Abstract
Multiple sclerosis patients treated with disease-modifying therapies experience varying immune responses to COVID-19 vaccinations. However, guidance regarding the impact of treatments on infection risks remains sparse. Integrating vaccine-based and long-term coronavirus infection-based antibody data, we calculated cumulative probabilities of breakthrough infection for untreated multiple sclerosis patients and patients treated with interferon, dimethyl fumarate, natalizumab, or teriflunomide undergoing alternative Pfizer-BioNTech BNT162b2 booster frequencies. Annual boosters appear to effectively reduce risks for untreated multiple sclerosis patients. For treated patients, booster vaccinations likely provide protection that is nearly equivalent to that obtained in untreated patients, with moderate treatment-specific increases and decreases in immunity.
Introduction
COVID-19 vaccination strategies have primarily focused on identifying optimal booster frequencies that sustain immunity within the general population. For patients with multiple sclerosis (MS) undergoing disease-modifying therapies (DMTs), booster vaccination may be an essential tool for safe care. However, the impact of DMTs on the immune system varies significantly; some therapies appear to permit a relatively normal vaccine response, while others can severely impair the ability to respond to vaccinations. Data on antibody levels following COVID-19 vaccination justify a concern that atypical responses to DMTs might entail distinct booster schedules to achieve sufficient protection. Antibody responses are likely to vary depending on the specific DMT regimen, underscoring the need for specific guidance regarding infection risks associated with alternate frequencies of boosting.
Methods
We obtained anti-S antibody levels in individuals vaccinated with Pfizer-BioNTech BNT162b2, from MS patients undergoing treatment with interferon, dimethyl fumarate, natalizumab, or teriflunomide, from a group with untreated MS, and from a control group without MS [1]. We scaled antibody levels [2] by imputing the control group peak relative to MS patients and normalizing these values relative to our expected BNT162b2 post-booster peak [3]. To predict the trajectory of antibody decline in these cohorts, we incorporated longitudinal waning data for anti-N and anti-S IgG antibodies from six coronaviruses—HCoV-OC43, HCoV-NL63, HCoV-229E, SARS-CoV-1, SARS-CoV-2, and MERS—into an ancestral and descendant states analysis [4],[5]. Using augmented logistic regression models, we assessed daily endemic infection probabilities without further interventions. These infection probabilities enabled us to calculate cumulative risks of breakthrough infection. We then compared these risks across booster schedules [3]—ranging from yearly variant-specific updates to tailored schedules for MS patients, either untreated or on continuous therapy.
Results
Annual booster vaccinations over a two-year period provided equivalent protection for individuals without MS and untreated patients with MS, reducing infection risk by more than half (28–29% risk without boosting versus 11% with annual boosting; Figure 1). Notably, MS patients receiving interferon therapy experienced an even greater benefit from annual boosters, with their risk of infection dropping to just 6%—a nearly four-fold reduction of the 23% risk without boosting. Patients on dimethyl fumarate, natalizumab, or teriflunomide faced a range of moderately elevated infection risks relative to untreated patients; without boosters, nearly one-third of patients undergoing dimethyl fumarate or natalizumab treatment could become infected over two years. Annual boosting reduced this risk by half, resulting in a 15% risk over the same period. Patients treated with teriflunomide would be more vulnerable to infection relative to untreated MS patients (22% versus 11%) based on their antibody response. However, this prediction regarding teriflunomide is complicated by its unmodeled antiviral properties [6].
Cumulative probabilities of breakthrough infection over a two-year time span for updated BNT162b2 booster doses after primary vaccination, over intervals of one, three, six, or twelve months (colored bands). The analysis compares the general population, untreated MS patients (n = 205), and MS patients receiving one of four disease-modifying therapies: interferon (n = 135), dimethyl fumarate (n = 161), natalizumab (n = 74), and teriflunomide (n = 56) [1].
Discussion
Our analysis indicates that patients undergoing several MS treatments exhibit risks of infection that are fairly similar to risks for the general population. We showed that frequent boosting in these populations correlates with better protection. Patients undergoing all treatments benefited substantially from booster vaccination. MS patients on dimethyl fumarate or natalizumab therapies appear to be at moderately higher infection risk and may warrant additional attentiveness to timely adherence as well as consideration for more intensive vaccination schedules. Teriflunomide-treated patients appeared to face the highest risk; a bi-annual booster schedule might be necessary for patients on teriflunomide to achieve comparable protection against infection.
Our analysis is based solely on the effects of these treatments on antibody levels, and the consequences of those antibody responses to vaccination on infection probability. Some DMTs increase or diminish the antibody response. For instance, interferon increases antibody response to booster vaccination and therefore in our analysis diminishes risk of infection for all booster schedules. On the other hand, long-term usage can induce neutropenia or lead to inflammation, which have adverse effects that increase risk of some infections, especially infections of bacterial or fungal origin. Nevertheless, studies of the effect of interferon treatment in MS have demonstrated—consistent with our result—that patients are not subject to any generalized increase in risk of viral infection [7].
Conversely, Teriflunomide has been shown to inhibit activated T and B cells [7], and decreases antibody response to booster vaccination [1]. This lower initial antibody response increases risk of infection for all booster schedules. However, studies of teriflunomide have also demonstrated that this therapy has antiviral properties against SARS-CoV-2 [6]. These unmodeled antiviral properties may lead to protection from infection, and therefore justify some concern that our results for teriflunomide may overstate the actual risk of infection at each booster vaccination frequency.
More potently immunosuppressive DMTs exist, including fingolimod, ocrelizumab, and rituximab. These DMTs can reduce the abundance of lymphocytes and alter various B-cell and T-cell traits, thereby impairing the ability to mount an effective immune response post-vaccination. Predicting the effects of these more immunosuppressive therapies poses even greater challenges. Limited data (n < 10; [1]) indicates that in MS patients under these therapies, levels of antibodies, even after vaccination, are lower than baseline unvaccinated antibody levels in non-MS populations. Our rates of antibody decline and probabilities of infection are based on those non-MS populations, in whom antibody levels do not decline below this baseline.
Consequently, there is no empirical basis enabling our projection of antibody levels subsequent to booster vaccination, nor to associate these low levels of antibody with probabilities of infection. In this context, ascertaining the effect of boosting is highly speculative. However, with such low levels of antibodies, risk of infection is high. Patients on fingolimod and ocrelizumab have been shown to have a reduced response to vaccination against SARS-CoV-2 [1],[8],[9], and research on cancer patients treated with rituximab has indicated that even monthly boosters would fail to confer substantial protection [10].
Our study uses infection data from fully endemic coronaviruses for long-term predictions, which means it is based on responses to evolving viruses and therefore is appropriate to scenarios involving regularly updated vaccines that target predominant strains. For the same reason, our analysis accounts for waning vaccine efficacy due to antigenic evolution. The infection probabilities we calculated are not based on early trial data, which has limitations for long-term predictions because the immune systems of early trial participants were largely naive to the virus, requiring significant immunological adaptation to develop effective cellular immunity. Our study does not account for antigenic changes occurring between vaccine production and deployment, which could reduce booster efficacy. Regardless, our findings provide crucial guidance for mitigating SARS-CoV-2 infections in MS patients undergoing various DMTs until extensive longer-term infection data becomes available in this distinct patient population.
Data Availability Statement
All data, code, and a detailed readme have been archived on Zenodo and are currently available: https://zenodo.org/doi/10.5281/zenodo.13384814.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
This work was supported by the National Science Foundation RAPID DEB 2031204 to JPT and AD and the National Science Foundation CCF 1918784 to JPT. The funding sources played no role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.
Acknowledgements
We thank members of the Townsend and Dornburg lab groups for helpful comments and support.
