Frequency and Potential Risk Factors Associated with the Development of Asymptomatic T2 Hyperintense Cervical Spine Lesions on MRI in Patients with Relapsing-Remitting Multiple Sclerosis

Yasser Fadlallah, Yujie Wang, Muhammad Taimur Malik, Fan Tian, Peter A. Calabresi, Izlem Izbudak, Yishang Huang, Kathryn C. Fitzgerald, Ellen M. MowrymedRxiv 2024 · DOI

TL;DR

This longitudinal study of 869 patients with relapsing-remitting multiple sclerosis (RRMS) found that asymptomatic cervical spine lesions are relatively rare and often coincide with new brain activity. However, Black/African American patients face a significantly higher risk (over 3x the odds) of developing these silent lesions, suggesting that personalized, rather than universal, spinal imaging protocols could improve monitoring efficiency and reduce healthcare costs.

What Problem Does This Paper Address?

In Multiple Sclerosis (MS) management, MRI is the gold standard for monitoring disease activity. While brain MRIs are routine, imaging the spinal cord—specifically the cervical spinal cord (c-SC)—is less consistent. Spinal scans are technically challenging, prone to artifacts, uncomfortable for patients, and expensive.

Current guidelines are ambiguous: they don't definitively recommend routine spinal imaging for clinically stable patients. This leaves a critical gap: we don't fully know how often "silent" (asymptomatic) lesions appear in the spine without accompanying symptoms or brain lesions. If these lesions are rare, routine scanning might be a waste of resources; if they are common or concentrated in specific groups, missing them could lead to undertreatment.

Methods

The researchers conducted a retrospective longitudinal study at the Johns Hopkins MS Center, analyzing data from 869 RRMS patients between 2014 and 2019.

Key approaches included: * Sequential Analysis: They tracked up to four sequential c-SC scans per patient. * Defining "Asymptomatic": Lesions were only counted if the patient was clinically stable (no new symptoms or relapses) at the time of the scan. * Statistical Modeling: The team used univariate and multivariable logistic regression to identify risk factors for the first scan. To account for the repeated nature of the data over time, they employed mixed-effects logistic regression, a robust method for longitudinal datasets. * Number Needed to Scan (NNS): They calculated the NNS to determine how many patients must be imaged to find a single isolated asymptomatic spinal lesion.

Datasets

The study utilized a clinical cohort from the Johns Hopkins MS Center.

  • Size: 869 patients with RRMS.
  • Demographics: Predominantly female (3.4:1 ratio), mean age 38.2, 67.9% White, and 22.1% Black/African American.
  • Scope: The dataset included 5 years of longitudinal follow-up, with patients contributing a median of 3 scans each.

Key Insights and Ideas

  • The Racial Disparity "Aha": The most striking finding was that Black/African American patients had significantly higher odds (OR = 3.26) of developing asymptomatic c-SC lesions. This aligns with other research suggesting a more aggressive disease course in this population and suggests that "one-size-fits-all" monitoring guidelines may exacerbate health disparities.
  • Low Yield of Isolated Activity: Roughly half of the patients who showed new spinal activity also had new lesions on their brain MRI. This suggests that for many, a brain MRI alone might be enough to trigger a treatment change, making the additional spinal scan redundant.
  • Diminishing Returns Over Time: The frequency of new asymptomatic lesions decreased across successive scans (from 12.1% at Scan 1 to 4.8% by Scan 4), while the "Number Needed to Scan" increased. This implies that the longer a patient remains stable on their current therapy, the lower the utility of routine spinal surveillance.
  • Confounding by Indication: Patients on higher-efficacy therapies appeared more likely to have lesions in the longitudinal model. The authors correctly identify this as "confounding by indication"—these patients were likely put on stronger drugs because they had more aggressive disease to begin with.

Key Results

  • Frequency: Asymptomatic c-SC lesions were found in 4.8% to 12.1% of scans.
  • Racial Risk: Black/African American individuals had an Odds Ratio (OR) of 3.26 (95% CI 1.79–5.93) for developing a lesion on the first scan compared to White patients.
  • NNS: The Number Needed to Scan to find one isolated spinal lesion (where the brain MRI was clear) was 18 at the first scan, but jumped to 55 by the fourth scan.
  • Longitudinal Odds: In the mixed-effects model, the odds of finding a new lesion dropped by ~70-80% in subsequent scans compared to the baseline scan.

Limitations and Future Work

The authors acknowledge several limitations:

  • Observational Design: As a retrospective study, it cannot prove causality, particularly regarding the effects of different medications (DMTs).
  • Tertiary Care Bias: The population at a specialized center like Johns Hopkins may have more complex or aggressive MS than the general population.
  • Radiology Reports: The study relied on existing clinical reports rather than a centralized re-reading of all raw images, which introduces potential variability in how lesions were identified.

Future Work: The findings motivate the development of predictive models or clinical decision support tools. Instead of scanning everyone, clinicians could use a patient's race, treatment history, and previous brain MRI results to calculate the "yield" of a spinal scan, potentially saving millions in healthcare costs while focusing resources on high-risk individuals.

Summary updated

Abstract

Abstract Background Multiple sclerosis (MS) is one of the most common non-traumatic neurologic disorders affecting young adults in the United States. Brain MRI is an important tool for monitoring disease activity and treatment efficacy in people with MS. Spinal cord (SC) imaging has been less consistently used for monitoring inflammatory disease activity, and the frequency of clinically silent breakthrough disease in the SC is still unknown. Given the particular vulnerability of the cervical spinal cord (c-SC) to inflammatory demyelination, it is important to evaluate the necessity of routine c-SC scanning or to ascertain appropriate strategies for its monitoring, considering the burden of each scan on a person with MS, including scanner discomfort and cost. Objective To determine how frequently follow-up magnetic resonance imaging (MRI) of the cervical spinal cord (c-SC) in patients with relapsing remitting multiple sclerosis (RRMS) reveals asymptomatic T2 hyperintense lesions, either in combination or in the absence of new MRI brain lesions, and to identify potential associated risk factors for developing such lesions. Methods Patients aged 18-65 years who were diagnosed with RRMS and were seen in longitudinal follow-up at the Johns Hopkins MS Center from January 1, 2014, to December 1, 2019, with an MRI brain and C spine performed during that period, were included. The results of up to four c-SC scans were considered during this study period. Asymptomatic new lesions were identified as new hyperintense T2 lesions, with or without enhancement, observed on MRI during routine follow-up surveillance. Univariate and multivariable-adjusted logistic regression (sex at birth, age, race, and current disease modifying therapy [DMT] category) were employed to identify factors associated with the development of an asymptomatic c-SC lesion for the first scan. Additionally, a mixed-effects logistic regression analysis was performed to assess factors associated with developing asymptomatic lesions across successive scans. Results A total of 869 individuals were included in the cohort, contributing a median of 3 (interquartile range: 2,4) scans per person. The proportion of incidental asymptomatic lesions identified ranged from 4.8 to 12.1% across the four scans analyzed in the study. Among those with new lesions in the c-SC, roughly half also showed concomitant new activity on brain MRI. The multivariate model was notable for Black/African Americans having higher odds of an asymptomatic new lesion (OR= 3.26, 95% CI 0.79, 5.93, p<0.0001), a result that persisted in mixed effects logistic regression models (OR = 1.73, 95% CI = 1.27, 2.35, p = 0.001). Higher-efficacy therapies were associated with higher odds of detection of a new lesion in the mixed-effects model, an association that was not present when considering just the first scan results as the outcome. Conclusion While the association of higher-efficacy therapy is presumed to be related to confounding by indication, Black/African American individuals with MS appear to be at higher risk of developing an asymptomatic c-SC lesion on MRI surveillance, which could suggest higher value of ordering asymptomatic screening imaging of the cord in this subpopulation. Ultimately, however, a very small percentage of the overall population of those with MS has a new cord lesion in the absence of symptoms, and half of those have a new lesion on screening brain MRI. These findings should motivate the creation and validation of predictive models to inform the utility cord imaging at a given timepoint for a given individual with MS, which could enhance healthcare quality and reduce costs.
BibTeX
@misc{fadlallah2024frequency,
  author = {Yasser Fadlallah and Yujie Wang and Muhammad Taimur Malik and Fan Tian and Peter A. Calabresi and Izlem Izbudak and Yishang Huang and Kathryn C. Fitzgerald and Ellen M. Mowry},
  title = {{Frequency and Potential Risk Factors Associated with the Development of Asymptomatic T2 Hyperintense Cervical Spine Lesions on MRI in Patients with Relapsing-Remitting Multiple Sclerosis}},
  year = {2024},
  doi = {10.1101/2024.09.24.24314299},
  howpublished = {medRxiv},
}
Figure