Abstract
Introduction:
Numerous immunomodulatory treatments exist for multiple sclerosis (MS), including those that deplete immune cells (e.g. anti-CD20 medications), relocate immune cells (e.g. natalizumab, S1P modulators), or modulate immune subsets (e.g. fumarates). All disease-modifying treatments (DMTs) can increase infection risk which could worsen with prolonged use.
Methods:
This is a retrospective, single-center, observational cohort study. We analyzed medical records of adult people with MS who took natalizumab, S1P modulators, fumarates or anti-CD20 medications for over two years between January 2013 and April 2021 at Yale. We identified severe infections (requiring hospitalization) and mild infections (identified through outpatient antibiotic prescriptions or chart reference to “infection”). We used a zero-inflated negative binomial regression to assess the effects of DMT use, treatment duration, and patient characteristics on infection likelihood and frequency, while controlling for biologic sex, body mass index, ambulatory status, Charlson Comorbidity Index (CCI), diagnosis, disease modifying therapy and treatment duration.
Results:
104 patients received natalizumab, 61 fumarates, 17 S1P modulators and 291 anti-CD20 medications, with significant baseline differences in age, diagnosis, duration of DMT use, and CCI. Mild infection rates did not differ across DMTs, but severe infections were more common in patients on fumarates. Patients with longer DMT duration or requiring a walking aid had higher mild infection rates, while those with progressive MS or on long-term fumarates had higher severe infection rates, even after controlling for other variables.
Discussion:
This study demonstrates how real-life practice patterns, patient factors and DMT choice can influence infection rates, differing from randomized trial patterns. Natalizumab appears safe over extended use, while fumarates were linked to more severe infections, potentially due to the clinical selection of patients with poorer baseline health. The duration of DMT use may predict mild infection rates.
Keywords: Disease modifying treatment, Infection risk, Multiple sclerosis, Long-term treatment
1. Introduction
Multiple sclerosis (MS) is an inflammatory central nervous system (CNS) disease that has proven to respond well to humoral and cell-mediated immunotherapy (Ransohoff et al., 2015). Left untreated, MS can result in significant morbidity and accumulated disability, thus in recent years there has been a push towards aggressive early treatment of patients with highly efficacious therapies such as natalizumab (approved in 2003), fingolimod (approved in 2009), or ocrelizumab (approved in 2017) (Spelman et al., 2021). As a result, many patients have now been on highly effective immunomodulatory medications for many years. Infection risk tends to increase with longer duration of immunosuppression, and it is not yet clear to what extent MS patients on long-term immunotherapy will experience real-world risks associated with these medications. Long-term outcome data can help guide clinical practice guidelines, and options for mitigating risk, such as extended dosage intervals or de-escalation approaches in older patients, are currently being explored (Vollmer et al., 2022; Bou Rjeily et al., 2024). It remains important to continue to track and better characterize infection risk posed by long-term use of MS immunotherapies. We conducted a retrospective cohort study to characterize infection risk in a real-world population of people with MS cared for at a single tertiary care center who were on natalizumab, fumarates, S1P modulators or anti-CD20 medications for more than two years.
2. Methods
2.1. Study design and data collection
We reviewed electronic medical records (EMR) of all people with MS within the Yale New Haven Health System who took natalizumab, S1P modulators, fumarates or anti-CD20 medications for more than two years during January 2013 to April 2021. The charts of patients on non-B-cell-depleting therapies were reviewed by two physicians (CML and TH). The charts of patients on anti-CD20 therapies were reviewed by a third physician (JP) and have been previously published; (Peters and Longbrake) for quality assurance these cases were re-reviewed (CML) to ensure inter-rater reliability in data collection methods. Cases were included if the DMT of interest had been used for over two years and there were greater than two follow-up visits with a neurologist. Patients with unknown duration of DMT use or with incomplete or absent neurologic follow-up were excluded (Supplemental Figure 1).
2.2. Study variables
We defined severe infections as those requiring hospitalization and intravenous (IV) antibiotic administration; these were identified through review of discharge summaries. To mitigate the potential confounder of lower infection rates during mandated lockdowns imposed by COVID-19, study follow-up was limited to Spring 2021. COVID-19 related hospitalizations and infections were only available for our non-B-cell depleting DMT cohort and were therefore excluded for the purposes of this analysis. We defined mild infections as those managed outpatient with oral antibiotics, anti-fungals, antivirals, or supportive care. Mild infections were identified through chart evidence of a prescription for oral antibiotics, antivirals, anti-protozoal or anti-fungal agents, antibiotic ear or eye drops, suppository anti-fungal agents or mention of an “infection” in the chart. Long-standing antimicrobial prescriptions for prophylactic infection suppression (in cases of recurrent urinary tract infections), antibiotics for acne treatment, and topical antibiotic powders and creams, such as nystatin powder, were excluded. We also collected current and past DMT use of natalizumab, fumarates, S1P and anti-CD20s, duration of DMT use, Charlson Comorbidity Index (CCI) (Charlson et al., 1987) and demographic data (Table 1). A binary variable of presence or absence of lymphopenia was collected for patients on fumarate medications. Patients were listed as having lymphopenia if they had 2 or more absolute lymphocyte count (ALC) readings <800/μL separated by >1 month (NIH: Common Terminology Criteria for Adverse Events (CTCAE) 2021).
Table 1.
Demographic data.
| Natalizumab (N = 104) | Fumarates (N = 61) | S1P modulator (N = 17) | Anti-CD20 (N = 291) | Total N = 456) | Between group differences | ||
|---|---|---|---|---|---|---|---|
|
| |||||||
| Diagnosis n (%) | RRMS | 92 (89.3) | 50 (82.0) | 12 (75.0) | 239 (82.1) | 395 (83.5) | X2=48.739 |
| SPMS | 10 (9.6) | 10 (16.4) | 3 (17.6) | 6 (2.1) | 29 (6.1) | p < 0.001 | |
| PPMS | 1 (1.0) | 1 (1.6) | 1 (5.9) | 16 (5.5) | 19 (4.0) | ||
| Other | 0 (0.0) | 0 (0.0) | 0 (0.0) | 30 (10.3) | 30 (6.3) | ||
| Gender n (%) | Female | 79 (76.7) | 50 (82.0) | 12 (75.0) | 200 (68.7) | 343 (72.5) | X2=5.764 |
| Male | 24 (23.1) | 11 (18.0) | 4 (23.5) | 91 (31.3) | 130 (27.5) | p = 0.124 | |
| Age | Average (std dev) | 47.82 (11.1) | 53.6 (12.2) | 51.4 (14.5) | 46.4 (13.0) | 47.9 (12.7) |
F = 5.876 p < 0.001 |
| Race n (%) | White | 76 (73.8) | 42 (68.9) | 14 (87.5) | 200 (68.7) | 332 (70.5) |
X2=9.228 p = 0.161 |
| Black | 16 (15.4) | 17 (27.9) | 1 (5.9) | 57 (19.6) | 91 (19.2) | ||
| Asian/Other | 11 (10.6) | 2 (3.3) | 1 (5.9) | 34 (11.7) | 48 (10.1) | ||
| Smoking Status n (%) | Current | 11 (10.6) | 8 (13.1) | 1 (5.9) | 41 (14.1) | 61 (13.0) | X2=3.927 |
| Former | 41 (39.8) | 25 (41.0) | 8 (47.1) | 98 (33.7) | 172 (36.5) | p = 0.687 | |
| Never | 51 (49.5) | 28 (45.9) | 7 (43.8) | 152 (52.2) | 238 (50.5) | ||
| Ambulation Status n (%) | No assist | 82 (79.6) | 36 (59.0) | 9 (52.9) | 233 (80.1) | 360 (76.4) |
X2= 19.779 p = 0.019 |
| Unilateral (cane) | 7 (6.7) | 11 (18.0) | 2 (11.8) | 21 (7.2) | 41 (8.7) | ||
| Bilateral (walker) | 5 (4.8) | 4 (6.6) | 2 (12.5) | 17 (5.8) | 28 (5.9) | ||
| Wheelchair | 9 (8.7) | 10 (16.4) | 3 (17.6) | 20 (6.9) | 42 (8.9) | ||
| CCI | Average (std dev) | 0.95 (1.28) | 1.95 (2.13) | 1.24 (1.44) | 0.84 (1.21) | 1.03 (1.43) |
F = 11.103 p < 0.001 |
| BMI | Average (std dev) | 28.4 (6.2) | 29.8 (8.9) | 28.3 (6.4) | 29.6 (7.9) | 29.29 (7.68) |
F = 0.829 p = 0.478 |
| Months on DMT | Average (std dev) | 86.9 (39.1) | 59.9 (25.4) | 79.4 (45.4) | 46.6 (15.5) | 58.3 (30.1) |
F = 69.000 p < 0.001 |
Demographic data with between group differences. Continuous variable group differences were assessed with analysis of variance (ANOVA) and percentage variables were assessed with Chi square test. Alpha was set to 0.05. Relevant abbreviations include: RRMS = relapsing remitting multiple sclerosis, SPMS = secondary progressive multiple sclerosis, PPMS = primary progressive multiple sclerosis, Other = all other diagnosis being treated with immunomodulatory treatments (including myelin oligodendrocyte glycoprotein antibody-associated disease, neuromyelitis optica spectrum disorder and TNF-alpha induced demyelination), std dev = standard deviation, CCI = Charlson Comorbidity Index, BMI = body mass index, DMT = disease modifying treatment.
2.3. Data analysis and modeling
Demographic data was assessed for between-group differences; continuous variables were analyzed using Analysis of Variance (ANOVA) and percentage variables with a Chi square test. Alpha was set to 0.05. A multivariate zero-inflated negative binomial (ZINB) regression was used to analyze the count outcome of number of infections. This model was chosen due to over-dispersion of the cohort, where variance exceeded the mean, and a large number of patients had zero infections. Separate models were generated for the primary endpoints of: (1) mild infections and (2) severe infections. ZINB models generate two outputs: the logistic output (zero-inflated component) and the count output (negative binomial component). The logistic (zero-inflated) component examines what covariates increase the probability of being in the “zero infection” (i.e. the over-dispersed) group, therefore this generates a binary outcome variable based on the presence/absence of infection. The count (negative binomial) component considers infection rates. Covariates of biologic sex, body mass index (BMI), ambulatory status, CCI (which includes age and baseline comorbidities), diagnosis, DMT and duration of DMT use were included.
The S1P modulator group was excluded from regression analyses due to low sample size. Secondary progressive and primary progressive MS were collapsed into a single variable of “progressive MS”, and use of a cane, walker or wheelchair were collapsed into a single variable of “walking aid”, due to low sub-group sample sizes. ZINB model performance measures were acceptable. Tolerance and Variance Inflation Factor (VIF) did not demonstrate multicollinearity. Incidence Rate Ratio (IRR) and Odds Ratio (OR) were generated by exponentiating co-efficients for statistically significant covariates from the count and zero inflated component respectively. OR indicates the odds of being in the “zero infection” group (i.e. an OR >1 would indicate a decreased risk of having an infection). IRR indicates the risk of having higher rates of infections with each incremental change in the independent variable.
A stratified analysis by drug group was conducted to further clarify the role of duration of DMT use on mild infections; a multivariate ZINB model, controlling for duration of DMT use, sex, BMI and CCI, was run for each drug group individually. An additional fumarate-only sub-group analysis was carried out to explore the relationship between lymphopenia and severe infections using a multivariate ZINB regression controlling for BMI, sex and CCI. Analysis was carried out with SPSS version 29.0.1.0. The study was approved by Yale University Institutional Review Board and was exempt under federal regulation 45 CFR 46.
3. Results
3.1. Demographic data
We reviewed 562 charts, and 456 met inclusion criteria: 104 patients were exposed to natalizumab, 61 to fumarates, 17 to S1P modulators and 291 to anti-CD20 medications. Seventeen patients were exposed to more than one DMT during the study period and thus exist in more than one group. Our cohort was comprised of 72.5 % females, 70.6 % white patients who were 86.9 % non-Hispanic identifying. Most patients could ambulate without assistance (76.3 %) and 83.5 % were diagnosed with relapsing MS. Half of patients were never smokers (Table 1). Significant between-group differences were identified in age, diagnosis, ambulation status, CCI and duration of time on DMT. Patients were on natalizumab for a longer duration than other DMTs (a mean of 86.9 months compared to 59.9 months on fumarates, 79.4 months on S1P modulators and 46.6 months on anti-CD20s). The mean CCI was significantly higher for patients on fumarates and S1P modulators (1.95 and 1.24 respectively, compared to 0.95 on natalizumab and 0.84 on anti-CD20s). The natalizumab group had the highest percentage of relapsing remitting MS patients (89.3 %) whereas the fumarate and S1P modulator groups had relatively more secondary progressive MS patients (16.4 % and 17.6 % respectively, compared to 1 % in the natalizumab group and 6.1 % in the anti-CD20 group). More patients in the fumarates and S1P groups required an ambulatory assistive device (41 % and 42 %, respectively) compared to the natalizumab (20.2 %) and anti-CD20 (23.5 %) groups.
3.2. Raw infection rates
In the natalizumab group, 7.7 % of all patients suffered at least one severe infection compared to 34.4 % of the fumarate group, 23.5 % of the S1P modulator group and 5.7 % of the anti-CD20 group. Males had between 0 and 8 mild infections and females had between 0 and 35. Both males and females had between 0 and 7 severe infections. Patients exposed to natalizumab had an incidence of 37.6 mild and 1.6 severe infections per 100-patient-years. The anti-CD20 group had 50.2 mild and 4.5 severe infections per 100-patient-years, while fumarate-exposed patients had 61.9 mild and 16.2 severe and the S1P modulator exposed patients had 60.2 mild and 5.7 severe infections per 100-patient-years. The most common infection types across all groups were urinary tract infections and respiratory tract infections / pneumonia (Table 2).
Table 2.
Severe infections by infection type and DMT group.
| Natalizumab | Fumarate | S1P | Anti-CD20 | |
|---|---|---|---|---|
|
| ||||
| Urinary tract infection | 7 | 16 | 4 | 14 |
| Respiratory infection | 3 | 12 | 2 | 11 |
| Skin Infection | 2 | 6 | 0 | 12 |
| Sepsis | 0 | 6 | 0 | 4 |
| Viral infection | 0 | 4 | 0 | 4 |
| Bone or Joint Infection | 0 | 2 | 0 | 3 |
| Tick borne illness | 0 | 0 | 0 | 2 |
| Gastrointestinal infection (eg. colitis) | 0 | 1 | 0 | 5 |
| Total | 12 | 47 | 6 | 55 |
Raw number of severe infections, suffered by patients who were found to have at least 1 severe infection during the study (n = 60).
3.3. Infection prevalence for patients on long-term DMT
For mild infections, sex and years on DMT were statistically significant. Males were more likely to have a mild infection compared to females (OR = 94.16, Z = 2.64, p = 0.008; see table 3 and 4). In the zero-inflated model only, with each year of use there was a decreased probability of having a mild infection (OR = 1.70, Z = 2.43, p = 0.015). Regarding severe infections, CCI was the only statistically significant covariate and higher CCI was associated with a higher risk of having a severe infection (OR=0.61, Z=−2.12, p = 0.034).
Table 3.
Zero inflated negative binomial regression.
| (A) MILD INFECTION | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||
| Log likelihood | −857.205 (DF=21) | |||||||||
|
|
||||||||||
| AIC | 1756.411 | |||||||||
|
|
||||||||||
| Number of iterations | 89 | |||||||||
|
|
||||||||||
| Covariates | Count component (negative binomial) i.e. infection rates | Logistic component (zero-inflated) i.e. infection prevalence | ||||||||
|
|
|
|||||||||
| Estimate | Standard Error | Z value | Significance | Estimate | Standard Error | Z value | Significance | |||
|
| ||||||||||
| Male1 | −0.845 | 0.170 | −4.976 | <0.001 | 4.524 | 1.713 | 2.640 | 0.008 | ||
| BMI | 0.005 | 0.008 | 0.619 | 0.536 | −0.74 | 0.078 | −0.955 | 0.340 | ||
| Walking aid2 | 0.734 | 0.164 | 2.081 | 0.037 | −15.622 | 112.046 | −0.139 | 0.991 | ||
| CCI | −0.021 | 0.045 | 0.470 | 0.638 | 0.618 | 0.328 | 1.882 | 0.060 | ||
| Progressive MS3 | 0.396 | 0.241 | 1.648 | 0.099 | 3.268 | 1.708 | 1.914 | 0.056 | ||
| Other3 | 0.326 | 0.241 | 1.355 | 0.175 | −10.389 | 79.338 | −0.131 | 0.990 | ||
| Years on DMT | 0.132 | 0.033 | 3.685 | <0.001 | 0.532 | 0.219 | 2.432 | 0.015 | ||
| Fumarate4 | 0.132 | 0.215 | 0.615 | 0.538 | −0.117 | 1.693 | −0.069 | 0.896 | ||
| Anti-CD204 | −0.052 | 0.174 | −0.300 | 0.764 | 1.182 | 1.442 | 0.820 | 0.412 | ||
|
| ||||||||||
| (B) SEVERE INFECTION | ||||||||||
|
| ||||||||||
| Log likelihood | −195.874 (DF=21) | |||||||||
|
|
||||||||||
| AIC | 433.747 | |||||||||
|
|
||||||||||
| Number of iterations | 29 | |||||||||
|
|
||||||||||
| Covariates | Count component (negative binomial) i.e. infection rates | Logistic component (zero-inflated) i.e. infection prevalence | ||||||||
|
|
|
|||||||||
| Estimate | Standard Error | Z value | Significance | Estimate | Standard Error | Z value | Significance | |||
|
| ||||||||||
| Male1 | 0.994 | 0.436 | 2.282 | 0.022 | 0.935 | 0.784 | 1.194 | 0.233 | ||
| BMI | 0.047 | 0.024 | .1938 | 0.053 | 0.048 | 0.039 | 1.230 | 0.219 | ||
| Walking Aid2 | 0.350 | 0.512 | 0.683 | 0.494 | −0.920 | 0.932 | −0.987 | 0.323 | ||
| CCI | −0.289 | 0.089 | −3.159 | 0.002 | −0.490 | 0.231 | −2.119 | 0.034 | ||
| Progressive MS3 | 1.368 | 0.513 | 2.668 | 0.008 | 0.009 | 1.041 | 0.009 | 0.993 | ||
| Other3 | 1.093 | 0.855 | 1.279 | 0.201 | 0.514 | 1.214 | 0.423 | 0.672 | ||
| Years on DMT | 0.016 | 0.009 | 1.878 | 0.060 | 0.004 | 0.148 | −0.027 | 0.978 | ||
| Fumarate4 | 1.300 | 0.642 | 2.025 | 0.043 | −1.584 | 0.899 | −1.762 | 0.078 | ||
| Anti-CD204 | 1.907 | 1.641 | 1.162 | 0.245 | −1.124 | 0.937 | −1.199 | 0.231 | ||
A zero-inflated negative binomial (ZINB) regression was performed. Count models examined the rate/frequency of infections in those susceptible and the zero-inflated component of the model examined the likelihood of being in the “zero infection” group. Covariates that are statistically significant in the model are bolded and highlighted. Alpha set at 0.05 for all p-values. BMI = body mass index, DMT = disease modifying treatment, CCI = Charlston Comorbidity Index.
Comparison group is female.
Comparison group is no walking aid.
Comparison group is relapsing remitting multiple sclerosis (RRMS).
Comparison group is natalizumab.
Table 4.
Incidence rate ratios (IRR) and Odds ratio (OR).
| Covariates | Mild Infection | Severe Infection | ||
|---|---|---|---|---|
|
|
|
|||
| Count (IRR) | Zero-Inflated (OR) | Count (IRR) | Zero-Inflated (OR) | |
|
| ||||
| Male1 | 0.43 | 94.16 | 2.70 | |
| Walking aid2 | 2.08 | |||
| CCI | 0.75 | 0.61 | ||
| Progressive MS3 | 3.93 | |||
| Months on DMT | 1.01 | 1.045 | ||
| Fumarate medication4 | 3.67 | |||
Incidence Rate Ratio (IRR) were generated by exponentiating coefficients for statistically significant covariates from the zero-inflated negative binomial (ZINB) model count component (Table 3). They were calculated to conceptualize effect size. Number of mild infections and number of severe infections are the two output variables. An IRR of 1 indicates no effect on outcome variable, <1 indicates negative relationship and >1 indicates positive relationship.
Comparison group is female.
Comparison group is no walking aid.
Comparison group is relapsing remitting multiple sclerosis (RRMS).
Comparison group is natalizumab.
3.4. Infection rates for patients on long-term DMT
Sex, ambulation status and years on DMT were statistically significant predictors of mild infection rates. Males had lower infection rates compared to females (IRR = 0.43, Z=−4.98, p < 0.001; see tables 3 and 4). More years on DMT were associated with an increasing rate of mild infection whereby each year on DMT led to a 21.3 % increased rate of mild infection compared to the prior year (IRR = 1.21, Z = 3.68, p < 0.001). Requiring a walking aid was also associated with an increased infection rate (IRR =1.41, Z = 2.08, p = 0.037).
For severe infections, use of long-term fumarate medication, biologic sex, CCI and having progressive disease were statistically significant. Long-term fumarate use was associated with a 3.7-fold increase in severe infections (IRR=3.69, Z = 2.03, p = 0.043). Being male was associated with a 2.7-fold increased rate (IRR=2.70, Z = 2.28, p = 0.022), and having progressive disease was associated with a 3.9-fold increased rate of severe infection (IRR=3.90, Z = 2.67, p = 0.008). Although there was a statistically significant relationship between CCI and severe infection rates, only 10.9 % of the cohort had a CCI>2 and this finding is thus considered of limited clinical significance.
3.5. Specific drug group analysis
ALC data was available for 57 fumarate-treated patients and 27 were found to have lymphopenia. The binary covariate of presence versus absence of lymphopenia was not a significant predictor of having severe infections (count model IRR=0.74, Z = 0.44, p = 0.51 and logistic model OR= 0.56, Z=−0.39, p = 0.70). Drug-group-stratified multivariate ZINB regression analysis identified years on DMT to be a predictor of increased rate of mild infections in the anti-CD20 group only (IRR =1.38, z = 0.418, p < 0.001). Interestingly, the natalizumab sub-group demonstrated an incremental decreased infection risk with longer duration of use (OR=2.01, Z = 2.07, p = 0.038).
4. Discussion
This paper endeavored to identify the real-world risk of long-term DMT use. We observed some expected and unexpected patterns in our cohort, but overall, there was low incidence of both mild and severe infections. We selected ZINB model to help mitigate overdispersion, since the majority of patients had no infection; this allowed us to examine both prevalence and rates of infection in our population. Several patient factors appeared to play a role in infectious risk. We found progressive MS to be associated with a nearly four times increased rate of severe infections, in keeping with other work suggesting infection rates are higher amongst those with progressive disease, regardless of DMT use (Knapp et al., 2022; Hu et al., 2024). The need for a walking aid was also associated with increased rates of infection. Interestingly, male sex was associated with a higher rate of severe infections after controlling for other covariates. The mechanism for this finding is not known, but others have observed similar phenomena (Wijnands et al., 2017). Paradoxically, while males were more likely to have any infection, we observed lower rates of mild infections in males compared to females. This likely reflects the variability in mild infection counts, the presence of outliers within the female subgroup, and the small number of male patients in our cohort and should be interpreted with caution.
We found that the duration of DMT was associated with an increased rate of mild, but not severe infections. Patients on natalizumab had the longest time on medication (86.9 ± 39.1 months) and consequently the longest-term follow-up, yet they exhibited the lowest raw rates of infection with 37.6 and 1.6 mild and severe infections identified per 100 person-years, respectively. Additionally, patients were observed to have a lower prevalence of mild infection with increasing duration of use. Overall, these findings bolster the assumption that for John Cunningham virus (JCV) antibody negative patients, the long-term infectious risk of natalizumab use is likely negligible. Our results are in line with a 10-year follow-up of the Tysabri Observation Program which found a 0.2 % incidence of opportunistic infections and 4.1 % incidence of any infection (Butzkueven et al., 2020). Unfortunately, there remains a sub-population of people with MS for whom natalizumab is relatively contraindicated due to prior JCV exposure or natalizumab neutralizing antibodies (Simpson et al., 1940). Natalizumab treatment also requires access to monthly infusions which can be geographically limiting. For patients that cannot use natalizumab, anti-CD20 or fumarate medications are generally considered safe, however, we unexpectedly identified more severe infections in patients treated with long-term fumarates compared to other DMTs.
Fumarates are thought to modulate the frequency and function of B and T cells, while exerting neuroprotective effects via Nrf2-dependant and independent pathways; they are generally considered among the safer MS DMTs from an infection standpoint (Yadav et al., 2019; Longbrake et al., 2018). For example, one analysis of 1429 patients on dimethyl fumarate compared to 3170 patients on ocrelizumab found the annualized infection rates to be lower on fumarates (Nicholas et al., 2022). The Monotherapy Safety and Efficacy Extension Study in MS (ENDORSE) followed patients for up to thirteen years on treatment and found a < 1 % incidence of opportunistic infections (Gold et al., 2022). In our cohort we observed a higher raw incidence of severe infections in fumarate-treated participants and this difference persisted when controlling for diagnosis, duration of treatment and medical comorbidities (IRR = 3.67, Z = 2.03 p = 0.043), which differs from other similar cohorts (Fox et al., 2019). As has previously been reported, the presence of lymphopenia was not predictive of frequency or likelihood of severe infection (Fox et al., 2019). Additionally, duration of fumarate use was not predictive of increased rate or prevalence of mild infections in a DMT group stratified sub-analysis. We suspect that baseline between-group differences may have influenced these results. Specifically, patients in the fumarate group were older, more likely to need ambulatory assistive devices, and had more comorbidities. One may speculate that this milieu of comorbidities, age and poor mobility could have influenced their primary neuroimmunologists to keep them on a more “benign” medication option, rather than switching them to a high efficacy, but theoretically riskier, treatment option. It is possible this group of patients would have been more susceptible to infection, regardless of DMT use or duration of treatment. We did not observe high rates of mild or severe infections among people treated with long-term anti-CD20 therapy (mean duration of use was 46.6 months). While this is reassuring, there was an association between rates of mild infections and increasing time of anti-CD20 use in subgroup analyses. Many people are on anti-CD20 medications for much longer than the duration of follow-up available in our cohort and it is possible that infectious complications would continue on this trajectory with additional years of treatment. Indeed, drug-induced complications predisposing these patients to infection (such as hypogammaglobulinemia) tend to increase linearly over time (Langer-Gould et al., 2024; Mears et al., 2023;Boleto et al., 2018). Additional work studying real world experiences of people treated with continuous anti-CD20 therapies for longer than 5 years is needed.
5. Limitations
Our study was limited by its sample size, single-center design, and the inherent noise of EMR retrospective data collection. Due to local practice patterns, we had fewer patients on S1P modulators and fumarates compared to natalizumab and anti-CD20 medications. While we focused on patients with long-term DMT treatments, our mean treatment duration ranged from 46 to 79 months. Even longer use of DMT might influence infection risk further. Finally, other patient characteristics that can influence susceptibility to infection may not have been captured in EMR data and therefore are not controlled for in our model.
6. Conclusion
For those eligible to use natalizumab, it continues to be a safe option with limited long-term infectious risk. Fumarate-treated patients in our cohort had higher rates of severe infection, which appeared to be mediated by patient factors including generally poorer health status, but not lymphocyte counts or duration of treatment. This may reflect selection bias, as prescribers tend to view fumarates as benign. Prolonged use of anti-CD20s was associated with increased rates of mild infections, but longer-term prospective and retrospective analysis of infectious complications, with a special focus on populations who are older, have multiple comorbidities and who are on medication for increasingly long time periods are needed. Overall, long-term immunomodulatory treatment for MS remains safe for most individuals with MS. Treatment decisions around DMT choice will continue to need to be a case-by-case decision based on patient factors and risk tolerance, and lifestyle interventions focused on other modifiable risk factors should not be neglected.
Supplementary Material
Funding support
This study was supported by an investigator-initiated grant from Biogen to EEL [US-TYS-12,210]. Biogen did not have a role in study design, data collection, analysis and interpretation of data, writing of the report or the decision to submit the article for publication.
Footnotes
Declaration of competing interest
EEL: Consulting for Genentech, Novartis, TG Therapeutics, Bristol Myers Squibb, EMD Serono, Genzyme. Research support from Biogen, Genentech. CML: None. JJP: None, TH: None
CRediT authorship contribution statement
Clare McGarvey Lambert: Writing – original draft, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Taimoor Hussain: Writing – review & editing, Data curation. John Peters: Writing – review & editing, Supervision, Project administration, Methodology, Data curation, Conceptualization. Erin E Longbrake: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis.
Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.msard.2024.106236.
References
- Boleto G, et al. , 2018. Predictors of hypogammaglobulinemia during rituximab maintenance therapy in rheumatoid arthritis: a 12-year longitudinal multi-center study. Semin. Arthritis Rheum. 48, 149–154. [DOI] [PubMed] [Google Scholar]
- Bou Rjeily N, Fitzgerald KC, Mowry EM, 2024. Extended interval dosing of ocrelizumab in patients with multiple sclerosis is not associated with meaningful differences in disease activity. Mult. Scler. J. 30, 257–260. [DOI] [PubMed] [Google Scholar]
- Butzkueven H, et al. , 2020. Long-term safety and effectiveness of natalizumab treatment in clinical practice: 10 years of real-world data from the Tysabri Observational Program (TOP). J. Neurol. Neurosurg. Psychiatry 91, 660–668. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Charlson ME, Pompei P, Ales KL, MacKenzie CR, 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis. 40, 373–383. [DOI] [PubMed] [Google Scholar]
- Fox EJ et al. Lymphocyte counts and infection rates Long-term fingolimod treatment in primary progressive MS Class of Evidence Criteria for rating therapeutic and diagnostic studies. 6, (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gold R, et al. , 2022. Multiple sclerosis MSJ journal long-term safety and efficacy of dimethyl fumarate for up to 13 years in patients with relapsing-remitting multiple sclerosis: final ENDORSE study results. Mult. Scler. J. 801–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hu Y, et al. , 2024. Hospital-treated infections and risk of disability worsening in multiple sclerosis. Ann. Neurol. 96, 694–703. [DOI] [PubMed] [Google Scholar]
- Knapp R, et al. , 2022. Serious infections in patients with relapsing and progressive forms of multiple sclerosis: a German claims data study. Mult. Scler. Relat. Disord. 68, 104245. [DOI] [PubMed] [Google Scholar]
- Langer-Gould A, Li BH, Smith JB, Xu S, 2024. Multiple sclerosis, rituximab, hypogammaglobulinemia, and risk of infections. Neurol. Neuroimmunol. Neuroinflammation 11, e200211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Longbrake EE, et al. , 2018. Dimethyl fumarate induces changes in B- and T-lymphocyte function independent of the effects on absolute lymphocyte count. Mult. Scler. J. 24, 728–738. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mears V, et al. , 2023. Predictors of hypogammaglobulinemia and serious infections among patients receiving ocrelizumab or rituximab for treatment of MS and NMOSD. J. Neuroimmunol. 377. [DOI] [PubMed] [Google Scholar]
- Nicholas JA, et al. , 2022. Dimethyl fumarate is associated with lower rates of infection and lower infection-related healthcare costs when compared with ocrelizumab. Mult. Scler. Relat. Disord. 63, 103921. [DOI] [PubMed] [Google Scholar]
- NIH: Common Terminology Criteria for Adverse Events (CTCAE). https://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm#ctc_40 (2021). [Google Scholar]
- Peters J & Longbrake EE Infection risk in a real-world cohort of patients treated with long-term B-cell depletion for autoimmune neurologic disease HHS Public Access. doi: 10.1016/j.msard.2022.104400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ransohoff RM, Hafler DA, Lucchinetti CF, 2015. Multiple sclerosis—a quiet revolution. Nat. Rev. Neurol. 11, 134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Simpson A, Mowry EM, Newsome SD, 1940. Early aggressive treatment approaches for multiple sclerosis. Curr. Treat. Options. Neurol. 23, 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Spelman T, et al. , 2021. Treatment escalation vs immediate initiation of highly effective treatment for patients with relapsing-remitting multiple sclerosis data from 2 different national strategies supplemental content CME Quiz at jamacmelookup.com and CME Questions page 1280. JAMa Neurol. 78, 1197–1204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Vollmer BL, Wolf AB, Sillau S, Corboy JR, Alvarez E, 2022. Evolution of disease modifying therapy benefits and risks: an argument for de-escalation as a treatment paradigm for patients with multiple sclerosis. Front. Neurol. 12, 799138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wijnands JMA, et al. , 2017. Infection-related health care utilization among people with and without multiple sclerosis. Mult. Scler. 23, 1506–1516. [DOI] [PubMed] [Google Scholar]
- Yadav SK, Soin D, Ito K, Dhib-Jalbut S, 2019. Insight into the mechanism of action of dimethyl fumarate in multiple sclerosis. J. Mol. Med. (Berl) 97 (4), 463–472. 10.1007/s00109-019-01761-5. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.