NeuroRad TractsNeuroradiology education
Search→Residents→Fellowship→Explore Research↗Anatomy Lab↗Neuroradiology Oral Boards→
Spine

Spine Infection: Disc-Centered, Then Hunt the Epidural

Fellowship-level neuroradiology teaching by E. Brooke Schrickel, MD. Open this topic in the interactive reading room →

Watch · concise explainerDisc first, then hunt the epidural

Core clinical idea

Pyogenic spondylodiscitis is disc-centered: confirm the disc-plus-paired-endplate pattern with endplate erosion, then reflexively hunt the epidural collection compressing the thecal sac — that complication is what turns a medical admission into a surgical emergency. Find the discitis, report the danger.

Bottom line

Confirm disc-plus-paired-endplate infection with erosion, then report the epidural/cord complication — that's the part that changes urgency.

Core workstation questions

  • Is this the disc + paired-endplate pattern, and is there true endplate erosion (vs reactive change)?
  • Is there an epidural collection compressing the thecal sac — phlegmon or abscess (DWI), where, how much?
  • Is the cord signal abnormal (venous congestion/ischemia) — the irreversible complication?
  • Is this collection truly epidural, or subligamentous (intact PLL) that I shouldn't upgrade?
  • Do the tempo and ESR/CRP fit pyogenic or granulomatous, and does my morphologic read agree?
  • Disc destroyed (pyogenic) or disc spared with subligamentous skip lesions and big paravertebral abscess (TB/fungal)?
  • Is there endplate (or facet subchondral) erosion — the line between infection and a degenerative/inflammatory mimic?
  • Do labs/tempo and plain films/CT agree with infection, or is this tumor (no fever, normal markers, destructive mass) or spondyloarthropathy?

What changes reporting / management

  • Commit pattern: T2-bright disc with lost intranuclear cleft, paired endplate edema (T1 dark/T2 bright), endplate erosion, height loss/alignment change, and disc/endplate/paravertebral enhancement.
  • Endplate erosion is the discriminator from degenerative mimics — make the erosion call explicitly.
  • Exclude epidural extension on every discitis; report presence, location (ventral/dorsal), extent, and degree of thecal-sac compression. Rim-enhancing + DWI-restricting = abscess; homogeneous enhancing tissue = phlegmon.
  • Cord T2 signal from epidural venous-plexus congestion/ischemia is a red flag and can be irreversible — escalate, don't sit on it.
  • Subligamentous collection (ligaments displaced but PLL intact) is NOT a true epidural abscess — don't upgrade; subdural abscess is rarer, ring-enhancing around the thecal sac, DWI-positive.
  • Routes: hematogenous most common, then nosocomial/iatrogenic (Pseudomonas/Klebsiella with instrumentation/catheter/discography/surgery), direct extension, CSF seeding.
  • Always compare with CT when available: a vacuum disc phenomenon (intradiscal gas) makes infectious discitis much less likely — don't call the gas 'fluid.' Marrow signal change accompanying a vacuum phenomenon is more often severe reactive/degenerative (Modic) change than infection; weigh the whole pattern before calling discitis-osteomyelitis.
  • TB/granulomatous pattern: vertebral signal change with preserved disc spaces, subligamentous anterior/posterior spread, skip lesions, large prevertebral/paravertebral abscess, possible gibbus deformity; fungal can look identical — explicitly search up and down for skip levels.
  • Facet septic arthritis = joint fluid PLUS subchondral erosion and perifacet inflammation (can track into the canal behind the thecal sac); fluid alone is not infection.
  • Arachnoiditis: nerve-root clumping, peripheral adhesion (empty-sac sign), root/intrathecal enhancement, loculated CSF; chronic disease can calcify (arachnoiditis ossificans).

Practical traps

  • Calling intact-PLL subligamentous spread an 'epidural abscess' (overstates surgical picture).
  • Missing a true compressive epidural abscess (understates urgency).
  • Treating cord signal in this setting as background rather than venous ischemia/infarct.
  • Anchoring on disc-destroying pyogenic morphology without checking tempo/ESR/CRP for a granulomatous, disc-sparing pattern.
  • Mistaking intradiscal gas (vacuum disc phenomenon) for fluid/infection — vacuum gas argues AGAINST active discitis.
  • Demanding disc destruction before calling infection — TB spares the disc, so disc sparing is a feature, not reassurance.
  • Calling facet fluid (or disc T2 signal) infection without subchondral/endplate erosion.
  • Reading multilevel lytic vertebrae as automatic metastases when coccidioidomycosis or another fungal osteomyelitis fits the host/region.
  • Over-calling Modic type 1 reactive endplate change (bright, edematous, enhancing, no erosion) as discitis.

Teaching pearls

  • Spondylodiscitis is disc-centered — paired endplate edema plus erosion is the commit pattern.
  • Endplate erosion is the line between infection and a degenerative mimic.
  • Every discitis read has to work the epidural space — that's where the surgical emergency hides.
  • Cord venous congestion/infarct from epidural disease is irreversible — don't sit on it.
  • Subligamentous (intact PLL) is not an epidural abscess — don't upgrade it.
  • TB spares the disc and skips levels subligamentously — disc sparing is a feature, not reassurance against infection.
  • A big paravertebral abscess with preserved discs and skip lesions is TB until proven otherwise.
  • Fluid in a facet (or a disc) isn't infection — erosion is.
  • No endplate erosion = degenerative (Modic 1), even if it's bright and enhances.
  • Multilevel lytic vertebrae can be coccidioidomycosis faking mets; no fever and normal markers point to tumor.

Source lectures

  • Spine Infections, Part 1
  • Spine Infections, Part 2

Educational material for radiology residents and neuroradiology fellows. Nothing here drives individual patient care, and it contains no patient data.