Introduction

Disease-modifying therapies (DMTs) for multiple sclerosis (MS) have proven efficacy in preventing relapses and slowing disability progression. However, these medications interfere with the immune system, which may lead to increased susceptibility to infections [1] and to reduced response to vaccinations [2].

The outbreak of the Coronavirus disease 2019 (COVID-19) pandemic has highlighted the potential impact of DMTs on infection risk of people with MS (pwMS). In particular, early evidence has shown that pwMS who are treated with anti-CD20 therapies are at increased risk of COVID-19-related morbidity [3]. Furthermore, these patients have demonstrated weaker humoral responses after COVID-19 infection, raising concerns about their response to future vaccination [4, 5].

On December 20, 2020, the Israeli vaccination campaign was launched, after the government had signed a contract with Pfizer. Since the beginning of the campaign, each adult MS patient was entitled to receive the BNT162b2 vaccine, a spike-encoding nucleoside-modified mRNA vaccine. On August 1, 2021, Israel began vaccinating patients with a third dose (booster).

The new vaccination was reported as safe for pwMS [6, 7] and pwMS seemed willing to receive it [8], however doubts remained regarding its efficacy due to the immunocompromised status of these patients. Several studies investigated the immune response to vaccination among pwMS treated with DMTs. Suboptimal humoral response has been reported for pwMS who were treated with ocrelizumab and fingolimod, a sphingosine-1-phosphate receptor modulator (S1PRM) [9]. Tortorella et al. reported similar findings, and noted that S1PRM treatment was also associated with a weak T-cell response [10]. On the other hand, adequate virus specific T-cell response to vaccination has been demonstrated for pwMS who were treated with anti-CD20 medication [11, 12]. Follow-up studies were in agreement with initial results [13, 14].

Epidemiological studies from Italy reported that low SARS-CoV-2 antibody level after the second vaccine dose predicted breakthrough COVID-19 infection among pwMS who were treated with DMTs during the delta and omicron waves [15]. Accordingly, breakthrough infection was more common among pwMS who were treated with fingolimod, as well as with anti-CD20s compared to all other DMTs [15, 16]. According to population based data from England, the incidence rate ratio of COVID-19 (incidence rate of infection among pwMS taking DMTs divided by incidence rate among the general population) increased after the start of mass vaccination for people who were treated with fingolimod and ocrelizumab, but not with other DMTs [17].

We aimed to further explore the risk of breakthrough COVID-19, despite vaccination, among pwMS who were treated with DMTs during the COVID-19 wave caused by the alpha variant in Israel.

Methods

A population-based retrospective cohort study, utilizing the computerized database of Clalit Health Services (CHS), the largest of four integrated healthcare organizations in Israel, which provides obligatory healthcare to more than 50% of the population. This insured population is geographically and socioeconomically diverse. CHS’s electronic medical record system extracts data from inpatient and outpatient care facilities, laboratories, and pharmacies. This database includes demographic information, diagnoses, laboratory measurements, and medication dispensing.

PwMS actively treated with DMTs were included in this study. Active treatment was identified by medication dispensing within 12 months, 9 months and 3 months prior to the beginning of follow-up for cladribine and alemtuzumab, ocrelizumab and all other DMTs, respectively. COVID-19 diagnosis was based on a polymerase chain reaction (PCR) test. PwMS who had COVID-19 prior to the beginning of follow-up were excluded. Severe COVID-19 was coded in CHS database according to the national institutes of health (NIH) definitions, namely the presence of one or more of the following indicators: oxygen saturation less than 94%, a ratio of arterial partial pressure of oxygen to fraction of inspired oxygen of less than 300 mmHg, a respiratory rate exceeding 30 breaths per min or the observation of lung infiltrates accounting for over 50% of the lung area (https://www.covid19treatmentguidelines.nih.gov).

Clinical data on age, gender, expanded disability status scale (EDSS), comorbidities, vaccination status, COVID-19 infection and severity were extracted from CHS database.

This study has been approved by the institutional review board (IRB) of the Lady Davis Carmel Medical Center. Because this was a retrospective cohort study, based on automatically and routinely collected administrative and clinical data in a coded database, informed consent has been waived by the IRB.

Statistical analysis

Multivariate stepwise cox proportional hazard regression models were used. In the first analysis, all pwMS were included, regardless of their vaccination status. In this model the dependent variable was COVID-19 infection. Independent variables were vaccination status (entered as a time-dependent variable), age, sex, EDSS, specific DMTs and various comorbidities. In the second analysis only pwMS who were vaccinated with 2 doses of BNT162b2 mRNA Covid-19 vaccine were included and the predictors were age, sex, EDSS, specific DMTs and comorbidities. The third analysis compared pwMS who were vaccinated with two doses, on each DMT separately, with unvaccinated pwMS. Independent variables were vaccination status (entered as a time-dependent variable), age, sex, EDSS and various comorbidities.

In the first and third analyses, which included vaccinated and unvaccinated PwMS, participants were followed since the commencement of the mass vaccination campaign on December 20, 2020. Participants were considered vaccinated in time-dependent analyses from 2 weeks after receiving the second vaccine dose. Importantly, COVID-19 infections, which had occurred after December 20, 2020 but before 2 weeks after receiving the second vaccine contributed to the hazard rate of the unvaccinated group. In the second analysis, which included only vaccinated pwMS, participants were followed from two weeks after receiving the second vaccine dose. The end of follow-up in all analyses was at the time of COVID-19 infection, the receipt of a third vaccine dose (booster), or until the end of August 2021, whichever came first. Patients who were vaccinated with only one dose were categorized as not vaccinated.

Statistical analyses were performed using SPSS (IBM, Chicago, Illinois).

A two-sided p-value of less than 0.05 was considered statistically significant in all analyses. Descriptive statistics are presented as means and standard deviations.

Results

A total of 2612 pwMS who were actively treated with DMTs were found in the database on December 20, 2020. Of whom, 101 patients were excluded due to a documented COVID-19 prior to the beginning of follow-up period. Among the included patients, 2123 (84.5%) received 2 vaccine doses, 41 (1.5%) received only 1 vaccine dose and 347 (14.0%) declined vaccination. The study participants’ baseline demographic and clinical characteristics are presented in Table 1 and the characteristics of the COVID-19-infected patients are presented in Table 2. The average age of patients at the beginning of follow-up was 46.2 ± 14.6, 70% were females. 137 patients developed COVID-19 infection during follow-up, 28% of whom were vaccinated. The mean duration of follow-up from the commencement of mass vaccination campaign was 239 ± 43 days. The mean duration of follow-up from 2 weeks after receiving 2 vaccine doses was 174 ± 31 days.

Table 1 Patients’ characteristics
Full size table
Table 2 Characteristics of COVID-19 cases during follow-up
Full size table

Risk factors for COVID-19 infection in multivariate analyses are reported in Table 3. Vaccination with 2 doses of mRNA BNT162b2 was protective from COVID-19, with a HR of 0.41 (95% CI 0.24–0.69) among pwMS who were treated with DMTs. Notably, treatment with ocrelizumab as well as with S1PRM was associated with higher rates of COVID-19 (Table 3).

Table 3 Multivariate Cox regression model evaluating risk factors for SARS-CoV-2 infection among pwMS (N = 2511)
Full size table

In the analysis that included only the vaccinated population, treatment with ocrelizumab, S1PRM and cladribine emerged as potential risk factors for breakthrough COVID-19 on stepwise multivariate analysis (Table 4). In the model that included all predictors, natalizumab was also associated with increased risk of breakthrough COVID-19 (Table 4; Fig. 1).

Table 4 Multivariate Cox regression model evaluating risk factors for SARS-CoV-2 infection among vaccinated pwMS (N = 2123)
Full size table
Fig. 1
figure 1

Cumulative incidence of breakthrough COVID-19 on various DMTs. Cumulative incidence of breakthrough COVID-19 on various disease modifying treatments (DMTs), among 2123 participants with multiple sclerosis, who were vaccinated with 2 doses. Follow up time was from 2 weeks after receiving the second vaccine dose, until COVID-19 infection, receipt of a third vaccine dose or the end of August 2021, whichever came first. S1PRM  Sphingosine-1-phosphate receptor modulators, Other DMTs include dimethyl fumarate, interferon beta, glatiramer acetate and teriflunomide. For statistics refer to Table 4

Full size image

On multivariate models that included unvaccinated and fully vaccinated pwMS on each DMT separately, a protective trend was noted for vaccination on all DMTs (0.09 < HR < 0.65), except for cladribine (HR = 1.1, P = 0.87). This protective trend was not statistically significant for pwMS on ocrelizumab, S1P receptor modulators and natalizumab (Table 5).

Table 5 Multivariate Cox regression models evaluating hazard ratios for SARS-CoV-2 infection among fully vaccinated pwMS on each DMT separately, compared to unvaccinated pwMS a
Full size table

Only 8 pwMS had a severe COVID-19 infection, just 2 of them were fully vaccinated. Their characteristics are presented in Table 6. There were no cases of COVID-19-related mortality during follow-up.

Table 6 Characteristics of the patients with severe COVID-19 despite vaccination
Full size table

Discussion

In this population-based study of pwMS who were actively treated with DMTs, mRNA vaccination against COVID-19 was protective. Although active treatment with ocrelizumab, S1PRM, cladribine and natalizumab was associated with increased risk of breakthrough COVID-19, the infection was generally mild, regardless of DMT.

The findings support the effectiveness of COVID-19 mRNA vaccine among pwMS who are using DMTs, with an HR of 0.41. The degree of protection is consistent with a previously reported HR of 0.51 during the delta wave in Italy [15], but much less than vaccine effectiveness of 92% that has been reported for the general Israeli population [18]. This difference in vaccine effectiveness may be due to disability and immune suppression of pwMS.

In the current study treatment with ocrelizumab and S1PRMs was found as a risk factor for COVID-19 infection in the post-vaccination era (Table 3). Since this analysis included vaccinated as well as unvaccinated pwMS, it is impossible to determine if the risk was due to breakthrough infection or owing to infection in those who had not been vaccinated. In line with our findings, treatment with anti-CD20 medications, has been found as an independent risk factor for COVID-19 in the pre-vaccination era [3, 19,20,21], while both S1PRM and anti-CD20 medications have been associated with COVID-19 after mass vaccination against COVID-19 took place [15,16,17]. The higher risk of infection is probably due to weakened humoral response to mRNA vaccines against SARS-CoV-2 for both S1PRM and anti-CD 20 medications, while pwMS on S1PRM had impaired cellular response as well [10, 14, 22, 23]. Diminished humoral response was also described after natural infection with SARS-CoV-2 on both anti-CD20 therapies and S1PRM [24, 25].

Upon inclusion of only pwMS who received 2 doses of mRNA vaccine, the increased risk with ocrelizumab and S1PRMs remained, while cladribine and natalizumab appeared to increase the risk of breakthrough COVID-19 as well (Table 4; Fig. 1). While adequate humoral response to mRNA vaccination on cladribine has been reported by some [9], others have noted diminished immune responses with this medication: post vaccination anti-Region-Binding-Domain median titter was lower in patients treated with cladribine compared to healthy people [10]. Interferon-γ-T-cell-specific responses were also found to be significantly lower in cladribine treated patients compared to healthy controls [10]. Moreover, a proportion of convalescent COVID-19 pwMS on cladribine did not develop IgG SARS-CoV-2 antibodies [13, 24, 26]. Data on the immunogenicity to COVID-19 vaccine among pwMS on natalizumab attest to preserved humoral and T-cell responses [27]. However, one study did show a significantly lower post-vaccine antibody levels in pwMS on natalizumab compared with untreated people [28]. Moreover, natalizumab may impair alpha-4-integrin mediated migration of monocytes into respiratory tissues during SARS-CoV-2 infection, which may contribute to increased risk of breakthrough COVID-19 despite adequate vaccine responses [29].

The aforementioned increased risk of breakthrough COVID-19 for people on ocrelizumab, S1PRM, natalizumab and cladribine must be interpreted with caution. One inherent limitation to the Cox regression approach is guarantee-time bias: events of COVID-19 were excluded from the analysis during the ‘guarantee’ time, from the start of follow-up on December 20, 2020, until the date of actual vaccination. Thus, guarantee-time differences between pwMS on different DMTs could potentially bias the results. To control for this potential bias, pwMS who were vaccinated with two doses, on each DMT separately, were compared with unvaccinated pwMS. In this analysis, vaccination status, on each DMT was entered as time-dependent variable, so that COVID-19 events after the start of follow-up and before vaccination were not excluded, but contributed to the hazard rate of the unvaccinated group [30]. Even after taking guarantee-time into account, the protective trend of vaccination was not statistically significant for people on cladribine, ocrelizumab, S1PRM and natalizumab (Table 5).

It is noteworthy and reassuring that probably due to vaccination COVID-19 among the participants of this study was generally mild. Only 2 severe infections were observed in vaccinated patients (Table 6) and no deaths were recorded. Indeed, the importance of vaccination is mainly to prevent severe infection. The paucity of severe breakthrough COVID-19 among vaccinated pwMS, regardless of DMT type was reported by others, and decreased significantly compared to the pre-vaccination era [15, 31]. As the virulence of SARS-CoV-2 variants decreases over time [32, 33], the odds of severe infection, despite vaccination, due to DMT use among pwMS is expected to decrease even further. Perhaps the retained ability to mount a vaccine-specific T-cell response on anti-CD20 therapies [11, 12, 34], might provide protection from severe disease by limiting viral replication to the upper respiratory tract [35]. In the case of S1PRM, it is possible that the diminished T-cell and humoral responses to vaccination that are detected in the peripheral circulation, do not adequately represent the capacity of the immune system to respond to severe infections, since lymphocytes are sequestered in lymph nodes, rather than abolished by these medications.

The COVID-19 pandemic introduced new uncertainties into clinical decision making with pwMS regarding the balance between efficacy and safety of DMTs. Our results support the notion that DMT choice in the COVID-19 era should still be guided first and foremost by MS disease activity, as COVID-19 vaccines seems effective in preventing severe infection, regardless of DMT. The signal of increased risk of breakthrough infection with ocrelizumab, S1PRM, natalizumab and cladribine call for considering personal risk of infection during therapeutic decision making, especially if MS is not highly active.

The strengths of this study are the use of population-based data, control of individual-level confounding factors, solid case definitions of COVID-19 by PCR testing and classification of COVID-19 severity according to NIH definitions, rather than by hospitalization status, which may be influenced by factors other than disease severity per se, such as background illnesses and use of immunosuppressive medications. Among the limitations are unknown antibody levels and the short follow-up time, dictated by the introduction of a mass vaccination campaign for a third vaccine dose (booster) and by the emergence of the delta wave. The short follow-up time, together with the limited number of participants resulted in small absolute numbers of breakthrough COVID-19 cases per DMT (Table 2). Small number of cases carry the risk of unstable results, as change of only few cases per DMT could have changed statistical significance. The short follow-up and small number of breakthrough infections also decreased the power of statistical analyses, explaining why protective trend of vaccination on some DMTs, demonstrated by HR < 1, did not reach statistical significance (Table 5).

In conclusion, vaccination against COVID-19 is effective for pwMS on various DMTs, significantly reduces the risk of infection, and particularly of severe infection. Although a signal of increased risk of breakthrough COVID-19, despite vaccination emerged with ocrelizumab, S1P receptor modulators, natalizumab and cladribine, the absolute number of infections was small, and COVID-19 was generally mild. Additional studies with longer follow-up times are needed to fully understand the risk of breakthrough COVID-19 and its consequences among pwMS who are treated with high-efficacy DMTs.

Data availability statement

Anonymized data will be shared by request from any qualified investigator.