Subjects
This was a retrospective cross-sectional study approved by Shanghai East Hospital (East Hospital Affiliated to Tongji University) Medical Ethics Committee (Shanghai, China). The clinical and radiographic data of 532 consecutive patients aged 18 to 60 years were reviewed. All patients came from the orthopedic clinic of Shanghai East Hospital between March 2012 and September 2018. They were diagnosed with LDH based on clinical presentations and radiological evidence including MRI. The diagnosis was made by Professor Jun Tan who has more than 20 years of experience in spine surgery. The diagnostic criteria include: 1. definite disc herniation shown on MRI; 2. back pain with or without radiculopathy and clinical signs that were consistent with radiographic findings [17]. Disc bulging was not included in this study. To minimize confounding variables that may affect the FJs, patients with the following diseases were excluded from this study: lumbar spondylolisthesis, scoliosis, severe central canal stenosis, and other congenital spinal deformities; and tumor, infection, rheumatoid arthritis, ankylosing spondylitis, and other systemic diseases. Additionally, patients with unsatisfactory imaging were also excluded from the study.
Imaging evaluation
MRI (in the neutral supine position) (Intera Achieva 3.0 T; Philips Healthcare, Best, the Netherlands) was performed for all patients. MRI sequence parameters: sagittal fast spin echo T1-weighted imaging (FSE T1WI): field of view (FOV) =300 mm, repetition time/echo time (TR/TE) =550 ms/11 ms; sagittal FSE T2WI: FOV = 300 mm, TR/TE = 3000 ms/100 ms; axial FSE T2WI: FOV = 200 mm, TR/TE = 3500 ms/100 ms; slice thickness: 3–5 mm.
LDH
The Michigan State University (MSU) classification [18] (Fig. 1) was used to evaluate LDH based on T2-weighted axial MRI slices. In this classification, the size of LDH is expressed as “1, 2, 3”, while the location of LDH is expressed as “A, AB, B, C”, which approximately corresponds to “central”, “paracentral”, “lateral” and “far lateral”. These subgrouping methods are based on an intra-facet line drawn transversely across the lumbar canal, to and from the medial edges of the right and left facet joint articulations. “1” and “2” are when the LDH extends less than or more than 50% of the distance from the non-herniated posterior aspect of the disc to the intra-facet line, and “3” is when the LDH extends beyond the intra-facet line. To define the location of the LDH, three points are placed along the intra-facet line, dividing it into four equal quarters; then, three vertical lines are drawn through these points, and four quadrants are created. “A” represents the left and right central quadrants, “B” represents the left and right lateral quadrants, “C” represents the area extending beyond the borderline of the lateral quadrants, and “AB” means that the furthest herniation is on the right and left lateral vertical lines. The level with the maximal herniation was selected for evaluation.
FJOA
Right and left FJOA were graded separately using the criteria introduced by Weishaupt et al. [19] and Kalichman L et al. [9] (Fig. 2): grade 0 (G0): normal; grade 1 (G1): joint space narrowing (< 2 mm) and/or mild osteophytes and/or hypertrophy; grade 2 (G2): joint space narrowing (1 mm) and/or moderate osteophytes and/or moderate hypertrophy and/or subchondral erosions; grade 3 (G3): joint space narrowing (bone to bone) and/or severe osteophytes and/or severe hypertrophy and/or severe subchondral erosions and/or subchondral cysts. Grade ≥ 2 was considered substantial FJOA.
FJ orientation and tropism
On an axial MRI slice that bisected the intervertebral disc, FJ angles relative to the sagittal plane were measured using the method described by Karacan I et al. [20] (Fig. 3). The FJ angle was defined as the angle between the reference line bisecting the base of the spinous process and the facet line connecting the margins of the superior articular process. Continuous FJ tropism was defined as the absolute difference between the left and right FJ angles.
Reliability of the assessment
All the evaluations were performed primarily by an experienced spine specialist (MJY) who was blinded to the patients’ details and to the study hypothesis. Before the formal assessment, he was trained by a senior radiologist (WPW) who specialized in the musculoskeletal system. First, 40 LDH segments (80 FJs) were randomly selected for evaluation and the interexaminer reliabilities for the spine specialist and the radiologist were calculated. Then, the spine specialist read another 40 segments on two separate occasions, and the intraexaminer reliability was obtained. Finally, the weighted Kappa values for interexaminer reliability were 0.845 and 0.883 for FJOA and the degree of LDH, respectively; the Kappa value for interexaminer reliability was 0.808 for the location of LDH. The weighted Kappa values for intraexaminer reliability were 0.885 and 0.934 for FJOA and the degree of LDH, respectively; the Kappa value for intraexaminer reliability was 0.904 for the location of LDH. These data represent good to excellent reproducibility. The FJ angle was measured by the above two doctors, and the mean values were used.
Statistical analysis
Because some patients in our study had multi-segment LDH and the data collected on those segments within one patient are not independent of each other, a mixed-effects ordinal logistic regression model was preferred to determine the potential factors associated with the severity of FJOA. Data analysis was conducted for the left and right sides, separately.
In this model, FJOA was treated as the response or dependent variable. The participant ID was used as random effect. Age, sex, BMI, level, facet tropism, facet orientation and LDH were treated as fixed effects. The difference between bilateral FJ angles can be classified into three types: left > right, right > left and left = right. All the LDH parameters were included in the same model. The cumulative logit function was used for ordinal response. An unstructured covariance structure was used for the statistical modeling. Statistical analysis was carried out using SPSS 23.0 software (SPSS Inc., Chicago, IL, USA); P < 0.05 was considered significant. The data are presented as the mean ± standard deviation.