Lumbar spinal brucellosis presenting as an atypical epidural abscess

Summary

Brucellosis is a zoonotic infection that affects the spine, typically presenting as spondylodiscitis. Predominantly, epidural involvement without significant concurrent disc or vertebral involvement is rare and may closely mimic non-infectious pathologies. We report a man in his 50s presenting with low back pain and lower limb radiculopathy. Initial MRI demonstrated an epidural-predominant ventral epidural collection at L4–L5 associated with a posterior annular fissure, without significant vertebral or disc involvement, mimicking a sequestered disc fragment. Diffusion-weighted imaging showed peripheral restriction, and post-contrast imaging demonstrated thick rim enhancement, favouring infection. Follow-up MRI at 10 days revealed new vertebral marrow oedema consistent with evolving spondylodiscitis. Blood cultures confirmed Brucella species. Despite antibiotic therapy, progressive neurological and radiological deterioration necessitated surgical decompression. Histopathology confirmed an infected epidural collection. This case illustrates a rare epidural-first presentation of spinal brucellosis and highlights MRI discriminators and the role of short-interval follow-up imaging in diagnostically challenging epidural lesions.

Background

Brucellosis, caused by Brucella species, is an endemic zoonotic infection in many regions, particularly the Middle East and Mediterranean basin.¹ The disease often presents with non-specific clinical features, contributing to delayed diagnosis. Musculoskeletal involvement is the most common focal manifestation, and the spine is the most frequently affected musculoskeletal site.^{2 3} Spinal brucellosis typically manifests as spondylodiscitis, most commonly involving the lumbar spine. Epidural involvement is well recognised but usually occurs as a secondary extension from adjacent vertebral body or intervertebral disc infection.^{3 4} In contrast, epidural-predominant disease as the initial imaging manifestation is uncommon and may pose a significant diagnostic challenge. Atypical epidural collections can closely mimic non-infectious conditions, including sequestered disc fragments or epidural haematomas, particularly when early imaging lacks the characteristic features of spondylodiscitis.^{5 6} Recent case reports continue to show that brucellar epidural abscess remains rare and diagnostically challenging, especially in endemic regions.^6–8 This report highlights an unusual epidural-first presentation of spinal brucellosis, with short-interval imaging demonstrating subsequent vertebral involvement, and emphasises the importance of integrating MRI evolution with microbiological and histopathological findings.

Case presentation

A man in his 50s with a history of dyslipidaemia and coronary artery disease, who had undergone coronary stent placement 10 years earlier and was maintained on long-term aspirin therapy, presented with progressive low back pain following heavy lifting. The pain radiated to the right lower limb and was associated with difficulty ambulating. The patient later experienced a transient episode of urinary incontinence. He denied bowel incontinence, persistent urinary retention, saddle anaesthesia, recent trauma or constitutional symptoms. He reported consuming unpasteurised milk. A chronological summary of the key clinical, microbiological and radiological events is provided in the clinical timeline (table 1).

Table 1. Chronological timeline of key clinical, microbiological and radiological events.

Timepoint	Key clinical, radiological and microbiological events
Day 0	Presentation with progressive low back pain and right lower limb radiculopathy following heavy lifting. Transient urinary incontinence noted. History of unpasteurised milk consumption disclosed.
Day 0–1	Neurological examination: mild right lower limb weakness (ankle dorsiflexion and extensor hallucis longus, power 4+/5). Afebrile and haemodynamically stable. ESR 55 mm/hour, CRP 86.4 mg/L, procalcitonin within normal limits. Initial lumbar MRI: ventral epidural collection at L4–L5 mimicking sequestered disc fragment; heterogeneous peripheral diffusion restriction and thick peripheral enhancement on advanced sequences.
Day 1–3	Blood cultures obtained. Infectious diseases consultation requested.
Day 3–5	Blood cultures confirmed Brucella species. Intravenous gentamicin (390 mg daily; weight-adjusted dosing) and oral doxycycline (100 mg two times per day) were initiated.
Day 10	Follow-up MRI: interval increase in epidural collection; new L4 vertebral body marrow oedema and patchy basivertebral enhancement consistent with evolving spondylodiscitis. Ongoing clinical deterioration.
Days 10–14	Progressive neurological and radiological deterioration despite antimicrobial therapy. Cauda equina compression worsening. Surgical decompression planned.
Surgery	L4–L5 posterior laminectomy and debulking of epidural collection. Intraoperative findings consistent with an epidural inflammatory/infected collection.
Postoperative	Histopathology: acute inflammation, granulation tissue, haemorrhage and necrotic debris; no malignancy. Fungal stains, Ziehl–Neelsen stain and acid-fast bacilli (AFB) culture negative. Surgical microbiology negative (attributed to preceding antimicrobial therapy). Commenced prolonged oral combination therapy: doxycycline 100 mg two times per day, rifampicin 600 mg once daily, TMP–SMX DS two tablets two times per day.
2-month follow-up	Patient ambulatory with no new neurological deficits. Mild residual lower limb weakness. Blood and Brucella cultures negative. Continuing prolonged antimicrobial therapy (minimum 6-month course planned).

Clinical timeline

The clinical timeline summarises the key clinical events, investigations, imaging findings and treatment during the patient’s hospital course.

Investigations

On examination, the patient was afebrile and haemodynamically stable. Neurological examination revealed mild right lower limb weakness, particularly involving ankle dorsiflexion and extensor hallucis longus power of 4+/5, with intact sensation and preserved reflexes. Laboratory investigations demonstrated elevated inflammatory markers, with an erythrocyte sedimentation rate of 55 mm/hour and C-reactive protein of 86.4 mg/L, which subsequently peaked at 113.3 mg/L during the early disease course. Procalcitonin remained low. Blood cultures were positive for Brucella species, establishing microbiological diagnosis of brucellosis in the appropriate clinical and epidemiological context. Additional serological testing was not pursued because the diagnosis had already been confirmed microbiologically. Repeat blood and Brucella cultures after initiation of antimicrobial therapy were negative.

Initial lumbar MRI demonstrated a ventral epidural collection at L4–L5, closely associated with a posterior annular fissure, without early disc signal abnormality or adjacent endplate oedema (figure 1). Additional MRI sequences further characterised the lesion: diffusion-weighted imaging (DWI)/apparent diffusion coefficient (ADC) supported restricted diffusion (figure 2), susceptibility-weighted imaging/T2* demonstrated peripheral susceptibility artefacts (figure 3) and post-contrast imaging showed thick peripheral enhancement (figure 4). Together, these features favoured an infected epidural process over a simple disc fragment or isolated haematoma. Follow-up MRI performed after 10 days showed mild interval enlargement of the collection with new L4 vertebral marrow oedema and patchy enhancement centred around the basivertebral venous region, consistent with evolving vertebral involvement (figure 5). Subsequent preoperative MRI during clinical deterioration demonstrated further progression of spondylodiscitis, enlargement of the epidural abscess, increased spinal canal stenosis and worsening cauda equina compression (figure 6). These findings led to surgical decompression.

Figure 1. Sagittal T1-weighted (A) and T2-weighted (B) images show an epidural fluid collection at L4–L5 (white arrow) associated with a posterior annular fissure of the L4–L5 disc (red arrow), mimicking an extruded disc fragment. Authors’ original images.

Figure 2. Sagittal T2-weighted fat-suppressed image (A) demonstrates preserved signal and morphology of the disc aside from posterior annular fissure. The vertebral bone marrow demonstrates no conspicuous signal abnormality aside from the posterior annular fissure. Diffusion-weighted imaging with ADC map (B) shows the epidural collection exhibiting heterogeneous, predominantly peripheral diffusion restriction. Authors’ original images.

Figure 3. Axial susceptibility-weighted imaging (SWI) (A) and sagittal T2* images (B) demonstrate predominantly peripheral susceptibility artefacts along the margins of the epidural collection. Authors’ original images.

Figure 4. Axial (A) and sagittal (B) post-contrast T1-weighted images show heterogeneous, predominantly thick peripheral enhancement of the epidural collection. Authors’ original images.

Figure 5. Follow-up sagittal short tau inversion recovery (STIR) (A) and post-contrast sagittal MRI (B) obtained at 10 days show mild interval increase in the epidural collection with new bone marrow oedema and patchy enhancement of the L4 vertebral body centred at the basivertebral vein (arrow). Authors’ original images.

Figure 6. Preoperative MRI during subsequent clinical deterioration. Sagittal T2-weighted (A) and post-contrast sagittal and axial images (B) show progression of spondylodiscitis, enlargement of the epidural abscess, increased spinal canal stenosis and further compression of the cauda equina nerve roots. Authors’ original images.

Differential diagnosis

The principal differential diagnoses were a sequestered disc fragment, epidural haematoma and pyogenic epidural abscess. A sequestered disc fragment was initially considered because of the ventral epidural location and close association with a posterior annular fissure. However, the presence of heterogeneous, predominantly peripheral diffusion restriction and thick peripheral enhancement made this diagnosis less likely. These features favoured a complex infected collection rather than a non-infected disc fragment.

Epidural haematoma was also an important consideration given the patient’s long-term antiplatelet therapy and the presence of peripheral susceptibility artefacts. However, the overall pattern did not favour a primary epidural haematoma. The combination of heterogeneous diffusion restriction, thick peripheral enhancement, positive Brucella blood cultures, subsequent vertebral marrow involvement and histopathological evidence of an infected epidural collection favoured infection over isolated haemorrhage.

Pyogenic epidural abscess was considered because of the imaging appearance. However, the subacute clinical course, absence of fever, exposure history of unpasteurised milk consumption and positive Brucella blood cultures supported brucellar rather than pyogenic infection.

Treatment

Following microbiological confirmation of Brucella species on blood cultures, antimicrobial therapy was initiated in consultation with the infectious diseases team. The patient received weight-adjusted intravenous gentamicin therapy (390 mg daily) for 14 days, followed by prolonged oral combination therapy with doxycycline 100 mg two times per day, rifampicin 600 mg once daily and trimethoprim–sulfamethoxazole double-strength two tablets two times per day.

Given the severity of infection, including bacteraemia, epidural abscess formation, neurological compromise and interval vertebral involvement, prolonged combination antimicrobial therapy was planned in consultation with the infectious diseases team to reduce the risk of relapse in complicated spinal brucellosis.

Despite antimicrobial therapy, the patient developed progressive neurological and radiological deterioration. Follow-up MRI demonstrated interval enlargement of the epidural collection with increasing cauda equina compression. Surgical decompression was subsequently performed via a posterior approach with L4–L5 laminectomy and debulking of the epidural collection. Intraoperatively, an epidural inflammatory collection was encountered and debulked.

A minimum treatment duration of 6 months was planned, with follow-up guided by clinical status, inflammatory markers and interval imaging.

Outcome and follow-up

A postoperative CT scan of the lumbar spine was performed a few days after surgery to evaluate the surgical bed and residual collection (figure 7). The imaging findings were consistent with expected early postoperative changes, with oedema and soft-tissue thickening at L4–L5 and a poorly visualised residual fluid collection. Histopathological examination of the surgical specimen confirmed an infected epidural collection, demonstrating acute inflammatory infiltrates with granulation tissue, haemorrhage and necrotic debris. No evidence of malignancy was identified. Special stains for fungal organisms, including Grocott methenamine silver and periodic acid–Schiff stains, and Ziehl–Neelsen stain for acid-fast bacilli were negative. Acid-fast bacilli culture was also negative. Surgical specimen microbiology was negative, likely reflecting the effect of preceding antimicrobial therapy. Although Brucella species were not isolated from the surgical specimen, histopathology confirmed an infected epidural collection, and the diagnosis was supported by positive blood cultures for Brucella species in the appropriate clinical and epidemiological context.

Figure 7. Postoperative CT of the lumbar spine. Sagittal (A) and axial (B) post-contrast CT showing expected early postoperative changes at L4–L5 following decompression and collection debulking, including oedema, soft-tissue thickening and a poorly visualised residual collection. Authors’ original images.

Postoperatively, the patient showed clinical improvement with preserved neurological function and mild residual weakness. Repeat microbiological assessment showed negative blood cultures and negative Brucella cultures after antimicrobial therapy had been initiated. C-reactive protein (CRP) decreased during follow-up, although interval fluctuation was noted during the period of clinical deterioration. At 2-month clinical follow-up, the patient was ambulatory with no new neurological deficits and remained on prolonged antimicrobial therapy.

Discussion

Spinal brucellosis typically presents as spondylodiscitis, with epidural involvement usually occurring by contiguous extension from adjacent vertebral or discal infection.^2–4 This case is notable because the initial MRI demonstrated an epidural-predominant collection at L4–L5 without radiologically apparent disc, endplate or marrow involvement, followed by new vertebral marrow oedema and enhancement on short-interval follow-up imaging. This radiological sequence may reflect haematogenous seeding of the valveless vertebral epidural venous plexus before osseous involvement became detectable on MRI, although occult early vertebral infection below imaging detectability cannot be entirely excluded.^{3 4}

The principal diagnostic challenge was distinguishing the lesion from a sequestered disc fragment and epidural haematoma. The ventral epidural location and association with an annular fissure initially favoured disc extrusion. However, heterogeneous predominantly peripheral diffusion restriction and thick peripheral enhancement supported an infected collection rather than non-infected disc material.^{5 9} Epidural haematoma was also considered because of the patient’s long-term antiplatelet therapy and the presence of peripheral susceptibility artefacts, but the overall multisequence imaging pattern, positive Brucella blood cultures, interval vertebral involvement and histopathological findings favoured infection over isolated haemorrhage.^{5 9} The subacute clinical course, absence of fever, low procalcitonin and history of unpasteurised dairy consumption further supported brucellar rather than pyogenic infection.^1–3

Negative surgical specimen microbiology did not exclude the diagnosis. Surgical decompression was performed after more than 10 days of targeted antimicrobial therapy, which likely reduced culture yield, particularly for Brucella species, a fastidious intracellular organism.³ The diagnosis was therefore established through integration of microbiological, radiological, histopathological and epidemiological findings.

Progressive cauda equina compression and neurological deterioration despite antimicrobial therapy provided clear indications for surgical decompression, consistent with published experience in brucellar epidural abscess.^{7 10} Prolonged combination antimicrobial therapy was selected because of bacteraemia, epidural abscess formation, neurological compromise, interval vertebral involvement and operative disease burden. A minimum 6-month treatment course was planned in line with evidence supporting prolonged or intensified therapy in complicated spinal brucellosis.^{11 12}

This case underscores the diagnostic value of short-interval MRI follow-up in diagnostically uncertain epidural collections. When imaging findings remain suspicious for infection but lack typical spondylodiscitis features, repeat MRI within 10–14 days may demonstrate interval vertebral involvement, clarify the diagnosis and guide timely surgical management.^{5 9} In endemic regions, Brucella species should remain in the differential diagnosis of atypical epidural collections even in the absence of fever or early vertebral signal abnormality.

Patient’s perspective.

“When I first came to the hospital, I thought I had only strained my back from heavy work. The pain in my leg kept getting worse, and I became worried when I started having difficulty passing urine. After the scans and tests, I was shocked to learn that I had a spinal infection related to drinking unpasteurised milk. Surgery was frightening, but I felt immediate relief afterward and gradually regained movement in my leg. The long course of treatment has been difficult, but I understand why it is necessary. I am now walking normally and slowly returning to my daily activities.”

[Translated from Arabic and edited by the authors for clarity. The patient has been made aware that the case report may be published and accessible online.]

Learning points.

• Spinal brucellosis can rarely manifest as an epidural-predominant pattern, in which an infected epidural collection precedes radiologically evident spondylodiscitis.

• In endemic regions, including the Middle East, Brucella should remain in the differential diagnosis of any atypical epidural collection, even in the absence of fever or early disc and endplate abnormalities. A dietary history—specifically unpasteurised dairy consumption—is a critical epidemiological clue.

• Key MRI features favouring an infected epidural collection over disc extrusion or haematoma include heterogeneous peripheral diffusion restriction on DWI, thick peripheral enhancement on contrast MRI and interval vertebral marrow oedema on follow-up imaging. Susceptibility artefacts should be interpreted in the context of the full imaging profile and not assigned haemorrhagic significance in isolation.

• Short-interval MRI follow-up (within 10–14 days) is essential in diagnostically uncertain epidural collections. Interval imaging may reveal evolving spondylodiscitis not apparent on the initial study, providing decisive diagnostic evidence and enabling timely clinical decision-making.

• Complicated spinal brucellosis with neurological compromise or progressive cauda equina compression despite antimicrobial therapy warrants surgical decompression. Prolonged combination antimicrobial therapy, individualised in consultation with infectious diseases specialists, is essential for disease control and relapse prevention.