De-Escalation of Disease-Modifying Therapies in MS: A Real-World Look at Outcomes

When considering de-escalation of disease-modifying therapies (DMTs) for patients with multiple sclerosis (MS), a personalized approach guided by a drug’s mechanism of action is critical, as switching from certain agents appears to carry a higher risk of relapse. So suggests a retrospective study from Cleveland Clinic published in Multiple Sclerosis and Related Disorders (2025;100:106552).

“This study provides valuable real-world data indicating that de-escalation of DMTs in MS should be approached thoughtfully and focused on individualized decision-making,” says senior and corresponding author Le Hua, MD, Director of the Multiple Sclerosis Program at Cleveland Clinic Lou Ruvo Center for Brain Health in Las Vegas. “The findings suggest that crucial considerations include the individual patient’s disease characteristics and the mechanism of action of the patient’s current DMT.”

Study rationale and design

For neurologists managing MS, optimal sequencing of DMTs has become increasingly complex, with more than 25 FDA-approved options now available. While expert consensus often favors early, highly effective treatment, questions persist about the appropriate duration of these potent therapies and when de-escalation or discontinuation might be considered. Discontinuation studies have shown increased disease activity in patients stopping therapy. However, de-escalation — i.e., transitioning to a lower-efficacy DMT — could serve as a bridge that balances disease suppression with reduction of risks associated with prolonged exposure to high-efficacy DMTs.

To explore the safety of de-escalation strategies, the researchers retrospectively analyzed medical records of patients with MS who underwent DMT de-escalation at Cleveland Clinic sites in Las Vegas, Nevada, and Cleveland, Ohio.

For purposes of the study, individual DMTs were categorized to efficacy levels as follows:

• High-efficacy therapies (HET): natalizumab and ocrelizumab
• Moderate-efficacy therapies (MET): S1P modulators and fumarates
• Low-efficacy therapies (LET): glatiramer acetate, interferons and teriflunomide

Two cohorts were then identified based on the nature of patients’ de-escalation:

• Cohort 1: patients de-escalating from HET to MET (n = 163)
• Cohort 2: patients de-escalating from MET to LET (n = 127)

The primary outcomes were annualized relapse rate (ARR) and the proportion of patients with new T2 or gadolinium-enhancing lesions on brain MRI. Statistical analysis used generalized linear mixed-effects models as well as noninferiority tests. A key component of the noninferiority assessment was a predefined margin of 0.08, indicating that the rate of disease events or proportion of MRI activity should not be more than 0.08 higher after de-escalation compared with the patient’s initial DMT. This margin was informed by a post hoc analysis of the DISCO-MS trial.

Key findings

Overall, the study was unable to demonstrate noninferiority for de-escalation (relative to the period before de-escalation) in either cohort, although insights emerged from each cohort’s de-escalation scenario.

Cohort 1 (HET to MET)

In cohort 1, the median age at de-escalation was 42.82 years, and disease duration was 11.28 years. The majority (83%) de-escalated from natalizumab, often due to safety concerns (67%). Patients remained on MET longer than on HET (mean of 3.68 vs. 2.57 years, respectively).

While 71% of cohort 1 patients remained relapse-free after de-escalation to MET, their ARR significantly increased from 0.06 on HET to 0.13 on MET. Similarly, the proportion of patients free from new T2 lesions decreased from 91% on HET to 73% on MET. The estimated rate ratio for relapses (3.935) and odds ratio for new T2 lesions (4.161) both exceeded the noninferiority thresholds.

Analysis of outcomes by individual DMTs revealed important nuances:

• Patients de-escalating from ocrelizumab to MET had favorable outcomes, with 100% remaining relapse-free after the switch (ARR decreased from 0.14 to 0) and 92% free from new T2 lesions.
• Patients de-escalating from natalizumab experienced increased disease activity, with only 64% remaining relapse-free and 70% free from new T2 lesions (ARR increased from 0.05 to 0.14).

Cohort 2 (MET to LET)

The median age at de-escalation in cohort 2, 52.02 years, was nearly 10 years higher than in cohort 1, and the 15.78-year median disease duration was also longer. Most patients in this cohort de-escalated from fumarates (75%), usually to glatiramer acetate (41%) or teriflunomide (38%). De-escalation was most often prompted by adverse effects, safety concerns or tolerability. Patients were on MET for a mean of 1.60 years and on LET for a mean of 2.57 years.

The ARR remained relatively stable after de-escalation (0.14 on MET vs. 0.13 on LET). However, noninferiority for ARR was not technically met. The share of patients free from new T2 lesions decreased slightly, from 73% on MET to 67% on LET, but this change was not statistically significant.

As in cohort 1, outcomes differed by DMT type:

• De-escalation from fumarates appeared to be safer, with 78% of patients remaining relapse-free and 74% free from new T2 lesions after the switch.
• De-escalation from S1P modulators was associated with a higher risk of MRI activity, with only 46% free from new T2 lesions after de-escalation.

Notably, within cohort 2, two factors significantly influenced outcomes:

• Each 10-year increase in age at de-escalation was associated with a 60.8% reduced risk of new T2 brain lesions.
• Having a relapse in the year before MET initiation was linked to an increased risk of new T2 lesions after de-escalating to LET.

Takeaways for practice and future research

In their study report, the researchers share several clinical insights from their findings:

1. Risk appears to be higher with cell-trafficking agents. De-escalating from natalizumab and S1P modulators (such as fingolimod) carries a higher risk of rebound disease activity. If a switch from these cell-trafficking therapies is desired, transitioning to an anti-CD20 therapy first may be a prudent strategy before de-escalation or discontinuation.
2. De-escalation may be safer with anti-CD20s and fumarates. De-escalation from ocrelizumab (an anti-CD20 therapy) and fumarates appears to be a safer strategy. For these agents, if a switch is required due to safety or tolerability concerns, de-escalation can be used as a bridge before considering full discontinuation.
3. Patient characteristics are key. When considering de-escalation from MET to LET, older age at de-escalation and the absence of a relapse in the year prior to MET initiation are favorable indicators. Accounting for these factors, along with individual disease characteristics, can foster better-informed decision-making with patients.

Additionally, the authors note that the stringent noninferiority margin of 0.08 might be excessively conservative for de-escalation scenarios. They point out that a slightly higher margin, such as 0.12 for ARR in cohort 2, would have demonstrated noninferiority.

The authors also note that their investigation is limited by its observational design, potential referral bias, nonstandardized follow-up, missing MRI data and a relatively small number of patients de-escalating from ocrelizumab.

“Despite these limitations, this study is a helpful step forward in understanding DMT de-escalation and promoting its application in a more personalized fashion,” Dr. Hua concludes. “Future prospective studies with larger cohorts, standardized protocols, and inclusion of disability outcomes and adverse event data can help further refine de-escalation guidance and identify the safest treatment sequencing strategies.”