949986

research-article20202020
MSJ0010.1177/1352458520949986Multiple Sclerosis JournalNG Caldito, A Shirani

MULTIPLE
SCLEROSIS MSJ
JOURNAL

Original Research Paper

Adverse event profile differences between Multiple Sclerosis Journal

1­–11

rituximab and ocrelizumab: Findings from DOI: 10.1177/

https://doi.org/10.1177/1352458520949986
1352458520949986
https://doi.org/10.1177/1352458520949986

the FDA Adverse Event Reporting Database © The Author(s), 2020.

Article reuse guidelines:
sagepub.com/journals-
permissions

Natalia Gonzalez Caldito , Afsaneh Shirani , Amber Salter and Olaf Stuve

Abstract Correspondence to:

O Stüve
Background: Rituximab and ocrelizumab are anti-CD20 monoclonal antibodies that have shown a Department of Neurology
marked reduction in multiple sclerosis (MS) inflammatory activity. However, their real-world safety pro- & Neurotherapeutics,
Immunology Graduate
file has not been adequately compared. Program, University of
Texas Southwestern Medical
Objective: To investigate the adverse event (AE) profile of rituximab and ocrelizumab reported to the Center, 6000 Harry Hines
Food and Drug Administration Adverse Event Reporting System (FAERS) database. Blvd., Dallas, TX 75390,
USA.
Methods: The FAERS database was filtered by indication (MS) and drug (rituximab or ocrelizumab). olaf.stuve@utsouthwestern.
Disproportionality analyses including but not limited to reporting odds ratio (ROR) were conducted to edu
Natalia Gonzalez Caldito
identify drug–AE associations. A signal was detected if the lower limit of the 95% confidence interval of Department of Neurology
ROR (ROR025) exceeded 1. & Neurotherapeutics,
Immunology Graduate
Results: There were 623 and 7948 reports for rituximab and ocrelizumab, respectively. The most fre- Program, University of
Texas Southwestern Medical
quent AEs with rituximab and ocrelizumab were infusion-related reaction (4.82%) and urinary tract Center, Dallas, TX, USA
infection (10.52%), respectively. The strongest drug–AE association for rituximab and ocrelizumab Afsaneh Shirani
were ear pruritus (ROR025: 47.53) and oral herpes (ROR025: 38.99), respectively. Ocrelizumab was Department of Neurological
Sciences, University of
associated with an almost two times higher frequency of infections than rituximab (21.93% vs 11.05%, Nebraska Medical Center,
Omaha, NE, USA
respectively).
Amber Salter
Conclusion: This study revealed differences in reporting AEs between rituximab and ocrelizumab. Infec- Division of Biostatistics,
Washington University
tions were reported more frequently with ocrelizumab. Although speculative, a potentially different or School of Medicine, St.
more extensive B-cell depletion by ocrelizumab might explain these findings. Additional pharmacovigi- Louis, MO, USA

lance studies need to be performed to better characterize differences in the AE profile in B-cell-depleting Olaf Stuve
Department of Neurology
therapies. & Neurotherapeutics,
Immunology Graduate
Program, University of
Texas Southwestern Medical
Keywords: Rituximab, ocrelizumab, CD20, adverse event profile, FAERS, multiple sclerosis Center, Dallas, TX, USA/
Neurology Section, VA
North Texas Health Care
Date received: 22 May 2020; revised: 18 May 2020; accepted: 23 July 2020. System, Dallas, TX, USA

Introduction B-cell activation and differentiation.5 A subset of T

Multiple sclerosis (MS) has a complex immunopatho- lymphocytes also expresses CD20. However, it is cur-
genesis.1,2 The innate and adaptive immune responses rently incompletely understood to what extend T-cell
are involved, and T cells as well as B cells play an depletion explains the beneficial clinical and paraclini-
important role in these immune cascades.2 The arma- cal impacts seen with anti-CD20 therapies in MS.6
mentarium for treating MS has significantly evolved
over the past two decades with the introduction of tar- Rituximab is a murine–human chimeric immunoglob-
geted monoclonal antibody (mAb) therapies.3 ulin G (IgG1) mAb that binds to the cell surface
antigen CD20.7 It is thought that rituximab reduces
The results of clinical trials with anti-CD20 (cluster of the number of circulating B cells via complement-
differentiation) mAbs have thus far provided the dependent cytotoxicity (CDC) and antibody-depend-
strongest evidence for a pathogenic role for B lympho- ent cellular cytotoxicity.8 In addition, rituximab may
cytes in MS.4 CD20 is an activated-glycosylated deplete B lymphocytes within cerebral perivascular
transmembrane phosphoprotein that co-mediates spaces directly, or through preventing the repopulation

journals.sagepub.com/home/msj 1
Multiple Sclerosis Journal 00(0)

of central nervous system (CNS) B cells from periph- Methods

eral compartments.9 In MS, the efficacy of rituximab
against placebo in patients with relapsing-remitting FAERS database
multiple sclerosis (RRMS) was first demonstrated in Established in 1968, the FAERS database is designed
the HERMES phase II study.10 Rituximab was also for post-marketing safety surveillance for all FDA-
tested against placebo in patients with primary- approved drugs and therapeutic biological products.15
progressive multiple sclerosis (PPMS) in the phase II/ It can also be used to identify new safety concerns not
III OLYMPUS trial, although the primary disability- previously identified in clinical trials. Furthermore, it
based outcome was not met.11 can be utilized for evaluating a manufacturer’s com-
pliance with reporting regulations and information
Ocrelizumab is a fully humanized anti-CD20 mAb sharing. The FAERS database contains information on
that was approved by the Food and Drug Administration AE and medication error reports submitted to the
(FDA) for relapsing forms of MS and PPMS.12,13 In FDA.15 Reports are submitted to the FDA by consum-
two identical phase III trials in patients with RRMS, ers, healthcare professionals, and manufacturers.
named OPERA I and OPERA II trials, ocrelizumab Although most of the reports submitted are from the
was compared to subcutaneous interferon beta-1a United States, any country can participate. Importantly,
(IFNβ-1a, Rebif®). The primary outcome was the consumers and healthcare professionals may report
annualized relapse rate, which was significantly lower directly to the manufacturer of a specific products,
with ocrelizumab in both studies.12 In another ground who is then required to forward these reports to the
breaking phase III trial called ORATORIO, ocreli- FDA. AE and medication errors are coded using terms
zumab was tested against placebo in people with in the Medical Dictionary for Regulatory Activities
PPMS revealing a modest but statistically significant (MedDRA) terminology.16 In this study, MedDRA
reduction in progression at 12 weeks.13 It subsequently version 22.1 was used.
became the first FDA-approved agent for PPMS.

Clinical trial efficacy and safety data suggest a favora- Study procedure
ble benefit-to-risk profile for anti-CD20 mAb in MS. Input data were obtained from the FAERS public
Infusion-related reactions were the most common AEs dashboard which is an interactive web-based tool for
reported with both HERMES and OLYMPUS trials. querying the FAERS database.15 A filter was applied
There was no difference in infection rates between the to only include reports in which the “suspect drug”
rituximab and placebo groups.10,11 Subsequent post- causing the AE was exclusively rituximab or ocreli-
marketing observational studies have corroborated a zumab, and the “indication for drug use” was exclu-
similar adverse event (AE) profile.14 Regarding ocre- sively MS (including all subtypes). All eligible reports
lizumab, OPERA I and II showed a similar AE profile included in the FAERS database since 1968 until the
for ocrelizumab as previously reported with rituxi- last quarter of 2019 were exported. Based on the
mab. However, in OPERA (I and II) and ORATORIO, MedDRA terminology,16 each AE is classified auto-
the frequency of infections was higher with ocreli- matically into an AE category in the FAERS database.
zumab compared to placebo and IFNβ-1a, respec- The AEs related to off-label use, product administra-
tively. The most common infections were upper tion, MS diagnosis, and pregnancy were excluded
respiratory tract infections. Of note, across both from the analysis.
OPERA (I and II) and ORATORIO trials, the fre-
quency of herpesvirus associated infections was
higher in the ocrelizumab arm than the comparison Statistical analysis
arm. Demographic data differences between rituximab
and ocrelizumab groups were studied using chi-
While much has been learnt about the efficacy and square test and Wilcoxon rank-sum tests in STATA
safety profile of rituximab and ocrelizumab in MS in version 13 (StataCorp, College Station, TX, USA).
phase II and III trials, there is paucity of studies on Differences between rituximab and ocrelizumab AE
comparative real-world safety profile of rituximab and categories were studied using chi-square test. For
ocrelizumab. The purpose of this study is to investi- signal detection, disproportionality analyses were
gate AEs reported for these agents using the Food performed. In pharmacovigilance, disproportionality
and Drug Administration’s Adverse Event Reporting analysis is a validated method that has been devel-
System (FAERS) database in the real-world practice oped to identify statistical associations between drugs
setting. and AEs (drug–AE associations or signals). Such

2 journals.sagepub.com/home/msj
 NG Caldito, A Shirani et al.

Table 1. Summary of basic demographic and clinical information.

Rituximab Ocrelizumab p-value

Total number of reports 623 7948
Age
Number of reports available 280 4510 <0.001a
Age (years); mean (SD), range 43.89 (12.66), 14–83 48.76 (12.45), 16–88
Sex
Number of reports available 506 7581 0.973b
Female, n (%) 361 (71.34) 5414 (71.42)
MS subtype
RRMS, n (%) 37 (5.95) 3163 (39.80) <0.001b
PMS, n (%) 35 (5.62) 1686 (21.21)
MS subtype not specified, n (%) 551 (88.44) 3099 (38.99)
AE severity
Non-serious, n (%) 219 (35.15) 3472 (43.68) <0.001b
Serious, n (%) 404 (64.85) 4476 (56.32)
AE resulting in death, n (%) 38 (5.75) 171 (2.11) <0.001b
Reporting source
Number of reports available 610 7942 <0.001b
United States, n (%) 393 (64.43) 5663 (71.30)
MS: multiple sclerosis; RRMS: relapsing remitting multiple sclerosis; PMS: progressive multiple sclerosis; SD: standard deviation;
AE: adverse events. Bold values indicate statistical significance (p<0.05).
aWilcoxon rank-sum test.
bChi-square test.

methods compare the observed count of a drug–AE Results

combination with an “expected” count. Unexpectedly
high reporting associations are identified as signals.17,18
Study population
In this study, proportional reporting ratio (PRR),
There were 623 and 7948 reports with rituximab and
reported odds ratio (ROR), and information compo-
ocrelizumab, respectively, in the FAERS database.
nent (IC) were computed. PRR and ROR can be
Demographic and clinical information are displayed
interpreted as relative risk and odds ratio, respec-
in Table 1. The mean age in the reports associated
tively. PRR and ROR are frequentist (non-Bayesian)
with rituximab was 5 years older than with the ocre-
statistics whereas IC is Bayesian.19 A two-by-two
lizumab reports (43.89 vs 48.76 years, p < 0.001).
contingency table is the framework for analyses
Progressive MS was more frequently found in the
(Supplementary Table 1a). The formulas used are reports associated with ocrelizumab than rituximab
defined in Supplementary Table 1b. In this section, (21.2% vs 5.6%). Yet, it needs to be noted that MS
only the scoring thresholds are given, and extensive subtype was not specified in 88.4% of the reports
details about the interpretation of these tests can be associated with rituximab versus 40.0% of reports
found elsewhere.19–21 χ2 (chi-square) with Yates’ cor- for ocrelizumab. Rituximab was associated with a
rection was calculated to test for independence in a higher proportion of reported serious AEs as com-
contingency table. Using PRR, a signal was detected pared to ocrelizumab (64.8% vs 56.3%, p < 0.001).
if the number of drug–AE combinations was 3 or Adverse events resulting in death were found in
more and the PRR was 2 or more associated with a 5.7% of rituximab reports versus 2.1% of ocreli-
χ2 value of 4 or more.21 For ROR, a signal was zumab reports (p < 0.001).
detected if the lower limit of the 95% confidence
interval (ROR025) exceeded 1.20 A signal was detected
using IC, when the lower limit of the 95% confidence Differences in adverse event categories between
interval (IC025) exceeded 0.19 Usually, when a drug- rituximab and ocrelizumab
AE association (signal) was observed with PRR, it As mentioned in section “Methods,” each AE was
was also presented with ROR and IC. automatically classified in an AE category. For

journals.sagepub.com/home/msj 3
Multiple Sclerosis Journal 00(0)

Table 2. Frequency of the adverse events classified by reactions groups.

Rituximab Ocrelizumab p-valuea

Total number of adverse events 1466 23,613
Total number of reports 623 7948
Blood and lymphatic system 42 (2.86%) 216 (0.91%) <0.001
Cardiac and vascular 69 (4.71%) 681 (2.88%) <0.001
Gastrointestinal, hepatobiliary, and endocrine 112 (7.64%) 1358 (5.75%) 0.003
General and administration site conditions 279 (19.03%) 4128 (17.48%) 0.130
Immune system 39 (2.66%) 264 (1.12%) <0.001
Infections and infestations 162 (11.05%) 5179 (21.93%) <0.001
Injury, poisoning, and procedural complications 58 (3.96%) 949 (4.02%) 0.906
Musculoskeletal and connective tissue 71 (4.84%) 1274 (5.4%) 0.362
Neoplasms (benign, malignant, and unspecified) 59 (4.02%) 302 (1.28%) <0.001
Nervous system and psychiatric 176 (12.01%) 3712 (15.72%) <0.001
Othersb 155 (10.57%) 2226 (9.43%) 0.146
Respiratory, thoracic, and mediastinal 121 (8.25%) 1840 (7.79%) 0.523
Skin and subcutaneous tissue 123 (8.39%) 1484 (6.28%) 0.001
aChi-square test.
bOthers include the following reactions groups (based on Medical Dictionary for Regulatory Activities terminology): investigations;
eye disorders; renal and urinary disorders; ear and labyrinth disorders; metabolism and nutrition disorders; pregnancy, puerperium,
and perinatal conditions; reproductive system and breast disorders; surgical and medical procedures; and congenital, familial, and
genetic disorders.

example, the AE pneumonia was classified in the Drug-adverse event associations (signals) with
infections category. There were a total of 1466 AEs rituximab
for rituximab and 23,613 AEs for ocrelizumab. The For this analysis, all the 623 reports associated with
total number of AEs for each drug was higher than rituximab were considered. A total of 321 unique
the total number of reports because a report could AEs were reported. The AEs included in this analysis
include more than one single AE. Of the 623 reports required to have a minimum frequency of 0.5%.
with rituximab, 496 had more than one reported AE, Table 3 summarizes the results of the disproportion-
and of the 7948 reports for ocrelizumab, 4832 had ality analysis.
more than one reported AE.
Of the AEs studied, despite a frequency of 0.8%, ear
There were several differences noted between rituxi- pruritus revealed the strongest signal with a PRR,
mab and ocrelizumab (Table 2). A pie chart for each ROR025, and IC025 of 113.88, 47.53, and 6.66, respec-
drug showing all the AEs distributed into AE catego- tively. This was followed by infusion-related reaction
ries is displayed in Figure 1. Ocrelizumab was asso- (4.82%) with a PRR, ROR025, and IC025 of 34.02,
ciated with an almost two times higher proportion of 24.73, and 5.08, respectively. In addition, throat irrita-
AEs in the infections category as compared to rituxi- tion (4.01%) and throat tightness (1.44%) demon-
mab (21.93% vs 11.05%, respectively, p < 0.001). strated a high signal. For throat irritation, PRR,
On the contrary, rituximab was associated with a sig- ROR025, and IC025 were 28.7, 20.01, and 4.83, respec-
nificant higher percentage of AEs in the blood and tively, and for throat tightness, PRR, ROR025, and
lymphatic system category as compared to ocreli- IC025 were 15.03, 7.89, and 3.86, respectively.
zumab (2.86% vs 0.91%, respectively, p < 0.001).
Rituximab was also linked to over three times higher Other AEs to highlight were malignant melanoma
proportion of AEs in the neoplasms category as com- (0.8%) and breast cancer (1.77%). Both revealed a
pared to ocrelizumab (4.02% vs 1.28%, p < 0.001, signal across the three methods. Malignant melanoma
respectively). Rituximab was associated with two demonstrated a PRR, ROR025, and IC025 of 18.28,
times higher frequency of AEs in the immune system 7.64, and 4.06, respectively, whereas breast cancer
category as compared to ocrelizumab (2.66% vs had a PRR, ROR025, and IC025 of 9.8, 5.49, and 3.28,
1.12%, p < 0.001, respectively). respectively. Another interesting AE was neutropenia

4 journals.sagepub.com/home/msj
 NG Caldito, A Shirani et al.

Figure 1. A pie chart for each drug displays the relative contribution of the different AE categories. The AE categories
have been automatically generated by the FAERS database and each AE has been automatically classified in its
correspondent AE category.
AE: adverse event.

(2.57%) which showed a signal across the three meth- respectively, for urinary tract infection, and 18.11,
ods as well (PRR, ROR025, and IC025: 6.75, 4.20, and 18.54, and 4.15, respectively, for nasopharyngitis.
2.80, respectively).
Among non-infection-related AEs, infusion-related
reaction (4.76%) showed one of the highest signals
Drug-adverse event associations (signals) with with a PRR, ROR025, and IC025 of 34.02, 32.15, and 5,
ocrelizumab respectively, followed by throat irritation (3.08%)
A total of 7948 reports associated with ocrelizumab with a PRR, ROR025, and IC025 of 22.22, 20.15, and
were considered in this analysis. There were 1271 4.39, respectively. Interestingly, MS relapses (4.1%)
unique AEs. Only a total of 42 AEs with a frequency demonstrated a high signal with a PRR, ROR025, and
higher than 0.5% were included in the analysis. IC025 of 16.78, 15.62, and 4.02, respectively.
Table 4 outlines the disproportionality analysis
results. Oral herpes with a reported frequency of
2.21% revealed the strongest signal across the three Discussion
different methods, with a PRR, ROR025, and IC025 of The results of this study using the FAERS database
44.35, 38.99, and 5.27, respectively. Supporting these suggest that rituximab and ocrelizumab have signifi-
results, herpes zoster (2.89%) showed one of the cant differences in reported AE profiles in the real-
highest signals, with PRR, ROR025, and IC025 of 16.39, world setting. When classified by AE categories, the
14.77, and 3.97, respectively. In addition, urinary tract frequency of AEs in the infections category was
infection (10.52%) and nasopharyngitis (9.79%) nearly two times higher with ocrelizumab as com-
showed high signals among non-herpetic infections. pared to rituximab. In addition, signals or drug-
PRR, ROR025, and IC025 were 20.11, 20.81, and 4.3, adverse event associations were identified between

journals.sagepub.com/home/msj 5
Multiple Sclerosis Journal 00(0)

Table 3. Associations of adverse events with rituximab.a

n (%) χ2b PRR ROR (ROR025) IC (IC025)

Total number of reports = 623
Infusion-related reaction 30 (4.82) 603.1 34.02 35.69 (24.73) 5.13 (5.08)
Pruritus 28 (4.49) 81.99 3.08 3.18 (2.18) 1.86 (1.8)
Fatigue 25 (4.01) 40.63 1.73 1.76 (1.18) 1.18 (1.12)
Throat irritation 25 (4.01) 687.3 28.7 29.85 (20.01) 4.89 (4.83)
Multiple sclerosis relapse 19 (3.05) 222.8 12.4 12.76 (8.08) 3.73 (3.64)
Rash 19 (3.05) 39.8 2.24 2.28 (1.44) 1.49 (1.4)
Dyspnea 16 (2.57) 17.65 1.2 1.21 (0.74) 0.8 (0.7)
Neutropenia 16 (2.57) 100.9 6.75 6.9 (4.2) 2.9 (2.8)
Pyrexia 15 (2.41) 22.73 1.65 1.66 (1) 1.14 (1.03)
Erythema 14 (2.25) 44.14 3.42 3.47 (2.04) 2.01 (1.9)
Fall 13 (2.09) 22.97 1.93 1.95 (1.13) 1.33 (1.2)
Urticaria 13 (2.09) 30.31 2.54 2.58 (1.49) 1.65 (1.53)
Nausea 13 (2.09) 8.467 0.73 0.72 (0.42) 0.33 (0.21)
Urinary tract infection 12 (1.93) 40.04 3.65 3.71 (2.09) 2.11 (1.97)
Breast cancer 11 (1.77) 98.03 9.8 9.96 (5.49) 3.43 (3.28)
Hypersensitivity 11 (1.77) 24.82 2.5 2.52 (1.39) 1.64 (1.49)
Chills 10 (1.61) 31.68 3.53 3.57 (1.91) 2.07 (1.91)
Throat tightness 9 (1.44) 120.5 15.03 15.24 (7.89) 4.03 (3.86)
Pneumonia 9 (1.44) 9.72 1.23 1.23 (0.64) 0.84 (0.67)
Sepsis 7 (1.12) 15.33 2.55 2.57 (1.22) 1.69 (1.47)
Dysphagia 6 (0.96) 13.23 2.64 2.65 (1.19) 1.75 (1.49)
Crohn’s disease 6 (0.96) 30.02 5.97 6.02 (2.69) 2.8 (2.54)
Malignant melanoma 5 (0.8) 73.94 18.28 18.42 (7.64) 4.36 (4.06)
Ear pruritus 5 (0.8) 459.5 113.88 114.79 (47.53) 6.97 (6.66)
PRR: proportional reporting ratio; ROR: reported odds ratio; IC: information component; IC025: lower limit of the 95% two-sided
confidence interval for IC.
aSignal detected when any of the following occurred: PRR > 2 and χ2 > 4, ROR
025 > 1, IC025 > 0.
bχ2 (chi-square) with Yates’ correction. Values in bold denote significant signals.

different infections such as oral herpes, nasopharyn- mAbs are categorized as Type I or Type II according
gitis, or urinary tract infection whereas none of these to their mode of CD20 binding and their primary
signals were seen with rituximab. Although specula- mechanism for depleting CD20-positive cells. The
tive, this might suggest that ocrelizumab—as a second- mechanism of B-cell depletion with type I anti-CD20
generation anti-CD20 mAb—may deplete B cells mAbs is mostly due to CDC activation as compared to
more extensively or differently, perhaps increasing type II anti-CD20 mAbs, which mediates direct cell
the risk for infections. However, the less frequent death through homotypic adhesion.23 Although rituxi-
serious reactions noted with ocrelizumab indicate mab and ocrelizumab have mainly CDC activity, ocre-
that these infections may be mild or moderate. lizumab is thought to have stronger antibody-dependent
cell-mediated cytotoxicity (ADCC) and relatively
Second-generation anti-CD20 mAb were primarily weaker CDC activities.22 Another characteristic that
developed to (1) decrease their immunogenicity, and rituximab and ocrelizumab share is that they interact
to (2) improve their efficacy and tolerability over with a very similar epitope on the CD20 tetramer mol-
first-generation anti-CD20 mAbs.22 Immunogenicity ecule. Specifically, both drugs bind to the large extra-
is often negatively correlated with efficacy and toler- cellular loop of CD20, and they share a very similar
ability, and the humanized molecular structure of the core epitope and contact region.23
newer generations of anti-CD20 mAbs is thought to
be a key factor in achieving these goals.3,22 Despite The higher efficacy of these second-generation anti-
their molecular differences, rituximab and ocreli- CD 20 mAbs has been proved in vitro although to
zumab share pharmacological properties. Anti-CD20 date it has not been confirmed in humans in vivo.24

6 journals.sagepub.com/home/msj
 NG Caldito, A Shirani et al.

Table 4. Associations of adverse events with ocrelizumab.a

n (%) χ2b PRR ROR (ROR025) IC (IC025)

Total number of reports = 7948
Urinary tract infection 836 (10.52) 16,567.06 20.11 22.36 (20.81) 4.3 (4.3)
Fatigue 833 (10.48) 3679.78 4.54 4.95 (4.61) 2.18 (2.17)
Nasopharyngitis 778 (9.79) 13,889.43 18.11 19.96 (18.54) 4.15 (4.15)
Headache 475 (5.98) 1159.33 2.51 2.6 (2.37) 1.32 (1.32)
Infusion-related reaction 378 (4.76) 12,630.94 34.02 35.67 (32.15) 5.01 (5)
Asthenia 374 (4.71) 1159.16 3.16 3.27 (2.94) 1.65 (1.65)
Malaise 363 (4.57) 1030.48 2.9 2.99 (2.69) 1.53 (1.52)
Pruritus 329 (4.14) 917.2 2.84 2.92 (2.62) 1.5 (1.5)
Multiple sclerosis relapse 326 (4.1) 5404.03 16.78 17.46 (15.62) 4.03 (4.02)
Pneumonia 299 (3.76) 940.53 3.2 3.28 (2.93) 1.67 (1.67)
Cough 298 (3.75) 1249.96 4.25 4.38 (3.9) 2.08 (2.07)
Pyrexia 295 (3.71) 735.35 2.54 2.6 (2.31) 1.34 (1.34)
Pain 279 (3.51) 491.34 1.8 1.83 (1.63) 0.85 (0.84)
Gait disturbance 277 (3.49) 1362.6 4.98 5.13 (4.55) 2.3 (2.3)
Nausea 274 (3.45) 318.32 1.2 1.21 (1.07) 0.26 (0.26)
Influenza 256 (3.22) 2624.38 10.37 10.68 (9.43) 3.34 (3.34)
Throat irritation 245 (3.08) 5364.09 22.22 22.89 (20.15) 4.4 (4.39)
Dyspnea 233 (2.93) 311.62 1.37 1.38 (1.22) 0.46 (0.45)
Herpes zoster 230 (2.89) 3721.9 16.39 16.85 (14.77) 3.98 (3.97)
Fall 225 (2.83) 580.94 2.62 2.67 (2.34) 1.39 (1.38)
Dizziness 224 (2.82) 310.63 1.42 1.43 (1.26) 0.51 (0.5)
Rash 218 (2.74) 431.29 2.02 2.05 (1.79) 1.01 (1)
Oropharyngeal pain 197 (2.48) 1558.04 8 8.18 (7.1) 2.97 (2.96)
Muscular weakness 194 (2.44) 1198.1 6.25 6.38 (5.53) 2.62 (2.61)
Infection 193 (2.43) 1012.74 5.31 5.42 (4.7) 2.39 (2.38)
Sinusitis 178 (2.24) 1210.1 6.88 7.01 (6.04) 2.76 (2.75)
Oral herpes 176 (2.21) 7618.17 44.35 45.34 (38.99) 5.28 (5.27)
URT infection 173 (2.18) 2331.9 13.66 13.94 (11.98) 3.71 (3.7)
Hypoesthesia 173 (2.18) 670.42 3.93 3.99 (3.43) 1.96 (1.95)
Pain in extremity 161 (2.03) 317.73 2.01 2.03 (1.74) 1 (0.99)
Bronchitis 150 (1.89) 997.13 6.73 6.84 (5.82) 2.72 (2.71)
Urticaria 131 (1.65) 258.98 2.01 2.03 (1.71) 1 (0.99)
Sepsis 127 (1.6) 455.38 3.64 3.68 (3.09) 1.85 (1.83)
Erythema 122 (1.53) 280.3 2.33 2.35 (1.97) 1.21 (1.2)
Paresthesia 120 (1.51) 253.42 2.15 2.16 (1.81) 1.09 (1.08)
Back pain 118 (1.48) 206.99 1.79 1.8 (1.5) 0.83 (0.82)
Depression 115 (1.45) 169.61 1.5 1.51 (1.26) 0.58 (0.57)
Flushing 114 (1.43) 389.01 3.46 3.5 (2.91) 1.77 (1.76)
Rhinorrhea 113 (1.42) 883.78 7.93 8.03 (6.67) 2.94 (2.93)
Vomiting 105 (1.32) 73.46 0.72 0.72 (0.59) –0.47 (–0.49)
Tremor 101 (1.27) 180.26 1.82 1.83 (1.5) 0.86 (0.84)
Neutropenia 52 (0.65) 87.29 1.72 1.72 (1.31) 0.77 (0.74)
PRR: proportional reporting ratio; ROR: reported odds ratio; IC: information component; IC025: lower limit of the 95% two-sided
confidence interval for IC; URT: upper respiratory tract.
aSignal detected when any of the following occurred: PRR > 2 and χ2 > 4, ROR
025 > 1, IC025 > 0.
bχ2 (chi-square) with Yates’ correction. Values in bold denote significant signals.

journals.sagepub.com/home/msj 7
Multiple Sclerosis Journal 00(0)

On average, studies have revealed that by week 2 for oral herpes and herpes zoster infections with
after infusion, rituximab achieved a CD19 B-cell ocrelizumab. This is corroborated by ORATORIO
depletion rate in peripheral blood of >95% whereas and OPERA I and II trials which showed a higher
ocrelizumab achieved a slightly higher rate of >99% frequency of herpes virus-related infections with
by week 2 after infusion.10,25 In addition to revealing ocrelizumab.12,13 Neither the results of our study nor
a more potent effect in B-cell depletion, by 6 months prior trials with rituximab in MS have reported oral
20% of patients with MS who received rituximab herpes or herpes zoster as an AE.10,11 In this study, the
started to repopulate as compared to 5% of the patients older age of patients treated with ocrelizumab
who received ocrelizumab.26 In addition to a postu- (48.76 years old as compared to 43.89 years old in
lated more potent effect, rates of immunogenicity those treated with rituximab) could potentially explain
appear to be lower in the newer anti-CD20 mAbs as the increased risk for infections that occurs with age,
evidenced by the lower incidence of human anti- in particular with herpes zoster.31 However, patients
human antibodies compared to that of rituximab.10,27 in the ocrelizumab arm of OPERA I and II trials had
Interestingly, in our study the reported frequency for a younger mean age of 37.1 years old, similar to the
infusion-related reaction was similar for rituximab rituximab studies; therefore, age cannot entirely
and ocrelizumab (4.82% vs 4.76%, respectively), but explain the association. It should be taken into
the proportion of AEs in the immune reactions cate- account that B cells are thought to be critical in the
gory was significantly higher for rituximab. This can immune responses against herpes simplex virus and
be explained because infusion-related reaction is clas- varicella zoster virus.32,33
sified into the injury, poisoning and procedural com-
plications rather than the immune system category in The immune system plays an important role in cancer
MedDRA. The AEs included in the immune system pathogenesis34 and with the development of more
category included hypersensitivity and anaphylactic effective disease modifying therapies, concerns may
shock which could indicate a more severe allergic arise for an increased risk of cancer over time in
reaction to the rituximab infusion.10,26 Another impor- patients with MS.35 There are scarce case reports sug-
tant immune system AE that anti-CD20 mAbs can gesting a possible relationship between increased risk
cause is hypogammaglobulinemia as low IgG levels of melanoma with rituximab.36 However, long-term
have been linked to an increased risk for infections.28 safety studies have not revealed any increased risk of
This is potentially pertinent to our observation of an invasive cancers with rituximab.35 On the contrary,
increased risk of reported infections with these agents. OPERA I and ORATORIO trials revealed a higher
Prior studies with rituximab have shown that the risk frequency of malignancies with ocrelizumab as com-
of developing hypogammaglobulinemia was directly pared to IFNβ-1a and placebo arms, respectively.12,13
related to a cumulative dose and lower baseline IgG Interestingly, the results of this study showed a drug-
levels.28 Remarkably, the rates of patients reaching a adverse event association between malignant mela-
IgG levels below lower limit of normal varied from noma and breast cancer with rituximab, but no signal
9% to 38% with rituximab28,29 as compared to 5.7% was observed with ocrelizumab. This may be partially
with ocrelizumab.30 Interestingly, our study did not explained due to over-reporting, although these find-
reveal any signals between rituximab or ocrelizumab ings warrant further evaluation in the context of can-
with the AE “hypogammaglobulinemia.” Very likely, cer epidemiology.
these data are not captured in current clinical practice
settings.28 Neutropenia is a less common but relevant AE that has
been reported with both rituximab and ocrelizumab.37,38
Prior clinical trials have demonstrated a favorable The classic presentation is neutropenia with a delayed
safety profile of rituximab and ocrelizumab, with infu- onset, usually >4 weeks since the last infusion.38 This
sion-related reaction and infections being the most is not an intuitive AE, as neutrophils lack the CD20
common AEs.10–13 It is important to note that in prior receptor.37 In order to explain neutropenia induced by
trials, the frequency of infections was significantly anti-CD20 mAbs, several theories such as immune-
higher in ocrelizumab as compared to placebo mediated mechanisms, silent infection, or neutrophil
whereas there was not a significant difference in the apoptosis triggered by the FAS/FAS ligand pathway
frequency of infections with rituximab as compared have been postulated.39 Our study found a signal for
to placebo.10,11,13 This finding is supported by the neutropenia as a drug–AE association with rituximab
results of our real-world analysis which showed a sig- and ocrelizumab, although the signal was not as
nificantly higher frequency of infections reported with strong with ocrelizumab as compared to rituximab
ocrelizumab as compared to rituximab. In addition, (PRR of 6.75 for rituximab vs 1.72 for ocrelizumab).
this study identified a signal (drug-AE association) Our observations are consistent with reported rates of

8 journals.sagepub.com/home/msj
 NG Caldito, A Shirani et al.

late onset neutropenia with rituximab (4.5%–6.5%) or drug–AE association with ocrelizumab. Although
and ocrelizumab (1%).13,40 speculative, a potentially more robust B-cell deple-
tion by ocrelizumab leading to more profound
Strengths and limitations of this study mainly include immunosuppression could explain these findings.
those associated with disproportionality analysis Additional pharmacovigilance studies are needed
studies.41 Some of them are intrinsic to the FAERS to explore and further characterize these findings.
database.17 In general, AEs are underreported in spon- Furthermore, these observations suggest that the
taneous reporting systems,42 and not all the AEs asso- AE profile of other second-generation anti-CD20
ciated with a specific drug are reported to the FDA. If mAb may also differ from those of rituximab and
a specific AE was not reported in a clinical trials pro- ocrelizumab.
gram, prescribers may initially not attribute it to that
agent. Furthermore, FAERS can contain duplicate Declaration of Conflicting Interests
reports for the same event and in certain occasions, The author(s) declared the following potential conflicts
misspelling and miswording may occur. Thus, given of interest with respect to the research, authorship, and/
the lack of a denominator, the incidence of a specific or publication of this article: Dr Afsaneh Shirani is cur-
AE cannot be accurately estimated. For the same rea- rently funded through a Physician Scientist Training
son, FAERS is not universally suitable to prove cau- Program Award by the University of Nebraska Medical
sality between a reported AE and a pharmacological Center and a Clinician Scientist Development Award
agent. Furthermore, limitations of a disproportional- by the National Multiple Sclerosis Society (USA). Dr
ity analysis compared to formal epidemiological stud- Olaf Stuve serves on the editorial boards of Therapeutic
ies include the limited detailed demographic data, Advances in Neurological Disorders. He has served on
duration of exposure, risk factors, and comorbidities. data monitoring committees for Genentech-Roche,
Therefore, these measures should be considered as Pfizer, and TG Therapeutics without monetary com-
exploratory to detect signals of unexpected or pensation. He has advised EMD Serono, Celgene,
unknown AEs that would warrant further investiga- Genentech, TG Therapeutics, and Genzyme. He cur-
tions. Another limitation of this study is that most of rently receives grant support from Sanofi Genzyme
the AEs in the FAERS database were reported in the and EMD Serono.
United States; therefore, generalization should be
done with caution. Furthermore, it is important to Funding
note that there could also be differences in AEs report- The author(s) received no financial support for the
ing between rituximab and ocrelizumab. For example, research, authorship, and/or publication of this article.
it is possible that AEs associated with rituximab were
not reported to the FDA, as it was not an FDA- Informed Consent
approved medication. In addition, an overreporting of Data were obtained from a publicly available de-iden-
AEs with ocrelizumab is conceivable, given the atten- tified and HIPAA complaint database; therefore, no
tion it received after its FDA approval. Another rele- informed consent was required.
vant factor to consider is the different dosages of
rituximab that have been administered to patients ORCID iDs
with MS, whereas ocrelizumab is given at the same Natalia Gonzalez Caldito https://orcid.
approved dosage. Consequently, it is not possible to org/0000-0002-5265-3404
accurately ascertain whether the reported AEs with Afsaneh Shirani https://orcid.org/0000-0002-8866
rituximab are dosage-dependent. Another aspect to -6426
consider is that insurance coverage or socioeconomi- Amber Salter https://orcid.org/0000-0002-1088
cal status may have influenced the differences in AEs -110X
between both drugs. The smaller number of rituximab Olaf Stuve https://orcid.org/0000-0002-0469-6872
reports as compared to ocrelizumab may limit gener-
alization. Ultimately, each report can have more than Supplemental Material
one AE, and the magnitude of effect of an AE of inter- Supplemental material for this article is available
est is difficult to precisely determine. online.

This is the first study comparing the reported AE pro-

files of rituximab and ocrelizumab using the FAERS
database. The frequency of reported infections was References
almost twice as high with ocrelizumab as compared 1. Compston A and Coles A. Multiple sclerosis. Lancet
to rituximab. Oral herpes revealed the highest signal 2008; 372: 1502–1517.

journals.sagepub.com/home/msj 9
Multiple Sclerosis Journal 00(0)

2. Ghasemi N, Razavi S and Nikzad E. Multiple 17. Sakaeda T, Tamon A, Kadoyama K, et al. Data
Sclerosis: Pathogenesis, symptoms, diagnoses and mining of the public version of the FDA adverse
cell-based therapy. Cell J 2017; 19(1): 1–10. event reporting system. Int J Med Sci 2013; 10:
796–803.
3. Greenfield AL and Hauser SL. B Cell therapy for
multiple sclerosis: Entering an era. Ann Neurol 2018; 18. Yoshimura K, Kadoyama K, Sakaeda T, et al. A
83(1): 13–26. Survey of the FAERS database concerning the
adverse event profiles of α1-adrenoreceptor blockers
4. Loma I and Heyman R. Multiple sclerosis:
for lower urinary tract symptoms. Int J Med Sci 2013;
Pathogenesis and treatment. Curr Neuropharmacol
10(7): 864–869.
2011; 9: 409–416.
19. Bate A, Lindquist M, Edwards IR, et al. A Bayesian
5. Stashenko P, Nadler LM, Hardy R, et al. neural network method for adverse drug reaction
Characterization of a human B lymphocyte-specific signal generation. Eur J Clin Pharmacol 1998; 54(4):
antigen. J Immunol 1980; 125: 1678–1685. 315–321.
6. Holley JE, Bremer E, Kendall AC, et al. 20. van Puijenbroek EP, Bate A, Leufkens HG, et al. A
CD20+inflammatory T-cells are present in blood comparison of measures of disproportionality for
and brain of multiple sclerosis patients and can be signal detection in spontaneous reporting systems for
selectively targeted for apoptotic elimination. Mult adverse drug reactions. Pharmacoepidemiol Drug Saf
Scler Relat Disord 2014; 3(5): 650–658. 2002; 11(1): 3–10.
7. Palanichamy A, Jahn S, Nickles D, et al. Rituximab 21. Szarfman A, Machado SG and O’Neill RT. Use
efficiently depletes increased CD20-expressing T of screening algorithms and computer systems to
cells in multiple sclerosis patients. J Immunol 2014; efficiently signal higher-than-expected combinations
193: 580–586. of drugs and events in the US FDA’s spontaneous
8. Beers SA, Chan CHT, French RR, et al. CD20 as reports database. Drug Saf 2002; 25(6): 381–392.
a target for therapeutic type I and II monoclonal 22. Du FH, Mills EA and Mao-Draayer Y. Next-
antibodies. Semin Hematol 2010; 47(2): 107–114. generation anti-CD20 monoclonal antibodies
9. Pieper K, Grimbacher B and Eibel H. B-cell biology in autoimmune disease treatment. Auto Immun
and development. J Allergy Clin Immunol 2013; 131: Highlights 2017; 8: 12.
959–971. 23. Klein C, Lammens A, Schäfer W, et al. Epitope
10. Hauser SL, Waubant E, Arnold DL, et al. B-cell interactions of monoclonal antibodies targeting CD20
depletion with rituximab in relapsing-remitting and their relationship to functional properties. Mabs
multiple sclerosis. N Engl J Med 2008; 358: 2013; 5(1): 22–33.
676–688. 24. Mössner E, Brünker P, Moser S, et al. Increasing
11. Hawker K, O’Connor P, Freedman MS, et al. the efficacy of CD20 antibody therapy through the
Rituximab in patients with primary progressive engineering of a new type II anti-CD20 antibody
multiple sclerosis: Results of a randomized double- with enhanced direct and immune effector cell-
blind placebo-controlled multicenter trial. Ann Neurol mediated B-cell cytotoxicity. Blood 2010; 115:
2009; 66(4): 460–471. 4393–4402.

12. Hauser SL, Bar-Or A, Comi G, et al. Ocrelizumab 25. Kappos L, Li D, Calabresi PA, et al. Ocrelizumab
versus interferon beta-1a in relapsing multiple in relapsing-remitting multiple sclerosis: A phase 2,
sclerosis. N Engl J Med 2017; 376: 221–234. randomised, placebo-controlled, multicentre trial.
Lancet 2011; 378: 1779–1787.
13. Montalban X, Hauser SL, Kappos L, et al.
Ocrelizumab versus placebo in primary progressive 26. Sorensen PS, Lisby S, Grove R, et al. Safety and
multiple sclerosis. N Engl J Med 2017; 376: efficacy of ofatumumab in relapsing-remitting
209–220. multiple sclerosis: A phase 2 study. Neurology 2014;
82: 573–581.
14. Scotti B, Disanto G, Sacco R, et al. Effectiveness
and safety of rituximab in multiple sclerosis: An 27. Baker D, Pryce G, James LK, et al. The ocrelizumab
observational study from Southern Switzerland. PLoS phase II extension trial suggests the potential to
improve the risk: benefit balance in multiple sclerosis.
ONE 2018; 13(5): e0197415.
medrxiv 2020; 2020.01.09.20016774.
15. FDA and Adverse Event Reporting System (FAERS)
28. Hallberg S, Boremalm M, Evertsson B, et al. Risk
Public Dashboard. FDA, www.fda.gov (accessed 2
of hypogammaglobulinemia in long-term treatment
April 2020).
with rituximab. ECTRIMS 2019, abstract 65, https://
16. MedDRA . Version 22.1, www.meddra.org (accessed onlinelibrary.ectrims-congress.eu/ (accessed 13 July
26 March 2020). 2020).

10 journals.sagepub.com/home/msj
 NG Caldito, A Shirani et al.

29. Ancau M, Berthele A and Hemmer B. CD20 36. Velter C, Pagès C, Schneider P, et al. Four cases of
monoclonal antibodies for the treatment of multiple rituximab-associated melanoma. Melanoma Res 2014;
sclerosis: Up-to-date. Expert Opin Biol Ther 2019; 24(4): 401–403.
19(8): 829–843.
37. Monaco WE, Jones JD and Rigby WFC. Rituximab
30. Derfuss T. Serum immunoglobulin levels and risk associated late-onset neutropenia-a rheumatology case
of serious infections in the pivotal Phase III trials of series and review of the literature. Clin Rheumatol
ocrelizumab in multiple sclerosis and their open- 2016; 35: 2457–2462.
label extensions. ECTRIMS 2019, abstract 65. https://
38. Zanetta C, Robotti M, Nozzolillo A, et al. Late onset
onlinelibrary.ectrims-congress.eu/ (accessed 13 July
absolute neutropenia associated with ocrelizumab
2020).
treatment in multiple sclerosis: A case report and
31. Varghese L, Standaert B, Olivieri A, et al. The review of the literature. J Neurol Sci 2020; 409:
temporal impact of aging on the burden of herpes 116603.
zoster. BMC Geriatr 2017; 17: 30.
39. Futosi K, Fodor S and Mócsai A. Neutrophil cell
32. Arvin AM. Humoral and cellular immunity to surface receptors and their intracellular signal
varicella-zoster virus: An overview. J Infect Dis transduction pathways. Int Immunopharmacol 2013;
2008; 197: S58–S60. 17: 638–650.
33. Deshpande SP, Kumaraguru U and Rouse BT. Dual 40. Cohen BA. Late-onset neutropenia following
role of B cells in mediating innate and acquired ocrelizumab therapy for multiple sclerosis. Neurology
immunity to herpes simplex virus infections. Cell 2019; 92: 435–436.
Immunol 2000; 202: 79–87.
41. Montastruc J-L, Sommet A, Bagheri H, et al. Benefits
34. Pandya PH, Murray ME, Pollok KE, et al. The and strengths of the disproportionality analysis
immune system in cancer pathogenesis: Potential for identification of adverse drug reactions in a
therapeutic approaches. J Immunol Res 2016; 2016: pharmacovigilance database. Br J Clin Pharmacol
4273943. 2011; 72(6): 905–908.
Visit SAGE journals online
35. Alping P, Askling J, Burman J, et al. Cancer risk for 42. Hazell L and Shakir SAW. Under-reporting of journals.sagepub.com/
home/msj
fingolimod, natalizumab, and rituximab in multiple adverse drug reactions: A systematic review. Drug
sclerosis patients. Ann Neurol 2020; 87(5): 688–699. Saf 2006; 29: 385–396. SAGE journals

journals.sagepub.com/home/msj 11