lable at ScienceDirect

Clinical Radiology 65 (2010) 974e981

Contents lists avai

Clinical Radiology

journal homepage: www.elsevierheal th.com/journals /crad

Original Paper

The MRI appearances of early vertebral osteomyelitisand discitisJ.A.T. Dunbar a, J.A.T. Sandoe a, A.S. Rao b, D.W. Crimmins c, W. Baig d, J.J. Rankine e, f,*

aDepartment of Microbiology, Leeds General Infirmary, Leeds LS1 3EX, UKbDepartment of Orthopaedics, Leeds General Infirmary, Leeds LS1 3EX, UKcDepartment of Neurosurgery, Leeds General Infirmary, Leeds LS1 3EX, UKdDepartment of Cardiology, Leeds General Infirmary, Leeds LS1 3EX, UKeDepartment of Radiology, Leeds General Infirmary, Leeds LS1 3EX, UKf Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK

article information

Article history:Received 4 November 2009Received in revised form5 March 2010Accepted 15 March 2010

* Guarantor and correspondent: J.J. Rankine,Leeds General Infirmary, Leeds LS1 3EX, UK. Teþ44 113 392 8241.

E-mail address: james.rankine@leedsth.nhs.uk

0009-9260/$ e see front matter � 2010 The Royal Codoi:10.1016/j.crad.2010.03.015

AIM: To describe the magnetic resonance imaging (MRI) appearances in patients witha clinical history suggestive of vertebral osteomyelitis and discitis who underwent MRI veryearly in their clinical course.MATERIALS AND METHODS: A retrospective review of the database of spinal infections from

a spinal microbiological liaison teamwas performed over a 2 year period to identify cases withclinical features suggestive of spinal infection and an MRI that did not show features typical ofvertebral osteomyelitis and discitis. All patients had positive microbiology and a follow up MRIshowing typical features of spinal infection.RESULTS: In four cases the features typical of spinal infection were not evident at the initial

MRI. In three cases there was very subtle endplate oedema associated with disc degeneration,which was interpreted as Modic type I degenerative endplate change. Intravenous antibiotictherapy was continued prior to repeat MRI examinations. The mean time to the repeatexamination was 17 days with a range of 8e22 days. The second examinations clearlydemonstrated vertebral osteomyelitis and discitis.CONCLUSION: Although MRI is the imaging method of choice for vertebral osteomyelitis and

discitis in the early stages, it may show subtle, non-specific endplate subchondral changes;a repeat examination may be required to show the typical features.

� 2010 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Introduction

Vertebral osteomyleitiswith discitis (VO) is an uncommonbut serious bacterial infection. Recently the incidence hasrisen markedly, which may be due to immunosuppressionand an ageing population.1 When VO is untreated outcomes

Department of Radiology,l.: þ44 113 392 3768; fax:

(J.J. Rankine).

llege of Radiologists. Published by

are poor. Mortality ranges from 2e20% and there are highrates of morbidity in survivors.2 Diagnosis is frequentlydelayed due to the indolent nature of the disease.3

A high clinical suspicion must be maintained as bothblood inflammatory markers and radiographs are known tobe insensitive diagnostic tools. Blood culture results willoften raise the suspicion of a spinal focus of infection.4,5

Staphylococcus aureus is the most common organism,accounting for between 55e80% of infections, althoughalmost any bacteria can be implicated. Treatment usuallyrequires 6 weeks of intravenous antibiotic therapy.6

Elsevier Ltd. All rights reserved.

Table 1Summary of the results of initial investigations of the four cases of vertebralosteomyelitis and discitis

Case Admission c-reactiveprotein (mg/l)

Admission white cellcount (�109/l)

Blood cultures(sets)

1 224 11.65 Enterococcusfaecalis (3)

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981 975

In recent years magnetic resonance imaging (MRI) hasproved itself to be an invaluable tool in the diagnosis of spinalinfection.7 One series showed a sensitivity of 97.7% fordetection of infection using MRI.8 The typical features ofspinal infection are contiguous involvement of two vertebraewith inflammatory changewithin the intervertebral disc, butthese are relatively chronic changes that may take weeks ormonths to develop. There is very little in the literature con-cerning the appearances of very early spinal infection withonly a few examples presented in one paper.9 This studypresents four cases where a clinical diagnosis of spinalinfectionwasmadewith supporting blood culture isolates. Ineach case early MRI images were not suggestive of infection,casting doubt on the initial diagnosis. In all four cases intervalexaminations confirmed the original clinical diagnosis of VO.

Materials and methods

The study was conducted in a large teaching hospitalwith a specialist spinal surgery unit and a dedicated spinalmicrobiological liaison team. The microbiology team havekept records of all cases of spinal infection over a 2 yearperiod. Cases where the initial MRI examination was notsuggestive of VO and repeat MRI was performed showingtypical features of VO were identified. The clinical notes andthe MRI images of these patients were reviewed.

All MRI examinations were performed using a Siemens1.5 T Magnetom Avanto (Siemens, Medical Solutions, For-cheim, Germany) using turbo spin-echo T2-weightedsequences and unenhanced and gadolinium-enhanced,Dotarem (gadoteric acid) 10 ml, T1-weighted sequences inthe sagittal and axial plane. In two cases, T2 fat-saturationsequences were used They were all reported by a specialistmusculoskeletal radiologist with 15 years of experience.MRI findings of spinal osteomyelitis were recorded. Thesewere bone oedema particularly related to the endplate,abnormal signal within the intervertebral disc, abnormalenhancement of the endplate and intervertebral disc andthe presence of abscess formation.

Institutional ethics review was not required for thisretrospective study. All patients gave written consent forpublication.

Results

Approximately 40 cases of haematogenous vertebralosteomyelitis and discitis are seen in our institution eachyear. Four cases were identified that had a clinical suspicionof VO where initial MRI was not suggestive of infection. Inall four cases intravenous antibiotic therapy was continuedprior to repeat MRI examinations. The mean time to repeatexamination was 17 days with a range of 8e22 days. Thesecond MRI images clearly demonstrated VO.

2 386 10.29 Staphylococcusaureus (2)

3 207 8.77 Staphylocccusaureus (2)

4 91 13.4 Streptococcusgordonii(3)

Case 1

A 76-year-old man was admitted with a 4 day history ofback pain, fever, and rigors. His general practitioner had

prescribed cefalexin for a presumed urinary tract infection3 days prior to admission. He had no history of back trauma.Previously he had been diagnosed with cervical osteoar-thritis, asbestosis, benign prostatic hypertrophy, and mitralregurgitation.

On examination he was mildly febrile with a tempera-ture of 37.5o centigrade. His pulse and blood pressure werewithin normal limits. He had a pan-systolic cardiac murmurbut cardiovascular and respiratory examinations wereotherwise unremarkable. He was tender over the L3eL5vertebrae. His baseline investigations are summarized inTable 1. His blood cultures grew Enterococcus faecalis.

Initial MRI images (Fig. 1) showed disc dehydrationaffecting the L3eL4 intervertebral disc with an adjacentextra-dural area of high T2-weighted signal that wasinterpreted as a disc extrusion of high water content. Thegadolinium-enhanced images showed a small focus ofenhancement of the posterior annulus of the L3eL4 andL1eL2 discs. A repeat MRI examination 18 days latershowed a clear discitis at L3eL4 and an epidural abscessextending from L1eL4 (Fig. 2).

He had a trans-thoracic echocardiogram, whichwas non-diagnostic, but a trans-oesophageal echocardiogramshowed mitral valve vegetation. Spinal infection wasconsidered to be secondary to endocarditis. He was treatedwith a 6 week course of intravenous amoxicillin andgentamicin. He was discharged pain free, mobile, andwithout neurological deficit.

Case 2

A 67-year-old man presented with severe low back painand leg weakness. He had a history of laminectomy fora prolapsed cervical disc 4 years earlier. On examination hewas clearly septic with fever, peripheral hypoperfusion, andreduced conscious level. He was tender over his L3eL4vertebrae. His investigations are summarized in Table 1. Hewas also found to have acute kidney injury with a creatinineof 350 mmol/l. His blood cultures were repeatedly positivefor Staphylococcus aureus despite appropriate antibiotictherapy.

His initial MRI examination (Fig. 3) showed L2eL3 andL3eL4 intervertebral disc prolapses and central canalstenosis. There was a very small amount of endplateoedema posteriorly at the L3eL4 level, but no enhancementof the intervertebral disc and the disc showed changes of

Figure 1 MRI images that were interpreted as L3eL4 disc extrusion of high water content. (a) Sagittal T2-weighted image. There is discdehydration affecting the L3eL4 intervertebral disc with an adjacent extra-dural area of high T2-weighted signal that was interpreted as a discextrusion of high water content. (b) T1-weighted and (c) T1-weighted post-gadolinium administration images. The gadolinium-enhanced imageshows a small focus of enhancement of the posterior annulus of the L3eL4 and L1eL2 discs. (d) Axial, T1-weighted, gadolinium-enhanced MRIsequences. There is ring enhancement around an area of intermediate signal intensity within the anterior aspect of the spinal canal, which wasinterpreted as extruded disc.

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981976

degeneration with dehydration. The endplate oedema wasinterpreted as Modic type 1 degenerative endplate change.A repeat examination performed 8 days later (Fig. 4)showed high fluid signal within the L3eL4 disc and anincrease in the endplate oedema. A small epidural abscesshad developed.

Two weeks after his presentation when his severe sepsishad improved his central canal stenosis was surgicallydecompressed. Pus drained from the paravertebral abscess

Figure 2 Same patient as Fig. 1 examined using MRI 18 days later. (a) Sagienhanced MRI sequences. VO at L3eL4 with enhancement of the posteri

grew Staphylococcus aureus. The patient was successfullytreated with 6 weeks of intravenous flucloxacillin and oralrifampicin.

Case 3

A 28-year-old woman was admitted with a 4 day historyof rapidly increasing low back pain and associated fevers.On the fifth day she had developed weakness in both legs

ttal, T2-weighted; (b) T1-weighted; and (c) T1-weighted, gadolinium-or aspects of the endplates and epidural abscess from L1 to L4.

Figure 3 MRI images interpreted as degenerative disc disease with disc protrusion and Modic type I change at L3eL4. (a) Sagittal, T2-weighted;(b) T1-weighted; and (c) T1-weighted, gadolinium-enhanced MRI sequences indicating that there are disc prolapses at L2eL3 and L3eL4 andcentral canal stenosis. There is a very small amount of endplate oedema posteriorly at the L3eL4 level, but no enhancement of the intervertebraldisc, and the disc showed changes of degeneration with dehydration.

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981 977

and urinary retention. She had a history of severe eczemafor which she was taking cyclosporin and had recently beentreated with low-dose oral prednisolone.

On examination she was febrile at 39 �C. She wastachycardic but normotensive. Her investigations are shownin Table 1. She also had acute kidney injury with a creatinineof 121 mmol/l. Her initial clinical diagnosis was spinalinfection later supported by two positive blood cultures

Figure 4 Same patient as Fig. 3 examined using MRI 8 days later. (a) Sagitenhanced MRI sequences showing VO at L3eL4 affecting the posterior asp

with Staphylococcus aureus. She was started on high-doseintravenous flucloxacillin.

Her initial MRI examination (Fig. 5) showed disc spacenarrowing at L1eL2 and L2eL3 with endplate Schmorl’snodes. The L2eL3 intervertebral disc showed disc dehy-dration and the L1eL2 disc showed a small amount ofnucleus pulposus of normal signal. There was very subtleoedema and diffuse contrast enhancement of the lower

tal, T2-weighted; (b) T1-weighted; and (c) T1-weighted, gadolinium-ect of the endplates, fluid centred on the disc, and an epidural abscess.

Figure 5 MRI images of a 28-year-old woman with a 4 day history of rapidly increasing low back pain and fever showing subtle oedema relatedto a Schmorl’s node. (a) Sagittal, T2-weighted; (b) sagittal, T1-weighted; and (c) sagittal, T1-weighted, gadolinium-enhanced MRI sequencesindicating that the L2eL3 intervertebral disc shows disc dehydration and the L1eL2 disc shows a small amount of nucleus pulposus of normalsignal. There is very subtle oedema and diffuse contrast enhancement of the lower aspect of the L1 vertebral body associated with a Schmorl’snode. There is no enhancement within the intervertebral disc. There is a Schmorl’s node in the superior endplate of L2 with fat signal around itsuggesting that this is a long-standing feature.

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981978

aspect of the L1 vertebral body associated with a Schmorl’snode. There was a Schmorl’s node in the superior endplateof L2 with a fat signal around it suggesting that this wasa long-standing feature.

In view of the clinical suspicion of VO, she continued toreceive intravenous flucloxacillin as if she had a spinalinfection. A repeat examination 22 days later showed very

Figure 6 Same patient as Fig. 5 examined using MRI 22 days later. (a) Ssagittal, T1-weighted, gadolinium-enhanced MRI sequences showing L1e

clear L1eL2 discitis and VO (Fig. 6). She made a rapidimprovement.

Case 4

This 48-year-old man was admitted with a 6 weekhistory of fever, malaise, and weight loss of over 10 kg. He

agittal, T2-weighted, fat-saturated; (b) sagittal, T1-weighted; and (c)L2 discitis and VO.

Figure 7 MRI images of a 48-year-old man with a 6 week history of fever, malaise, and weight loss. (a) Sagittal, T2-weighted, fat-saturated; (b)sagittal, T1-weighted; and (c) sagittal, T1-weighted, gadolinium-enhanced MRI sequences showing subtle bone enhancement at L5/S1 with poordefinition of the central portion of the S1 endplate. (d) Right para-sagittal, T2-weighted, fat-saturated and (e) right para-sagittal, T1-weighted,gadolinium-enhanced MRI sequences showing a focal area of bone oedema and endplate enhancement.

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981 979

had associated diarrhoea and vomiting. He was know tohave severe aortic regurgitation and was awaiting aorticvalve replacement after prophylactic dental treatment hadbeen completed. On examination he was febrile andtachycardic but normotensive. He had a diastolic aortic

Figure 8 Same patient as Fig. 7 examined using MRI 19 days later. (a) Ssagittal, T1-weighted, gadolinium-enhanced MRI sequences showing thaendplate erosion and anterior cortical vertebral body destruction.

murmur but no peripheral stigmata of endocarditis. Initialinvestigations are listed in Table 1. Three sets of bloodcultures all grew Streptococcus gordonii.

The patient’s transthoracic echocardiogram showedvegetation on the aortic valve. He was started on high-dose

agittal, T2-weighted, fat-saturated; (b) sagittal, T1-weighted; and (c)t the endplate oedema has progressed and there is now evidence of

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981980

benzyl-penicillin. Initially he appeared to respond well totherapy but 3e4 days into his admission he complained ofpain and reduced range of movement in his right hip andnumbness over the right buttock. MRI on day 6 showeda small focal area of endplate oedema present on two of theright para-sagittal images at the L5/S1 level (Fig. 7). The L5/S1 intervertebral disc showed degenerative changes but noevidence of inflammation. The appearances were inter-preted as a focal area of Modic type I degeneration. Anti-biotics were continued and on day 25 the MRI examinationwas repeated (Fig. 8). The endplate inflammation hadworsened and there was irregularity of the endplates. Adiagnosis of VO was made.

The patient continued to have a very stormy clinicalcourse with emergency laparotomy to repair a rupturedmesenteric artery mycotic aneurysm. Following a pro-longed stay in the intensive therapy unit, he hada successful aortic valve replacement on day 50. He remainsin hospital over 80 days after his initial presentation but isnow making satisfactory progress. He has no back pain orneurological deficit.

This patient’s discitis and vertebral osteomyelitis isassumed to be a secondary embolic phenomenon from hisaortic valve endocarditis.

Discussion

MRI is reported to be highly sensitive for the diagnosis ofVO. The present cases confirmMRI imagesmay be equivocalearly in the course of infection. It is imperative wherea history is consistent with VO, particularly if supported bypositive blood cultures, that intravenous antibiotic therapyis continued and a repeat MRI performed.

The initial MRI images were abnormal in all of these cases,but were not suggestive of spinal infection. When the MRIimageswerereviewedretrospectively therewasclearevolutionof VO. The atypical findings could be reflective of the earlypresentation and diagnosis of each case. Their successfulmanagement was the result of a multidisciplinary teamapproach, including spinal surgery, radiology, and clinicalmicrobiology. In previous studies there has generally beena longer time from first symptoms to MRI, with a delay indiagnosisof4monthsbeingtypical.10Thisdelaymayhavegivenabiasedexpectationof thesensitivityofMRI, aspreviouslymostpatients underwent MRI well into their clinical course.

Spinal infection starts as an osteomyelitis of the endplateand then progresses to the well-recognized features ofdiscitis. Three of the present cases showed very subtleendplate oedema associated with disc degeneration, whichis frequently seen in Modic type I degenerative endplatechange. A recent study demonstrated that in 36% of cases ofModic type I change the intervertebral disc showed a para-doxical increased in the T2-weighted signal, an appearancethat can mimic VO.11 Endplate oedema is also reported inassociation with Schmorl’s nodes.12 It is important torecognize that these may be the very earliest signs of VOand follow up these changes with a repeat MRI when thereis a clinical suspicion of spinal infection. A progression in

the appearances is very suggestive of infectionwith markedchanges occurring in as little as 8 days.

In one of the present cases a small epidural abscessrelated to the posterior annulus was misdiagnosed as anextruded disc. Extruded disc material can have a fluidcontent forming a disc cyst.13 This case also showed someenhancement of the annulus of two of the discs; a findingthat is not specific to infection and also found in annulartears. Contrast enhancement of the epidural space has alsobeen demonstrated in association with annular tears 14

causing further confusion with infection. In retrospect, theextent of the enhancement of the disc would have beenunusual for an annular tear in the absence of a history ofdiscectomy at this level. In all three cases the use of gado-linium has helped to demonstrate these very early, albeitnon-specific, changes of VO. T2-weighted, fat-saturatedsequences, or short tau inversion recovery (STIR) sequences,are likely to be as sensitive at detecting bone oedema asunenhanced and gadolinium-enhanced T1-weightedsequences, as demonstrated in cases 3 and 4. The greatestvalue of gadolinium is probably in the detection of earlyenhancement of the disc as demonstrated in Fig. 1.

The pathophysiology of infecting organisms could affectthe MRI appearances. It is logical to assume that moreinvasive organisms might produce more widespread ormore rapidly progressive MRI abnormalities. However, inthe present series disease progression was evident withboth virulent pathogens (Staphylococcus aureus) and lessvirulent organisms (Enterococcus and oral Streptococcus.)We have insufficient data to relate rapidity of change tospecific pathogens. Some organisms, for example, Strepto-coccus milleri, have a higher propensity to pus formation.

Repeat MRI was found to be useful in all of the presentcases. There are no published data to guide the timing ofrepeat MRI, but the range in the present series from 8e22days was sufficient in each case to demonstrate convincingprogression. Where clinical suspicion remains high despitea non-diagnostic early MRI, it is reasonable to treat for VOand follow up the patient with repeat MRI.

Acknowledgements

This study was part supported by the National Institutefor Health Research.

References

1. Darouiche RO. Spinal epidural abscess. N Eng J Med 2006;355:2012e20.2. Butler JS, Shelly MJ, Timlin M, et al. Nontuberculous pyogenic spinal

infection in adults: a 12-year experience from a tertiary referral centre.Spine 2006;31:2695e700.

3. Davis DP, Wold RM, Patel RJ, et al. The clinical presentation and impactof diagnostic delays on emergency department patients with spinalepidural abscess. J Emerg Med 2004;26:285e91.

4. An HS, Seldomridge JA. Spinal infections: diagnostic tests and imagingstudies. Clin Orthopaed Relat Res. 2006;444:27e33.

5. Sapico FL, Montgomerie JZ. Vertebral osteomyelitis. Infect Dis Clin N Am.1990;4:539e50.

6. Berbari EF, Steckleberg JM, Osmon DR. Osteomyelitis. In: Mandell GL,Dolin R, Bennett JE, editors. Principles and practice of infectious diseases.

J.A.T. Dunbar et al. / Clinical Radiology 65 (2010) 974e981 981

6th ed. Philadelphia, PA: Elsevier, Churchill Livingstone; 2005. p.1322e32.

7. Rankine JJ. MRI of spinal infection. Curr Orthopaed 2004;18:426e33.8. Ledermann HP, Schweitzer ME, Morrison WB, et al. MR imaging

findings in spinal infections: rules or myths. Radiology 2003;228:506e14.

9. Hong SH, Choi J, Lee JW, et al. MR imaging assessment of the spine:infection or imitiation? RadioGraphics 2009;29:599e612.

10. Rankine JJ, Barron DA, Robinson P, et al. Therapeutic impact ofpercutaneous spinal biopsy in spinal infection. Postgrad Med J 2004;80:607e9.

11. Luoma K, Vehmas T, Grönblad M, et al. Relationship of Modic type 1change with disc degeneration: a prospective MRI study. Skeletal Radiol2009;38:237e44.

12. Wu HT, Morrison WB, Schweitzer ME. Edematous Schmorl’s nodes onthoracolumbar MR imaging: characteristic patterns and changes overtime. Skeletal Radiol 2006;35:212e9.

13. Khalatbari K, Ansari H. MRI of degenerative cysts of the lumbar spine.Clin Radiol 2008;63:322e8.

14. Saifuddin A, Mitchell R, Taylor BA. Extradural inflammation associatedwith annular tears: demonstration with gadolinium-enhanced lumbarspine MRI. Eur Spine J 1999;8:34e9.