Correlation between the fatty infiltration of paraspinal muscles and disc degeneration and the underlying mechanism

A retrospective study was conducted on 109 patients with chronic LBP. Between June 2015 and July 2021, a total of 94 patients with lumbar disc herniation (age range, 17–63 years) were included. The inclusion criteria were as follows: 1) single-level lumbar disc herniation at L4/L5 in MRI and CT; 2) radiological examination matching with clinical symptoms (LBP and sciatica); 3) conservative treatment (medical treatment and/or physiotherapy) was ineffective for at least 3 months. The exclusion criteria were as follows: spinal fractures, trauma, spinal cord injuries, spinal infections, spinal tumors, lumbar kyphosis, pregnancy, previous lumbar surgery, scoliosis, spondylolisthesis, comorbidities (diabetes, osteoporosis, uremia, stroke, rheumatoid arthritis, myasthenia gravis, and malignant tumors). Fifteen patients with nonspecific chronic LBP were also selected (Pfirrmann Grade I). The study was approved by the Institutional Review Board of Shanghai Ruijin Hospital.

There were two main reasons for choosing only the L4/L5 segment in this study: 1) the primary site of disc herniations in the lumbar spine occurs at L4/L5 [23]; 2) the L4/L5 disc level plays an important role in the stabilization of the lumbar spine by paraspinal muscles, especially in the early stage of spinal degeneration [18].

The MRI scans we used were from the Signa HDx 1.5 T MRI system (GE Healthcare, Milwaukee, WI, USA). T2-weighted fast spin-echo (FSE) MRI scans (TR 2550 ms/TE 110 ms; field of view, 320 × 320 mm; matrix, 250 × 324; section thickness, 4 mm; number of excitations, 2) were used to examine the degeneration of the lumbar intervertebral discs and to analyze the fatty infiltration of paraspinal muscles [18, 24‐26]. T2-weighted sagittal lumbar spine MRIs were used to evaluate LDD at the L4/L5 disc level by using the Pfirrmann classification (Fig. 1) [27]. Three spine surgeons who were blinded to the clinical data independently assessed the grade of the L4/L5 disc. The intra-class correlation coefficient (ICC) was used to determine the inter/intra-observer agreement using the Pfirrmann grading system. Grade I did not exist in any patients with disc herniation, but it existed in patients with nonspecific chronic LBP. Pfirrmann grades of the patients and their corresponding T2-weighted axial images are shown in Fig. 1.

The T2-weighted axial images we collected were used to measure the cross-sectional area (CSA) and fatty infiltration of the paraspinal muscles at the L4-L5 disc level. Muscles were segmented on the axial slice at the L4-L5 disc by locating lines on sagittal images based on previous studies [28‐30]. The L5-S1 level was excluded because the axial-cutting gantry was obstructed by the iliac crest and the muscular anatomy is different from that of the upper level [30]. Moreover, the inclination of L5 is often so large that the CSA of paraspinal muscles cannot be appropriately visualized [31]. Quantitative measurement of the paraspinal muscles was obtained from axial T2-weighted images using Image J software (version 1.52, National Institutes of Health, Bethesda, Maryland, US). According to previous studies [2, 25, 26, 30], we manually defined the regions of interest (ROI) per slice for the bilateral paraspinal musculature (Fig. 2). The posterior layer of the thoracolumbar fascia was used as the border of the ES. Measurements were obtained by using a highly reliable thresholding technique, which is based on the difference in signal intensity between the muscle and fat tissue. The fat infiltrated area was measured using the pseudocoloring technique through Image J. In this technique, bright pixels of fat tissue were colored red (Fig. 2). Subsequently, the area of the red region in the muscle compartment was calculated based on the scale of the image. The muscle measurements of interest were as follows: total muscle CSA (TMCSA) and fat cross-sectional area (FCSA). The ratio of FCSA to TMCSA (FCSA/TMCSA) was calculated as an indicator of muscle composition (fatty infiltration). To compensate for the influence of body shape and size on paraspinal muscle CSA, the relative CSA (RCSA) was calculated by dividing the muscle CSA by the CSA of the L4 vertebral body endplate (VBE) at the inferior level. The relative percentage of asymmetry in CSA was calculated using the following formula: [(L – S)/L)] × 100, where L is the larger side and S is the smaller side [25]. The high reliability of the paraspinal muscle thresholding technique of ImageJ software was already validated [32]. All measurements were obtained by two independent spine surgeons with more than 5 years of experience. The ICC was used to assess the inter/intra-observer agreement.

From December 2021 to March 2022, 16 patients with L4/5 degenerative disk diseases were selected. We used the visual analogue scale (VAS), Oswestry disability index (ODI), and Japanese Orthopaedic Association Scores (JOA) to evaluate low back pain during hospitalization. According to Kjaer’s study [24], the fatty infiltration of multifidus was rated as grade 0 if normal condition (< 10%); grade 1 for slight fat infiltration (10–50%), and grade 2 for severe fat infiltration (> 50%). Grade 0 did not exist in our patents. The number of patients with Kjaer grade 1 was 11, and the number of patients with Kjaer grade 2 was 5.

To preliminarily explore whether inflammation was associated with fatty infiltration and disc degeneration, we defined the lumbar disc of Pfirrmann grade II as the mild-degeneration group and the lumbar disc of Pfirrmann grade IV as the severe-degeneration group, the fatty infiltration of MF in Pfirrmann grade IV was significantly higher than that in grade II (Fig. 3). Between May 2021 and July 2021, eight individuals with chronic LBP were recruited for this study. The patients, aged 26–66 years, underwent discectomy for disc herniation. Multifidus and lumbar disc specimens were collected at the time of surgery. All of the patients permitted us to use the discarded tissue for research and signed informed consent forms. There were three patients in the mild-degeneration group and five patients in the severe-degeneration group. There was no significant difference in patients’ characteristics between the two groups.

During surgery, the intervertebral disc was obtained from the operated segment. Samples of the multifidus adjacent to the surgical segment were taken from the side ipsilateral to the disc herniation. The samples were washed three times with PBS, quickly placed in liquid nitrogen, and then stored at − 80 °C.

Total RNA was isolated using TRIzol reagent (Invitrogen, Cat.15596026, USA) in accordance with the manufacturer’s instructions. Complementary DNAs (cDNA) were obtained by using the reverse-transcription kit (Takara, Cat.RR036A, Japan). Subsequently, cDNAs were served as templates, and certain sequences among them were amplified by the qPCR kit with SYBR Green Mastermix (Takara, Cat. RR420A, Japan). The relative expression was calculated as 2^−ΔΔCt according to Ct values. The genes and their respective primer pairs are listed in Table 1.

Total protein was detected by using enhanced BCA protein assay kit (Beyotime, Cat. P0010S, China). The level of TNF was determined by using the ELISA kit (Beyotime, Cat.PT518, China). All experiment procedures followed the manufacturer’s instructions.

Continuous variables were expressed as the mean ± standard deviation and were analyzed using SPSS 25.0 software (SPSS Inc., Chicago, IL, USA). Categorical variables were compared using Fisher’s exact test. Parametric and non-parametric continuous variables were compared among groups using one-way ANOVA and the Kruskal–Wallis test, respectively. We performed post hoc analysis using the Bonferroni method for continuous variables [33]. The Mann–Whitney U test or independent-samples t-test was also used for comparison between the two groups. To assess the correlation among LDD, age, gender, BMI, and fatty infiltration of paraspinal muscles, we performed Spearman correlation and multivariate linear regression analysis (since the Pfirrmann grading is an ordinal variable). Correlations > 0.7 were considered very strong; 0.5–0.7, strong; 0.3–0.5, moderate; and < 0.3, weak [26]. The ICC was calculated using a two-way random model, characterized by absolute agreement to quantify measurement reliability. Statistical significance was set at α < 0.05.

The inter-observer agreement of the Pfirrmann classification was excellent, with an ICC of 0.974 [0.965 − 0.981], the intra-rater reliability values for the evaluation of discs were also excellent, with an ICC of 0.979 [0.969–0.989], which suggested that the measurement was reliable. We evaluated 109 patients presenting with chronic LBP (mean age, 42.06 ± 12.04 years; duration of symptoms, 30.11 ± 33.97 months;) and L4/L5 disc degeneration grade included grade I (15, 13.7%), grade II (9, 8.3%), grade III (54, 49.5%), grade IV (26, 23.9%), and grade V (5, 4.6%). Further demographic details of the patients with different Pfirrmann grades are shown in Table 2. There was no statistical significance in patients’ characteristics among groups.

The inter-observer agreement in assessing musculature were excellent (MF: ICC = 0.989, ES: ICC = 0.990, PS: ICC = 0.934). The intra-rater reliability values for the evaluation of musculature were also excellent (MF: ICC = 0.984, ES: ICC = 0.994, PS: ICC = 0.96). For analyzing the fatty infiltration of paraspinal muscles, we measured the TMCSA and the fatty infiltration of MF, ES, and PS in different Pfirrmann grades. The results showed that the TMCSA, RCSA, and CSA asymmetry of paraspinal muscles among different grades was not significantly different (Table 3). However, with an increase in Pfirrmann grades, the fatty infiltration of MF and ES significantly increased (Fig. 4A, B). Fatty infiltration of PS in Pfirrmann III, IV, and V patients also significantly increased compared with that in Pfirrmann I patients (Fig. 4C).

Then, we utilized the Spearman correlation analysis and multivariate linear regression to further confirm the relationship between fatty infiltration of paraspinal muscles and LDD. The results suggested that the Pfirrmann grade had a strong positive correlation with fatty infiltration of MF (Rho = 0.57, p < 0.001) and had a moderate positive correlation with fatty infiltration of ES (Rho = 0.49, p < 0.001) and PS (Rho = 0.31, p < 0.05). Moreover, the multivariate linear regression analysis verified that the Pfirrmann grade, as the first weighting factor, was independently associated with fatty infiltration of MF (β coefficient = 0.515; p < 0.0001) and ES (β coefficient = 0.418; p < 0.001), and as the second weighting factor in PS (β coefficient = 0.206; p < 0.001).

Consistent with previous findings [5, 24, 26], age showed a strong association with fatty infiltration of the MF (Rho = 0.523, p < 0.001) and ES (Rho = 0.512, p < 0.001), and a weak association with PS (Rho = 0.287, p < 0.05). BMI was not related to fatty infiltration of MF, ES, and PS (Rho =  − 0.027, 0.018, 0.104, respectively; p > 0.05). TMCSA of paraspinal muscles in men was larger than that in women, but fatty infiltration in females was higher than that in males, as shown in Table 4. Moreover, the TMCSA of paraspinal muscles decreased with age and fatty infiltration of the paraspinal muscles increased with age, as shown in Table 5 and Fig. 5. The number of each age group was as follows: 17–30 years group (n = 23), 31–40 years group (n = 27), 41–50 years group (n = 26), 51–63 years group (n = 33).

For further exploring the respective weights of these factors, we performed the multivariate linear regression analysis. We found that gender and age were respectively associated with fatty infiltration of MF, ES, and PS (MF: male sex: β coefficient =  − 0.224; p < 0.05; age: β coefficient = 0.316; p < 0.001; ES: male sex: β coefficient =  − 0.269; p < 0.001; age: β coefficient = 0.354; p < 0.001, and PS: male sex: β coefficient =  − 0.156; p = 0.084; age: β coefficient = 0.266; p < 0.05). BMI had no relation with fatty infiltration of paraspinal muscles (MF: BMI: β coefficient =  − 0.094, p = 0.196, and ES: BMI: β coefficient =  − 0.04, p = 0.596, and PS: BMI: β coefficient =  − 0.058, p = 0.535). The results of Spearman correlation analysis and multivariate linear regression analysis were summarized in tables as additional files.

The RCSA of men in PS was larger than that of women, but the muscle CSA asymmetry did not show a significant difference between men and women. The RCSA of MF and PS decreased with age; however, the muscle CSA asymmetry did not change with age.

Consistent with previous study [24, 34], we found that the VAS (7.60 ± 0.55) and ODI (23.80 ± 6.50) in Kjaer grade 2 were significantly higher than those in Kjaer grade 1(VAS: 4.45 ± 1.21, ODI: 13.00 ± 5.37), the JOA(17.64 ± 3.14)in Kjaer grade 1 was also significantly higher than that in Kjaer grade 2 (JOA: 11.00 ± 3.40) [Fig. 6]. The results suggested that LBP was associated with fatty infiltration of the multifidus.

TNF expression of the lumbar disc in the severe-degeneration group was significantly higher than that in the mild-degeneration group (Fig. 7). There were no differences in the expression levels of IL-1β, IL-6, IL-8, NOS2, and TGF-β between the two groups (Table 6).

TNF expression of the multifidus was not significantly different between the mild-degeneration group and the severe-degeneration group (Fig. 7). There were also no differences in the expression levels of IL-1β, IL-6, IL-8, NOS2, and TGF-β between the two groups (Table 7).

Based on the result of qPCR, we detected the concentration of TNF in the lumbar disc and multifidus by using ELISA. We found that the protein expression of TNF of the lumbar disc and multifidus in the severe-degeneration group was significantly higher than that in the mild-degeneration group (Fig. 8).