Computed tomography in axial spondyloarthritis

Abstract

Purpose of review

Computed tomography (CT) is increasingly being used in ankylosing spondylitis (AS) for imaging the spine and sacroiliac joint (SU). We review new insights to diagnosis and evaluation revealed by the use of CT.

Recent findings

Studies using low-dose CT in AS to detect syndesmophytes can image the entire spine, but semiquantitative scoring of the scans by human readers decreases the reliability and validity of this method. The thoracic spine is the segment most involved with syndesmophytes. Syndesmophytes are not randomly distributed around the vertebral rim but have preferred locations, which vary with the vertebral level and may be related to biomechanics. Examination of SU on abdominal CT scans has found structural changes of sacroiliitis in up to 35% of patients with inflammatory bowel disease. The significance of monosodium urate crystal deposition in the pelvis of axial spondyloarthritis patients without coexisting gout is uncertain.

Summary

Low-dose CT is a promising tool in AS. Studies of biomarkers or medications and their relations with syndesmophyte progression should take the thoracic spine into account. Abdominal CT scans are useful for detecting changes related to sacroiliitis.

INTRODUCTION

Imaging plays a central role in the clinical management of ankylosing spondylitis (AS). Ascertaining sacroiliitis visually is essential to the diagnosis of AS [1]. Monitoring structural damage in the spine is also possible only through imaging techniques. Traditionally, these tasks have been performed using plain radiography because of its relatively low radiation exposure, low cost and wide availability. Investigators have increasingly become more concerned with the limitations of radiography. These limitations mainly stem from using a two-dimensional technique to visualize complex three-dimensional structures, with the resulting problems of loss of three-dimensional information and superimposition of extraneous anatomical objects on the same two-dimensional image. Although MRI incurs no radiation exposure and is now recommended for early examination of inflammatory changes in the sacroiliac joint (SIJ), the absence of signal from cortical bone has limited its use for the visualization of structural damage, particularly for syndesmophytes in the spine. To overcome these limitations, investigators have in recent years increasingly taken advantage of the three-dimensional imaging capabilities of computed tomography (CT). We review the most recent developments in the growing role of CT in AS. We first highlight the use of low-dose CT as an alternative to conventional CT, and then review the latest findings made possible by the use of CT in the imaging of the spine and the SIJ.

LOW-DOSE COMPUTED TOMOGRAPHY

Although concerns about radiation exposure have traditionally limited the use of CT in AS, low-dose CT, which has benefitted from substantial advances in scanner technology such as iterative reconstruction, has recently emerged as a promising alternative. De Bruin et al. [2^■] used low-dose CT to scan the entire spine from C2 to S1, providing 23 intervertebral disk spaces (IDSs) for analysis, for an effective dose of only about 4 mSv [3^■■]. In comparison, Tan et al. [4^■], using conventional CT, scanned from T5 to L4 (11 IDSs), incurring an effective dose of about 8 mSv. Using low-dose CT, de Bruin et al. more than doubled the scanning range while halving the radiation dose. However, they also noted that the radiation exposure was still one order of magnitude higher than with radiography, when adding doses from cervical, lumbar and pelvic radiographs.

Diekhoff et al. [5] also used low-dose CT to scan the SIJ in a study where it was used as the gold standard for sacroiliitis, and against which MRI and radiography were compared. Remarkably, in this study, the mean radiation exposure for low-dose CT was similar to the mean radiation exposure for radiography (0.51 mSv compared to 0.52 mSv), whereas radiography was found to vastly underestimate the number of structural SIJ lesions. Further reduction of radiation dose can be obtained by scanning the SIJ obliquely with the gantry suitably tilted [6]. Li et al. reported that oblique scanning of SIJ could be made to avoid the ovaries, and that fewer slices were needed to cover the whole SIJ compared to the usual axial scanning. However, gantry tilt is a feature that may not be available on some recent bulkier dual energy CT scanners.

It should be noted that neither de Bruin et al. nor Diekhoff et al. validated low-dose CT against conventional CT. Traditionally, the trade-off for lower radiation dose has been higher noise, but recent technical improvements such as iterative reconstruction seem to have greatly improved noise reduction. However, without proper experiments, it is unknown if full-dose conventional CT would have higher sensitivity and specificity for lesion detection compared to the current implementation of low-dose CT. Using conventional CT, Tan et al. [7,8] fully quantified syndesmophytes in terms of volume, height and location. In de Bruin et al.’s study, syndesmophytes were assessed by human readers using a semiquantitative scoring system similar to the modified Stoke AS Spinal Score (mSASSS) [2^■]. This approach discards the direct data provided by the CT, and introduces a source of error by using human readers. It is not known if low-dose CT would be as accurate as conventional CT for full quantification of syndesmophytes.

Monitoring of structural damage in AS can best be investigated in the spine through the growth of syndesmophytes. Because of the low sensitivity to change of radiography and mSASSS, at least 2 years are required to detect reliable progression. It has been hoped that CT could help reduce this time interval. De Koning et al. [3^■■] reported the detection of considerably more new and growing syndesmophytes over a 2-year period using low-dose CT compared to radiographs. Unfortunately, data for a 1-year period were not available. Eighty-four percentage of the 37 patients studied exhibited increases in their CT scores over a period of 2 years, compared to 46% who had an increase in mSASSS. However, the proportion of patients with changes beyond the smallest detectable change was similar by low-dose CT semiquantitative score and mSASSS (32 versus 27%, respectively). On the less exacting measure of any syndesmophyte growth, 90–96% of patients had progression by low-dose CT, compared to 56–66% of patients by mSASSS based on two readers. Consensus between the readers was poor, with only 50% agreement on which vertebral levels showed syndesmophyte progression on the low-dose CT scans. Six percentage of patients were excluded from analysis because of poor quality of the low-dose CT scans. In comparison, using conventional CT and fully quantitative measurement, 79% of patients had a syndesmophyte volume increase over 2 years (70% beyond measurement error), and 73% of patients had a syndesmophyte volume increase over 1 year (55% beyond measurement error), and no patient needed to be excluded for technical reasons [7].

COMPUTED TOMOGRAPHY IMAGING OF THE SPINE

Syndesmophytes in the thoracic spine

The thoracic spine is particularly difficult to visualize using plain radiography. Overlying ribs and lungs and smaller disk space make the identification of syndesmophytes challenging, which led to the exclusion of the thoracic spine from the mSASSS. CT, which provides excellent visualization of the entire spine, has opened the thoracic spine for investigation (Fig. 1). Tan et al. [4^■] investigated the lower half of the thoracic spine together with the lumbar spine (T5-T6 to L3-L4). They found more syndesmophytes and bridging in the thoracic than in the lumbar spine. Syndesmophytes were often seen in the thoracic spine even in the absence of lumbar syndesmophytes. These findings have been corroborated by De Bruin et al., who reported that the thoracic spine was the most involved segment. In addition, over the course of 2 years, the lower half of the thoracic spine exhibited the greatest syndesmophyte progression [2^■,3^■■]. These results have important implications for studies of potential biomarkers of spinal fusion or of the effects of medication on syndesmophyte progression, which are currently performed using radiography excluding the thoracic spine as the standard. Relating systemic biomarkers in blood or urine to radiographic changes while omitting the spine region where most abnormalities occur would not be expected to provide valid results.

FIGURE 1.

Patient with a lumbar mSASSS of 0 and syndesmophytes (arrows) at multiple levels in the thoracic spine on CT scan. (a) Lateral lumbar radiograph. (b) Coronal slice. (c) and (d) Sagittal slices.

Three-dimensional localization of syndesmophytes around the vertebral rim

CT allows the precise localization of syndesmophytes around the vertebral body rim. The question of whether syndesmophytes arise randomly around the rim or have preferred locations can therefore be investigated. Tan et al. [8] plotted the mean syndesmophyte distribution around the vertebral rim for vertebral levels from T10–T11 to L3–L4. There was marked preferential involvement of the posterolateral rim for levels from T10–T11to T12–L1, with a more uniform distribution for L1–L2 to L3–L4 (Fig. 2). De Bruin et al. [2^■] also addressed the question of syndesmophyte spatial distribution, although they could only partition the rim into four quadrants (anterior, posterior, left and right). In the cervical spine and superior third of the thoracic spine, ‘anterior’ was the preferred location. In the rest of the thoracic spine, the preferred location became ‘right’; then, both ‘right’ and ‘left’ at the thoracolumbar junction and then ‘anterior’ again in the lumbar spine. The variation of the preferred syndesmophyte location along the spine could reflect variation in mechanical stresses.

FIGURE 2.

Three-dimensional surface reconstructions of a lumbar spine CT scan from one patient showing four consecutive intervertebral disk spaces (from T11–T12 to L2–L3) with syndesmophytes (in white). (a) Lateral view of T11–T12 with bridging syndesmophytes at posterolateral locations. (b) Lateral view of T12–L1 with syndesmophytes at posterolateral locations (bridging on one side). (c) and (d) Anteroposterior view of L1–L2 and L2–L3 with syndesmophytes at posterolateral and anterolateral locations.

Zygapophyseal joint involvement in ankylosing spondylitis

Although visualization of the zygapophyseal joint on radiographs is even more challenging than the visualization of syndesmophytes, CT offers a clear unobstructed view of the posterior elements of vertebrae including the zygapophyseal joint. Using CT, Tan et al. [9] investigated the zygapophyseal joint joint from T10–T11 to L3–L4 in patients with AS, in particular, the relationship between zygapophyseal joint fusion and syndesmophytes. They found that zygapophyseal joint fusion was common, affecting about one-half of patients. It was most often bilateral and occurred more commonly in the thoracic levels than the lumbar levels. Syndesmophytes and even bridging were likely to occur at vertebral levels without zygapophyseal joint fusion, suggesting that abnormal bone formation at the disk space commonly occurs before zygapophyseal joint fusion. Furthermore, only 0.4% of 271 zygapophyseal joints without fusion at baseline developed new fusion over 2 years. Also, syndesmophytes and bridging contributed more to limitations in spinal mobility than zygapophyseal joint fusion. For these reasons, syndesmophytes are likely to remain a more useful imaging marker of spinal fusion progression in AS than zygapophyseal joint fusion.

COMPUTED TOMOGRAPHY IMAGING OF THE SACROILIAC JOINT

Diagnosing sacroiliitis on abdominal computed tomography scans

Some AS patients may have abdominal CT scans for medical reasons. Abdominal CT scans usually have lower resolution than scans performed specifically for the examination of the SIJ. The reconstructed slices are axial rather than oblique, and contrast is often used, all of which may affect the sensitivity of abdominal CT for abnormalities in the SIJ. Two groups investigated the predictive value of such opportunistic scans for screening or detecting SIJ abnormalities. Chan et al. [10] compared specific SIJ abnormalities in a training set of 12 patients with AS and 12 patients without AS who underwent abdominal CT for clinical indications, followed by study of a test set of 34 patients with AS and 34 patients without AS. Ankylosis, number of erosions and depth of juxtaarticular sclerosis were examined for the ability to differentiate AS patients from controls. The definition that had the best discrimination was the presence of either ankylosis or three or more erosions, which had both a specificity and sensitivity for AS of 0.91, and a positive likelihood ratio of 10.3. It is important to note that the median duration of AS among patients in this study was 19 years, so advanced changes in the SIJ were likely. It is unknown if the measure would perform similarly in early patients, for whom screening is more relevant.

Melchior et al. [11,12] compared SIJ findings on abdominal CT to those on SIJ CT in 58 patients who had SIJ CT performed to verify the diagnosis of spondyloarthritis before the start of tumor necrosis factor-α inhibitor treatment. They defined SIJ structural abnormalities on CT analogously to the modified New York radiographic classification, with erosions, joint space narrowing or widening, partial ankylosis or complete ankylosing scored as positive if present on two consecutive slices. Compared to SIJ CT, abdominal CT had a sensitivity of 0.71 and specificity of 1.0, with a positive predictive value of 1.0. These results suggest that abdominal CT is a useful tool to screen for SIJ structural abnormalities.

Prevalence of sacroiliac abnormalities in related diseases using computed tomography

De Kock et al. [13] studied 80 patients with Crohn’s disease who had abdominal CT scans. They compared the presence of features of sacroiliitis to controls without inflammatory bowel disease. Erosions were seen in 35% of patients with Crohn’s disease and inflammatory back pain, 10% of patients with Crohn’s disease but without inflammatory back pain and 0% of 40 control individuals. Prevalences were similar for sclerosis. SIJ ankylosis was seen in 8, 8 and 3% of the patients, respectively.

In an application of their newly developed screening tool, Chan et al. [14^■] studied the prevalence of sacroiliitis on abdominal CT in 233 patients with Crohn’s disease, 83 patients with ulcerative colitis and 108 controls. Sacroiliitis was present in 15, 16.9 and 5.6% of patients in the three groups, respectively. Few of the patients with inflammatory bowel disease and SIJ abnormalities had been referred to a rheumatologist, suggesting that screening could help facilitate appropriate care.

Klang et al. [15] studied 484 young adults with low back pain who had lumbar spine CT scans, mostly to evaluate possible disk disease, on which the entire SIJ was visible. SIJs were scored analogously to the modified New York criteria. Definite sacroiliitis (grade 2 or higher) was found in 3.3% of the patients, whereas suspicious changes were found in an additional 10.2%. SIJs were not commented on in the original radiology reports in most cases, whereas five patients with grade 2 changes were misdiagnosed.

Use of computed tomography in the differential diagnosis of sacroiliac joint abnormalities

Panwar et al. [16] used CT scans to investigate characteristic imaging features that could help differentiate sacroiliitis associated with spondyloarthritis from sacroiliitis associated with gout. Eleven of 52 patients with gout who had SIJ CT had SIJ structural abnormalities. Large multilobular erosions, occasionally associated with soft tissue densities suggestive of tophi, were typical in gout, as was the absence of sclerosis, whereas erosions were smaller and associated with deeper sclerosis in patients with spondyloarthritis.

Normal growth-related changes in the SIJ of children and adolescents may present particular challenges for diagnosing sacroiliitis [17]. Using CT scans of trauma patients younger than age 18 who had no signs of spondyloarthritis, Zejden et al. found 17% of patients had joint facet defects more than 3 mm, which could simulate erosions. They also noted ossified nuclei in the SIJ spaces or juxta-articular regions in 77% of patients age 13 or older. Because ossified nuclei are accompanied by bone turnover, these could mimic inflammation on MRI.

Monosodium urate crystal deposits in axial spondyloarthritis patients

Using dual energy CT, Zhu et al. [18^■] were able to observe depositions of monosodium urate (MSU) crystal in the SIJ of 30% of patients with axial spondyloarthritis without coexisting gout. Unfortunately, no age-matched control group was scanned to determine the normal level of MSU crystal deposition in healthy individuals. Higher MSU volumes were significantly associated with a higher radiographic grade of the SIJ, after adjustment for duration of spondyloarthritis. As this was a cross-sectional study, it is not clear if MSU deposition may promote structural changes in the SIJ in patients with AS.

CONCLUSION

Advances in scanner technology have made lowdose CT a promising tool for the visualization of the spine and SIJ in AS. CT has shown that more extensive syndesmophytes and more syndesmophyte growth occur in the thoracic spine than in the cervical or lumbar spine. CT allows the localization of syndesmophytes to be studied and offers a clear unobstructed view of the zygapophyseal joint. The nonrandom localization of syndesmophytes around the vertebral rim could suggest a role for mechanical factors in their formation. Abdominal CT scans may be useful to detect structural abnormalities related to sacroiliitis in patients at risk. Longitudinal studies are needed to determine the potential role of SIJ CT in measuring the progression of structural damage at this site.

KEY POINTS.

• Low-dose CT is a promising tool for the study of AS.

• More existing and more growing syndesmophytes have been observed in the thoracic than the cervical or lumbar spine.

• Syndesmophytes do not arise at random locations around the vertebral rim but have preferred locations.

• Existing abdominal CT scans should be used to diagnose sacroiliitis in patients at risk.

• CT is the modality of choice for visualizing the finer bone structure of the SIJ.