When an incidental MRI finding becomes a clinical issue

This study reviewed the medical records of lumbar spine MR examinations of 581 adult patients, performed within a 3-year period between January 2016 and December 2018. Initially, these patients were referred to the tertiary referral center to undergo an MRI work-up because of non-radicular low back pain. Inclusion criteria were: a) patients with low-grade disc degeneration, grade 2 or 3 according to the Pfirrmann classification [12], and without any other abnormalities, b) patients with combined Modic vertebral endplate degeneration (Modic types 1–3) and facet joint osteoarthitis, including possible additional disc degeneration and herniation according to the lumbar disc nomenclature of the North American Spine Society [12‐15], c) patients with sero-negative, HLA B27-positive axial spondyloarthritis according to the Assessment of SpondyloArthritis international Society criteria, including lumbar involvement with or without sacroiliitis, as well as possible additional disc and vertrebral degeneration [16] and d) patients with bacterial, histologically proven, non-tuberculous spondylodiscitis, including possible additional disc and vertebral degeneration. Exclusion criteria included: vertebral fractures, spondylolysis, severe scoliosis and congenital vertebral abnormalities, Baastrup syndrome with interspinous fluid collection, edema of the paraspinal musculature or myositis, previous spinal trauma, previous spinal surgery or interventional therapy, spondylodiscitis with involvement of the posterior vertebral elements or with abscess formation, pregnancy and medical conditions, such as tumor diseases, venous thrombosis, cardiac, renal, hepatic and skin diseases. These stringent criteria were chosen to exclude potential sources of edema in the paravertebral soft tissues, changes associated with edema of the cutis into the subcutaneous fat tissue, and internal disorders that may produce local or generalized edema, anasarca, ascites or pulmonary congestion.

Based on the aforementioned criteria a total number of 279 study individuals were included and 302 patients were excluded. Of the 279 patients there were 79 patients with low-grade disc degeneration, 101 with combined Modic degeneration and facet joint degeneration, 53 with axial spondyloarthritis and 46 with infectious spondylodiscitis.

All examinations of the lumbar spine were performed on a 3.0 T MR unit (Philips Achieva; Philips Medical Systems, Best, the Netherlands) using a 16-channel spine coil. All MR scans were performed with the patient in the supine position and head first on the scanner table. Sagittal MR images were oriented along the posterior edge of the vertebral bodies and covered the lumbar spine in a craniocaudal direction from the upper endplate of vertebral body Th12 to at least vertebral body S2. In the dorsoventral direction, the lumbar spine was visualized, including at least 5 cm of antevertebral soft tissue and the skin of the dorsum, and in the lateral direction, the lumbar spine was visualized including the transverse process on both sides. The standard MR protocol included the following sequence relevant to this study: a sagittal short tau inversion recovery (STIR) sequence (repetition time 4250–7000 ms, echo time 60–80 ms, flip angle 90°, slice thickness 3 mm, matrix 528 × 528 mm, duration ca. 3:58 min).

All the MR examinations were anonymized and randomly presented on a picture archiving and communication system (IMPAX, Agfa HealthCare GmbH, Bonn, Germany) to an MRI specialist (S. F. N., 16 years experience in MRI), who was not aware of any patient data. For the assessment of the interreader agreement, a radiological fellow (U. S. N., 6 years experience in MRI) evaluated the same images in an independent reading session, who was also unaware of any patient data. For the assessment of the intrareader agreement, the image data were re-evaluated by the MRI specialist 4 weeks after the first evaluation.

The whole image stack of the sagittal STIR sequence was used to confirm or to rule out the presence of PLSE, which was defined as diffuse STIR-hyperintense signal changes within the deep posterior lumbar subcutaneous tissue that did not reach the superficial fat layers [1, 2, 4‐11]. PLSE was also defined by its location along the fascia plane next to the thoracolumbar fascia and posterior to the spinous process, at the midline, and extending laterally [1, 2, 4‐11]. Any edema was graded according to its craniocaudal extension with respect to the vertebral bodies from 1–5 (i.e., from one to five vertebral levels), and was consequently classified as monosegmental, one vertebral level, or multisegmental, two or more vertebral levels (Fig. 1).

Statistical analysis was performed using software (IBM SPSS Statistics 25.0 for Windows; SPSS, Armonk, NY, USA). A P-value equal to or less than 0.05 was considered to indicate a significant difference. First, for each patient group the mean value, standard deviation, and range of the patients’ age and BMI, defined by the formula weight (kg)/height (m²), were recorded, and the differences were tested for statistical significance using a one-way analysis of variance. These demographic data were extracted from the patients’ MRI safety questionnaire. Second, the number and the percentage of female and male patients were recorded, and the differences were tested for statistical significance using a χ²-test. Third, for each patient group the number and percentage of obese (defined as BMI ≥30 kg/m²) and nonobese patients, with and without PLSE, were recorded using a cross-tabulation, and the overall differences were tested for statistical significance using Fisher’s exact test. Fourth, for each patient group a multiple logistic regression analysis was performed to test the relationship between PLSE and the patients’ BMI, including age and sex, for statistical significance. Fifth, an ordinal regression analysis (logit model) was performed to test the relationship between the PLSE extension and the patients’ BMI, including age and sex for statistical significance. Finally, the intrareader and interreader agreement for the assessment of PLSE was assessed using weighted κ statistics.

The study population consisted of a total number of 279 individuals, including 124 females and 155 males (age range 18–82 years, mean age 49.8 years, mean weight 78.1 kg, weight range 41–125 kg). Table 1 gives details for each patient group, the patients’ age and BMI as well as the gender distribution. Statistically significant differences for the patients’ age (P < 0.001), BMI (P = 0.001), and sex (P = 0.002) were identified. Table 2 details the specific MRI findings for each patient group. Of the 79 patients with disc degeneration, 13 (16.5%) had monosegmental, and 66 (83.5%) multisegmental changes. Of the 101 patients with combined Modic and facet joint degeneration, 5 (5.9%) had monosegmental, and 96 (94.1%) multisegmental changes. Of the 53 patients with axial spondyloarthritis, 18 (34.0%) had monosegmental, and 35 (66.0%) multisegmental changes. Of the 46 patients with infectious spondylodiscitis, 43 (93.5%) had monosegmental, and 3 (6.5%) multisegmental changes.

Overall, PLSE was found in 83/279 (29.7%) patients and was not depicted in 196/279 (70.3%) patients (Fig. 2 and 3). In detail, PLSE was found in 11/79 (13.9%) patients with low-grade disc degeneration, in 37/101 (36.6%) with combined Modic and facet joint degeneration, in 7/53 (13.2%) with axial spondyloarthritis, and in 28/46 (60.9%) with infectious spondylodiscitis (Table 3). Furthermore, PLSE was found in 21.1% of nonobese and in 72.3% of obese patients (P = 0.000) (Table 3). Of a total of 83 patients with PLSE, only 4 (4.8%) patients demonstrated a monosegmental PLSE extension. The vast majority, 79/83 patients (95.2%), demonstrated a multisegmental PLSE extension as follows: 10/11 (90.9%) in disc degeneration, 34/37 (91.9%) in Modic and facet joint degeneration, 7/7 (100%) in axial spondyloarthritis and 28/28 (100%) in infectious spondylodiscitis. For each group, there was a statistically significant relationship between PLSE and the BMI (P = 0.000–P = 0.031), except for infectious spondylodiscitis (P = 0.054) (Table 4). For each group, there was a statistically significant relationship between the PLSE extension and the patients’ BMI (P = 0.000–P = 0.049) (Table 4). There was no statistically significant relationship between PLSE and its extension and the patients’ age and sex, except for the patients’ age in combined Modic and facet joint degeneration (P = 0.013) (Table 4).

For the assessment of PLSE, there was excellent intrareader (weighted κ = 1) and interreader agreement (weighted κ = 1).