SAPHO Syndrome: Imaging Findings of Vertebral Involvement

BACKGROUND AND PURPOSE: Imaging findings in patients with a combination of synovitis, acne, pustulosis, hyperostosis, and osteitis (SAPHO) are often misinterpreted as discitis/osteomyelitis or metastases, resulting in multiple biopsies and delayed diagnosis. We have incidentally noted a semicircular morphology in vertebral body imaging in several cases of SAPHO syndrome with vertebral involvement. Our goal was to evaluate the prevalence of this distinctive morphology in these patients. MATERIALS AND METHODS: A retrospective review of patients with SAPHO syndrome diagnosed between July 1998 and August 2013 was conducted. A descriptive analysis of MR imaging, CT, radiography, bone scanning, and PET imaging was performed for the presence and distribution of vertebral body signal intensity or attenuation changes and/or enhancement; contiguous vertebral body involvement; vertebral body collapse; endplate irregularity; disc space, facet, and spinous process involvement; subligamentous thickening; and paraspinal soft-tissue involvement. RESULTS: Eighteen patients (16 women [89%]; mean age, 52.9 years) with SAPHO and spine involvement were included. Contiguous involvement of ≥2 vertebral bodies was found in 16 patients (89%), with a curvilinear or “semicircular” pattern involving portions of adjacent vertebral bodies in 10 (63%, P = .14). Most intervertebral discs demonstrated absence of abnormal T2 hyperintensity (73%) and enhancement (89%). Subligamentous thickening was present in 12 (67%). Paraspinal soft-tissue involvement was present in 6 (33%). CONCLUSIONS: SAPHO syndrome should be included in the differential diagnosis in a patient with a curvilinear or semicircular pattern of vertebral involvement, contiguous vertebral body involvement, and absence of intervertebral disc edema and enhancement.

T he association of bone disease and chronic cutaneous pustular lesions has been observed since the 1960s, but it was not until 1987 that Chamot et al 1 first used the acronym SAPHO to describe this rare group of chronic, relapsing, inflammatory osteoarticular disorders commonly associated with skin manifestations. SAPHO was proposed to refer to a combination of synovitis, acne, pustulosis, hyperostosis, and osteitis as a heading for these syndromes. Cutaneous lesions are characterized by palmoplantar pustulosis, acne conglobata, and/or hidradenitis suppurativa. A wide variety of bone and joint manifestations has been described.
SAPHO syndrome shares some overlapping features with other spondyloarthropathies, including reactive arthritis, psoriatic arthritis, spondyloarthropathy associated with inflammatory bowel disease, and idiopathic ankylosing spondylitis. In children, the disease most commonly presents as a recurrent multifocal osteomyelitis, favoring the long bone metaphysis. This presentation is in contradistinction to that in adults in whom the anterior chest wall, including the sternoclavicular and manubriosternal junctions, is most commonly affected. [1][2][3][4] The spine is also frequently involved in adults. Findings of spinal involvement on conventional radiographic imaging have been reported to include vertebral body osteosclerosis, paravertebral ligament ossification, hyperostosis, and discovertebral junction lesions. Although plain radiographic findings of vertebral involvement have been welldescribed, the radiology literature has only a limited number of case reports and small case series describing imaging findings in detail with other modalities, specifically MR imaging. [5][6][7][8][9][10] These advanced imaging findings are often misinterpreted as discitis/ osteomyelitis or metastases. 11,12 In our experience, this misinter-pretation can lead to unnecessary biopsies, other invasive procedures, and, ultimately, delayed diagnosis.
Anecdotally, we have commonly observed a "semicircular" pattern of contiguous vertebral body involvement localized to either the anterior or posterior vertebral bodies of the middle segments and adjacent anterior or posterior endplates of the surrounding vertebral bodies. This was first reported by Peffers et al. 13 The goal of the study was to evaluate the imaging findings in a series of patients with vertebral involvement as part of SAPHO syndrome, with specific attention paid to the prevalence of this unique semicircular morphology in the vertebral bodies.

Patients
Institutional review board approval was obtained, and informed consent was waived for this retrospective review. This study complied with all Health Insurance Portability and Accountability Act requirements. A search of clinical notes and the radiology information management system for patients with the possible diagnosis of SAPHO or any combination of the terms "synovitis," "acne," "pustulosis," "hyperostosis," and "osteitis" between July 1998 and August 2013 identified 66 patients (Fig 1). Of these, 36 patients had a final clinical diagnosis of SAPHO meeting at least 1 of the 4 criteria proposed by Benhamou et al 14 : 1) osteoarticular manifestations of acne conglobata, acne fulminans, or hidradenitis suppurativa; 2) osteoarticular manifestations of palmar plantar pustulosis; 3) hyperostosis involving either the anterior chest wall, spine, or limbs with or without dermatosis; and 4) chronic recurrent multifocal osteomyelitis with or without dermatosis. Eighteen patients were further excluded because they did not have imaging of the spine.

Imaging and Analysis
We retrospectively reviewed the radiographic imaging findings, including MR imaging, CT, bone scintigraphy, and PET/CT. The images were evaluated by 2 experienced neuroradiologists in consensus (A.L.K. with 15 years, J.M.M. with 7 years). Biopsies, including location, number, and histopathologic and microbiologic culture results, were also recorded.
MR imaging was evaluated for vertebral body signal-intensity changes and/or enhancement; contiguous vertebral body involvement; the presence of vertebral body collapse; endplate irregularity; disc space involvement including enhancement; facet and spinous process involvement; subligamentous thickening; and paraspinal soft-tissue involvement. MR imaging sequences, imag-ing parameters, and magnet strength varied because many of the scans came from outside facilities. Vertebral body signal intensity was evaluated on T1and T2-weighted MR images and on gadolinium-enhanced T1-weighted images when available. The semicircular pattern of enhancement was recorded if present. The intervertebral discs were evaluated for enhancement, irregularity, associated fluid, and narrowing or widening. The number and laterality of the facet joints with signal abnormalities and/or enhancement and the presence of spinous process signal changes were recorded. We looked for ligamentous thickening of the anterior or posterior longitudinal ligaments and ligamenta flava, including thickening that extended to uninvolved vertebral bodies. When there was masslike soft-tissue involvement, the location, maximal diameter (in millimeters), and longitudinal extent of the abnormal tissues were recorded.
CT was evaluated for vertebral body sclerosis or lysis. Radiotracer uptake, if present, was recorded for bone scintigraphy and PET/CT examinations.

Statistical Analysis
Statistical analysis was performed by using the JMP software package (Version 9.0; SAS Institute, Cary, North Carolina). The Fisher exact test was used to determine the association between the presence of vertebral involvement and sex, the semicircular pattern of involvement, disc space narrowing, and endplate irregularities. Only 2-tailed tests were used. A P value of Ͻ .05 was significant.

Vertebral Lesions
A total of 104 vertebral bodies were involved in 18 patients (Table). One patient had a single level of involvement (6%), while the remaining 17 patients had a median of 5 vertebral levels involved (range, 2-13 levels; mean, 5.8 Ϯ 3.4 levels). The thoracic spine was most commonly involved (n ϭ 14, 78%), with Ն2 separate thoracic segments involved in 57% of patients (n ϭ 8/14). The lumbosacral spine was the next most commonly involved (n ϭ 7, 39%), followed by the cervical spine (n ϭ 3, 17%). Multiple regions of the spine were involved in 7 patients (39%). Contiguous involvement of Ն2 vertebral bodies was found in 16 patients (89%), with a curvilinear or semicircular pattern involving the anterior or posterior portions of adjacent vertebral bodies in 10 of these patients (63%, P ϭ .14) (Fig 2A, -B). MR imaging was available for 16 patients (89%). Many patients came to our institution with imaging from other facilities. Therefore, there was variation in the MR imaging field strength, sequences, and imaging parameters. T1-weighted images were available in all cases; 1 patient (6%) did not have T2-weighted imaging. Vertebral body low-signal intensity on T1-weighted images was seen in 15/16 patients (93%) with corresponding highsignal intensity on T2-weighted images in 12 patients (75%), compatible with bone marrow edema (Fig 2A, -B). One patient (6%) had mixed hypo-/hyperintensity on T1-weighted images, while 3 patients (18%) had mixed hypo-/hyperintensity on T2weighted images, suggestive of concomitant bone marrow edema and cancellous bone sclerosis. Gadolinium-based contrast was administered in 9 patients in whom MR imaging was available (56%), with 100% of patients demonstrating enhancement in the same distribution as the signal changes described above. Vertebral body corner erosions were not observed in any patient.
Corresponding noncontrast CT imaging was available in 14 of the 16 patients in whom MR imaging was available. CT demonstrated sclerosis in the areas of MR signal abnormality in 100% of patients, regardless of the pattern of the MR imaging signal changes (Fig 2C). No lytic lesions were identified.
Nine patients had technetium Tc99m methylene diphosphonate 3-phase wholebody bone scans to correlate with the spinal MR imaging. All 9 patients demonstrated increased radiotracer on MR imaging in the vertebral bodies affected (Fig 2D). There were no discordant areas of vertebral body uptake between the 2 modalities. Extravertebral findings on bone scans are described in further detail below.
In 2 of the 18 (11%) patients with spinal involvement who did not have an MR imaging, the number, level, and contiguity of vertebral lesions were determined solely by nuclear medicine bone scanning in 1 patient and with lumbar spine radiographs in the second. In the first patient, the bone scan demonstrated radiotracer uptake in the noncontiguous T5, T9, T11, and L1 vertebral bodies and right sacroiliac joint. In the second, lumbar spine radiographs demonstrated attenuated vertebral body sclerosis from L3 to the sacrum and partial fusion of the left sacroiliac joint.
Thirteen of 18 patients (72%) had plain radiographs. One patient had normal findings on plain radiographic imaging (8%). Twelve patients had vertebral body sclerotic changes, including shiny corners (92%). Despite abnormal findings on MR imaging, no abnormal FDG uptake was seen in the 2 patients with whole-body FDG-PET/CT.

Disc Spaces
MR imaging demonstrated at least 1 intervening or adjacent disc space that was narrowed in 8 of 16 (50%) patients, with irregular endplates observed in 7 patients (43%). The combination of disc space narrowing with endplate irregularity was noted in 5 patients who had MR imaging (62%; P ϭ .11). High signal on T2-weighted images in the disc was seen in 5 patients (31%). Disc space enhancement on gadolinium-enhanced T1-weighted images was observed in 2 of 9 patients receiving gadolinium (22%). The combination of disc space narrowing, abnormal T2 hyperintensity, and endplate irregularity was observed in 3 patients (18%), while only 1 patient (11%) had all 4 findings: disc space narrowing, abnormal T2 hyperintensity, endplate irregularity, and disc space enhancement. Note the absence of abnormal signal within the intervertebral disc spaces. C, Sagittal CT image shows associated sclerosis (arrows) corresponding to levels of abnormal increased signal on MR imaging. D, Bone scan, posteroanterior view, demonstrates focal areas of increased radiotracer uptake within thoracic vertebral bodies and the right sternoclavicular joint (arrow). E, Previously undiagnosed plantar pustulosis was evident on physical examination. Increased radiotracer uptake 10

Soft Tissues
Thickening of the anterior longitudinal ligament was present in 12 patients (80%) and extended to otherwise uninvolved vertebral bodies in 3 patients (25%). No other spinal ligaments were involved. Paraspinal soft-tissue involvement consisted of masslike soft-tissue thickening with enhancement (Fig 4). This was observed in 6 patients (37%), with a range of 1-6 vertebral levels involved. Masslike paraspinal soft-tissue involvement did not exceed 15 mm in maximal diameter in any case (range, 2-15 mm).

Pathology/Microbiology
Twelve patients (67%) underwent biopsy; 7 patients had a single biopsy, 3 patients had 2 biopsies, and 2 patients had 3 biopsies, totaling 19 biopsies. The vertebral body was the most commonly biopsied location (n ϭ 11), followed by the paravertebral soft tissues (n ϭ 5) and disc interspace (n ϭ 1). In 2 patients with spinal involvement, a concomitant sternal lesion was biopsied. Pathology was negative for malignancy in all biopsy specimens (n ϭ 19), including 2 patients (11%) with a known history of malignancy. Microbiologic cultures were also negative for infection in all 8 patients who had microbiologic testing performed.

Inflammatory/Infectious Markers
C-reactive protein levels were available in 13 of 18 patients (72%), with elevated levels in 8 patients (62%). Sedimentation rate levels were available in 16 of 18 patients (89%), with elevated levels in 6 patients (38%). White blood cell counts were available in 17 of 18 patients (94%), with all patients having normal levels.

DISCUSSION
A curvilinear or a semicircular pattern of contiguous vertebral body involvement localized to either the anterior or posterior vertebral bodies of the middle segments and adjacent anterior or posterior endplates of the surrounding vertebral bodies was found in most of our patients (63%) with SAPHO and spinal involvement. The high prevalence of this semicircular pattern of vertebral body signal alteration and enhancement may help to  MR images in a 69-year-old woman with cervical and thoracic back pain. A, Sagittal T2weighted fat-suppressed image of the cervical spine demonstrates paravertebral soft-tissue thickening (arrows), extending from the C3-C4 interspace to T3. Note the abnormally high T2-weighted signal within the C5-C6 disc (arrowhead). B, Sagittal T1-weighted gadolinium-enhanced fat-suppressed image demonstrates corresponding enhancement of the paravertebral soft tissue (arrows) with disc space enhancement (arrowhead). C, Axial T1-weighted gadoliniumenhanced fat-suppressed image demonstrates enhancement of the thickened paravertebral soft tissues (arrows). D, Axial T2-weighted fat-suppressed image further demonstrates the extent of the paravertebral soft-tissue thickening (arrows). differentiate SAPHO syndrome from metastases, which tend to be randomly distributed throughout the spine. When combined with a relatively low prevalence of abnormal disc space T2-signal and enhancement, the high prevalence of the semicircular pattern may also be helpful to radiologists in cases in which spinal infection is being considered. Correlation with CT may be beneficial because sclerosis in the areas of MR signal abnormality was present in 100% of our patients, regardless of the pattern of the MR signal changes. Recognition of these findings by radiologists should prompt clinical consultation with dermatology and/or rheumatology to assess typical skin and other musculoskeletal manifestations.
Osteosclerosis of the vertebral bodies in SAPHO syndrome has been described by Leone et al 15 and may progress to produce diffuse and generalized sclerosis with development of hyperostosis in more chronic cases. Prior case reports [16][17][18] have shown the utility of [ 18 F] FDG-PET/CT to differentiate active from healed chronic inflammatory lesions because PET/CT shows increased uptake only in lesions with active inflammation. The chronic osteosclerotic changes seen in our patients are hypothesized to represent a quiescent phase of the disease, which likely accounts for the normal FDG uptake in the 2 patients who had PET/CT scans in our series.
In contrast to a study performed by Laredo et al, 6 our patients were more likely to have involvement at contiguous levels. In addition, while our patients had frequent involvement of vertebral body corners, we did not identify erosion in any case. Because we are a tertiary referral center, our patients may have presented at a later clinical stage. In fact, nearly all our patients presented after other manifestations such as skin lesions or sternoclavicular involvement, both of which are considered highly specific, were evident.
The frequency of spondylodiscitis in SAPHO syndrome has been reported to be 9%-32%. 10 Prior studies have shown chronic sterile nonspecific inflammation from intervertebral disc biopsies in patients with SAPHO syndrome, equivalent to the Andersson lesion, which has been identified in 4.5% of patients with ankylosing spondylitis. 5,10 Others have suggested that spondylodiscitis is due to Propionibacterium acnes. 19 Although disc space involvement with fluidlike signal on T2-weighted images and disc space enhancement have been previously described in patients with SAPHO syndrome, 5,7 our study showed that most intervertebral discs in patients with SAPHO syndrome demonstrated the absence of both fluidlike signal on T2-weighted images (69%) and disc space enhancement (78%). The absence of these findings may be helpful in distinguishing SAPHO involvement from spinal infection.
We found single-level vertebral involvement rare, with contiguous involvement of Ն2 levels in nearly 90%, often with Ն5 levels involved. This finding is in contrast to cases of spondyloarthropathy, in which it is uncommon to see contiguous involvement. 6 Nachtigal et al 7 previously described the paravertebral soft tissues in patients with SAPHO involvement of the spine possibly showing abnormal signal intensity. We found similar results, with thickening and enhancement of the anterior longitudinal ligament present in 37% of patients with SAPHO syndrome. These findings further complicate the differential diagnosis of SAPHO syndrome versus infection but may help to differentiate it from spondyloarthropathies, which, to our knowledge, have only been reported to affect the supraspinous ligament, interspinal ligaments, and ligamenta flava. 20 Additionally, in the 2 patients with vertebral body involvement determined solely on nuclear medicine bone scans or plain radiographs, both demonstrated unilateral sacroiliac joint involvement. Depasquale et al 11 and Leone et al 15 have both described increased sclerosis and hyperostosis, particularly at the iliac side of the joint, in patients with SAPHO. When seen in patients with features of moderate sacroiliitis, this presentation was thought to be highly suggestive of SAPHO syndrome and helped to differentiate it from other spondyloarthropathies.
Peffers et al 13 first described a subset of patients with nonbacterial osteitis involving the thoracic spine. Spinal involvement in our study had a strong female predominance and preferentially affected the thoracic spine. This is in contrast to osteomyelitis and discitis, in which males are affected twice as often as females and the lumbar spine is most commonly affected. 21 MR imaging of the spine and sacroiliac joints has played a key role in patients with spondyloarthropathies and has led to improved understanding of the course of the disease and to an earlier diagnosis. 20 We suggest an imaging protocol consisting of at least a sagittal T1-weighted turbo spin-echo sequence, sagittal fat-saturated T2-weighted turbo spin-echo sequence or STIR sequence with high resolution, and a sagittal fat-saturated gadolinium-enhanced T1-weighted sequence for evaluation of patients with suspected SAPHO syndrome, spondyloarthropathy, or osteomyelitis/discitis.
Our study had a number of limitations. This was a retrospective review without a control group, and images were reviewed in consensus rather than independently by the 2 neuroradiologists, which might limit the reproducibility. Additionally, the number of patients was small; however, SAPHO syndrome is an uncommon entity, and ours is one of the largest studies to date. The MR imaging protocols were not standardized. For example, not every patient had a fat-saturated sequence, 1 patient lacked a T2weighted sequence, and 7 of 16 patients (44%) did not receive intravenous contrast material. Variation in MR imaging field strength and imaging protocols was a relatively minor limitation because signal change was noted on precontrast T1-weighted images in all patients. Finally, biopsy and microbiologic results were not available in half of the patients, perhaps due to the presence of other highly specific anterior chest wall and/or skin manifestations.
Additional studies involving larger numbers of patients with SAPHO syndrome and comparison groups of other infectious and noninfectious spinal inflammatory conditions are needed to determine whether this semicircular pattern is a unique spinal finding in patients with SAPHO syndrome.

CONCLUSIONS
In a patient with a curvilinear or semicircular pattern of contiguous vertebral involvement, sclerosis along ligamentous attachment sites, and the absence of intervertebral disc abnormal T2 hyperintensity and enhancement, SAPHO syndrome should be included in the differential diagnosis and clinical consultation with dermatology and/or rheumatology should be considered. This may prevent the imaging findings from being misinterpreted as discitis/osteomyelitis or metastases, with subsequent potential reduction in the number of unnecessary biopsies and delayed diagnoses.