UCL Université Catholique de Louvain Faculté de Médecine Cliniques Universitaires Saint-Luc

Mémoire de fin de spécialisation en radiologie

Spondylodiscites infectieuses « précoces » ou atypiques : diagnostic en IRM Early infectious spondylodiscitis: MR imaging findings

Docteur Sandrine BOSMANS Année académique 2006-2007 Promoteur Prof. F. LECOUVET

Early infectious spondylodiscitis: MR imaging findings

1) Introduction 2) Materials and methods Patient characteristics MR imaging studies MR image analysis • Bone - Bone Marrow - Endplates • Disks - Height - Signal intensity - Enhancement - Cleft • Soft tissues - Paraspinal - Epidural

3 7

3) Results

11 • •

•

4) 5) 6) 7) 8)

Discussion Conclusions Tables Bibliography Figures

Bone - Bone Marrow - Endplates Disks - Height - Signal intensity - Enhancement - Cleft Soft tissues - Paraspinal - Epidural 15 20 21 30 32

Je tiens à remercier les Professeurs J. Malghem et B. Vande Berg, pour leur dévouement au cours de ma formation en radiologie ostéo-articulaire. Ce travail est le fruit d’une collaboration nourrie avec l’équipe de radiologie osseuse du CHU de Lille, particulièrement les Professeurs A. Cotten et X. Demondion. Qu’ils soient ici remerciés. Merci à Françoise et Martine, pour leur sourire et leur disponibilité. Enfin, ma plus vive gratitude au Professeur F. Lecouvet, pour l’encadrement exemplaire au quotidien des radiologues en formation et pour les conseils précieux dans la rédaction de ce travail. S. B. Juin 2007

2

Early infectious spondylodiscitis: MR imaging findings SUMMARY

Purpose : To identify the key imaging features on MR studies of the spine performed in atypical or early stages of infectious spondylodiscitis (IS). Materials and methods : We reviewed MR studies of the spine collected in 17 patients in two academic MSK radiology departments. These files were prospectively kept because of the discretion of MR imaging abnormalities suggestive of IS, or because some diagnostic residual doubt was present after the first examination. All studies included at least sagittal T1- and T2weighted images, and transverse T1 and/or T2 images; contrast injection was performed in 14 cases. The positive diagnosis of IS was confirmed at imaging and microbiological follow-up. Eight MR features validated in the literature as commonly seen in IS were reviewed and recorded in terms of frequency, location, extent. The association of several of these signs was also evaluated. These signs were: for the disk: narrowing, disappearance of the “nuclear cleft”, high signal intensity areas on T2 images, enhancement on post-gadolinium T1 images; for adjacent vertebrae: bone marrow edema (BME), endplate erosion; for soft tissues: paravertebral fat and/or epidural fat infiltration or abscess. Results : The frequencies of the MR imaging features were : disk (n=17) : narrowing, 2 (12%); disappearance of the “nuclear cleft”, 3 (18%); high signal on T2 images, 11 (65%); enhancement on post-gadolinium T1 images, 3 (18%); for adjacent vertebrae (n=34): BME, 32 (94%), endplate erosion, 11 (32%); for soft tissues (n=17): paravertebral fat infiltration or abscess, 13 (76,5%); anterior epidural fat infiltration or abscess, 8 (47%). The extent of these signs was very subtle in the vast majority of cases. Six disks presented a completely normal MR appearance. Out of the 8 investigated MR features, only 1 was present in 1 patient; 3 in 3 patients; 4 in 7 patients; 5 in 4 patients; and 6 in 1 patient. No patient presented more than 6 concurrent features. Conclusions : Several MR findings typical for IS may be absent or very subtle in early stages of the disease. The presence of BME adjacent to the disk and involvement of the paraspinal soft tissues are the most frequent MR abnormalities that should be useful to make this difficult diagnosis. A normal MR appearance of the disk does not rule out the diagnosis of IS. 3

INTRODUCTION: BACKGROUND AND PURPOSE

Infectious spondylodiscitis (IS) (also called vertebral osteomyelitis) is a septic condition that involves the disk and adjacent vertebrae (17,22). It represents 2% to 7% of all cases of osteomyelitis. It occurs most frequently in the lumbar region, followed by thoracic and cervical locations (8,15). It occurs most commonly in adults with a mean age between 60 and 70 years. IS most often results from hematogenous dissemination of bacteria to bone. It usually begins by the hematogenous spread of micro-organisms to richly vascularized bone adjacent to disk (focus in the anterior subchondral bone) via nutrient arteries or Batson’venous plexus. The infection than usually spreads by extension across the disk space or along longitudinal spinal ligaments to involve the adjacent vertebral bodies (29). Other causes include trauma, extension of infection from adjacent structures, and complications of spine surgery and other interventional procedures. Staphylococcus Aureus and Staphyloccocus Epidermidis are the most frequent causative bacteria, in spontaneous and iatrogenic IS, respectively (28). Predisposing conditions include extra-spinal infection (streptococcal infections in patients with endocarditis, Haemophilus in patients with meningitis,...), urinary tract instrumentation, hemodialysis, IV drug abuse (Pseudomonas Aeruginosa), cancer and diabetes mellitus (7).

Spinal infection is a significant cause of morbidity (2,6,27). Despite advances of antibiotic treatment, the incidence is not decreasing due at least in part to an increase in ‘atrisk’ populations, namely the eldery, to the use of immunosuppressive therapies, and to the increasing frequency of spinal interventional procedures.

A long delay before diagnosis is not unusual because of the discretion or the lack of specificity of the clinical symptoms. Significant alteration of the biological markers of inflammation is usually present, but lacks specificity (2,7). Imaging plays a cardinal role in the positive diagnosis of the disease. Radiographs remain the first line imaging modality. However, initial radiographs may be completely normal and radiographic findings typically lag several weeks behind clinical symptoms (2,27). The earliest radiographic signs include blurring of the endplates and narrowing of the disk space, followed by trabecular bone destruction in the adjacent vertebrae. The bone destruction secondary to osteomyelitis is not detectable on plain films before 40% to 50% of the involved 4

region is destroyed, which usually takes at least 2 weeks after the beginning of the infection (9,30). This accounts for the limited sensitivity (68-82%) and specificity (78%) of radiographs reported in the literature (2,17). Computed tomography (CT) may be helpful in the diagnosis of IS. This technique may reveal hypodense areas within the disk and allows better detection of endplate and adjacent trabecular bone destruction, and of paravertebral soft tissues involvement. But it provides limited information on the bone marrow and involvement of the spinal canal (30). Radionuclide bone scans have been suggested as effective diagnostic tools (technetium 99 mHDP bone scanning shows a sensitivity of 90% and a specificity of 78%, combination of gallium 67 scanning and bone scan shows a sensitivity of 92% and a specificity of 100%) (17). But the findings are often nonspecific, since increased uptake may be observed in infection, but also in trauma, tumor or degenerative disk disease. Moreover, scintigraphic techniques may be negative in early aggressive disease, time-consuming (Gallium 67), and lack sufficient spatial resolution to differentiate the anatomic structures (19,28).

Magnetic resonance (MR) imaging is currently the modality of choice for the evaluation and diagnosis of suspected spinal infection with a sensitivity of 96% and a specificity of 93%, especially in the early phase (13,20,30). MR is more sensitive and specific than nuclear medicine techniques or plain radiography, provides a greater degree of confidence in the diagnosis, and is likely to show diagnostic findings earlier than other imaging techniques (2,15,17). Advantages of MR imaging include the capability of multiplanar imaging and of direct evaluation of the bone marrow, paraspinal soft tissues, and content of the spinal canal, i.e. the spinal cord, nerve roots, and epidural spaces (25). Differentiation of degenerative and metastatic disease from infection is much more confident on MR images than on radionuclide bone scans or radiographic examinations (17).

Numerous studies have described the MR imaging patterns seen in IS. Morphologic and signal intensity alterations of the disk, vertebrae and adjacent soft tissues indicative of spinal infection have been widely described (8,9,30). A careful literature review identified eight signs commonly observed on MR images in patients with IS. These signs are reduced disk height, disk areas of high signal intensity on T2-weigthed images, disk enhancement on post-contrast T1-weighted images, loss of visibility of the “nuclear cleft”, endplate erosion, bone marrow edema (BME) in adjacent

5

vertebral bodies, and paravertebral or epidural soft tissues infiltration or abscess (11) (Figs 1) (1,4,5,11,13,15,17,20,28). In our experience, and as incidentally mentioned by several authors, some of these typical MR imaging features may be absent or very subtle in the early stages of the disease, and the MR imaging “picture” may be incomplete, making the diagnosis difficult, doubtful or even impossible (4,8,23,28,11,15,16).

In order to identify the most frequent and reliable MR features seen in these early or atypical cases, we reviewed MR studies considered negative, non conclusive, or at least ambiguous - due to the lack or discretion of disk and surrounding structures abnormalities on MR imaging - and in which the diagnosis of IS was later confirmed by imaging, biological and microbiological follow-up. We analyzed early MR studies prospectively collected in two academic musculoskeletal radiology departments in 17 patients. This study reports the frequency, location, extension and association of the MR imaging features seen in these "difficult" cases in order to identify the most useful or early findings to make a confident diagnosis in this situation.

6

MATERIALS AND METHODS: PATIENT CHARACTERISTICS (Table 1) The study included 17 patients, 10 men and 7 women, with a mean age of 60 years (range, 24-83 years), who underwent MR examinations of the spine between 2002 and 2006 in two academic hospitals. These files were prospectively collected because the diagnosis of IS had initially been considered as doubtful, difficult, or missed. In all cases, IS was proven at imaging, clinical and microbiological follow-up. The clinical symptoms leading to the MR investigation were back pain (n=17), fever (n=7), neurological symptoms (unilateral sciatic pain) (n=2), fatigue (n=5), anorexia or alteration of the general condition (n=4). The duration of the clinical evolution before the initial MR examination was performed ranged from 5 to 60 days (mean, 17 days), and was less than 15 days in 12 patients. Predisposing factors for spine infection included diabetes mellitus (n=4), renal transplant (n=1), recent dental care (n=1), urinary tract infection (n=2), bone marrow autograft for multiple myeloma (n=1), breast cancer treated by chemotherapy (n=1) and ovarian cancer with febrile neutropenia (n=1). There was no predisposing factor in 6 cases (n=6).

Review of the report of this initial MR examination revealed that the imaging diagnosis of IS was missed in 2 patients, suspected in 8, and considered as very likely in 7. The anatomical levels of infection were T11/T12 (n=1), T12/L1 (n=3), L1/L2 (n=2), L2/L3 (n=2), L3/L4 (n=5), L4/L5 (n=3), L5/S1(n=1) (Table 2). All patients had follow-up MR imaging studies performed 8 to 60 days (mean, 22 days) after the initial MR study, and these second MR studies confirmed the diagnosis of IS in all cases. Biological findings also confirmed the diagnosis, showing consistent increase in serum CRP levels, ranging from 15 to 70 mg/dl (mean, 32mg/dl). The causative organism was isolated in all 17 patients, by blood culture (n=9) and by culture of disk material aspiration or biopsy (n=8). The following bacteria were cultured: Staphylococcus aureus (n = 7), Coagulase-negative Staphylococcus (n = 5), Streptococcus (n= 2), Escherichia Coli (n= 2), Serratia Marcescens (n=1).

MR IMAGING STUDIES All initial and follow-up MR imaging studies of the lumbar spine were performed on 1.5 T superconducting systems (Philips Intera 1.5, Best, The Netherlands, or Siemens Magnetom 1.5T, Erlangen, Germany) using a phased array coil. 7

All 17 initial MR imaging studies included sagittal T1-weighted spin-echo sequences. Sagittal T2-weighted fast-spin-echo MR images were performed in 5 patients, sagittal fat suppressed T2-weighted (or STIR) MR images in 8 patients, and both sequences in 4 patients. Transverse T1-weighted and T2-weighted sequences were obtained in 15 and 7 patients, respectively. Injection of contrast material was performed in 14 patients. Sagittal contrast-enhanced T1weighted MR images were obtained in 14 patients, sagittal fat-saturated T1-weighted images in 4, and axial T1-weighted in 5.

MR IMAGE ANALYSIS All initial MR examinations were reviewed in consensus by two musculoskeletal radiologists with more than ten years experience (one from each university), and one senior resident in radiology. All MR sequences were available by the time of this reading. The readers evaluated and quantified each of the 8 features identified in our literature review of MR findings in IS: bone marrow edema within adjacent portions of the vertebral bodies, erosions of adjacent vertebral endplates, height, appearance (presence of a “nuclear cleft”) and signal intensity of the intervertebral disks, and involvement (infiltration or abscess) of paravertebral and epidural soft tissues.

Bone Bone Marrow Edema (BME) Signal intensity alterations considered as indicative of IS were low signal intensity of the vertebral bone marrow on T1-weighted MR images and high signal intensity on T2-weighted (fat-saturated) MR images (17,24,28). These signal intensity changes were assessed in comparison to adjacent vertebral bodies considered as normal. The extent of BME areas was considered as diffuse, when involving more than 75% of the vertebral body, or as focal, when involving less than 75% of the vertebral body. When focal, antero-posterior, cranio-caudal and transverse extent (deducted from the number of sagittal slices where it was visible) of this abnormal signal was measured as a percentage of the length of the vertebral endplate in the antero-posterior direction, of the vertebral height in the sagittal direction, and of the transverse diameter of the vertebral body in the transverse plane. Volume of the BME was quantified as a percentage of the volume of the vertebra by calculation from these measurements.

8

The location of BME in the sagittal and in the transverse plane was precised as being either anterior or posterior, and lateral or central.

Endplates Both vertebral endplates adjacent to the infected spinal segment were evaluated on T1-, T2-, and contrast-enhanced T1-weighted MR images. These endplates were qualified as intact or eroded when showing interruption or destruction of the linear low signal of the endplates. Sagittal and transverse extent of this destruction was measured in millimeters and as a percentage of the length or width of the vertebral endplate; its transverse extent was deducted from the number of sagittal slices where it was visible.

Disks Disk height A the level of infection, the disk height was graded as normal, narrowed less than 50% or narrowed more than 50%, taking as reference the height of adjacent disks, which were considered as normal or showing degenerative height loss.

Signal intensity Signal intensity of the infected disk was qualified on T1- and T2-weighted MR images as being either identical, lower, or higher in comparison to that of the adjacent disks. When higher on T2-weighted images, this signal was further qualified as being either lower than that of fluid (CSF), or equivalent to that of fluid. This abnormal high signal was considered as focal when it was visible on less than 75% of the antero-posterior length of the disk, and diffuse when it involved at least 75% of the disk or the entire disk. When focal, its location was defined as anterior, central or posterior in the sagittal plane. Sagittal and transverse extent of the high signal intensity area was measured as a percentage of the antero-posterior length of the disk and of its transverse diameter (deducted from the number of sagittal slices where it was visible).

Enhancement On sagittal contrast-enhanced T1-weighted MR images, the disk was defined as showing no enhancement, focal enhancement or diffuse enhancement. Sagittal extent of the enhancement

9

was graded as a percentage of the antero-posterior length of the disk and its transverse extent was deducted from the number of sagittal slices where it was visible. When focal, its location was defined as anterior, central or posterior in the sagittal plane.

Cleft The low signal “nuclear cleft” normally seen in the central area of the disk was also evaluated on sagittal T2-weighted MR images. The “nuclear cleft sign” was considered positive when this cleft was absent in the infected disk and present in the adjacent non infected disks.This nuclear cleft sign was considered “not applicable” if the adjacent non infected disks had no visible clefts on T2-weighted MR images.

Soft tissues The presence of paraspinal and/or epidural involvement by the infectious process was evaluated and qualified as being either a phlegmon or an abscess. A phlegmon was defined by the presence of abnormal low signal intensity on T1-weighted MR images within the paraspinal or epidural fat, intermediate to high signal on T2-weighted MR images, with diffuse enhancement and no avascular area on contrast-enhanced T1weighted images (17,28). An abscess was defined by the presence of low signal intensity on T1-weighted MR images, with central high signal intensity on T2-weighted MR images, and a central avascular area with peripheral rim enhancement on contrast-enhanced T1-weighted MR images (17,20,28). The location of paraspinal soft tissues abnormalities was further defined as being either anterior, lateral or antero-lateral on all available sequences. Their antero-posterior, sagittal and transverse extent were measured in mm on all available sequences.

Associations The number of associated "typical" MR features was calculated for each patient, to determine the most frequent associations of these signs. Disk enhancement was excluded from this analysis because only 14 patients underwent gadolinium administration.

10

RESULTS: Analysis of the eight MR features suggestive of IS in our series had the following results. Bone Bone Marrow edema (BME) (Figs 2-3) The MR characteristics of the BME observed in the vertebral bodies (n = 34) of the 17 patients summarized in Table 3. Frequency : among the seventeen patients, BME was observed in both adjacent vertebral bodies in 15 cases (88%), and in only one vertebral body in 2 cases (12%). In terms of sequences, T1-weighted images showed BME as low signal intensity in all cases. STIR or fat-saturated T2-weighted images, available in 12 patients, showed BME as high signal intensity in all 12 patients. FSE T2-weighted sequences, available in 9 patients, showed this BME as high signal intensity in 5 and were considered negative in 4; of these four patients, two had fat-saturated T2 sequences that revealed the BME. In terms of vertebral bodies (n=34), areas of low signal on T1-weighted MR images were observed in 32 vertebral bodies (94%) and absent in 2 (6%); areas of high signal on T2weighted MR images were observed in 28 cases (82%) and absent in 6 cases (18%). The extent of BME, evaluated in the 34 vertebral bodies on T1-weighted MR images, was considered as diffuse in 6 vertebral bodies (17.5%), focal in 26 (76.5%), and absent in 2 (6%). When focal, it involved less than 30% of the volume of the vertebral body in 23 (67.5%), with a volume ranging from 1 to 75 % (mean, 14.3%) of the vertebral volume. The extent of BME, evaluated among the 34 vertebral bodies on T2-weighted MR images, was considered as diffuse in 5 vertebral bodies (15%), focal in 23 (67.5%), and absent in 6 (17.5%). When focal, it involved less than 30% of the vertebra volume in 20 (59%), ranging from 1 to 75% (mean, 14.0%) of the vertebral volume. Predominant location of the BME in the sagittal plane was anterior in 17 vertebral bodies (50%), posterior in 6 (17.5%) and diffuse in 9 (26.5%). In the transverse plane, it was central in 14 vertebral bodies (41%), right-sided in 4 (12%), left-sided in 3 (9%) and diffuse in 11 (32%).

Endplate erosions (Table 4) (Fig 4) Frequency : among the 17 patients, 7 (41%) had normal endplates adjacent to the infected level, and 10 (59%) presented focal erosions within endplates (9 within only one endplate, and one within both adjacent endplates). In terms of endplates (n=34), 23 were intact (68%) and 11 presented erosions (32%). 11

All 11 erosions were visible on T1-weighted MR images, 7 were visible on T2-weighted MR images, and only 2 were visible on contrast-enhanced T1-weighted MR images. The extent of these erosions in the antero-posterior plane ranged from 5 to 18 mm (mean, 10 mm) ; in terms of percentage, this extent ranged from 10 to 50 % of the endplate (mean, 30%). In the transverse plane, this extent ranged from 4 mm to 35 mm (mean,13 mm); in terms of percentage, this extent ranged from to 10 to 75 % of the transverse diameter of the endplate (mean, 25 %). Most of these erosions were visible on only one or two sagittal slices. Location Erosions involved the superior endplate of the involved spinal segment in 3 cases, and the inferior endplate in 8 cases. In the sagittal plane, erosions were located in the anterior third of the endplate in 6 cases, in its central third in 2 cases, and in its posterior third in 3 cases. In the transverse plane, erosions were central in 5 cases, left-sided in 2 cases and right-sided in 4 cases.

Disks (Table 5) Height (Figs 5-7) The height of the involved disk was considered as “completely” normal in 9 cases (52%). Narrowing of the involved disk was observed in 8 cases (48%), but in 6 cases (36%), the reduced disk height was similar to that of adjacent disks, considered as degenerative. In these cases, reduced disk height was thus not considered as an indicative sign of IS. Hence, the disk height was considered as “really” reduced and as indicative of IS in only 2 cases (12%). In these cases, the height loss was more than 50 % compared to the height of adjacent normal disks.

Signal intensity (Figs 8, 9) On sagittal T1-weighted MR images, no abnormality was noted within any involved disk. On sagittal T2-weighted MR images, six infected disks (35%) did not show any area of increased signal intensity. Among these disks, 4 (23%) were “strictly normal”, 2 presented reduced disk height, but similar to the height of adjacent degenerative disks. Of these 6 cases, enhanced T1-weighted images were available in 5 and none showed any enhancement area within the disk. Eleven infected disks (65%) presented areas of increased signal intensity on T2-weighted images. In 10 cases, this high signal was focal (anterior in 4, central in 1, posterior in 5), and limited to the peripheral aspect of the annulus fibrosus in 5 cases. These abnormalities were 12

subtle, visible only on one to three sagittal slices (thickness ranging from 4 to 15mm). It was diffuse and involved the nucleus in only one case. Sagittal T1-weighted images obtained after contrast material administration were available in 14 patients. Ten disks (71,5%) showed absolutely no enhancement. Focal areas of disk enhancement were observed in four cases (28,5%). These areas were very limited, peripheral, anterior in 2 cases, posterior in 2, visible on one to two sagittal slices (thickness ranging from 4 to 7 mm). These areas of enhancement did not correspond to the areas of high signal intensity seen on T2-weighted MR images.

Cleft (Fig 7) Visualization or not of the intranuclear cleft was also considered. This sign was evaluable in the 17 patients, present (negative) in 14 patients (82%) and absent (positive) in 3 (18%).

Soft tissues Sixteen patients (94%) had MR signs of extension of the disco-vertebral infection to the soft tissues. In eight, this extension only involved paraspinal soft tissues; in two epidural soft tissues only; and in six, both paraspinal and epidural soft tissues. There was no soft tissue involvement in one patient (6%).

Paraspinal (Fig 8) Involvement of the paraspinal soft tissues was present in 14 patients (82%). The lesion was categorized as a phlegmon in 11 cases (79%) and as an abscess in 3 cases (21%) Extent The extent of the lesion ranged from 4 to 20 mm (mean, 9.5 mm) in the antero-posterior direction, from 10 to 60 mm (mean, 40 mm) in the sagittal plane, and from 10 to 40 mm (mean, 23mm) in the transverse plane. Location Location of involvement of the paraspinal soft tissues was anterior in 2 cases (12%), anterolateral and lateral in 6 cases (35%), lateral in 2 cases (12%), anterior, antero-lateral and lateral in 4 cases (23.5%). Abnormalities of the signal of the psoas muscle were noted in 6 cases (35%). Among these 6 cases, an abscess was present in 2 (12%).

13

Epidural (Figs 3,6,9) Eight patients (47%) had evidence of anterior epidural involvement. In 5 cases (29%), this involvement was considered as a phlegmon, and in 3 cases (18%) as an abscess. Extent Anteroposterior extent of this epidural involvement ranged from 3 to 8 mm (mean, 3.6 mm ). Its sagittal extent ranged between 10 and 50 mm (mean, 30 mm). Its transverse extent ranged from 4 to 16.5 mm (mean, 10 mm) Associations Table 7 summarizes the number of associated MR features in our patient population. In our 17 patients, none presented the complete “MR imaging picture” or semiology, with these seven commonest signs present simultaneously; one patient presented six signs (6%), 3 patients had 5 signs (17%), 7 patients had 4 signs (41%), 5 patients had 3 signs (29%), none had 2 signs, and finally 1 patient had one sign (6%) consisting in isolated BME. Thirteen patients (76.5%) only presented 1 to 4 signs.

14

DISCUSSION MR imaging is the modality of choice for the diagnosis of infectious spondylodiscitis (IS) (12,13,17,20,22). Typical MR findings have been widely reported. Our study began with a literature review that identified the most commonly reported MR features observed in “established” IS. These 8 signs and their frequency are listed hereafter and in table 8. Bone marrow edema (BME) in vertebral bodies adjacent to the infected disk space has a sensitivity ranging from 86 to 100% (3,4,11,15,17). Endplate erosions have a sensitivity ranging from 80% to 95% (3,4,11). Reduced disk height has a sensitivity ranging from 52% to 93% (3,11,28). Areas of increased signal intensity within the disk on T2-weighted MR images have a sensitivity ranging from 67 to 93% and areas of disk enhancement on post-contrast T1weighted MR images have a sensitivity ranging from 94 to 100% (3,4,11,15,17). Disappearance of the “intranuclear cleft” has a sensitivity ranging from 83 to 87% (11,17). Finally, paravertebral and epidural soft tissue involvement have a sensitivity ranging from 57 to 95% (3,4,11,28). For Ledermann, who compared the sensitivity of these signs in a series of 46 patients with proven IS, BME was an almost consistent finding and almost diffusely involved both vertebral bodies adjacent to the infected disk. Additional criteria with good to excellent sensitivity included involvement of paraspinal or epidural tissue, contrast enhancement of the disk, high or fluid-equivalent signal intensity of the disk on T2-weighted MR images, and erosions of the vertebral endplates mostly visible on T1-weighted MR images (11).

Increasing awareness of IS and of patients at risk for this disease, its increasing incidence and increasing availability of MR imaging has resulted in earlier imaging of this condition (8). Negative MR studies have been reported in the very early course of the disease (2,21). Several authors incidentally mentioned the possible absence of several of these typical MR features in early stages (Table 9) (4,8,11,15,16,20,23). Herein, we focused on a series of 17 cases of “early” or "atypical" IS investigated at MR imaging. These cases were kept because the initial MR study had raised diagnostic difficulties, doubt or even mistakes due to of the atypical appearance, incomplete presentation and/or discretion of the MR signs observed. In all 17 patients, IS was proven by aspiration or biopsy of the lesion itself or by blood cultures, and by follow-up MR studies. The purpose of

15

this study was to identify the most useful diagnostic features for a reliable and confident diagnosis of these early or difficult presentations.

This study had three main results. First, our evaluation of the eight most common signs found in the literature showed that the frequency of the majority of these signs in our study was by far inferior to the reported frequency in established IS (Table 8). This was not the case for the presence of BME, which was present in 2 vertebral bodies adjacent to the infected disk in 15/17 patients (88%) and in one vertebral body in the remaining 2 patients (12%), and for the involvement of soft tissues (paraspinal and/or epidural), which was observed in 16/17 patients (94%). This parallels the diagnostic value of these signs underlined by Lederemann (see above). The other signs were by far less common in our study. Endplate erosions were observed in 10 patients (59%) and in 11 of the 34 endplates (32%). Reduced disk height was observed in 2 cases (12%), disk areas of high signal intensity on T2-weighted MR images were observed in 11 cases (65%), areas of enhancement were observed in 4 of the 14 patients who underwent gadolinium injection (28.5%), disappearance of the “intranuclear cleft” was observed in 3 cases (18%).

A second observation was the discretion of the typical MR features described in the literature when they were present in our series. The study of these discrete findings highlights the most useful sequences and anatomical areas to check for a correct diagnosis of early IS. In most cases, the areas of BME were very focal, involving less than 30% of the vertebral body volume in 23/34 (67,5%), and BME was diffuse in only 6/34 (17.5%), involving more than 75% of the vertebral body volume. In his study, Ledermann observed that the extent of BME was diffuse in 60% of patients and involved less than one-third of the vertebral body in 17% (11).These differences suggest that our cases represent very early stages of IS with only limited involvement of the vertebral bodies. Of note, BME was visible as low signal intensity on T1-weighted images in all patients. It was consistently visible as a high signal intensity on T2-weighted images used with fat suppression (12 cases), whereas it was visible on T2-weighted images in only five of the 9 patients who underwent this sequence. This superiority of fat suppression in T2 sequences to detect BME has been emphasized (2,4,14,28,29). Location of the BME in the sagittal plane in our series was predominantly anterior. This observation is in agreement with the welknown history of IS which begins near anterior vertebral angles, due to their rich arterial vascularization, and later spreads to the rest of the 16

vertebra, to disk and adjacent vertebra, which explains the more extensive MR imaging characteristics of “established IS” (12,27,29,30). In two patients, BME only involved one vertebra. This parallels the observation by several authors of "single segment vertebral osteomyelitis", which is most frequent in tuberculous spondylodiscitis, and is quite exceptional in pyogenic infections (8,23). As in other series, we observed that these patients went on to develop more typical disease patterns on follow-up MR imaging, with involvement of both adjacent vertebrae. Endplate erosions were observed in less than one third of vertebral endplates. When present, there erosions most frequently involved the anterior aspect of the endplate, and had a limited sagittal extent, involving less than 30% of the antero-posterior vertebral body length, and visible on only one sagittal slice in 7 cases. Most of them were visible on T1-weighted MR images only, which is in agreement with the previous observation of the limited value of T2 and post-contrast T1-weighted MR images (4,20,25). Areas of increased, CSF-like, disk signal intensity were seen on T2-weighted MR images in 11 cases (65%), but very focal in ten, with a “punctifom” appearance, most often limited to a portion of the annulus; only one patient had diffuse high signal involving both the nucleus and annulus. Areas of disk enhancement on post-contrast images were not frequent (4/14, 28.5%), and were also very limited, with an extent that was inferior to 30% of the antero-posterior diameter of the disk, or a visibility on only one or two sagittal slices. Most importantly, the low frequency and limited extent of disk abnormalities in our study demonstrate that patients with IS may present without disk involvement (6 cases in our series, 35%) or with only very limited alterations at MR imaging, i.e., with preservation of a normal disk height, a preserved “intranuclear cleft”, and absence of any area of high signal intensity on T2-weighted images and of enhancement after gadolinium injection. This underlines the possible observation of "spondylodiscitis with a normal or almost normal disk appearance". Soft tissue involvement was present in 16 of our 17 patients, but was most frequently minor or moderate. Fourteen patients (82%) had evidence of infiltration of the paraspinal soft tissues. The mean extent of these abnormalities ranged from 10 to 40 mm. The location of this soft tissue involvement was lateral in all cases, antero-lateral in 10 cases and anterior in 5 cases, which emphasizes the importance of carefully checking these areas, especially on the most lateral of sagittal slices and on axial images. Eight patients (47%) had evidence of epidural soft tissue involvement and the extent of this involvement was limited, with a mean antero-posterior diameter of 3.6 mm, mean transverse extent of 10 mm, and a more important sagittal extent, with a mean of 30 mm. 17

As third observation and as a consequence of the previous points, the MRI “picture” or semiology observed in our patients with “early IS” was very incomplete, since only few of the typical MR features of IS were present simultaneously. We have studied the association of the 7 of the 8 most common signs reported in the literature: BME within two adjacent vertebral bodies, endplate erosion, reduced disk height, disk high signal intensity on T2-weighted MR images, disappearance of the intranuclear cleft and involvement of paraspinal or epidural soft tissues. Disk enhancement was excluded from this analysis because only 14 patients underwent gadolinium administration. Of note, in our study, only 4 out of the 14 patients who underwent gadolinium injection presented a focal disk enhancement; this is in contrast with the good sensitivity of this sign reported by Ledermann (11). The most frequent association observed was that of BME and soft tissue involvement. This is not surprising since these two signs were the most common “individual” signs (table 8) The patients who had 3 or 4 signs presented BME, soft tissue involvement, endplate erosions and/or disk high signal intensity on T2-weighted MR images and/or disappearance of the intranuclear cleft. Patients who had 5 or 6 signs presented in addition to these signs a reduced disk height. One patient presented only one sign, isolated BME; in that case, clinical and biological findings had been crucial for the correct diagnosis of IS by the time of the initial MR study.

Medical records of our 17 patients were kept prospectively between 2002 and 2006 in 2 academics hospitals, and MR examinations were reviewed retrospectively. This retrospective analysis, and the “bicentric” origin of the data constitute the first limitation of our study, explaining some heterogeneities in MR imaging protocols; for example, the lack of T2-weighted fat-saturated and contrast-enhanced T1-weighted sequences respectively in 5 and in 3 cases. The small patient population (17) represents a second limitation. However, early spondylodiscitis appears to be a rare entity, as proven by the 4-year time delay necessary to collect this series, by including patients from 2 university hospitals meeting the inclusion criteria (CHRU Lille and UCL Brussels). This parallels the reported low frequency of these atypical or "early stages" IS (4,8,9,11,15,16,23,20). To our knowledge, the current series of "early stage" or atypical IS seems to be the largest in the literature.

In daily MR imaging practice, the most important problem raised by these atypical or early IS may be the differential diagnosis from degenerative disk disease (DDD), since both 18

conditions may share several features, i.e. BME adjacent to the disk, endplate erosions, disk narrowing, discrete areas of high signal intensity on T2 images or enhancement within the disk on post-contrast T1 images (18, 26). Our study did not aim at comparing the MR characteristics of IS and DDD in 2 patient populations because this has already been done (3,25). Previous papers have shown that, whereas individual MR features were insufficient, the association of several signs was confidently indicative of DDD: well-delimited juxtadiscal marrow changes combining fat- and edema-like signal changes, intradiscal vaccum, absence of CSF-like high signal within the disk, and observation of “repetitive” similar findings at adjacent disk levels. On the contrary, the observation of soft tissue involvement should be considered pathognomonic for IS, as it was also found in 16 of our 17 patients, and is never reported in DDD. In case of residual doubt, correlative Xrays or CT, and biological findings may help differentiate erosive DDD from IS, and follow-up MR imaging will show rapid evolution of the abnormalities over several days in case of IS.

In conclusion, our series of patients with IS was atypical by the very incomplete MR imaging findings and by the discretion of the typical signs when they are present. This limited frequency and discretion of MR abnormalities compared, as well as the short delay between initial symptoms and MRI suggest the very early stage of the disease in our patients. Our results show that these early stages of IS may present with only few or with discrete abnormalities within the disk, vertebral body and soft tissues, and that most useful early MR signs for the diagnosis of "early" IS are BME and paraspinal or epidural soft tissue involvement. This study also highlights the unequal value of the different sequences to detect subtle abnormalities within bone, disk and soft tissue, and the anatomical locations to check carefully on MR images to detect these abnormalities. Finally, it shows that the absence of several MR features should not rule out the diagnosis of (early) IS, for example a preserved discal “cleft”, absence of disk high T2, or of enhancement on post-contrast T1 images. Most impressive findings were the possible observation of IS with completely normal (6/17) or almost normal disk appearance, and limitation of the abnormalities to one vertebral body.

19

CONCLUSIONS Our series highlights the possible atypical appearance of IS, presumably at an early stage, due to the absence or discretion of some of the commonly reported "typical" MR imaging features of IS, which might complicate confident diagnosis. Knowing this should lead to careful screening of subtle discal and juxtadiscal bone and soft tissue abnormalities on the different sequences, and to the recognition of these sometimes discrete signs and their possible association. The importance of BME and of soft tissue abnormalities, even very limited, is cardinal. Close attention should be paid to endplate erosions and disk signal alterations, even tiny. Finally, the possibility that the disk may appear completely or almost normal and that BME may be confined to only one vertebral body in these early stages must be kept in mind. Facing an incomplete MR “picture”, the diagnosis of "early" IS should be at least suggested, and further microbiological work-up and MR imaging follow-up should be advocated in case of any doubt.

20

TABLES: Table 1 : Patient Characteristics

Patient Age/ # Gender

Clinical Symptoms

Predisposing factors

Symptoms duration before first MRI

Serum Positive

Levels Diagnosis

Causative organism

Delay of follow-up MRI

CRP levels confirmation

(mg/dl)

(Days)

(Days)

(1)

52/M

(2)

24/F

(3)

83/M

(4)

60/M

(5)

43/M

(6)

72/M

(7)

50/M

(8)

61/M

(9)

69/M

(10)

60/F

(11)

74/F

(12)

72/F

(13)

68/F

(14)

57/M

(15)

71/F

(16)

65/M

(17)

45/F

Acute back pain Back pain – fever Back pain Back pain – Fever at night

Renal transplant

10

L4/L5

BC

E.Coli

22,7

30

(-)

14

L3/L4

DA

S.Epidermidis

(-)

60

(-) Prostatic endoscopic resection

14

T12/L1

DA

S.Aureus

(-)

30

10

L3/L4

BC

Strepto enterococcus

38

14

20

L4/L5

DA

S.Epidermidis

57

14

10

L3/L4

DA

S.Epidermidis

70

21

30

T11/T12

DA

S. Aureus

34

15

11

L3/L4

BC

Streptoccus Mitis

(-)

30

60

L2/L3

DA

S.Aureus

67

10

15

L2/L3

BC

E.Coli

(-)

15

15

L1/L2

BC

S.Aureus

26

21

5

L4/L5

BC

S. coagulase negative

15

28

40

L5/S1

BC

Serratia marcescens

24

14

10

L1/L2

DA

S.Aureus

(-)

14

40

T12/L1

BC

S.Epidermidis

3

21

14

T12/L1

DA

S.Aureus

12

14

14

L3/L4

BC

S.Aureus

17

30

Back pain – Dental care Fever Back pain – Neurologic (-) deficit Back painDiabetes fatigue, AGC mellitus Back painBone marrow Fever autograft Back pain(-) AGC Back painDiabetes Fever mellitus Back painUrinary tract High Fever infection Back painNeutropeniaAGC-Fever Breast neoplasia Back painDiabetes Neurological mellitus deficit Back pain (-) Back pain Ovarian cancer after neutropenia exercises Diabetes Back pain mellitus Acute back (-) pain

Abbreviations : BC = blood culture ; DA = disc aspiration ; AGC = alteration of general condition 21

Table 2 : Distribution of infected disks in the study population

Number of patients

5 4 3 2 1 0

T11/T12

T12/L1

L2/L3

L3/L4

L4/L5

L5/S1

Spinal Levels

Bars represent the number of involved spinal segments.

22

Table 3 : MR characteristics of BME in 17 patients (34 vertebrae) with “early” IS

•

Frequency : Vertebrae - present on T1 32/34 (94%) - present on FSE T2 28/34 (82%) and/or on T2 Fat Sat Patients involvement of 2 vertebral bodies 15/17 (88 %) involvement of only 1 vertebral body 2/17 (12 %)

•

Distribution - diffuse - focal - ≤ 30 % V.V. (vertebral volume)

6/34 (17.5%) 26/34 (76.5%) 23/34 (67.5%)

Location in sagittal plane - anterior - posterior - anterior + posterior

17/34 (50%) 6/34 (17.5%) 9/34 (26.5%)

Location in transverse plane: - central - marginal - diffuse

14/34 (42%) 7/34 (20%) 11/34 (32%)

Extent (on T1-weighted MR images) - antero-posterior (% endplate) - sagittal (% anterior wall) - transversal (% endplate) - volume (% vert.volume)

15-100% 15-100% 30-100% 1-100%

•

•

•

(mean, 56.9%) (mean, 51.1%) (mean, 58.4%) (mean, 27.1%)

23

Table 4 : MR characteristics of erosions within vertebral endplates (VE) adjacent to the infected disks in 17 patients with “early” IS

•

Frequency : Vertebral endplate erosions - present (visible T1-T2-T1Gd) 11/34 (32%) (11 ;7 ;2) - absent 23/34 (68%) Patients : 7/17 : no erosion (41%) 10/17 : erosions (59%)

•

Distribution - diffuse - focal

•

•

•

0/34 (0%) 11/34 (32%) (≤ 30% V.E : 8 ; ≤ 60% : 11/11)

Location in sagittal plane - superior VE - inferior VE - anterior third - middle third - posterior third

3/34 ( 9%) 8/34 (24%) 6/34 (18%) 2/34 (6%) 3/34 ( 9%)

Location in transverse plane - central - marginal

5/34 (15%) 6/34 (18%)

Extent - antero-posterior

-

transverse

< 30% VE 8/34 (24%) < 60% VE 3/34 (9%) (mean %, 30%; mean mm,10 mm) < 30% VE 9/34 (26%) < 60% VE 1/34 (3%) > 60% VE 1/34 (3%) (mean %, 25%; mean mm, 13 mm)

24

Table 5 : MR characteristics of disk abnormalities in 17 patients with “early” IS.

•

Height - Similar to adjacent (normal or degenerative) disks 15/17 (88%) - “Really” reduced 2/17 (12%) (> 50% height loss, compared to adjacent normal or degenerative disks)

•

Areas of high signal intensity on T2-weighted images Frequency : present 11/17 (65%) Distribution - focal 10/17 (59%) - diffuse 1/17 (6%) Location - ant . 4 - cent. 1 - post. 5 Extent - ≤ 1/3 disk length 6/11 (54%) - ≤ 2/3 disk length 10/11 (91%)

•

Areas of enhancement on post-contrast T1-weighted images (n=14) Frequency - absent 10/14 (71,5%) - present 4/14 (28,5%) Distribution - focal 4 - diffuse 0 Location - ant 2 - cent 0 - post 2 Extent - ≤ 1/3 disk length 4/4 (100%)

•

Cleft - Evaluable and positive sign (disappearance) : 3/17 (18%) - Evaluable and negative sign (preservation) : 14/17 (82%)

25

Table 6 : Soft tissue abnormalities at MR imaging in 17 patients with “early” IS.

Paraspinal fat involvement •

•

•

Frequency - absent - present

3/17 (18%) 14/17 (82%) phlegmon 11/14 (85%) abscess 3/14 (23%)

Location - anterior - anterior and antero-lateral - lateral - anterior, antero-lateral, lateral Extent (thickness) - antero-posterior - sagittal - transverse

2/17 (12%) 6/17 (35%) 2/17 (12%) 4/17 (23.5%)

4 - 20 mm (mean, 9.5 mm) 10 - 60 mm (mean, 40 mm) 10 - 40 mm (mean, 23 mm)

Epidural involvement •

•

Frequency - absent - present

9/17 (53%) 8/17 (47%) phlegmon 5/8 (29%) abscess 3/8 (18%)

Extent -

antero-posterior sagittal transverse

3 - 8 mm (mean, 3.6 mm) 10 - 50 mm (mean, 30 mm) 4 - 16.5 mm (mean, 10 mm)

26

Table 7: Association of MR Features in 17 patients with early IS

Number of associated MR features

Number of patients presenting the association

7

0/17

6

1/17 (6 %)

5

3/17 (17 %)

4

BME + paraspinal fat + erosion + disk high T2 BME + paraspinal fat + epidural fat + erosion 7/17 (41 %) BME + epidural fat + erosion + disk high T2 BME + paraspinal fat + epidural fat + disk high T2 BME + paraspinal fat + disk high T2 + loss of cleft

(2) (1) (1) (1) (2)

3

BME + paraspinal fat + disk high T2 BME + epidural fat + disk high T2 5/17 (29 %) BME + paraspinal fat + epidural fat BME + paraspinal fat + erosion

(1) (1) (1) (2)

2

0/17

1

1/17 (6 %)

0

0/17

MR imaging features

(-) BME + paraspinal+ epidural fat + erosion + disk narrowing + loss of cleft (1) BME + paraspinal fat + erosion+ disk narrowing + disk high T2 BME + paraspinal fat + epidural fat + erosion +disk high T2

(1) (2)

(-) BME

(1)

(-)

Note : only 7 features are considered; disk enhancement (observed in 4/14 patients who received contrast material injection) is not considered in this table.

27

Table 8 : Most common MR features seen in IS, after literature search and comparison with our results

Literature 96 % Ledermann < 1/3 : 11% < 2/3 : 26% 1) Bone Marrow Edema Diffuse: 59% within 2 adjacent vertebral 100% Champsaur bodies 86% Thrush 95% Dagirmajian 96% Meyers 84% Ledermann 2) Endplate erosions 80% Champsaur 95% Dagirmanjian 52% Lederemann 3) Reduced Disk height 93% Champsaur 64% Thrush 93% Lederemann 67% Champsaur 4) Disk high signal intensity 95% Dagirmanjian on T2- weighted images 87% Modic 85% Meyers 95% Ledermann focal in 7% 5) Disk enhancement on postrim in 68% diffuse in 20% contrast T1-weighted images 100% Champsaur 94% Dagirmanjian 6) Disappearance of the 83% Ledermann “nuclear cleft” 87% Modic >90% Ledermann 7) Paravertebral soft tissue 95% Dagirmanjian involvement 79% Thrush 57% Champsaur 8) Epidural soft tissue <90% Ledermann involvement 79% Thrush

Our group (17 patients) 88% (15/17 patients) 94% (32/34 endplates) < 1/3 : 68% < 2/3 : 9% Diffuse: 17% (2 patients : only 1 vertebra) 32% (11/34)

12% (2/17)

65% (11/17), focal in 10 (59%), diffuse in one (6%)

28% (4/14), focal in 100%

18% (3/17)

82% (14/17)

47% (8/17)

28

Table 9: Atypical magnetic resonance findings in IS, as reported in the literature and comparison to our results.

Atypical MR imaging features Frequency in the literature (%) in IS 2/25 (8) Gillams 2/26 (7.5) Post No involvement of disk 2/39 (5) Dagirmanjian 2/39 (5) Dagirmanjian Normal disk on T2 weighted 6/26 (23) Post(but gado+) MR images 1/27 (4) Meyers 2/39 (5) Dagirmanjian No disk enhancement 2/16 (12.5) Shih Involvement of one vertebral 1/46 (2) Ledermann body and adjacent disk 3/27 (11) Meyers 1/25(4) Gillams Involvement of one vertebra 1/17 (6) Thrush 1/46 (2) Ledermann Intact endplate 2/39(5) Dagirmanjian 1/46 (2) Ledermann Epidural inflammation only

Our group (17 patients)(%) 6/17 (35)

6/17(35) 10/14 (71.5) 1/17 (6) 2/17(12) 7/17(41) 0

29

BIBLIOGRAPHY 1. Aguila L.A., Piraino D.W., Modic M.T., Dudley A.W., Duchesneau P.M., Weinstein M.A. The Intranuclear Cleft of the Intervertebral Disk : Magnetic Resonance Imaging. Radiology 1985; 155: 155-158. 2. Bird P.A., ShnierR., Edmonds J. Questionning the Sensitivity of Magnetic Resonance Imaging in Early Septic Spondylodiscitis J.Clin.Rheumatology 2001; 7 (3): 184-7 3. Champsaur P., Parlier-Cuau C.,Juhan V.,Daumen-Legre V.,Chagnaud C. and al Differential diagnosis of erosive intervertebral osteochondrosis vesus infectious discitis J.Radiol. 2000; 81(5): 516-22 4. Dagirmanjian A., Schils J., McHenry M., Modic M.T. MR Imaging of Vertebral Osteomyelitis Revisited. AJR 1996; 167: 1539-1543. 5. De Roos A., Kressel H., Spritzer C., Dalinka M. MR Imaging of Marrow Changes Adjacent to Endplates in Degenerative Lumbar Disk Disease. AJR 1987; 149: 531-534. 6. Frank Andreas M., Trappe A.E. The role of Magnetic Resonance Imaging (MRI) in the diagnosis of spondylodiscitis Neurosurg.Rev 1990; 13: 279-283 7. Friedman J.A., Maher C.O., Quast L.M. and all. Spontaneous Disc Space Infections in Adults Surg Neurol 2002; 57:81-6 8. Gillams A.R., Chaddha B., Carter A.P. MR Appearances of the Temporal Evolution and Resolution of Infectious Spondylitis. AJR 1996; 166:903-907. 9. Kamani I., Syed I., Saifuddin A., Green R., MacSweeney F. Vertebral osteomyelitis without disc involvement. Clin radiology 2004; 59: 881-891 10. Kothari N.A., Pelchovitz A.J., Meyer J.S. Imaging of Musculoskeletal Infections. Radiologic Clinics of North America 2001; vol 39; 653-670. 11. Ledermann H.P., Schweitzer M.E., Morrison W.B., Carrino J.A. MR Imaging Findings in Spinal Infections : Rules or Myths? Radiology 2003;228:506-14 12. Longo M., Granata F.,Ricciardi G.K., Gaeta M., Blandino A. Contrast-enhanced MR imaging with fat sat suppression in adult-onset septic spondylodiscitis Eur Radiol 2003; 13: 626-637. 13. Maiuri F., Iaconetta G., Gallicchio B., Manto A., Briganti F. Spondylodiscitis: Clinical and Magnetic Resonance Diagnosis Spine 1997; 22: 1741-1746. 14. Mellado JM, Perez del Palomar L, Camins A, Salvado E, Ramos A, Sauri A. MR imaging of spinal infection: atypical features, interpretive pitfalls and potential mimickers. Eur Radiol. 2004;14):19801989. 15. Meyers S.P., Wiener S.N. Diagnosis of Hematogenous Pyogenic Vertebral Osteomyelitis by Magnetic Resonance Imaging Arch Intern Med 1991; 12: 1087-1093. 16. Michael A.S., Mikhael M.A. Spinal Osteomyelitis: Unusual Findings on Magnetic Resonance Imaging Comput Med Imaging Graph 1988; 12: 329-31. 17. Modic M.T., Feiglin D.H., Piraino D.W., et al. Vertebral Osteomyelitis : Assessment Using MR Radiology 1985; 157: 157-166. 30

18. Modic M.T., Masaryk T.J., Ross J.S., Carter J.R. Imaging of Degenerative Disk Disease Radiology 1988; 168: 177-186. 19. Nolla-Sole JM, Mateo-Soria L, Rozadilla-Sacanell A, Mora-Salvador J, Valverde-Garcia J, RoigEscofet D.Role of technetium-99m diphosphonate and gallium-67 citrate bone scanning in the early diagnosis of infectious spondylodiscitis. A comparative study. Ann Rheum Dis. 1992;51(5):665-667. 20. Post M.J., Sze G., Quencer R.M., Eismont F.J., Green B.A., Gahbauer H. Gadolinium-Enhanced MR in Spinal Infection. J Comput Assist Tomogr 1990; 14: 721-729. 21. Schindler O.S., Wilson-MacDonald J. Acute MRI; changes in infectious discitis: report on two cases Eur Spine J 1995; 4: 360-361 22. Sharif H.S. Role of MR Imaging in the Management of Spinal Infections AJR Am J Roentgenol 1992; 158: 1333-1345. 23. Shih T.T., Huang K-M., Hou S-M. Early Diagnosis of Single Segment Vertebral Osteomyelitis-MR Pattern and its Characteristics Clin Imaging 1999; 23: 159-67. 24. Stäbler A., Doma A.B., Baur A., Krüger A., Reiser M. Reactive Bone Marrow Changes in Infectious Spondylitis: Quantitative Assessment with MR Imaging Radiology 2000; 217: 863-868 25. Stäbler A., Reiser M.F. Imaging of Spinal Infection Radiol Clin North Am 2001; 39: 115-135. 26. Stäbler A., Weiss M., Scheidler J., Krödel A., Seiderer M., Reiser M. Degenerative disk vascularization on MRI: correlation with clinical and histopathologic findings. Skeletal radiol 1996; 25: 119-126. 27. Tali T.E. Spinal Infections. Eur Journal of radiology 2004, 50: 120-33Thrush A., Enzmann D. MR Imaging of Infectious Spondylitis. AJNR Am J Neuroradiol 1990; 11: 1171-1180. 28. Thrush A., Enzmann D. MR Imaging of Infectious Spondylitis AJNR Am J Neuroradiol 1990; 11: 1171-1180. 29. Tyrrell P.N.M., Cassar-Pullicino V.N., McCall I.W. Spinal Infection. Eur Radiol 1999; 9: 1066-77. 30. Varma R., Lander P., Assaf A. Imaging of Pyogenic Infectious Spondylodiscitis Radiologic Clinics of North America 2001; 39-2:203-213

31

FIGURE LEGENDS Fig 1 : Typical MR appearance of an “established” infectious spondylodiscitis (IS). Sagittal T1 (a), T2 with fat suppression (b) and contrast enhanced T1(c) images show the typical and "complete" picture of IS. Bone involvement combine extensive bone marrow edema (BME) within to adjacent vertebral bodies, and endplate erosions (EE). The disk shows significant height loss (HL), absent central "cleft" (AC) on the T2 image (a cleft is visible in the adjacent disks), high signal intensity on T2 (HI) and areas of enhancement (CE) on postcontrast T1 images. Paraspinal (PS) and epidural (ED) soft tissues show mass effect. Fig 2 : Bone marrow edema (BME). Sagittal T1 (a), T2 with fat suppression (b) and contrast enhanced T1(c) images show discrete anterior BME. Limited areas of low signal intensity are visible on the T1 image near the anterior vertebral corners adjacent to the L4-L5 disk; they show high signal intensity on T2 and enhancement on post-contrast images (arrows in a,b,c). Note the infiltration of the paraspinal soft tissues (arrowheads). Fig 3 : BME : discrete posterior involvement and superiority of fat suppressed T2-weighted images over “simple” T2 images. Sagittal T1 (a), T2 (b) and T2 with fat suppression (c) images show discrete areas of low signal intensity on the T1 images (arrows in a); almost no abnormality is seen on the FSE T2 images (b), whereas areas of BME becomes evident on the fat saturated T2 image (arrows in c). Note discrete infiltration of the epidural soft tissue (arrowhead in a and b), discrete (< 50%) loss of the disk height, and focal high signal intensity within the posterior aspect of the disk (curved arrow in b and c). Fig 4 : Endplate erosion. Sagittal T1 (a), T2 with fat suppression (b) and contrast enhanced T1(c) images. Focal interruption of the anterior segment of the superior vertebral endplate is evident on the T1image (arrow in a), but is much less evident on the corresponding T2 and contrast-enhanced T1 images. Note extensive BME within the both adjacent vertebral bodies (arrowheads), discrete infiltration of the anterior paraspinal soft tissues (curved arrows in a and b), and almost normal disk appearance. Fig 5 : Disk abnormalities Sagittal T1 (a), T2 with fat suppression (b), and contrast enhanced T1(c) images. Predominant involvement of the L3-L4 disk : height loss superior to 50%, disappearance of the central "cleft", high signal intensity involving the central portion of the disk (nucleus). Discrete BME infiltration of the superior aspect of the L4 vertebral body (arrowhead in a) showing central high signal intensity area on T2 images (arrowhead in b) and central avascular area on postcontrast image (arrowhead in c) : small intravertebral abscess. Note discrete areas of enhancement within the posterior aspect of the disk on the post-contrast image (curved arrow in c). Note absence of evident abnormalities within L3 and paraspinal soft tissues. Heterogeneity in the bone marrow signal intensity was due to recent bone marrow transplant for multiple myeloma. Fig 6 : Evident disk enhancement after contrast injection. Sagittal T1 (a) and post-contrast T1 (b) images show evident enhancing areas within the disk (arrowheads in b). Note extensive BME within adjacent vertebral bodies (arrows in a) and involvement of the epidural soft tissues, considered as a phlegmon (absence of central avascular area) (curved arrow in b). Fig 7 : Disk abnormalities: loss of the central nuclear cleft and increased signal on T2 images illustrated in two different patient.

32

Sagittal fat-saturated T2 images (a) and FSE T2 (b) in two different patients show disappearance of the central cleft within an infected disk : L1-L2 in (a) and T12-L1 in (b) (arrows), and increased disk signal intensity (diffuse in a, focal anterior in b). Note associated BME within the adjacent vertebrae (arrowheads in a) and epidural abscess in a (curved arrow in a) and limited BME in b (arrowhead in b). Fig 8 : Paraspinal soft tissue involvement (with abscess of the psoas muscle). Transverse T1 (a), T2 (b) and post-contrast T1 (c) images show infiltration of the left lateral paraspinal soft tissues (arrowheads in a); T2 and post-contrast T1 images show central fluid-like area without enhancement after injection (arrows in b and c) typical for an abscess within the proximal portion of the psoas muscle. Note the concurrent presence of an anterior epidural abscess (curved arrow in b and c). Fig 9 : Anterior epidural abscess. Sagittal T1 (a), TSE T2 (b) and post-contrast T1 (c) images show discrete narrowing of the T12-L1 disk (arrow in a,b,c) and BME within the adjacent vertebral bodies (arrowheads in a). Extensive ascending epidural abscess is seen on T2 and post-contrast T1 images only (curved arrow in b and c).

33