Reading Spine MRI

Timelines of Degenerative Cervical Myelopathy

Why we follow these patients, how often, and what we are actually watching for.

Every number below is drawn from a cited primary study. Time axes are in months.

1How long until a compressed cord becomes myelopathic?

199 patients with MRI cord compression but no myelopathy, followed 2–12 years. 22.6% went on to develop myelopathy. The question this panel answers: how soon, and who first?

Terms used in this panel

Radiculopathy
Dysfunction of a single nerve root — as opposed to myelopathy, which is dysfunction of the cord. Bednarik required all three: typical radicular pain or paraesthesia always present; any motor, sensory or reflex finding confined to one dermatome or myotome matching that pain; and correlation with the level of root compression on imaging. Root problems are segmental and one-limb; cord problems are below-the-level and plurisegmental.
SEP
somatosensory
evoked potential
Stimulate a peripheral nerve (median at the wrist, tibial at the ankle) and record the volley at successive points as it ascends — Erb's point → cervical cord → brainstem → sensory cortex. Tests the dorsal column pathway. A cord lesion is called when cervical or cortical responses are absent or delayed while the peripheral response is normal, which localises the problem to the cord rather than the nerve.
MEP
motor
evoked potential
The mirror image. Transcranial magnetic stimulation of motor cortex, recording the muscle response (abductor digiti minimi, abductor hallucis); stimulation at the nerve root is then subtracted. The difference is central motor conduction time — how long the signal takes to descend through the cord. Tests the corticospinal tract.
Reading the report
The localising logic is the same in both: an abnormal central response with a normal peripheral response (N9 for median SEP, N22 for tibial) places the lesion in the cord, not the nerve. If the report does not state that the peripheral response was normal, the central finding is uninterpretable — call the lab.
Why they predict
SEP and MEP interrogate the long tracts, ascending and descending, and can detect cord dysfunction before the patient has clinical myelopathy — which is exactly why they forecast conversion. Note though that radiculopathy alone carries as much weight (RR 3.68) as either study (RR 2.91–3.21), and costs nothing but an examination.
Ordering guidance — from the authors of the study this panel is built on: electrophysiology is "highly recommended, not only in patients with MRI-documented asymptomatic cord compression, but also in patients with radiculopathy or back pain before MRI examination. The presence of 'central' myelopathic SEP or MEP abnormalities would justify further MRI examination." [1]
So the order runs both ways: compression on MRI → stratify with SEP/MEP; or radiculopathy and neck pain with no MRI yet → SEP/MEP → a central abnormality justifies obtaining the MRI. Evoked potentials need no waiting period after symptom onset. TMS-based MEP is not widely available outside academic centres; SEP is — and radiculopathy alone still outweighs either.

Every patient leaves the same origin. Each path is one presentation, and stops at the month by which 1 in 4 of that group has become myelopathic — the higher the path climbs, the longer that took.

How long until a compressed cord becomes myelopathic? The origin — month 0 Cord compression on MRI. No myelopathy yet. (n = 199) Every path leaves from here. What separates them is what the patient already has on day one — a starting label, not an outcome. The event being timed Onset of myelopathy — cord symptoms appear, mJOA falls by at least 1 point. First symptoms recorded: gait disturbance · clumsy hand syndrome · Lhermitte's sign · spastic paresis (usually lower paraparesis) · plurisegmental sensory loss 22.7 months compression + abnormal SEP 23.6 months compression + radiculopathy 23.7 months compression + abnormal MEP 31.7 months compression + T2 hyperintensity (n.s.) 48.4 months compression only — whole cohort 0 12 24 36 48 60 all three arrive together at ~23 months months from detection to onset of myelopathy
Read this as two different things happening. Radiculopathy, an abnormal SEP and an abnormal MEP are already present at month 0 — they are how you label a patient the day you meet them. What happens later, at the marked point, is something else entirely: that patient develops myelopathy — gait disturbance, clumsy hands, Lhermitte's, spastic paraparesis — and the mJOA drops. Root symptoms are the predictor. Cord symptoms are the event.

A patient who walks in with compression and radiculopathy reaches that event roughly twice as fast as the cohort as a whole (23.6 vs 48.4 months). And the thing that predicts it is clinical: not one imaging measurement predicted conversion — Pavlov ratio (P = 0.49), compression ratio (P = 0.15), cord cross-sectional area (P = 0.96), number of stenotic levels (P = 0.30). [1] The scan tells you compression exists. The examination tells you who is going to convert, and roughly when.

2How long until the damage stops being reversible?

167 surgical patients with pre- and post-operative MRI. Mean symptom duration at the time of surgery, grouped by what the cord looked like on the post-operative scan.

Reading the figure
Three curves, one per group, all leaving the same point — symptom onset, the dot at the origin. Each curve ends where that group came to surgery, and its endpoint on the horizontal axis is that group's mean symptom duration: 17.0, 24.7, 39.6 months. The MRI at each endpoint is what the cord looked like on the post-operative scan, with the finding and how much function came back beside it. Only the horizontal position is data — the height a curve climbs stands for cord injury accumulating, it is not a measured quantity.
"Recovery ratio"
⚠️ Not the share of patients who improved. It is the share of lost function regained — (post-op mJOA − pre-op mJOA) ÷ (18 − pre-op mJOA), averaged across the group. So 72% means these patients got about 72% of what they had lost back, not that 72% of them got better.
The trend
P = 0.003: the further right a curve runs before it ends, the less reversible the cord signal at its endpoint and the less function regained — 72% → 51% → 36%, monotonic across all three.
Where patients actually land
Across four cohorts, patients arrive for surgery having been symptomatic a mean of 26–30 months — which puts their endpoint just past the orange curve's (24.7 months) and heading for the red one's (39.6 months). The average patient is already past the point at which cord signal change still reverses.
How long until the damage stops being reversible? 17.0 months of symptoms at surgery Signal change resolved on the post-operative MRI 72% of lost function regained 24.7 months of symptoms at surgery T2 hyperintensity persists on the post-operative MRI 51% of lost function regained 39.6 months of symptoms at surgery T1 hypointensity — irreversible on the post-operative MRI 36% of lost function regained 0 12 24 36 48 months of myelopathic symptoms at the time of surgery
What this teaches. This is the clearest evidence in the literature that delay is a decision with a biological cost. Patients whose cord signal normalised after surgery had been symptomatic for 17 months; those left with irreversible T1 change had waited 39.6 months. Post-operative resolution of T2 predicted the best outcomes (RR 72% vs 47%, P = 0.04) — but resolution of T1 predicted nothing (38% vs 26%, P = 0.36). [2]

Where the average patient's curve ends is the uncomfortable part. Across four cohorts patients arrive for surgery having been symptomatic a mean of 26–30 months [2,3,4,5] — which lands them past the 24.7-month mark where T2 change stops resolving, and on the road to the 39.6-month mark where it becomes T1. The average patient in these studies was referred too late to get the best available result. Only 11% of T2 hyperintensity resolves after decompression at all.

3How long does recovery take after surgery?

Three checkpoints after decompression — what the trials measured at each, and a line you can actually say to a patient.

Recovery after surgery: what has happened by 6, 12 and 24 months What these scores mean mJOA 18-point scale (hands, walking, sensation, bladder). 18 = normal. Recovery ratio share of lost function regained, not share of patients who improved. SF-36 physical — general quality of life. 6 months First routine post-operative MRI. After ventral (front-of-neck) surgery, 41% developed dysphagia — half of those still have it at this point. CSM-S randomised trial, n = 163 What you can say “Swallowing trouble is common early on, and about half of it has settled by now.” 12 months mJOA up 3.2 points — 63% of lost function regained. SF-36 physical +5.9 ventral vs +6.2 dorsal (P = 0.84) — approaches equal. Arvin, n = 57 · CSM-S randomised trial, n = 163 What you can say “Most of the improvement you are going to get, you already have.” 24 months No decline between year one and year two. How much returns is set by the pre-op cord signal: 72% if the signal resolves 51% if T2 hyperintensity persists 36% if T1 hypointensity remains CSM-S randomised trial, n = 163 · Kato, n = 167 What you can say “What you have at a year, you keep — and how much came back was decided before we operated.” 0 6 12 18 24 months after surgery

4So how often should we actually see them?

The schedule below is derived from the three panels above, not from convention.

How often should we actually see them? Every interval below is derived from the three figures above, not from convention. Patient What you are watching for Follow-up interval Moderate or severe myelopathy (mJOA ≤ 14) Nothing — this patient is not a surveillance patient. 2017 AO Spine: strong recommendation for surgery Refer now Compression + radiculopathy, or abnormal SEP / MEP Conversion to myelopathy — it arrives about twice as fast. 25th-percentile conversion 23.6 months · Wilson: strong recommendation to counsel surgery Every 6 months Mild myelopathy (mJOA 15–17), managed non-operatively A 1-point mJOA drop — that is the meaningful change in mild disease, not 2. MCID is 1 point mild / 2 moderate / 3 severe · 2025 AO Spine Every 6 months Compression, no myelopathy, no radiculopathy New myelopathic symptoms. Repeat the MRI only if the examination changes. 25th-percentile conversion 48.4 months · no imaging measure predicted conversion Every 12–24 months
The one-sentence rationale for surveillance. We follow these patients because conversion to myelopathy is common enough to matter (23% by ~4 years) but unpredictable from the scan — and because the window in which cord injury is still reversible closes somewhere between roughly 17 and 40 months of symptoms. Surveillance exists to catch the conversion early enough that the patient is still on the left of Panel 2. What we are following is the examination; the MRI confirms what the examination tells us.
References
  1. Bednarik J, Kadanka Z, Dusek L, et al. Presymptomatic spondylotic cervical myelopathy: an updated predictive model. Eur Spine J. 2008;17:421–431. (n = 199; 22.6% converted; median follow-up 44 months)
  2. Kato S, Nouri A, Reihani-Kermani H, et al. Postoperative resolution of MRI signal intensity changes and the associated impact on outcomes in degenerative cervical myelopathy. Spine. 2018;43:824–831. (n = 167)
  3. Nouri A, Martin AR, Kato S, et al. The relationship between MRI signal intensity changes, clinical presentation, and surgical outcome in degenerative cervical myelopathy. Spine. 2017;42:1851–1858. (n = 419; mean symptom duration 26.9 mo)
  4. Tetreault L, Kopjar B, Nouri A, et al. The modified Japanese Orthopaedic Association scale. Eur Spine J. 2017;26:78–84. (mean symptom duration 26.9 mo, range 0.25–432)
  5. Arvin B, Kalsi-Ryan S, Mercier D, et al. Preoperative MRI is associated with baseline neurological status and can predict postoperative recovery in CSM. Spine. 2013;38:1170–1176. (n = 57; mean symptom duration 30.3 mo; 1-yr mJOA recovery ratio 63.3%)
  6. Wilson JR, Barry S, Fischer DJ, et al. Frequency, timing, and predictors of neurological dysfunction in the nonmyelopathic patient with cervical spinal cord compression, canal stenosis, and/or OPLL. Spine. 2013;38(22S):S37–S54. (systematic review; ~8% at 1 yr, 23% at median 44 mo)
  7. Fehlings MG, Evaniew N, Ter Wengel PV, et al. AO Spine clinical practice recommendations for diagnosis and management of degenerative cervical myelopathy. Global Spine J. 2025;15(5):2585–2593.
  8. Fehlings MG, Tetreault LA, Riew KD, et al. A clinical practice guideline for the management of patients with degenerative cervical myelopathy. Global Spine J. 2017;7(3 Suppl):70S–83S.
  9. Ghogawala Z, Terrin N, Dunbar MR, et al. Effect of ventral vs dorsal spinal surgery on patient-reported physical functioning in patients with cervical spondylotic myelopathy: a randomized clinical trial. JAMA. 2021;325(10):942–951.

Prepared for the MANS 2026 Summer Meeting APP session. All figures constructed from full-text data in the cited papers.