Table 4 Detailed results analysis for non-pooled data

1.0 Paraspinal muscle morphology in patients with lumbar disc herniation – assessed with imaging:

1.0.1 Patients serving as own controls

 • 3 studies assessed the TCSA of the PMM [15, 16, 37]:

  ○ [15]: acute patients: PMM was larger on side of LDH regardless of measurement relationship to LDH; not statistically significant.

  ○ [16]: acute and chronic patients at LDH level only: PMM CSA was insignificantly smaller on side of LDH for both groups.

  ○ [37]: chronic patients: PMM was smaller on side of LDH regardless of relationship to level of LDH; statistically significant at L4/5, L5/S1 (P < 0.05). Median % reduction of TCSA on side of LDH averaged 8.5% (P < 0.05) at L4/5, L5/S1.

 • 2 high quality studies assessed the TCSA and FCSA of the ESM [15, 41]:

  ○ [15]: no significance to these differences at any level.

  ○ [41]: significantly smaller FCSA measures at L5/S1 (level below LDH) (P = 0.04), and significantly smaller ratios at L4/5, L5/S1 (P = 0.04, 0.007); TCSA on side of LDH was larger at the level above LDH (P = 0.05).

 • 2 studies assessed MRI signal intensity of the LMM and ESM: one including acute [15], one chronic [41] patients. Both studies demonstrated:

  ○ significant increase in mean LMM signal (i.e., more fat) on side of LDH at the level below (P = 0.014 [15] and 0.04 [41]); no consistent or significant differences noted at or above the level of LDH.

  ○ higher mean ESM signal on side of LDH at the level of LDH in acute patients (P = 0.017) and the level below LDH (S1) in chronic (P = 0.02).

 • 1 moderate quality study assessed the combined echo intensity of the LMM and ESM [42], and reported statstically non-significant results.

 • 1 high quality study compared the affected to non-affected side to assess median TCSA and MLD of the LMM against duration of symptoms and severity of NR compression [35], reporting:

  ○ non-significant results for TSCA.

  ○ MLD larger on side of LDH across all duration groups (P < 0.05); MLD progressively enlarged as duration increased (P = 0.021).

  ○ non-significant results for MLD across all severity groups.

 • 1 high quality study assessed FCSA and MLD ratios of the LMM, and their relationship to various clinical measures [18]:

  ○ no significant relationship found between FCSA or MLD ratios and severity of NR compression, symptom duration, or motor deficit.

  ○ 1 moderate quality study assessed TCSA, FCSA, and FCSA:TCSA ratios in LDH patients without radiculopathy [38], reporting no significant differences.

1.0.2 Compared to healthy controls without LDH or radiculopathy

 • 4 LDH studies included a healthy control group [37,38,39, 42]:

  ○ [37]: measured side-to-side difference in TCSA of the PMM in a control group: no difference found between sides. Control group median TCSA was smaller than both LDH groups from L3/4 – L5/S1, but no statistical comparison was made between groups.

  ○ [38]: compared TCSA, FCSA, FCSA:TCSA ratio, and involved:uninvolved side FCSA ratios (IS:US) of control group LMM to patients with LDH – with and without radiculopathy.

   ▪ TCSA smaller at L5/S1 on side of LDH in both patient groups (P < 0.05); FCSA significantly smaller in both patient groups at L4/5, L5/S1 (P < 0.05).

   ▪ FCSA:TCSA ratio smaller in both patient groups at L3/4, L4/5 (P < 0.05).

   ▪ IS:US ratio for radiculopathy group significantly smaller than controls at L4/5, L5/S1 (P < 0.01), and when all levels were combined (P < 0.05).

   ▪ IS:US ratio abnormal in 79% of radiculopathy cases and 10% of controls (P < 0.01), but not between control group and uninvolved side of LDH.

  ○ [39]: assessed amount of combined fat infiltration of LMM and ESM (presumably bilaterally, but not defined). Fat infiltration was greater in the LDH group at all levels (P < 0.05 at L2/3; P < 0.001 from L3/4 – L5/S1).

  ○ [42]: assessed echo intensity of LMM and ESM combined, with no difference noted between any groups.

1.0.3 Compared to low back pain patients without LDH or radiculopathy

 • 3 LDH studies included LBP comparison groups; 1 high quality [36], 1 moderate quality [40], and 1 low quality [43]:

  ○ [36]: compared TCSA and quantitative gradings of LMM, PVM, PMM, QLM for LDH with unilateral or bilateral radiculopathy to chronic LBP only patients.

   ▪ smaller TCSA of right QLM only noted in the LDH group (P = 0.01).

   ▪ higher grades of fat infiltration more prevalent in LDH group at all locations except PMM (P range: 0.02 – 0.04).

   ▪ NB: 8 patients in the comparison group also had facet arthrosis, but no leg pain.

  ○ [40]: used point-of-contact calculations (Cavalieri approximation principle) to compare muscle:fat ratios of LMM, ESM, & PMM in single or multi-level LDH patients to same ratios in LBP patients (unknown symptom duration) with single or multi-level degenerative disc disease; individual muscles were combined bilaterally. No difference in ratios found between patient groups for any muscle at any level.

  ○ [43]: used quantitative muscle grading to assess the LMM bilaterally from L3/4 – L5/S1; greater atrophy, and more severe atrophy, reported in LDH group at all levels (P < 0.01). NB: no analyses made regarding side of LDH; unknown if differences in symptom chronicity present requiring adjustment.

1.0.4 Compared to LDH patients without radiculopathy

 • 1 study compared the TCSA, FCSA, FCSA:TCSA ratio, and IS:US ratio (FCSA) of the LMM between LDH patients with and without radiculopathy [38]:

  ○ IS:US ratio for radiculopathy group smaller than LDH-only group at L4/5, L5/S1 (P < 0.01), and with all levels combined (P < 0.05).

  ○ IS:US ratio abnormal in 24% of LDH-only cases vs. 79% of radiculopathy cases (P < 0.01).

  ○ no differences in the remaining CSA measures were noted.

1.1 Paraspinal muscle morphology in patients with lumbar disc herniation – assessed with biopsy:

 • 3 studies assessed mean fiber type diameter and distribution for the ESM [45] and the LMM [44, 46]:

 ○ [45]: compared the affected to non-affected side and found no difference for any measures.

 ○ [44]: compared to deceased controls: Type I fiber distribution in LMM higher for males with LDH (P<0.05); type I fiber diameter larger in males (P<0.05) and females (P < 0.01) with LDH; type IIA and IIB fiber diameter larger in male LDH patients (P < 0.05). NB: not adjusted for differences in age or sex.

 ○ [46]: compared to deceased controls: Type I fiber diameter was significantly larger in male LDH patients (P < 0.01).

 • 2 studies assessed mean muscle strength factor (MSF), one for the LMM and ESM [45], and one for the LMM only [47]:

 ○ [45]: non-significant results for both muscle groups.

 ○ [47]: type II fiber MSF was lower on the LDH side (P < 0.05).

 • 2 studies compared % frequencies of fiber type grouping and small angular fibers in the LMM [22, 33]:

 ○ [22]: higher grouping frequency on side of LDH at the level below LDH (L5): 27.6% vs.10.3%; higher angular fiber frequency noted on side of LDH at L5: 20.7% vs. 3.4% (no P values).

 ○ [33]: higher grouping frequency on side of LDH at L5 (level below LDH): 35% vs. 6%; higher angular fiber frequency noted on side of LDH at L5: 41% vs. 24% (no P values).

 • 1 study, using patients as their own control, measured mean fiber type CSA of the LMM, and fiber CSA with or without a +SLR [47], noting:

 ○ significantly and consistently smaller CSA for both fiber types on affected side of LDH (P < 0.05); this became more pronounced when considering +SLR patients only (P < 0.01), but non-significant with −SLR patients only.

 • 1 moderate quality study assessed mean atrophy/hypertrophy factors for type I & II fibers, and mean % core targetoid and moth-eaten change in the LMM against cadaveric controls [46]:

 ○ an increase in core-targetoid presence was found in male (P < 0.01) and female patients (P < 0.001), with higher type I fiber hypertrophy factor in male patients (P < 0.01) and an increase in moth-eaten change in female patients (P < 0.001).

1.3 Paraspinal muscle morphology in patients with facet arthrosis – assessed with imaging:

 • 4 studies assessed the association of facet arthrosis with paraspinal muscle density (2 using the same general population data set) [(31,32),49,50]:

 ○ [31]: reported severe arthrosis (grade 3) at L4/5 consistently associated with greater fat infiltration of the LMM and ESM (P range: 0.0002 – 0.056).

 ○ [32]: identified associations between reduced paraspinal muscle density and arthrosis [AOR: 3.68 [1.36 – 9.97] (LMM); 2.80 [1.10 – 7.16] (ESM)].

 ○ [49]: assessed L4/5, with AORs showing associations between arthrosis and muscle density ratios (P range: 0.001 – 0.009 (LMM); 0.002 – 0.01 (ESM)), as well as arthrosis and higher fat infiltration grades (P <0.0001 (LMM & ESM)). No associations found between arthrosis and mean muscle density only.

 ○ [50]: demonstrated negative correlations between LMM, PMM, and Longissimus muscle density and facet arthrosis with univariate analysis (P <0.0001 for all muscles), but non-significant correlations following multivariate analysis.

   ▪ NB: this study grouped all measures rather than assessing muscle density or arthrosis by spinal level.

 • 1 study assessed arthrosis in relation to CSA and the muscle-fat index (MFI) [51], identifying:

 ○ smaller CSA and higher MFI (i.e., higher fat content) at all levels with arthrosis present (P < 0.001); increased MFI was independently associated with arthrosis at all levels (P range: <0.001 – 0.005).

 ○ differences in CSA asymmetry were greater for CSA with arthrosis present at L5/S1 only (P range: <0.001 – 0.005).

1.4 Paraspinal muscle morphology in patients with canal stenosis – assessed with imaging:

 • 4 studies assessed the association of canal stenosis with paraspinal muscle fat inflitration (2 using the same general population data set) [(31,32),52,53]:

 ○ [31, 32]: after adjusting for age, sex, and/or BMI, spinal stenosis was not associated with altered CT muscle density for the LMM or ESM (N = 15).

 ○ [52]: showed associations of stenosis with increased CT muscle density in LMM, ESM, and PMM in 165 patients with confirmed clinical symptoms of stenosis (P range = 0.036 to < 0.001).

   ▪ NB: the lower fat content in the stenosis group may be the result of the muscles being measured above the level of reported stenosis.

 ○ [53]: assessed fat infiltration with MRI at the level of stenosis, with a greater fat infiltration ratio in the stenosis group (N = 40; P = 0.004).

 • 3 studies assessed the relationship of stenosis to muscle atrophy [20, 52, 53]:

 ○ [20]: reported a reduction in FCSA of LMM between stenosis and healthy control groups (P = 0.04), but not stenosis and LBP groups.

 ○ [52]: demonstrated greater FCSA of the ESM in male (P = 0.011) and female (P = 0.014) stenosis patients, and PMM in male stenosis patients (P = 0.042), but no significant difference for the LMM. NB: acquired measurements at L3, which could account for conflicting results with other studies.

 ○ [53]: evaluated muscle asymmetry and TCSA measures of the LMM at L4/5;with greater asymmetry (P = 0.006) and lower TCSA:body ratios (P = 0.006) reported in the stenosis group.

 • 1 study used MR spectroscopy to assess extramyocellular lipids (EMCL) and intramyocelluar lipds (IMCL) [54]:

 ○ when assessing EMCL content (i.e, the fat tissue deposits visible on standard MR imaging), no difference was found between the stenosis and CLBP groups; this is in agreement with the study by Yarjanian above [20].

 ○ IMCL content was higher in the CLBP vs. stenosis group (P =< 0.01).

   ▪ NB: IMCL deposits not are seen on standard imaging, being more likely to increase due to metabolic changes associated with inactivity; significantly higher VAS scores in the CLBP group would be a more important contributor than potential NR compression.

   ▪ NB: this study also analysed postural changes between each group but found no differences; alignment was not a confounding factor in this analysis.