with potential confounders (e.g., nutritional status, genetic background) remaining uncontrolled; (2) Conclusions rely solely on imaging parameters without support from functional clinical data (e.g., muscle strength, pain scores, quality of life), while measurements remain subject to subjective interpretation; (3) Limited sample size and absence of histopathological validation preclude definitive conclusions.

To address these limitations, we propose: (1) a multicenter prospective cohort study with ≥ 3-year follow-up to establish causality; (2) integration of standardized patient-reported outcomes (ODI/VAS) and AI-based image analysis; (3) expanded recruitment with histological validation where feasible.

Conclusion

In summary, this study establishes that DS progression involves closely linked degeneration of the paraspinal muscles and facet joints, demonstrating a coordinated “musculoskeletal unit” deterioration. The L4–S1 segments were identified as the biomechanical focus, with the L4/5 multifidus LCSA/GCSA serving as a highly sensitive screening marker. Patients with lower BMD, reduced muscle parameters (LCSA/GCSA, MF CSA, ES CSA), or larger facet joint measurements (FJA, FO Length) show increased DS risk. These findings support integrating L4/5 paraspinal muscle evaluation into clinical practice and establishing targeted muscle rehabilitation as a core therapeutic strategy to improve spinal support and slow disease progression.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Jiangkai Yu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing. Cong Zhang: Data curation, Formal analysis, Investigation, Writing—review & editing.Yan Zhou: Resources, Supervision, Writing—review & editing. Chengming Li: Data curation, Formal analysis, Project administration, Supervision, Writing—review & editing.

Funding

The research was supported by Grants from the National Natural Science Foundation of China (82202768), Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Construction Funds (GSP-LCYJFH20), the Fundamental Research Funds for the Central Universities (2242020K40156), Open Project Programme of the Key Base for Standardized Training for General Physicans, Zhongda Hospital, Southeast University (ZDZYJD-QK-2022-16), Jiangsu Entrepreneurship and Innovation Doctors (202030338).

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethical Committee of the Zhongda Hospital, Southeast University (Approval No:2022ZDSYLL406-P01) and conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all patients. This study is a retrospective analysis and not a clinical trial. Clinical trial number: not applicable. Written informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable. No identifiable images or personal/clinical details requiring separate publication consent are presented in this manuscript.

Competing interests

The authors declare no competing interests.