The spinal cord is not spared in multiple sclerosis (MS) – most patients show pathological changes that can influence the course of the condition. This makes spinal cord imaging a key tool of both the diagnostic and prognostic assessment of MS [1].
Spinal Cord Imaging as a Cornerstone in Diagnosis
What is the role of spinal cord imaging in the diagnostic investigation of MS? We pose this question to Professor Massimo Filippi of the “Vita-Salute” San Raffaele University, Milan, Italy. “Spinal cord imaging is extremely important for different reasons,” Prof Filippi tells us. “First and foremost, MS lesions are frequently found in the spinal cord. In more than 90% of patients, spinal cord abnormalities can be detected over the course of the disease. At onset, spinal cord lesions are already present in 40-50% of patients. MS lesions in the spinal cord have very characteristic features: they are small, typically not exceeding two vertebral segments in length, and they do not involve the entire transverse area of the cord. This makes them quite distinct from spinal cord lesions that occur in other conditions, including other inflammatory demyelinating diseases of the central nervous system (CNS). For example, in neuromyelitis optica spectrum disorders (NMOSD), lesions are very long, extensive, and bulky, and they may even cause swelling of the cord. Spinal cord lesions are also uncommon in CNS vasculitis, and unlike in the brain of individuals over 40-45 years of age, nonspecific spinal cord lesions are essentially never seen. All of these factors provide strong reasons to perform complete, meticulous spinal cord imaging during the diagnostic workup – not least because the latest 2024 revisions of the McDonald criteria reaffirm that one of the typical sites for establishing an MS diagnosis (now increased to five with the addition of the optic nerve) is the spinal cord. Thus, identifying spinal cord lesions is also important for demonstrating the spatial dissemination (DIS) of the condition. While it may not be essential if lesions have already been detected in other typical locations, it nonetheless provides valuable information for both diagnosis and differential diagnosis. Under the new criteria, the involvement of 4 out of 5 topographies is sufficient to establish a diagnosis of MS. Consequently, spinal cord lesions may serve as the decisive diagnostic factor, conferring a high degree of certainty and allowing for the early initiation of high-efficacy therapy. I believe that spinal cord imaging should always be performed as part of the diagnostic process, particularly when differential diagnosis poses a challenge. Spinal cord imaging has a dual value: it can either provide confirmatory evidence or prompt a reassessment of the diagnosis when the findings are not as expected – that is, when they are not typical of MS.”
Spinal cord lesions in people with MS are most often found in the lateral or dorsal white matter, particularly in the cervical region, though rarely confined to it [2]. Grey matter damage occurs evenly across all spinal levels, whereas white matter damage is concentrated in the upper cervical cord [1]. Axonal loss is extensive – even in normal-appearing white matter – showing about a 46% reduction in density [1]. Damage is greatest in the lateral columns and minimal in the anterior columns [1]. Brain and spinal cord MRI features in the diagnostic workup can lower the chance of misdiagnosing, helping differentiate between MS, and other inflammatory and non-inflammatory neurological conditions [3].
“As with adults, spinal cord imaging is essential during the diagnostic and prognostic phase in children with suspected MS”, Professor Mara Rocca, from the “Vita-Salute” San Raffaele University, says, “Spinal cord imaging allows for an accurate differential diagnosis from other paediatric neurological conditions. It is particularly important to distinguish between paediatric MS and MOGAD, as spinal cord lesions in MOGAD differ markedly: they are usually more extensive, span longer segments, and often involve the lower spinal cord, including the conus medullaris and cauda equina.”
As in adults, spinal lesions in children with MS occur most likely in the cervical spinal cord. More than 30% of children with MS exhibit at least one cervical lesion [4]. Spinal cord lesions in children with MS tend to be positioned more peripherally and affect the white matter tracts [5]. On the contrary, in paediatric MOGAD, spinal cord involvement is typically characterised by longitudinally extensive lesions affecting the lower spinal segments, marked grey matter involvement producing the ‘H-sign’, and frequent leptomeningeal enhancement [5]. (For an in-depth review of the spinal cord clinical features in NMOSD and MOGAD, see our Spotlight article.)
Prognostic Insights from Spinal Cord Imaging
In MS, spinal cord imaging also has high prognostic value. Damage to the grey matter of the cervical spinal cord is an important predictor of disability progression [6]. Cervical spinal cord MRI measures, particularly grey matter cross-sectional area, account for Expanded Disability Status Scale (EDSS) scores and sensorimotor impairment more effectively than brain MRI measures [7]. A large part of disability progression is driven by spinal cord damage [8]. Both spinal cord lesions and grey matter atrophy are closely associated to a patient’s clinical condition [8]. Spinal cord damage is a strong predictor of poorer 5-year clinical outcomes, with cortical volume loss playing a lesser yet contributory role [9].
“Even a single spinal cord lesion in a patient with MS is a significant indicator of an unfavorable prognosis.” Prof Rocca discusses with us the role of spinal cord imaging in guiding treatment decisions, “One lesion occurring before the age of 30 is associated with a substantially higher risk of severe disability and a more rapid progression to a secondary progressive form of the disease. This insight strongly influences therapeutic decisions [10].”
Spinal cord atrophy is also associated with chronic brain inflammation [10]. A large multicentre study showed that progression independent of relapse activity (PIRA) is linked not only to paramagnetic rim lesions (PRLs) but also to spinal cord atrophy [10].
“Assessing spinal cord damage is crucial for identifying patients with PIRA.” Professor Rocca continues, “Data consistently show that patients who develop PIRA earlier exhibit greater spinal cord atrophy than those without this phenomenon. Notably, an accelerated loss of spinal cord volume may allow identification – up to four years in advance – of patients likely to progress towards a secondary progressive phenotype. Such early detection could provide critical warnings well before clinical monitoring alone would reveal changes.”
Prof Filippi tells us, “The spinal cord is a relatively small structure, yet it contains all ascending sensory and descending motor fibers of the CNS. Given the high density of neural pathways, even a small lesion in this strategically important location can have a much greater impact than a lesion in many cerebral areas. It is comparable to a traffic incident in a major intersection, where the disruption is disproportionately greater than in a less frequented area. Knowing whether the spinal cord is already affected at disease onset provides important prognostic information. Spinal cord lesions are associated with a higher risk of disease progression, making early intervention especially important in these patients. When such lesions are present at disease onset, an even more protective approach is warranted to preserve long-term neurological function and quality of life.”
Spinal Cord Imaging in Monitoring: Still a Marginal Player
“Spinal cord MRI is essential for diagnosis, essential for prognosis, and should be used judiciously in the monitoring phase.” Prof Filippi says, “We generally perform a follow-up brain MRI between 6 and 12 months to assess whether the prescribed therapy is maintaining its efficacy. Since it is not feasible to perform all possible MRI sequences at every follow-up, it is important to clarify the rationale for repeating a spinal cord MRI in the context of patient monitoring. One possible reason is to determine whether new lesions have developed, which could indicate treatment failure. In such cases, if a brain MRI already reveals new lesions, the answer is clear, and a spinal cord MRI is not necessary. If no cerebral lesions are detected, the likelihood of finding spinal cord lesions – given its smaller size – is lower, and a spinal MRI should therefore be performed only with a clear rationale. For example, it may be warranted if the patient is not responding optimally to therapy and a reassessment of prognosis is needed. In some cases, a spinal cord lesion may be absent or not detectable at onset, only to develop later. Moreover, new symptoms suggestive of other pathologies may also prompt repeat imaging of the spinal cord. If there is suspicion of a spinal process of potentially different etiology, this is clearly an indication for follow-up imaging.”
“Monitoring the spinal cord is technically challenging,” Prof. Rocca notes. “Standardisation is difficult because patients are often scanned on different MRI machines, introducing measurement variability – a critical issue for a small structure like the spinal cord. This is particularly important, as identifying a new lesion may lead to consideration of a therapy change. Furthermore, spinal cord lesions are usually symptomatic, whereas many brain lesions are silent. This makes regular brain MRI essential, while routine spinal cord imaging is less critical during follow-up.”
***
Written by Stefania de Vito
Special thanks to Professor Massimo Filippi (“Vita-Salute” San Raffaele University; IRCCS San Raffaele Hospital, Milan, Italy) and to Professor Mara Rocca (“Vita-Salute” San Raffaele University; IRCCS San Raffaele Hospital, Milan, Italy) for their insights.
References
[1] Gass A et al. Lancet Neurol. 2015; 14(4): 443-454.
[2] Weier K et al. Mult. Scler. J. 2012; 18(11): 1560-1569.
[3] Rocca MA et al. J. Neurol. 2025; 272(6): 388.
[4] Margoni M et al. Ann. Clin. Transl. Neurol. 2025.
[5] Fadda G et al JAMA network open 2021; 4(10): e2128871 – e2128871.
[6] Azzimonti M et al. J. Neurology 2025; 272(3): 228.
[7] Morozumi T et al. Mult. Scler. J. 2024; 30(8): 1004-1015.
[8] Hugh K, Miller DH, and Ciccarelli O. Nat. Rev. Neurol. 2015; 11(6): 327-338.
[9] Rocca MA et al. JNNP 2023; 94(1): 10-18.
[10] Cagol A et al. Neurology 2024; 102(1): e207768.