The
new england journal
of
medicine
n engl j med 376;3 nejm.org January 19, 2017
221
The authors’ full names, academic de-
grees, and affiliations are listed in the Ap-
pendix. Address reprint requests to Dr.
Hauser at the Weill Institute for Neuro-
sciences, Department of Neurology, Uni-
versity of California, San Francisco, 675
Nelson Rising Ln., San Francisco, CA
94158, or at stephen . hauser@ ucsf . edu;
or to Dr. Kappos at University Hospital
Basel, University of Basel, Petersgraben
4, CH-4301 Basel, Switzerland, or at
ludwig . kappos@ usb . ch.
* A complete list of investigators in the
OPERA I and OPERA II trials is provided
in the Supplementary Appendix, available
at NEJM.org.
Drs. Hauser and Kappos contributed
equally to this article.
This article was published on December 21,
2016, at NEJM.org.
N Engl J Med 2017;376:221-34.
DOI: 10.1056/NEJMoa1601277
Copyright © 2016 Massachusetts Medical Society.
BACKGROUND
B cells influence the pathogenesis of multiple sclerosis. Ocrelizumab is a human-
ized monoclonal antibody that selectively depletes CD20+ B cells.
METHODS
In two identical phase 3 trials, we randomly assigned 821 and 835 patients with
relapsing multiple sclerosis to receive intravenous ocrelizumab at a dose of 600 mg
every 24 weeks or subcutaneous interferon beta-1a at a dose of 44 μg three times
weekly for 96 weeks. The primary end point was the annualized relapse rate.
RESULTS
The annualized relapse rate was lower with ocrelizumab than with interferon beta-1a
in trial 1 (0.16 vs. 0.29; 46% lower rate with ocrelizumab; P<0.001) and in trial 2 (0.16
vs. 0.29; 47% lower rate; P<0.001). In prespecified pooled analyses, the percentage of
patients with disability progression confirmed at 12 weeks was significantly lower
with ocrelizumab than with interferon beta-1a (9.1% vs. 13.6%; hazard ratio, 0.60;
95% confidence interval [CI], 0.45 to 0.81; P<0.001), as was the percentage of patients
with disability progression confirmed at 24 weeks (6.9% vs. 10.5%; hazard ratio, 0.60;
95% CI, 0.43 to 0.84; P = 0.003). The mean number of gadolinium-enhancing le-
sions per T
1
-weighted magnetic resonance scan was 0.02 with ocrelizumab versus
0.29 with interferon beta-1a in trial 1 (94% lower number of lesions with ocreliz-
umab, P<0.001) and 0.02 versus 0.42 in trial 2 (95% lower number of lesions,
P<0.001). The change in the Multiple Sclerosis Functional Composite score (a com-
posite measure of walking speed, upper-limb movements, and cognition; for this
z score, negative values indicate worsening and positive values indicate improve-
ment) significantly favored ocrelizumab over interferon beta-1a in trial 2 (0.28 vs.
0.17, P = 0.004) but not in trial 1 (0.21 vs. 0.17, P = 0.33). Infusion-related reactions
occurred in 34.3% of the patients treated with ocrelizumab. Serious infection oc-
curred in 1.3% of the patients treated with ocrelizumab and in 2.9% of those
treated with interferon beta-1a. Neoplasms occurred in 0.5% of the patients
treated with ocrelizumab and in 0.2% of those treated with interferon beta-1a.
CONCLUSIONS
Among patients with relapsing multiple sclerosis, ocrelizumab was associated with
lower rates of disease activity and progression than interferon beta-1a over a period
of 96 weeks. Larger and longer studies of the safety of ocrelizumab are required.
(Funded by F. Hoffmann–La Roche; OPERA I and II ClinicalTrials.gov numbers,
NCT01247324 and NCT01412333, respectively.)
ABSTR ACT
Ocrelizumab versus Interferon Beta-1a
in Relapsing Multiple Sclerosis
S.L. Hauser, A. Bar-Or, G. Comi, G. Giovannoni, H.-P. Hartung, B. Hemmer,
F. Lublin, X. Montalban, K.W. Rammohan, K. Selmaj, A. Traboulsee,
J.S. Wolinsky, D.L. Arnold, G. Klingelschmitt, D. Masterman, P. Fontoura,
S. Belachew, P. Chin, N. Mairon, H. Garren, and L. Kappos,
for the OPERA I and OPERA II Clinical Investigators*
Original Article
The New England Journal of Medicine
Downloaded from nejm.org on March 9, 2021. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
n engl j med 376;3 nejm.org January 19, 2017
222
The
new england journal
of
medicine
D
espite the availability of several
disease-modifying treatments for relaps-
ing forms of multiple sclerosis, patients
often continue to have clinical and subclinical
disease activity, and neurologic disability continues
to accrue. Thus, there is a need for more effective
treatments with acceptable safety profiles.
1-3
B cells are thought to influence the underly-
ing pathogenesis of multiple sclerosis by means
of antigen presentation,
4
autoantibody produc-
tion,
5,6
cytokine regulation,
4
and the formation
of ectopic lymphoid aggregates in the meninges,
which possibly contribute to cortical demyelin-
ation and neurodegeneration.
7,8
Ocrelizumab is a
humanized monoclonal antibody that selectively
targets CD20, a cell-surface antigen that is expressed
on pre-B cells, mature B cells, and memory B cells
but not on lymphoid stem cells and plasma cells.
9
Humanized anti-CD20 antibody was designed to
reduce immunogenicity, which was shown in a
phase 2 study.
10
Ocrelizumab binds to the large
extracellular loop of CD20 with high affinity,
selectively depleting CD20-expressing B cells
11,12
while preserving the capacity for B-cell recon-
stitution and preexisting humoral immunity.
13,14
B-cell depletion is achieved by means of several
mechanisms, including antibody-dependent cell-
mediated phagocytosis, antibody-dependent cell-
mediated cytotoxicity, complement-dependent cy-
totoxicity, and induction of apoptosis.
15
On the basis of results from previous phase 2
studies of the chimeric anti-CD20 antibody ritux-
imab
16
and ocrelizumab,
10
we undertook two
phase 3, multicenter, randomized, double-blind,
double-dummy, active-controlled, parallel-group
trials (OPERA I and OPERA II) to investigate the
efficacy and safety of ocrelizumab, as compared
with subcutaneous interferon beta-1a, in patients
with relapsing multiple sclerosis. The two trials
used identical protocols but were conducted in-
dependently at nonoverlapping trial sites (see the
Supplementary Appendix, available with the full
text of this article at NEJM.org).
Methods
Trial Oversight
The sponsor, F. Hoffmann–La Roche, designed
the trials in consultation with members of the
OPERA I and OPERA II steering committee. Data
were collected by the site investigators, queries
were responded to by site personnel, and the data
were analyzed by the sponsor; the aggregated and
individual results of the participants were reviewed
by the sponsor and steering committee. An inde-
pendent data and safety monitoring committee
reviewed ongoing safety data and provided guid-
ance on trial continuation, modification, or ter-
mination (see the Study Oversight section in the
Supplementary Appendix).
All the authors vouch for the accuracy and
completeness of the data and for the fidelity of
the trial to the protocol. A subgroup of authors,
which included academic authors and authors who
are employees of the sponsor, drafted the manu-
script, and all the authors approved the final
version and made the decision to submit the
manuscript for publication. Medical-writing as-
sistance was funded by the sponsor. The trial
was conducted in accordance with the Interna-
tional Conference on Harmonisation guidelines
for Good Clinical Practice
17
and the Declaration
of Helsinki.
18
Patients
Key eligibility criteria included an age of 18 to
55 years; a diagnosis of multiple sclerosis (accord-
ing to the 2010 revised McDonald criteria
19
); an
Expanded Disability Status Scale (EDSS) score of
0 to 5.5 at screening (scores range from 0 to 10.0,
with higher scores indicating a greater degree of
disability
20
); at least two documented clinical re-
lapses within the previous 2 years or one clinical
relapse within the year before screening; magnetic
resonance imaging (MRI) of the brain showing
abnormalities consistent with multiple sclerosis;
and no neurologic worsening for at least 30 days
before both screening and baseline (day 1 trial
visit). The key exclusion criteria were a diagnosis
of primary progressive multiple sclerosis, previ-
ous treatment with any B-cell–targeted therapy or
other immunosuppressive medication as defined
in the protocol (available at NEJM.org; also see
the Additional Methodology Details section in
the Supplementary Appendix), and a disease dura-
tion of more than 10 years in combination with
an EDSS score of 2.0 or less at screening. All the
patients provided written informed consent.
Trial Design
In the OPERA I trial, patients from 141 trial sites
across 32 countries underwent randomization be-
tween August 31, 2011, and February 14, 2013.
In the OPERA II trial, patients from 166 trial sites
The New England Journal of Medicine
Downloaded from nejm.org on March 9, 2021. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
n engl j med 376;3 nejm.org January 19, 2017
223
Ocrelizumab vs. Interferon Beta-1a in Relapsing MS
across 24 countries underwent randomization
between September 20, 2011, and March 28, 2013.
Patients were randomly assigned, in a 1:1 ratio,
to receive ocrelizumab at a dose of 600 mg by
means of intravenous infusion every 24 weeks,
administered as two 300-mg infusions on days
1 and 15 for the first dose and as a single 600-mg
infusion thereafter, or interferon beta-1a at a dose
of 44 μg (Rebif, EMD Serono), administered sub-
cutaneously three times weekly throughout the
96-week treatment period (Fig. S1 in the Supple-
mentary Appendix). Patients in each group re-
ceived a matching subcutaneous or intravenous
placebo, as appropriate. All the patients received
one 100-mg dose of intravenous methylpredniso-
lone before each infusion. Prophylaxis with anal-
gesic or antipyretic agents and an antihistamine
was recommended, but the decision to use these
medications was left up to the infusion center.
Adjustment of the infusion rate and treatment of
symptoms during infusion were permitted in or-
der to manage infusion-related reactions.
Randomization was performed centrally with
the use of an independent interactive Web-response
system. Each trial center had separate treating
and examining investigators, all of whom were
unaware of the treatment assignments through-
out the trial. The examining investigator conducted
the neurologic assessments, including the Mul-
tiple Sclerosis Functional Composite (a compos-
ite quantitative measure, expressed as a z score,
of walking speed, upper-limb coordinated move-
ments, and cognition; for this z score, negative
values indicate worsening and positive values
indicate improvement)
21
and the EDSS. The EDSS
assessment and data collection were captured
with the use of a real-time, electronic data-entry
system in conjunction with an algorithm and
central consistency check and feedback on the
basis of expert review. MRI scans were analyzed
centrally at an MRI reading center by personnel
who were unaware of the treatment assignments.
Details are provided in the protocol, including
the statistical analysis plan, and in Table S9 in
the Supplementary Appendix.
Trial Procedures and End Points
The primary end point was the annualized relapse
rate by 96 weeks, which reflects the number of
relapses meeting the prespecified criteria that were
observed per person-year of follow-up (see the
Supplementary Appendix). There were 10 hierar-
chically ordered secondary end points: the pro-
portion of patients with disability progression
confirmed at 12 weeks in a pooled time-to-event
analysis of both trials through week 96, in which
disability progression was defined as an increase
from the baseline EDSS score of at least 1.0 point
(or 0.5 points if the baseline EDSS score was >5.5)
that was sustained for at least 12 weeks; the total
(cumulative) mean number of gadolinium-enhanc-
ing lesions identified on T
1
-weighted MRI of the
brain at weeks 24, 48, and 96; the total number
of new or newly enlarged hyperintense lesions
on T
2
-weighted MRI of the brain at weeks 24, 48,
and 96; a pooled analysis of the proportion of
patients with disability improvement confirmed
at 12 weeks through week 96, which was defined
as a reduction from the baseline EDSS score of at
least 1.0 point (or 0.5 points if the baseline EDSS
score was >5.5) that was sustained for at least
12 weeks in patients with a baseline EDSS score
of at least 2.0; a pooled time-to-event analysis
of the rate of disability progression confirmed at
24 weeks through week 96; the total number of
new hypointense lesions on T
1
-weighted MRI
of the brain at weeks 24, 48, and 96; the change
in the Multiple Sclerosis Functional Composite
score from baseline to week 96; the percentage
change in brain volume from week 24 to week 96;
the change in the physical-component summary
score of the Medical Outcomes Study 36-Item
Short-Form Health Survey (SF-36, on which scores
range from 0 to 100, and higher scores indicate
better physical health–related quality of life) from
baseline to week 96; and the proportion of pa-
tients with a baseline EDSS score of at least 2.0
who had no evidence of disease activity (defined
as no relapse, no disability progression con-
firmed at 12 weeks or at 24 weeks, no new or
newly enlarged lesions on T
2
-weighted MRI, and
no gadolinium-enhancing lesions on T
1
-weighted
MRI) by week 96. The analysis of percentage
change in brain volume was performed with the
use of SIENA/X software.
22
Additional secondary
end points were the pharmacokinetics, pharmaco-
dynamics, and immunogenicity of ocrelizumab;
and the safety profile of ocrelizumab.
Statistical Analysis
We performed efficacy analyses in the intention-
to-treat population (all the patients who under-
went randomization) or, for the end point of no
evidence of disease activity, in a modified inten-
The New England Journal of Medicine
Downloaded from nejm.org on March 9, 2021. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
n engl j med 376;3 nejm.org January 19, 2017
224
The
new england journal
of
medicine
tion-to-treat population that excluded patients who
were withdrawn from the trial for reasons other
than efficacy failure or death and who had no
evidence of clinical disease activity at the time of
treatment discontinuation in the trial. The an-
nualized relapse rate was analyzed with the use
of a negative binomial model testing for treatment
differences between ocrelizumab and interferon
beta-1a, with adjustment according to geograph-
ic region and baseline EDSS score. A significant
result at a two-sided alpha of 0.05 would show
the superiority of ocrelizumab with regard to a
lower annualized relapse rate than that observed
with interferon beta-1a.
The sample size for each trial was based on
an estimated annualized relapse rate of 0.165 in
the ocrelizumab group and 0.33 in the interfer-
on beta-1a group. Using a two-sided t-test, we
calculated that a sample of 400 patients per
group would provide the trials with 84% statisti-
cal power to maintain a type I error rate of 0.05
and to detect a 50% lower rate with ocrelizumab
than with interferon beta-1a (assuming a with-
drawal rate of approximately 20%).
According to the statistical analysis plans of
the individual trials, 10 secondary efficacy end
points were prespecified to be tested in a hierar-
chical order at a two-sided alpha of 0.05 (see the
Supplementary Appendix). Seven end points of
this hierarchy were to be tested in each individ-
ual trial, and three end points (disability pro-
gression confirmed at 12 weeks and at 24 weeks
and disability improvement confirmed at 12 weeks)
were to be assessed in the pooled data set. From
the first P value that was above 0.05, all subse-
quent P values in the predetermined hierarchy
were considered to be nonconfirmatory (i.e., de-
scriptive only). (See the Statistical Analysis section
in the Supplementary Appendix.)
All patients who received any study treatment
were included in the safety population. All data
collected during the double-blind, double-dummy
treatment period and the safety follow-up were
included in the main safety analyses. Data from
patients who entered the safety follow-up earlier
than week 96 were included in this analysis
from the time that they entered the safety fol-
low-up until week 96. Safety outcomes are re-
ported for the individual trials with the exception
of herpesvirus infections and neoplasms, for
which pooled data are presented because of low
incidences.
Results
Patients
Overall, 1656 patients underwent randomization
(intention-to-treat population), with 821 patients
in the OPERA I trial and 835 in the OPERA II
trial. The demographic and disease characteristics
at baseline were similar in the assigned groups in
the two trials (Table 1). In the OPERA I trial, 366
of 410 patients (89.3%) in the ocrelizumab group
and 340 of 411 (82.7%) in the interferon beta-1a
group completed the 96-week treatment; in the
OPERA II trial, 360 of 417 patients (86.3%) and
320 of 418 (76.6%), respectively, completed the
96-week treatment (Fig. S2 in the Supplementary
Appendix). There was no interaction between
treatment group and trial, which allowed the
pooling of data for the prespecified planned hi-
erarchical analysis (Table S9 in the Supplemen-
tary Appendix). In the pooled analysis, which
included 827 patients treated with ocrelizumab
and 829 treated with interferon beta-1a, all the
primary and secondary end points significantly
favored the ocrelizumab group over the interferon
bet a-1a g roup.
Efficacy
Relapses
Clinical, MRI, and patient-reported outcomes are
summarized in Table 2. The primary end point,
the annualized relapse rate at 96 weeks, in the
OPERA I trial was 0.16 in the ocrelizumab group,
as compared with 0.29 in the interferon beta-1a
group (difference, 0.14 annualized relapses [dif-
ferences are based on unrounded data]). In the
OPERA II trial, the annualized relapse rate was
0.16 in the ocrelizumab group, as compared with
0.29 in the interferon beta-1a group (difference,
0.14 annualized relapses) (Table 2). These findings
indicate a 46% lower annualized relapse rate with
ocrelizumab in the OPERA I trial and a 47% lower
rate with ocrelizumab in the OPERA II trial
(P<0.001 for both comparisons).
Disability
In the prespecified pooled analysis, the percent-
age of patients with disability progression con-
firmed at 12 weeks was 9.1% in the ocrelizumab
group, as compared with 13.6% in the interferon
beta-1a group (40% lower risk with ocrelizumab;
hazard ratio, 0.60; 95% confidence interval [CI],
0.45 to 0.81; P<0.001) (Fig. 1A). Over the 96-week
The New England Journal of Medicine
Downloaded from nejm.org on March 9, 2021. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.
n engl j med 376;3 nejm.org January 19, 2017
225
Ocrelizumab vs. Interferon Beta-1a in Relapsing MS
Characteristic OPERA I Trial OPERA II Trial
Ocrelizumab
(N = 410)
Interferon Beta-1a
(N = 411)
Ocrelizumab
(N = 417)
Interferon Beta-1a
(N = 418)
Age — yr 37.1±9.3 36.9±9.3 37.2±9.1 37.4±9.0
Female sex — no. (%) 270 (65.9) 272 (66.2) 271 (65.0) 280 (67.0)
Geographic region — no. (%)
United States 105 (25.6) 105 (25.5) 112 (26.9) 114 (27.3)
Rest of the world 305 (74.4) 306 (74.5) 305 (73.1) 304 (72.7)
Time since symptom onset — yr 6.74±6.37 6.25±5.98 6.72±6.10 6.68±6.13
Time since diagnosis — yr 3.82±4.80 3.71±4.63 4.15±4.95 4.13±5.07
No. of relapses in previous 12 mo 1.31±0.65 1.33±0.64 1.32±0.69 1.34±0.73
No previous disease-modifying therapy —
no./total no. (%)†
301/408 (73.8) 292/409 (71.4) 304/417 (72.9) 314/417 (75.3)
Previous disease-modifying therapy —
no./total no. (%)‡
107/408 (26.2) 117/409 (28.6) 113/417 (27.1) 103/417 (24.7)
Interferon 81/408 (19.9) 86/409 (21.0) 80/417 (19.2) 75/417 (18.0)
Glatiramer acetate 38/408 (9.3) 37/409 (9.0) 39/417 (9.4) 44/417 (10.6)
Natalizumab 0/408 1/409 (0.2) 1/417 (0.2) 0/417
Fingolimod 1/408 (0.2) 0/409 4/417 (1.0) 0/417
Dimethyl fumarate 1/408 (0.2) 0/409 0/417 0/417
Other 2/408 (0.5) 3/409 (0.7) 1/417 (0.2) 1/417 (0.2)
Mean EDSS score§ 2.86±1.24 2.75±1.29 2.78±1.30 2.84±1.38
No. of gadolinium-enhancing lesions on T
1
-weighted MRI
— no./total no. (%)
0 233/405 (57.5) 252/407 (61.9) 252/413 (61.0) 243/415 (58.6)
1 64/405 (15.8) 52/407 (12.8) 58/413 (14.0) 62/415 (14.9)
2 30/405 (7.4) 30/407 (7.4) 33/413 (8.0) 38/415 (9.2)
3 20/405 (4.9) 16/407 (3.9) 15/413 (3.6) 14/415 (3.4)
≥4 58/405 (14.3) 57/407 (14.0) 55/413 (13.3) 58/415 (14.0)
No. of lesions on T
2
-weighted MRI 51.04±39.00 51.06±39.90 49.26±38.59 51.01±35.69
Volume of lesions on T
2
-weighted MRI — cm
3
10.84±13.90 9.74±11.28 10.73±14.28 10.61±12.30
Normalized brain volume — cm
3
1500.93±84.10 1499.18±87.68 1503.90±92.63 1501.12±90.98
* Plus–minus values are means ±SD. The intention-to-treat population included all the patients who underwent randomization. There were no
significant differences in the baseline characteristics between groups in each trial and between the two trials. A full listing of countries in-
volved in the trials is provided in the Supplementary Appendix. Data on the number of relapses within the previous 12 months were missing
for 1 patient in the interferon beta-1a group in the OPERA I trial and for 1 patient in each group in the OPERA II trial. Data on the number
and volume of lesions on T
2
-weighted MRI were missing for 2 patients in the ocrelizumab group and for 3 in the interferon beta-1a group in
the OPERA I trial and for 3 in the ocrelizumab group and 2 in the interferon beta-1a group in the OPERA II trial. Data on the normalized
brain volume were missing for 4 patients in the ocrelizumab group and for 7 in the interferon beta-1a group in the OPERA I trial and for 3 in
the ocrelizumab group and 4 in the interferon beta-1a group in the OPERA II trial.
† Data include patients who were untreated with any disease-modifying therapy in the 2 years before screening. The inclusion criteria did not
select for untreated patients.
‡ Data on previous treatment were collected only for the 2 years before screening. Patients could be counted in several categories. Treatment
with cyclophosphamide, azathioprine, mycophenolate mofetil, cyclosporine, or methotrexate within 2 years before screening was an exclu-
sion criterion. Patients treated with natalizumab were eligible for the trial only if the duration of treatment with natalizumab was less than
1 year. Other medications were intravenous immune globulin, mycophenolate mofetil, and azathioprine (protocol deviation if ≤24 months
before screening).
§ Scores on the Expanded Disability Status Scale (EDSS) range from 0 to 10.0, with higher scores indicating worse disability.
20
Data were
missing for one patient in the interferon beta-1a group in the OPERA I trial.
Table 1. Demographic and Disease Characteristics of the Patients at Baseline (Intention-to-Treat Population).*
The New England Journal of Medicine
Downloaded from nejm.org on March 9, 2021. For personal use only. No other uses without permission.
Copyright © 2017 Massachusetts Medical Society. All rights reserved.