Introductıon

Baastrup’s disease, known as ‘kissing spines’, is characterized by the close approximation and contact of the adjacent spinous processes, leading to edema, cystic changes, sclerosis, flattening, and enlarged articulating bone surfaces. This condition also involves the formation of an adventitious bursa in the interspinous region and, occasionally, epidural cysts or midline fibrotic masses [1, 2]. It typically affects the lumbar spine, usually involving only one level and predominantly manifesting at the L4–L5 level with additional degenerative changes. It tends to occur more frequently in individuals over 70, with no specific gender predilection [3, 4]. Baastrup’s disease is regarded as a pathological syndrome of clinical relevance related to an increased degree of lordosis and repetitive strain on the interspinous ligaments, involving the formation of nociceptors leading to pain or as a component in the spectrum of the anticipated degenerative changes of the spine associated with advancing age [5]. While the clinical significance of Baastrup’s disease remains ambiguous, and controversy persists regarding its classification as a nociceptor, symptomatic patients report midline pain specifically at the level of the pathological interspinous ligament, which is aggravated during extension and relieved during flexion [6]. The treatment of symptomatic Baastrup’s disease generally includes a conservative approach utilizing analgesics and non-steroidal anti-inflammatory drugs, as well as percutaneous infiltrations involving corticosteroids and local anesthetics as an alternative method [4, 7, 8]. Moreover, surgical interventions like bursa excision or osteotomy are also regarded as therapeutic measures [7, 9].

While X-rays provide essential clues in the diagnosis, computed tomography (CT) is highly effective in revealing the close approximation and contact of adjacent spinous processes, sclerosis, flattening and enlargement of the articulating surfaces, and accompanying degenerative changes. Magnetic Resonance Imaging (MRI), however, stands as the preferred and superior imaging method for identifying lumbar interspinous bursitis, which is characterized by a fluid signal between opposing spinous processes, as well as bone, soft tissue, and interspinous ligament edema at the affected level. Furthermore, MRI can depict contrast enhancement around the soft tissues, indicating active disease, inflammation, and associated findings such as epidural cysts and midline fibrotic masses.

In the literature, few studies have explored the prevalence of Baastrup’s disease in diverse populations, including young athletes and heavy automotive vehicle drivers, and have reported varying values ranging between 6.3 and 81.3% for different age groups [3, 10, 11]. Notably, the singular study that employed MRI to investigate the prevalence and its association with degenerative changes lacked the inclusion of the short tau inversion recovery (STIR) sequence, which is highly sensitive to fluid and edema, in the standard protocol [6]. Our study aims to establish the precise prevalence of Baastrup’s disease and its association with degenerative-related findings using MRI, making it the first to incorporate the STIR sequence into conventional pulse sequences. We also aimed to explore the impact of varying levels of expertise and the degree of interreader agreement among radiologists with different levels of experience in diagnosing Baastrup’s disease using MRI. This aspect has yet to be studied in any previous literature.

Materıals and methods

Study design and patient population

Local ethics committee approval was obtained for this retrospective study, and the requirement for informed consent was waived (Approval number: 60116787-020-415581). A search within the radiology information system was carried out to identify routine unenhanced lumbar spine MRI scans performed between January 2021 and January 2022, and the study included 375 patients (73% females, 27% males; age range 10–88 years old; mean, 48; median, 50) who had undergone these examinations due to back or leg pain (lumbago and sciatica). Patients with rheumatologic diseases (such as ankylosing spondylitis and rheumatoid arthritis), benign or malignant lesions (i.e., metastases, plasmacytoma, schwannoma), spondylodiscitis, recent trauma history with acute or chronic fractures, or prior spinal surgery were excluded. Contrast-enhanced MR examinations were also not included. Furthermore, MR images were assessed by a neuroradiologist with 28 years of experience, and inadequate examinations that included only some pulse sequences, including the STIR sequence, or met the minimum technical requirements for appropriate evaluation were excluded.

Image acquisition and MRI protocol

All MRI scans were conducted using a 1.5-T scanner (Ingenia; Philips Healthcare) equipped with a dStream Posterior coil. The scanner had a gradient power of 45 mT/m on each axis and a maximum slew rate of 200 mT/m/sec. All MRI scans comprised the following pulse sequences: T1-weighted sagittal fast spin echo (FSE) without fat suppression, T2-weighted sagittal FSE without fat suppression, T2-weighted axial FSE without fat suppression parallel to the disc, and sagittal short tau inversion recovery (STIR). The pulse sequence parameters are summarized in Table 1.

Table 1 Spinal MRI sequence parameters
Full size table

Image analysis

A fourth-year radiology resident who completed training in both neuroradiology and musculoskeletal radiology and a radiologist with 28 years of experience in neuroradiology independently reviewed spine MR images on a PACS (Picture Archiving and Communications System) workstation using a structured report form. At every intervertebral lumbar disc level, the assessment of MR images included Baastrup’s disease, abnormalities in disc contour (bulge and herniation), disc degeneration (spondylosis), alterations in endplate marrow signal (Modic changes), osteoarthritis of the facet joints, central canal stenosis, and sagittal displacement (anterolisthesis, retrolisthesis).

The diagnosis of Baastrup’s disease was established when any of the following features, namely lumbar interspinous bursitis with fluid-like signals observed between consecutive spinous processes, hyperintense signal changes of spinous processes, or hyperintensity of interspinous ligaments on STIR images, were identified (Fig. 1).

Fig. 1
figure 1

Sagittal STIR sequence images of three distinct patients who underwent MRI examination for lumbago. In the leftmost image a flattening, enlargement, and bone marrow edema of spinous processes at L2–L5 levels, with the most pronounced changes observed at the L4 level (blue arrow), are evident. The middle image b shows hyperintense signals in the interspinous ligament of the L4–5 level (white arrow), indicating soft tissue edema. The STIR image of the third patient c reveals a fluid signal in the interspinous region at the L4–L5 level (red arrow), consistent with bursitis

Full size image

Disc contour abnormalities were categorized as either bulge or herniation, following definitions outlined in the multidisciplinary standard nomenclature [12]. Bulges were characterized by diffuse contour abnormalities covering 50% or more (180°) of the disc circumference, while herniations were identified as focal contour irregularities encompassing less than 50% or 180° of the disc circumference. Detailed subtype designations were not employed for this context, including protrusion, extrusion, and sequestrated disc. In disc evaluation, no disc height loss was considered normal, and disc height decrease was considered grade 3 and 4 disc degeneration according to the Pfirrmann classification [13]. The endplate bone marrow signal abnormalities were assessed according to Modic’s classification [14]. Subtype designations were intentionally omitted in this assessment. Instead of categorizing endplate marrow signal alterations based on superior versus inferior endplate involvement, they were differentiated by disc level. Facet joints were evaluated and categorized as normal or degenerated [15]. The scoring of left and right facet joints was conducted collectively, with the rating primarily influenced by the more severe findings. Normal facet joints exhibited a uniform joint space ranging from 2 to 4 mm, devoid of osteophytosis or subchondral bone reaction. Facet osteoarthritis was documented if any of the following, or a combination thereof, were observed: osteophytes, narrowing of the joint space, hypertrophy of the articular process, subarticular bone edema, or subchondral cysts. Stenosis of the central canal was characterized by an anteroposterior dimension measuring less than 10 mm on axial images aligned parallel to the intervertebral disc. Sagittal images were utilized to detect anterolisthesis and retrolisthesis, and their presence or absence was recorded as a binary value.

Statistical analysis

Findings specific to each level (including Baastrup’s disease, disc contour abnormalities, disc degeneration, endplate signal abnormalities, facet osteoarthritis, central canal stenosis, retrolisthesis, and anterolisthesis) were combined at the respective levels for comprehensive analysis. The analysis unit was based on each level for level-specific findings and on each patient for generalized features. Statistical analyses were performed using SPSS (Statistical Package for the Social Sciences) software version 25.0 (SPSS Inc). Kolmogorov–Smirnov and Shapiro–Wilk tests examined the suitability of data to normal distribution. The presence of Baastrup’s disease was assessed with MRI findings through the use of Χ2 analysis for dichotomous variables, including bulge, herniation, disc degeneration, endplate signal changes, facet osteoarthritis, central canal stenosis, anterolisthesis, and retrolisthesis. The Kappa coefficient was used to determine the interreader agreement. Kappa value ranges were accepted and interpreted: less than 0.20 as absent; 0.21–0.40 as minimal agreement; 0.41–0.60 as moderate agreement; 0.61–0.80 as substantial agreement; and 0.81–1 as perfect agreement. All statistical tests were calculated two-sided, and a P value lower than 0.05 was accepted as statistically significant.

Results

Table 2 presents the frequency distributions of MRI findings across 1875 intervertebral levels in the cohort of 375 patients. Baastrup’s disease, evaluated with the STIR sequence, was identified in 296 levels, corresponding to 37.6% (141 out of 375) of the patients. In cases where Baastrup’s disease was observed, it manifested at various levels in 71.6% (101 out of 141) instances. Among these, the majority (59.4% or 60 out of 101) exhibited involvement at two levels, while the minority (40.6% or 41 out of 101) showed Baastrup’s disease across three or more levels. In the remaining 40 patients, Baastrup’s disease was determined to be at a single level.

Table 2 Distribution of lumbar spinal MRI findings
Full size table

Disc bulge was observed at 717 levels in 65.3% (245 out of 375) of the patients. Disc herniation was noted in 174 levels, representing 41.9% (157 out of 375) of the patient cohort. Disc degeneration manifested in 143 levels, accounting for 25.6% (96 out of 375) of the patients. Endplate signal changes (Modic alterations) in the endplate were identified at 565 levels, encompassing 54.9% (206 out of 375) of the patient cohort. Facet osteoarthritis manifested in 402 levels, accounting for 41.6% (156 out of 375) of the patients. Central canal stenosis was observed in 188 levels, affecting 25.1% (94 out of 375) of the patients. Anterolisthesis was present in 56 levels, accounting for 13.9% (52 out of 375) of the patient cohort. Similarly, retrolisthesis was identified in 42 levels, representing 9.1% (34 out of 375) of the patients.

Baastrup’s disease was predominantly observed at the L4–L5 level (35.5% or 105 out of 296), followed by occurrences at the L5–S1 (29.4% or 87 out of 296) and L3–L4 levels (24.7% or 73 out of 296). The remaining 31 levels of Baastrup’s disease consist of 26 out of 296 (8.8%) and 5 out of 296 (1.7%) involvements at the L2–L3 and L1–L2 levels, respectively. Additionally, spinal stenosis, anterolisthesis, and retrolisthesis were most frequently identified at the L4–L5 level, while bulging, herniation, disc degeneration, endplate signal intensity changes (Modic), and facet osteoarthritis exhibited a higher prevalence at the L5–S1 level.

Out of the 141 patients assessed, representing 37.8% of the total, manifestations of Baastrup disease were observed, affecting at least one spinal level. The prevalence of kissing spines exhibited a noticeable rise with advancing age, escalating decade by decade and reaching its peak at 87.5% in individuals aged 80 years and above. For a detailed breakdown of Baastrup’s disease frequency for each decade, please refer to Fig. 2

Fig. 2
figure 2

Frequency of Baastrup’s disease by decade

Full size image

Baastrup’s disease demonstrated a statistically significant correlation with disc bulging (P = 0.0012), disc herniation (P = 0.0033), disc degeneration (P = 0.0013), endplate signal intensity changes (P = 0.034), facet osteoarthritis (P = 0.0041), central canal stenosis (P = 0.005), and anterolisthesis (P = 0.0049). However, no statistically significant associations were observed between Baastrup’s disease and gender (P = 0.468) and retrolisthesis (P = 0.167).

In the conclusive Baastrup diagnoses, relying on the STIR sequence, there was complete agreement between the radiology resident and the neuroradiologist, boasting 28 years of experience. The Kappa value was determined to be 0.87.

Discussıon

Our study showed that Baastrup’s disease, diagnosed on the STIR sequence, was present in 37.8% of the 375 MRI examinations. Associations with degenerative changes, such as bulging, herniation, disc degeneration, Modic changes, facet osteoarthritis, spinal stenosis, and anterolisthesis, were also noticed, while no correlation with gender or retrolisthesis was found. Furthermore, an increase in frequency associated with age was observed, reaching 87.5% among patients aged 80 years and older. The L4–L5 level exhibited the highest prevalence of involvement. In instances where Baastrup’s disease was evident, it was more commonly distributed across multiple levels rather than confined to a single level, in contrast to the patterns observed in some published studies in the literature [3, 6]. Finally, in diagnosing Baastrup’s disease, complete agreement was observed among radiologists, regardless of their experience level.

As mentioned earlier, apart from some case reports, only a few studies have addressed the prevalence of Baastrup’s disease, with varying percentages ranging from 6.3 to 81.3% in different age groups and populations [3, 6, 10, 11]. In the literature, Maes and colleagues stand as the sole group to utilize MRI exclusively for diagnosing Baastrup’s disease, reporting a prevalence of 8.2%, a figure notably lower compared to the present study, which also examines the association of degenerative changes with Baastrup’s disease [6]. In our study, an association with degenerative changes was similarly observed; however, it is believed that the utilization of the STIR sequence, known for its heightened sensitivity to fluid compared to conventional MRI sequences, has significantly enhanced diagnostic sensitivity in detecting Baastrup’s disease. Furthermore, our study employed a more inclusive criterion for diagnosing Baastrup’s disease, including the assessment of interspinous bursitis. In contrast, Maes et al. [6] focused on visualizing the bursa for disease identification. Finally, Maes et al. [6] did not include information on the age distribution of the patients, making a direct comparison of figures difficult.

Baastrup’s disease is believed to emerge due to repetitive strain on the interspinous ligaments, leading to their degeneration and collapse, ultimately permitting contact between opposing spinous processes. As observed in our study, a higher incidence of Baastrup’s disease in older age group and a strong association with other degenerative changes, such as loss of disc height and spondylolisthesis, which can contribute to the approximation of spinous processes and absence of characteristic midline pain specifically at the level of the pathological interspinous ligament in most of patients may suggest Baastrup’s disease as a part of degenerative changes related to aging rather than a distinct clinical entity [5]. Furthermore, MRI is often considered oversensitive to detect findings related to Baastrup’s disease. As a result, patient’s pain symptoms may not consistently align with a diagnosis of Baastrup’s, nor does a Baastrup’s diagnosis always explain patient’s pain symptoms [6]. For all these reasons, including the poor surgical outcomes published in a few previous studies, Kwong et al. and some other authors propose renaming this entity as the Baastrup phenomenon instead of referring to it as Baastrup’s disease. [3, 16,17,18]. However, the utilization of CT in the study of Kwong et al. which has limited soft tissue resolution leading to impaired assessment of some degenerative changes like disc degeneration and an inability to visualize interspinous bursitis, makes this suggestion questionable. Furthermore, despite its low incidence, the presence of Baastrup’s disease in patients without degenerative changes, as also discussed in the same study, and some case reports showing surgical decompression as a beneficial therapeutic option, suggest that this proposition warrants additional studies [7, 19, 20]. All in all, we believe that although the results of our study suggest Baastrup’s disease as a clinical entity in the spectrum of degenerative processes, further research is needed to elucidate the clinical significance and exact role of Baastrup’s disease as an independent nociceptor.

To our knowledge, this is the first and only study examining interobserver agreement in diagnosing Baastrup’s disease, and we observed complete agreement between readers, regardless of their experience. The STIR sequence, which enables even less experienced readers to identify Baastrup’s disease-related changes, may have significantly contributed to this high level of agreement.

Lastly, it is noteworthy that radiologists often emphasize anterior and middle column pathologies while overlooking diseases in the posterior column, including Baastrup’s disease. An inexperienced radiologist unfamiliar with this condition might mistakenly diagnose interspinous bursitis as infection or neoplasm or misinterpret ligament and bone edema as manifestations of rheumatologic diseases. Furthermore, someone who is unaware of the strong association of Baastrup’s disease with degenerative processes may excessively mention related findings, potentially leading to misguided clinical management of the patients. Therefore, awareness of imaging findings of Baastrup’s disease in all imaging modalities is essential for the correct diagnosis of this disease.

There are some limitations to our study. First, this was a retrospective study, and it was performed in a single center using a single MRI scanner. Additionally, our study population consisted of a symptomatic patient group with lumbago and sciatica, needing a control group, making it unfeasible to pinpoint the exact prevalence of Baastrup’s disease in the general population. Lordosis believed to be a potential factor or contributor to Baastrup’s disease, was not intentionally assessed due to the supine and non-weight-bearing position during the MRI examination, which could result in a reduction of lordosis. Future studies should incorporate lumbar radiographs conducted in a standing position to evaluate lordosis’s association with Baastrup’s disease thoroughly. Lastly, the absence of a reference standard for confirming interspinous bursitis and the need for detailed characteristics regarding patient’s low back pain complicate the evaluation and relation of MRI findings with the correct clinical context. The association between Baastrup’s disease and clinical findings, particularly pain, remains unclear, warranting caution when considering it as the primary cause of back pain, especially in older patients with spinal degenerative changes.

Conclusion

In summary, among individuals experiencing symptoms of lumbago and sciatica, findings related to Baastrup’s disease are prevalent rather than rare, exhibiting a gradual increase with advanced age on lumbar MRI, particularly noticeable on the STIR sequence. Additionally, Baastrup’s disease, identified through MRI, is linked to the manifestation of degenerative changes. Finally, using the STIR sequence facilitates a consensus among radiologists, regardless of their experience level.