Articles

Adalimumab for prevention of uveitic flare in patients with

inactive non-infectious uveitis controlled by corticosteroids
(VISUAL II): a multicentre, double-masked, randomised,
placebo-controlled phase 3 trial
Quan Dong Nguyen, Pauline T Merrill, Glenn J Jaffe, Andrew D Dick, Shree Kumar Kurup, John Sheppard, Ariel Schlaen, Carlos Pavesio, Luca Cimino,
Joachim Van Calster, Anne A Camez, Nisha V Kwatra, Alexandra P Song, Martina Kron, Samir Tari, Antoine P Brézin

Summary
Background Non-infectious uveitis is a potentially sight-threatening ocular disorder caused by chronic inflammation Published Online
and its complications. Therapeutic success is limited by systemic adverse effects associated with long-term August 16, 2016
http://dx.doi.org/10.1016/
corticosteroid and immunomodulator use if topical medication is not sufficient to control the inflammation. We S0140-6736(16)31339-3
aimed to assess the efficacy and safety of adalimumab in patients with inactive, non-infectious uveitis controlled by
See Online/Comment
systemic corticosteroids. http://dx.doi.org/10.1016/
S0140-6736(16)31327-7
Methods We did this multicentre, double-masked, randomised, placebo-controlled phase 3 trial at 62 study sites in Ocular Imaging Research and
21 countries in the USA, Canada, Europe, Israel, Australia, and Latin America. Patients (aged ≤18 years) with inactive, Reading Center (OIRRC),
Omaha, NE, USA
non-infectious intermediate, posterior, or panuveitic uveitis controlled by 10–35 mg/day of prednisone were randomly
(Prof Q D Nguyen MD); Rush
assigned (1:1), via an interactive voice and web response system with a block size of four, to receive either subcutaneous University Medical Center,
adalimumab (loading dose 80 mg; biweekly dose 40 mg) or placebo, with a mandatory prednisone taper from week 2. Chicago, IL, USA (P T Merrill MD);
Randomisation was stratified by baseline immunosuppressant treatment. Sponsor personnel with direct oversight of Duke University, Durham, NC,
USA (Prof G J Jaffe MD);
the conduct and management of the study, investigators, study site personnel, and patients were masked to treatment
University of Bristol, Bristol Eye
allocation. The primary efficacy endpoint was time to treatment failure, a multicomponent endpoint encompassing Hospital, Bristol, UK
new active inflammatory chorioretinal or inflammatory retinal vascular lesions, anterior chamber cell grade, vitreous (Prof A D Dick MD); National
haze grade, and visual acuity. Analysis was done in the intention-to-treat population. This trial is registered with Institute for Health Research
(NIHR) Biomedical Research
ClinicalTrials.gov, number NCT01124838.
Centre (Prof A D Dick),
Moorfields Eye Hospital and
Findings Between Aug 10, 2010, and May 14, 2015, we randomly assigned 229 patients to receive placebo (n=114) or University College London,
adalimumab (n=115); 226 patients comprised the intention-to-treat population. Median follow-up time was 155 days Institute of Ophthalmology,
London, UK (C Pavesio MD);
(IQR 77–357) in the placebo group and 245 days (119–564) in the adalimumab group. Treatment failure occurred in
Wake Forest Baptist Medical
61 (55%) of 111 patients in the placebo group compared with 45 (39%) of 115 patients in the adalimumab group. Time Center, Winston-Salem, NC,
to treatment failure was significantly improved in the adalimumab group compared with the placebo group (median USA (Prof S K Kurup MD); Lions
not estimated [>18 months] vs 8·3 months; hazard ratio 0·57, 95% CI 0·39–0·84; p=0·004). The 40th percentile for Medical Eye Bank of Eastern
Virginia, Eastern Virginia
time to treatment failure was 4·8 months in the placebo group and 10·2 months in the adalimumab group. No Medical School and Virginia Eye
patients in either group had opportunistic infections (excluding oral candidiasis and tuberculosis). No malignancies Consultants, Norfolk, VA, USA
were reported in the placebo group whereas one (1%) patient in the adalimumab group reported non-serious (J Sheppard MD); Austral
squamous cell carcinoma. The most common adverse events were arthralgia (12 [11%] patients in the placebo group University, Buenos Aires,
Argentina (A Schlaen MD);
and 27 [23%] patients in the adalimumab group), nasopharyngitis (16 [17%] and eight [16%] patients, respectively), Arcispedale Santa Maria Nuova
and headache (17 [15%] patients in each group). IRCCS, Reggio Emilia RE, Italy
(L Cimino MD); University
Interpretation Adalimumab significantly lowered the risk of uveitic flare or loss of visual acuity upon corticosteroid Hospitals Leuven, Leuven,
Belgium (J Van Calster MD);
withdrawal in patients with inactive, non-infectious intermediate, posterior, or panuveitic uveitis controlled by AbbVie Deutschland,
systemic corticosteroids. No new safety signals were observed and the rate of adverse events was similar between Ludwigshafen, Germany
groups. These findings suggest that adalimumab is well tolerated and could be an effective treatment option in this (A A Camez MD,
patient population. An open-label extension study (NCT01148225) is ongoing to provide long-term safety data for Prof M Kron PhD); AbbVie,
North Chicago, IL, USA
adalimumab in patients with non-infectious uveitis. (N V Kwatra PhD, A P Song MD,
S Tari MD); and Université Paris
Funding AbbVie. Descartes, Hôpital Cochin, Paris,
France (Prof A P Brézin MD)

Introduction restrict their long-term use in the treatment of Correspondence to:

Dr Quan Dong Nguyen,
Uveitis and its associated complications account for intermediate, posterior, or panuveitic forms.4–6 Guidance Ocular Imaging Research and
roughly 10–15% of preventable blindness in developed from the Standardization of Uveitis Nomenclature Reading Center (OIRRC), Omaha,
countries.1–3 Corticosteroids are the mainstay of uveitis (SUN) working group supports the use of systemic NE 68198, USA
treatment, but ocular and systemic adverse effects corticosteroid-sparing drugs in patients on chronic [email protected]

www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3 1

 Articles

Research in context
Evidence before this study The French Uveitis Network did a multicentre study of
We searched PubMed for articles published up to 160 patients with refractory uveitis treated with anti-TNFα
March 20, 2016, in any language, for drugs or agents that have drugs (infliximab and adalimumab); the overall response rate
been used for the treatment of non-infectious uveitis. was 93% at 12 months. However, most of these publications are
Our search terms were “non-infectious uveitis”, “anti-TNF”, case reports, case series, or open-label studies. An adequate, well
“immunosuppression”, and “biologics”. We identified various controlled study of the efficacy and safety of anti-TNF therapy is
publications on the use of anti-tumour necrosis factor (TNF) absent in the present literature. Previous clinical trials that were
agents in the treatment of various types of anterior, initiated to evaluate therapeutic potential for inactive,
intermediate, posterior, or panuveitic uveitis. Several non-infectious uveitis have either not achieved their primary
publications reported the effectiveness of anti-TNF drugs endpoint or were prematurely terminated for unknown reasons.
(infliximab and adalimumab) in the treatment of uveitis. Some
Added value of this study
of the diseases for which adalimumab is currently indicated,
VISUAL II is a multicentre, double-masked, randomised,
such as juvenile idiopathic arthritis, ankylosing spondylitis, and
placebo-controlled phase 3 trial assessing the efficacy and
psoriatic arthritis can present with uveitis. There have been
safety of adalimumab in patients with inactive non-infectious
reports of efficacy of adalimumab in paediatric patients with
intermediate, posterior, or panuveitis controlled by
juvenile idiopathic arthritis-associated or idiopathic uveitis.
corticosteroids. We did this study at 62 sites in 21 countries,
A retrospective study in patients with refractory chronic uveitis
representative of the global diversity of the study population.
showed that adalimumab effectively controlled inflammation in
To our knowledge, this is the first study to have achieved its
35% of patients refractory to previous treatment with infliximab
prespecified primary endpoint (time to treatment failure) and
or etanercept. In a prospective open-label pilot study of
show promise in the treatment of inactive non-infectious
19 patients with various uveitic diagnoses, adalimumab
uveitis in patients dependent on chronic oral corticosteroids
significantly reduced inflammation in 63% of patients with
(≥10 mg/day) to maintain disease inactivity.
complete resolution of cystoid macular oedema in 55% of eyes
after 1 year. A multicentre study of 131 patients with a mean age Implications of all the available evidence
of 27 years also showed that adalimumab significantly improved Adalimumab significantly lowered the risk of uveitic flare or
anterior chamber and vitreous inflammation with the ability to visual acuity loss in patients with steroid-dependent inactive,
taper corticosteroids. In an open-label study of infliximab, 77% non-infectious intermediate, posterior, or panuveitic uveitis.
of patients with refractory autoimmune uveitis achieved clinical No new safety signals were identified and the safety profile of
success by week 10. In the open-label uncontrolled RHAPSODY adalimumab was similar to that of other approved indications.
study in patients with ankylosing spondylitis, adalimumab These findings suggest that adalimumab could be well
decreased the rate of acute anterior uveitis flares by 51%. In a tolerated and offered as an effective treatment option for
prospective study, adalimumab reduced anterior chamber and patients with inactive, non-infectious uveitis who are at risk of
vitreous inflammation, improved visual acuity, and reduced the the long-term side-effects of corticosteroids.
corticosteroid burden in patients with refractory uveitis.

corticosteroid treatment with quiescent (inactive) inflammatory conditions, such as rheumatoid arthritis,
disease; the ability to achieve a reduction in corticosteroid psoriasis, ankylosing spondylitis, Crohn’s disease,
dose below a clinically meaningful threshold while ulcerative colitis, hidradenitis suppurativa, and juvenile
maintaining inactive disease is a key determinant of idiopathic arthritis.13 Several prospective studies,
treatment success.6 Few currently approved non- including the VISUAL I clinical trial, have shown the
corticosteroid immunomodulatory drugs for uveitis can efficacy and safety of anti-TNF drugs (infliximab and
provide long-term control of the disease.7,8 Globally, adalimumab) in the treatment of chronic and refractory
there is an unmet need that warrants pursuit of uveitis and in reducing corticosteroid use.14–19
additional effective treatments in steroid-dependent Treatment of uveitis has two major goals: to achieve
patients with non-infectious uveitis who are at risk of quiescence in an eye with active intraocular
long-term corticosteroid side-effects. inflammation (the focus of the VISUAL I trial), and to
Tumour necrosis factor α (TNFα) is a proinflammatory prevent recurrence of intraocular inflammation and
cytokine produced by various cells, including reduce side-effects of long-term corticosteroid use in
macrophages and neutrophils.9–12 Adalimumab (Humira; patients with a history of uveitic flare controlled by oral
AbbVie, North Chicago, IL, USA) is a recombinant corticosteroids (≥10 mg/day). We did the VISUAL II to
human immunoglobulin (IgG1) monoclonal antibody assess the efficacy and safety of adalimumab in
that binds specifically to TNF and neutralises its preventing reactivation of non-infectious intermediate,
biological function.13 The safety and efficacy profile of posterior, and panuveitic uveitis controlled by
adalimumab spans more than 13 years for various corticosteroids.

2 www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3

 Articles

Methods starting at week 1 for the duration of the study. Patients

Study design and participants received 10–35 mg/day of oral prednisone at baseline
We did this multicentre, double-masked, randomised, and, from week 2, all patients underwent a mandatory
placebo-controlled phase 3 trial at 62 study sites in prednisone taper to 0 mg by week 19. The schedule of
21 countries in the USA, Canada, Europe, Israel, study procedures is described in the appendix (p 15).
Australia, and Latin America. Presence or absence of inflammatory chorioretinal or
Eligible patients were individuals aged 18 years and older retinal vascular lesions was determined by dilated
with inactive non-infectious intermediate, posterior, or indirect ophthalmoscopy. Anterior chamber cell counts
panuveitic uveitis. Key inclusion criteria were inactive and cataracts were assessed with slit-lamp biomicroscopy
uveitis inactivity for at least 28 days before the baseline at every study visit; cell counts were graded according to
visit, and use of 10–35 mg/day oral prednisone to maintain SUN criteria whereas cataracts were graded with the
inactive disease. We defined inactive uveitis as eyes without Age-Related Eye Disease Study lens opacity grading
active inflammatory chorioretinal or retinal vascular system.20,22 Vitreous haze was assessed with dilated
lesions, an anterior chamber cell grade of 0·5+ or less indirect ophthalmoscopy and graded with SUN-adapted
(SUN criteria; score range, 0–4+),20 or a vitreous haze grade National Eye Institute criteria.20,21 Macular oedema was
of 0·5+ or less (National Eye Institute criteria adapted by assessed with optical coherence tomography (Stratus
SUN).20,21 To demonstrate corticosteroid dependency, optical coherence tomography [Carl Zeiss Meditec, Jena,
patients were required to have a documented history of Germany], Cirrus high-definition optical coherence
having at least one disease flare within 18 months of the tomography [Carl Zeiss Meditec, Jena, Germany], or
screening visit. Additionally, this flare should have Spectralis [Heidelberg Engineering, Heidelberg,
occurred during or up to a maximum of 28 days after Germany]; appendix pp 3–5).
tapering off the oral corticosteroid therapy. We excluded Clinic visits were scheduled at screening; baseline; and
patients who were receiving more than one weeks 2, 4, and roughly every 4 weeks thereafter. Patients
immunosuppressive drug (not including corticosteroids) were assessed until treatment failure was determined or
within the last 28 days before the baseline visit; the dose of until treatment completion. The maximum duration of
the one concomitant drug had to be stable for at least treatment was 80 weeks or when the 106th treatment
28 days before baseline and within the dose range failure occurred. Beginning at or after week 2 and at
(appendix pp 6–12). Additional exclusion criteria were every subsequent visit thereafter, treatment failure was See Online for appendix
corneal or lens opacity that precluded visualisation of the determined if any of the following criteria were met in at
fundus or that might have required cataract surgery during least one eye: new active, inflammatory chorioretinal or
the study, and isolated anterior or infectious uveitis or any inflammatory retinal vascular lesions (investigator-
condition that would preclude safe participation in the determined by clinical examination or ancillary testing,
study or interfere with study assessments. The appendix such as fluorescein angiography); worsening of best-
provides a complete list of inclusion and exclusion criteria. corrected visual acuity by 15 letters or more; a two-step
The study protocol (appendix) was approved by the increase in anterior chamber cell grade; or a two-step
responsible ethics committees and internal review boards increase in vitreous haze grade relative to baseline.
at each study site and the study was done in compliance
with the Declaration of Helsinki, Good Clinical Practice Outcomes
guidelines, and applicable local regulations. All patients The primary efficacy endpoint was time to treatment
provided written informed consent. failure. We assessed nine ranked secondary endpoints in
prespecified hierarchical order: (1) change in anterior
Randomisation and masking chamber cell grade in each eye; (2) change in vitreous
Patients were randomly assigned (1:1), via an interactive haze grade in each eye; (3) change in best-corrected visual
voice and web response system with a block size of four, acuity (logMAR) in each eye; (4) time to optical coherence
to receive adalimumab or placebo. Randomisation was tomographic evidence of macular oedema in at least one
stratified by baseline immunosuppressant treatment. All eye; (5) percentage change in central retinal thickness
sponsor personnel with direct oversight of the conduct (as defined by centre point thickness) in each eye;
and management of the study (with the exception of (6) change in National Eye Institute Visual Functioning
those providing study treatments), investigators, study Questionnaire-25 (VFQ-25) composite score; (7) change in
site personnel, and patients were masked to treatment VFQ-25 distance vision subscore; (8) change in VFQ-25
allocation throughout the treatment period. near vision subscore; and (9) change in VFQ-25 ocular
pain subscore. With the exception of the fourth ranked
Procedures endpoint, all ranked endpoints were analysed comparing
Adalimumab and placebo were supplied in prefilled baseline with the final or early termination visit.
syringes and administered subcutaneously. Patients in Adverse events were monitored throughout the study
the adalimumab group received an 80 mg baseline and reported from the first dose of study drug until
loading dose followed by 40 mg doses every other week 70 days after the last dose of study drug or until patients

www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3 3

 Articles

were enrolled into a separate extension study. Serious assumptions were done with East5 (version 5.2.0.0).
adverse events were recorded from the time of informed A sample size of about 220 patients was needed to
consent. Adverse events were tabulated with the Medical achieve roughly 96 treatment failure events.
Dictionary for Regulatory Activities (version 17.0) and An independent data monitoring committee set up at
categorised by system organ class and preferred terms. the beginning of the trial independently monitored and
Adalimumab immunogenicity was assessed at multiple assessed data until the end of the study. At each
times throughout the study. committee meeting, the committee undertook a
comprehensive review and assessment of the cumulative
Statistical analysis safety data. The committee met roughly every 6 months,
We assumed an overall treatment failure rate of 30–35% or at a frequency determined by the members, to render
at 6 months, with an expected treatment effect their recommendation for study termination,
corresponding to an absolute difference of 15% between continuation, or amendment. The independent data
treatment failure rates in the adalimumab and placebo monitoring committee analyses were done by an external
groups. A conservative assumption was that treatment statistics vendor (Axio Research, Seattle, WA, USA) to
failures would begin to occur after 2 months because of enable the study sponsor to remain masked to the results
prednisone taper. A pooled dropout rate of 35% over of the study. The committee met eight times and did not
12 months was also assumed. On the basis of these identify any safety issues that would necessitate either a
assumptions, 84 to 107 treatment failures were sufficient temporary hold or an early termination of the study.
for a two-sided significance level of 5% in a log-rank test. Protocol deviations were monitored via assessment of
This calculation assumed 80% power and an average inclusion and exclusion criteria at study entry and
accrual rate of three patients per month in the first throughout the study. We analysed efficacy endpoints in
28 months and 16 patients per month thereafter. A series the intention-to-treat population, which comprised all
of calculations with different sample sizes using the patients randomly assigned to treatment.
event rate, recruitment rate, and dropout rate The primary endpoint of time to treatment failure was
compared between groups with a log-rank test; a
337 patients assessed for eligibility proportional hazards model with treatment as a factor
was fitted to estimate the hazard ratio (HR) and 95% CI.
As additional prespecified exploratory endpoints, time to
108 excluded
88 because of criteria violations
treatment failure due to each component of the primary
1 because of criteria violations endpoint was analysed similarly.
and other reasons Testing of ranked secondary endpoints was done
8 withdrew consent
11 for other reasons in hierarchical order and nominal p values for
between-group differences were provided. Changes in
anterior chamber cell grade, vitreous haze grade, best-
229 randomly assigned corrected visual acuity, and central retinal thickness were
compared between groups by ANOVA, with treatment as
a factor adjusted for clustered observations within a
114 assigned to receive placebo 115 assigned to receive adalimumab
patient—ie, a repeated-measures ANOVA was used to
114 received placebo 115 received adalimumab account for correlation between measurements from
both eyes of a patient. Analysis of central retinal thickness
used the optical coherence tomography machine type as
3 excluded from analysis
1 had incomplete efficacy source data an additional factor. Time to optical coherence
2 had general GCP compliance issues tomographic evidence of macular oedema on or after
week 2 was compared between groups with a log-rank
95 completed the study* 101 completed the study*
test, excluding patients with pre-existing macular
16 discontinued the study† 14 discontinued the study oedema at baseline. Changes in VFQ-25 composite score
7 had adverse events 10 had adverse events and subscores were compared between groups by
3 were lost to follow-up 2 withdrew
3 withdrew 2 for other reasons ANOVA, with treatment as a factor. For analysis of
3 because of lack of efficacy secondary variables, with the exception of time to optical
3 for other reasons
coherence tomographic evidence of macular oedema, we
imputed missing data with last observation carried
111 included in intention-to-treat 115 included in intention-to-treat forward.
analysis analysis
114 included in safety analysis 115 included in safety analysis
We did safety analysis in the safety population, which
comprised patients who received at least one dose of
study drug. Adverse events were presented as events per
Figure 1: Trial profile
GCP=Good Clinical Practice. *Patients who had treatment failure or reached 80 weeks of treatment without 100 patient-years to avoid confounding by between-group
treatment failure. †Some patients had more than one reason for discontinuation. differences in duration of exposure to study treatment.

4 www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3

 Articles

All statistical tests were two sided at a significance level

Placebo group Adalimumab
of 0·05. Statistical analyses were done by the study (n=111) (n=115)
sponsor with SAS version 9.2. This trial is registered with
Sex
ClinicalTrials.gov, number NCT01124838.
Female 72 (65%) 66 (57%)
Male 39 (35%) 49 (43%)
Role of the funding source
Race
The funder of the study contributed to study design, and
White 93 (84%) 96 (83%)
participated in data collection, data analysis, and data
Black or African American 8 (7%) 6 (5%)
interpretation, and preparation and approval of the report
Asian 3 (3%) 3 (3%)
for publication. All authors had access to study data,
Other 5 (5%) 9 (8%)
reviewed and approved the final report, and take full
responsibility for the accuracy of the data and statistical Age (years)

analysis. The corresponding author had full access to all Mean (SD) 42·2 (14·0) 42·9 (12·9)

the data in the study and had final responsibility for the Range 20–79 18–75
decision to submit for publication. Type of uveitis
Intermediate 30 (27%) 17 (15%)
Results Posterior 34 (31%) 39 (34%)
Between Aug 10, 2010, and May 14, 2015, we randomly Panuveitis 46 (41%) 57 (50%)
assigned 229 patients to receive placebo (n=114) or Intermediate and posterior 1 (1%) 2 (2%)
adalimumab (n=115); 226 patients comprised the Diagnosis
intention-to-treat population (figure 1). 54 (24%) patients Idiopathic 40 (36%) 29 (25%)
had important reportable protocol deviations, including Birdshot choroidopathy 15 (14%) 15 (13%)
criteria violations; receipt of excluded concomitant Multifocal choroiditis and panuveitis 2 (2%) 5 (4%)
treatment; receipt of wrong treatment or incorrect dose Vogt-Koyanagi-Harada 25 (23%) 26 (23%)
of adalimumab, placebo, or prednisone; or development Sarcoid 14 (13%) 18 (16%)
of withdrawal criteria without being withdrawn (appendix Behçet’s 6 (5%) 10 (9%)
p 14). No patients received a treatment to which they Other 9 (8%) 12 (10%)
were not randomised for the entire period; therefore, all Affected eye
patients were analysed as allocated for both safety and Left 3 (3%) 2 (2%)
efficacy analyses. Baseline characteristics were similar Right 4 (4%) 1 (1%)
between groups; almost half of patients had panuveitic Both 104 (94%) 112 (97%)
uveitis (table 1). Mean patient age was 42·5 years Duration of uveitis (months)
(SD 13·4), and mean duration of uveitis was 66 months Mean (SD) 62·9 (67·7) 59·5 (64·5)
(61; table 1). Median follow-up time was 155 days Range 4–394 2–381
(IQR 77–357) in the placebo group and 245 days (119–564) Number of flares in past 12 months
in the adalimumab group. 0–1 46 (41%) 48 (42%)
We recorded an early and sustained separation of the 2 40 (36%) 43 (37%)
treatment failure curves between the adalimumab and ≥3 25 (23%) 24 (21%)
placebo groups (figure 2). Treatment failure occurred in Concomitant immunomodulators at baseline
61 (55%) of 111 patients in the placebo group compared
Azathioprine 11 (10%) 3 (3%)
with 45 (39%) of 115 patients in the adalimumab group.
Cyclosporine 11 (10%) 15 (13%)
Time to treatment failure was significantly improved in
Methotrexate 14 (13%) 19 (17%)
the adalimumab group compared with the placebo group
Mycophenolate mofetil 17 (15%) 17 (15%)
(43% risk reduction; median not estimated [>18 months;
more than half the adalimumab-treated patients did not Data are n (%), unless otherwise indicated.
have treatment failure] vs 8·3 months; HR 0·57, 95% CI
Table 1: Demographics and baseline characteristics
0·39–0·84; p=0·004; figure 2). The 40th percentile for
time to treatment failure was 4·8 months in the placebo
group and 10·2 months in the adalimumab group. adalimumab for all ranked secondary variables except
Patients given adalimumab met fewer treatment failure change from baseline in VFQ-25 near vision subscore
criteria than did those given placebo (figure 3). (table 2).
Hierarchical testing of the nine ranked secondary In exploratory analyses of the four prespecified reasons
variables was stopped after the first ranked endpoint for treatment failure, the largest between-group
because no statistically significant difference was shown difference in proportions of patients with treatment
between groups; p values provided for ranked secondary failure was shown for visual acuity (21% in the placebo
endpoints are exploratory in nature (table 2). Results group and 9% in the adalimumab group; figure 3). Time
were numerically, albeit non-significantly, in favour of to treatment failure due to visual acuity was significantly

www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3 5

 Articles

100 Placebo The incidence of adverse and serious adverse events

Adalimumab was similar between treatment groups (table 3). The
most frequently reported adverse events were injection-
80
site reactions and allergic reactions (table 3). Serious
Treatment failure (%)

Prednisone taper infections occurred at a similar rate between groups

60
HR 0·57, 95% CI 0·39–0·84; p=0·004 (table 3). One malignancy (non-serious squamous cell
carcinoma) was reported in the adalimumab group, and
40
one event and three events of latent tuberculosis were
reported in the placebo and adalimumab group,
20 respectively (table 3). No patients had opportunistic
infections (excluding oral candidiasis and tuberculosis),
0 active tuberculosis, lupus or lupus-like reactions, or
0 2 4 6 8 10 12 14 16 18 20
Time (months)
demyelinating disorders (table 3).
Number at risk
Placebo 111 89 66 50 41 33 26 21 19 17 0
Seven (6%) patients in the placebo group and ten (9%)
Adalimumab 115 105 82 67 58 51 42 37 34 30 0 patients in the adalimumab group discontinued study
drug due to adverse events. Adverse events leading to
Figure 2: Kaplan–Meier plot of treatment failure for any reason
patient discontinuation in the adalimumab group were a
HR=hazard ratio.
positive test for Mycobacterium tuberculosis complex
(n=4), pulmonary sarcoidosis (n=2), and bronchitis,
A neutropenia, hepatic stenosis, dermatitis, and worsened
50 Placebo
Adalimumab migraine (n=1 each). At baseline, 60 (27%) patients were
pseudophakic. Six (5%) patients in the placebo group and
40 two (2%) patients in the adalimumab groups, developed
Proportion of patients (%)

cataracts during the study. Overall, two (2%) patients in

30 the placebo group and one (1%) patient in the
adalimumab group had cataract surgery or YAG-laser
20 capsulotomy during the study, but did not withdraw. One
death due to aortic dissection and cardiac tamponade
was reported post-treatment in a patient in the
10
adalimumab group (table 3); the investigator deemed
these events as unrelated to study drug. Six (5%) patients
0 had anti-adalimumab antibodies during the study. Five
One Two Three Four
Number of reasons (83%) of these patients had treatment failure at weeks 13,
16, 16, 24, and 31 weeks, respectively; median time to
B
treatment failure was not estimable for patients with no
30
antibodies against adalimumab (n=109) because more
Reasons for treatment failure (%)

than half of these patients did not have treatment failure.

20
Discussion
Our findings show that adalimumab significantly
10 reduced the risk of uveitic flare or loss of visual acuity in
patients with inactive, non-infectious intermediate,
posterior, or panuveitic uveitis, as shown by an early and
0 sustained separation of the adalimumab and placebo
Chorioretinal or retinal Anterior chamber cells Vitreous haze grade Visual acuity
vascular lesion
treatment failure curves. Median time to treatment
failure for adalimumab-treated patients, although not
Figure 3: Causes of treatment failure estimable, was significantly longer than for placebo-
Number of reasons for treatment failure (A). Individual reasons for treatment failure (B). treated patients. Furthermore, patients given
adalimumab met fewer treatment failure criteria than
improved in the adalimumab group compared with the did those given placebo. Risk of treatment failure based
placebo group (67% risk reduction; figure 4). Time to on logMAR best-corrected visual acuity was reduced by
treatment failure due to new active inflammatory roughly two-thirds in patients in the adalimumab group
chorioretinal or inflammatory retinal vascular lesions, compared with those in the placebo group. Rates of
increases in anterior chamber cell grade, and increases treatment failure based on active inflammatory lesions,
in vitreous haze grade did not differ significantly between anterior chamber cell grade, or vitreous haze were
groups, but were numerically lower in the adalimumab numerically, albeit not significantly, lower in the
group (figure 4). adalimumab group than in the placebo group.

6 www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3

 Articles

Placebo group (n=111) Adalimumab group (n=115) Mean difference (95% CI) p value

n* Mean (SD) n* Mean (SD)

1 Change in anterior chamber cell grade

Left eye 110 0·57 (1·00) 115 0·41 (0·97) ·· ··
Right eye 110 0·53 (0·96) 115 0·40 (0·93) –0·14 (–0·37 to 0·08) 0·218†
2 Change in vitreous haze
Left eye 110 0·33 (0·73) 115 0·16 (0·60) ·· ··
Right eye 110 0·27 (0·61) 115 0·18 (0·60) –0·13 (–0·28 to 0·01) 0·070†
3 Change in logMAR best-corrected visual acuity
Left eye 110 0·06 (0·24) 115 0·01 (0·25) ·· ··
Right eye 110 0·02 (0·20) 115 –0·01 (0·17) –0·04 (–0·08 to 0·01) 0·096†
4 Time to optical coherence tomographic evidence 95 Median not estimable 90 Median not estimable HR 0·75 (0·34 to 1·69)‡ 0·491§
of macular oedema on or after week 2 (months)
5 Percentage change in central retinal thickness
Left eye 107 6·4% (20·7) 114 4·5% (29·8) ·· ··
Right eye 108 7·7% (28·9) 113 5·4% (34·8) –2·3 (–8·5 to 3·8) 0·451¶
6 Change in VFQ-25 total score 109 1·24 (10·7) 115 3·36 (11·7) 2·12 (–0·84 to 5·08) 0·16||
7 Change in VFQ-25 distance vision subscore 109 0·76 (16·3) 115 2·64 (17·2) 1·88 (–2·53 to 6·29) 0·40||
8 Change in VFQ-25 near vision subscore 109 3·98 (17·4) 115 3·88 (18·3) –0·10 (–4·81 to 4·61) 0·97||
9 Change in VFQ-25 ocular pain subscore 109 2·87 (17·2) 115 3·42 (21·3) 0·56 (–4·56 to 5·68) 0·83||

With the exception of endpoint 4 (time to optical coherence tomographic evidence of macular oedema), data represent change from baseline to final or early termination
visit. HR=hazard ratio. VFQ-25=Visual Functioning Questionnaire-25. *Number of patients with non-missing values. †From ANOVA, with treatment as a factor adjusted for
clustered observations. ‡HR of adalimumab versus placebo from proportional hazards regression with treatment as factor. §Log-rank test. ¶From ANOVA, with treatment
and type of optical coherence tomographic machine as factors adjusted for clustered observations. ||From ANOVA, with treatment as factor.

Table 2: Ranked secondary efficacy variables

Most of the measurable effect of adalimumab was on uveitis showed that adalimumab effectively controlled
the best-corrected visual acuity component of the primary inflammation in 35% of patients refractory to previous
efficacy endpoint. Although the effect of adalimumab on treatment with infliximab or etanercept. In a prospective
the other inflammatory components of the primary open-label pilot study24 of 19 patients with various uveitic
endpoint was not significant, the improvement in best- diagnoses, adalimumab significantly reduced
corrected visual acuity is likely to be through its effect on inflammation in 63% of patients with complete resolution
multiple aspects of inflammation within the eye, some of cystoid macular oedema in 55% of affected eyes after
of which might not have been included in the 1 year. In another non-comparative open-label prospective
multicomponent endpoint. The inflammatory mani- study25 of 31 patients with refractory uveitis, 68% of patients
festations in patients with vision loss that might partly had a clinical response at 10 weeks and 39% had a
have been the cause of the loss were increases in anterior sustained response at 50 weeks. A multicentre prospective
chamber cell grade and vitreous haze grade (≥1), new study14 of 131 patients with a mean age of 27 years also
inflammatory or chorioretinal vascular lesions, retinal showed that adalimumab significantly improved anterior
thickening, and cataracts. The cross-sectional study by chamber and vitreous inflammation with the ability to
Dick and colleagues,2 based on population insurance taper corticosteroid use. The French Uveitis Network did a
data, provides supportive evidence that presence of multicentre observational study26 of 160 patients with
chronic low-grade inflammation in this population refractory uveitis treated with anti-TNFα drugs (infliximab
worsens visual outcomes. and adalimumab); the overall response rate was 93% at
The efficacy results of the present trial were in accordance 12 months.
with those of previous studies. In the multicentre, double- The low adalimumab immunogenicity in the present
masked, randomised controlled VISUAL I trial,19 study was within the range of rates reported in other
adalimumab significantly reduced the risk of treatment disease states.13 Moreover, the safety profile of adalimumab
failure by 50% compared with placebo in patients with in this study was similar to that in other approved
active non-infectious uveitis. In both VISUAL I (active indications, for which the rates of adverse events, serious
disease) and VISUAL II (inactive disease), adalimumab adverse events, and discontinuations due to an adverse
halved risk to failure and almost doubled time to failure. A event were similar in both adalimumab and placebo
retrospective study23 in patients with refractory chronic groups,13 and no new safety signals were detected.27,28

www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3 7

 Articles

A Vitreous haze B New lesions

100 Placebo
Adalimumab

80
Treatment failure (%)

60
HR 0·79, 95% CI 0·34–1·81; p=0·589 HR 0·55, 95% CI 0·26–1·15; p=0·105

40

20

C Anterior chamber cells D Best-corrected visual acuity

100

80
Treatment failure (%)

60 HR 0·70, 95% CI 0·42–1·18; p=0·180 HR 0·33, 95% CI 0·16–0·70; p=0·002

40

20

0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Time (months) Time (months)
Number at risk
Placebo 111 89 66 50 41 33 27 21 19 17 0 111 89 66 50 41 33 26 21 19 17 0
Adalimumab 115 105 82 67 58 51 42 37 34 3 0 115 105 82 67 58 51 42 37 34 3 0

Figure 4: Kaplan–Meier plot of treatment failure due to vitreous haze (A), new lesions (B), anterior chamber cells (C), or best-corrected visual acuity (D)
HR=hazard ratio.

Placebo group (n=114; Adalimumab group

been prematurely terminated for unknown reasons.29–31
71·0 person-years) (n=115; Thus, to our knowledge, VISUAL II is the first phase 3 trial
94·5 person-years) of a non-steroidal systemic drug in patients with quiescent
Any adverse event 642 (905) 831 (879) disease to have reached its prespecified primary endpoint
Adverse event leading to death* 0 2 (2·1) (time to treatment failure) and show promise in treatment
Serious adverse event 10 (14·1) 13 (13·8) of patients with inactive non-infectious uveitis controlled
Adverse event leading to discontinuation of study drug 7 (9·9) 11 (11·6) by chronic oral corticosteroids (≥10 mg/day).
Serious infectious adverse event 2 (2·8) 3 (3·2) Strengths of this study were the trial design, relatively
Injection-site reactions 16 (22·6) 36 (38·1) large study population, range of uveitis diagnoses, and
Malignancies† 0 1 (1·1) multicomponent primary endpoint. The composite
Opportunistic infections (excluding oral candidiasis and 0 0
primary endpoint assessed various facets of the disease,
tuberculosis) ranging from anterior to posterior segments of the eye,
Active tuberculosis 0 0 and facilitated detailed assessment of treatment response
Latent tuberculosis 1 (1·4) 3 (3·2) and efficacy, given that inflammation does not always
Demyelinating disease 0 0 manifest as one endpoint (eg, vitreous haze).
Lupus-like reaction 0 0 Furthermore, all patients had a mandatory corticosteroid
Allergic reactions (including angio-oedema, anaphylaxis) 8 (11·3) 5 (5·3) taper to zero, enabling assessment of the corticosteroid-
sparing effect of adalimumab.
Data in parentheses are events are per 100 person-years. *One death due to two fatal events of aortic dissection and There were limitations to the interpretation of the
cardiac tamponade (day 54 [18 days after last adalimumab dose]); these events were deemed not related to treatment.
†One event of non-serious squamous cell carcinoma (day 210, resolved on day 215; adalimumab treatment was not secondary endpoints given that the magnitude of the
interrupted). treatment effect was diluted because only a few patients
had treatment failure due to one of the four components.
Table 3: Adverse events (safety population)
Thus, the magnitude of mean change was small for these
endpoints. A floor effect could have been possible
Previous clinical trials that were done to evaluate the because most patients started with reasonably good
therapeutic potential for inactive, non-infectious uveitis visual acuity and minimum inflammation; detection of a
have either not achieved their primary endpoint or have change might have been particularly difficult because

8 www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3

 Articles

more than half of patients in the adalimumab group consultant for Xoma, Servier, and Santen; and has served on advisory
never achieved treatment failure. Although range of boards for Xoma, Servier, Santen, Alcon, and Bausch & Lomb. JVC has
served on advisory boards for AbbVie and MSD, and has served as a
uveitis diagnoses was a strength of our study, it could consultant for MSD. AS declares no competing interests.
also be a potential limitation because no information was
Acknowledgments
provided about which disease groups (with their This study was funded by AbbVie. We thank Friederike Mackensen for her
recognised heterogeneity) were actually responsive to contribution to the study conception and design, and data analysis and
therapy. Due to the difficulty in recruitment of patients in interpretation. Dr Mackensen passed away during the development of the
a rare disease cohort with multiple competing studies, no manuscript. We thank Amin Kherani, Andrew Logan, Antonio Ciardella,
Justus Garweg, Chloe Gottlieb, David Scales, De Schryver, Eric Fortin,
restriction on the number of recruiting sites was Farzin Forooghian, G Pertile, Hiroshi Goto, Jennifer Thorne,
imposed—an additional weakness. Furthermore, the Kenichi Namba, Koh-Hei Sonoda, Koju Kamoi, Lourdes Arellanes-Garcia,
study was not statistically powered to analyse a differential Lucia Kuffova, Lyndell Lim, Maria Pia Paroli, Marta Misiuk-Hojlo,
efficacy among the different causes of uveitis. Masahiro Zako, Michal Kramer, Nicholas Beare, Nobuhisa Mizuki,
Noriyasu Hashida, Pablo Franco, René Cervantes-Castañeda,
Studies or clinical trials intended for the treatment of Robert Wang, Rui Proença, Sanjay Kedhar, Sarju Patel, Sofia Androudi,
uveitis face various challenges. Uveitis is a heterogeneous Talin Barisani-Asenbauer, Theresa Larson, Thomas Flynn, Thomas Ness,
group of conditions characterised by intraocular Toshiaki Abe, Toshikatsu Kaburaki, and Yan Guex-Crosier for their
inflammation. Most uveitis syndromes are individually contribution and assistance for the study. Gaurav Patki (AbbVie,
North Chicago, IL, USA) provided medical writing assistance.
rare, but for taxonomic and clinical convenience are
commonly clustered according to their anatomical References
1 Durrani OM, Tehrani NN, Marr JE, Moradi P, Stavrou P, Murray PI.
classification, despite the wide range of systemic and Degree, duration, and causes of visual loss in uveitis.
clinical associations they represent. Another challenge Br J Ophthalmol 2004; 88: 1159–62.
encountered in any uveitis trial is the scarcity of high- 2 Dick AD, Tundia N, Sorg R, et al. Risk of ocular complications in
patients with noninfectious intermediate uveitis, posterior uveitis,
quality outcome measures. Currently, vitreous haze or panuveitis. Ophthalmology 2016; 123: 655–62.
grade, as defined by Nussenblatt,32 is a surrogate endpoint 3 Tomkins-Netzer O, Talat L, Bar A, et al. Long-term clinical outcome
for disease activity that is accepted by the US Food and and causes of vision loss in patients with uveitis. Ophthalmology
2014; 121: 2387–92.
Drug Administration for clinical trials. This score uses a 4 Becker MD, Smith JR, Max R, Fiehn C. Management of
subjective ordinal scale of cloudiness of the vitreous sight-threatening uveitis: new therapeutic options. Drugs 2005;
humour, but has substantial interobserver variability. 65: 497–519.
5 Cervantes-Castaneda RA, Jawed N, Foster CS. Immunosuppressive
Our findings suggest that adalimumab could be a well therapy for noninfectious uveitis. Retin Physician 2007; 4: 30–38.
tolerated and effective treatment option for patients with 6 Jabs DA, Rosenbaum JT, Foster CS, et al. Guidelines for the use of
inactive, non-infectious intermediate, posterior, or immunosuppressive drugs in patients with ocular inflammatory
panuveitis who are at risk of the long-term side-effects of disorders: recommendations of an expert panel. Am J Opthamol
2000; 130: 492–513.
corticosteroids. An open-label extension study 7 Okada AA, Goto H, Ohno S, Mochizuki M; for the Ocular Behçets
(NCT01148225) is ongoing to provide long-term safety data Disease Research Group of Japan. Muticenter study of infliximab
for adalimumab in patients with non-infectious uveitis. for refractory uveoretinitis in Behçet disease. Arch Ophthalmol 2012;
130: 592–98.
Contributors 8 Heiligenhaus A, Zurek-Imhoff B, Roesel M, Hennig M,
QDN, PTM, SKK, JS, AS, CP, LC, JVC, and ADD participated in the Rammrath D, Heinz C. Everolimus for the treatment of uveitis
conduct of the study, including selection, treatment, and follow-up of refractory to cyclosporine A: a pilot study.
patients; data interpretation; and preparation and critical review of the Graefes Arch Clin Exp Ophthalmol 2013; 251: 143–52.
report. QDN, GJJ, and APB participated in the conception and study 9 Beckham JC, Caldwell DS, Peterson BL, et al. Disease severity in
design, analysis and interpretation of data, and preparation and critical rheumatoid arthritis: relationships of plasma tumor necrosis
review of the report. AAC, NVK, APS, MK, and ST participated in the factor-alpha, soluble interleukin 2-receptor, soluble CD4/CD8 ratio,
analysis and interpretation of data, and the preparation and critical review neopterin, and fibrin D-dimer to traditional severity and functional
of the report. All authors provided a final review and approved measures. J Clin Immunol 1992; 12: 353–61.
the manuscript. 10 Brennan FM, Feldmann M. Cytokines in autoimmunity.
Curr Opin Immunol 1992; 4: 754–59.
Declaration of interests 11 Rabinovich CE. Use of tumor necrosis factor inhibitors in uveitis.
AAC, MK, APS, NVK, and ST are AbbVie employees and all hold AbbVie Curr Opin Rheumatol 2007; 19: 482–86.
stock or options. QDN has served on the scientific advisory boards for 12 Turan B, Gallati H, Erdi H, Gurler A, Michel BA, Villiger PM.
AbbVie, Santen, XOMA, Bausch & Lomb, and Genentech uveitis studies, Systemic levels of the T cell regulatory cytokines IL-10 and IL-12 in
and chairs the steering committee for the VISUAL, EYEGUARD, and Bechcet’s disease; soluble TNFR-75 as a biological marker of disease
SAKURA studies. APB has served on advisory boards and as a activity. J Rheumatol 1997; 24: 128–32.
consultant for AbbVie. GJJ has served as a consultant for AbbVie. 13 Burmester GR, Panaccione R, Gordon KB, McIlraith MJ,
ADD has served on advisory boards for AbbVie. PTM has served on the Lacerda AP. Adalimumab: long-term safety in 23 458 patients from
steering committee for the VISUAL studies and has served as consultant global clinical trials in rheumatoid arthritis, juvenile idiopathic
for Santen. SKK has been an advisor or steering committee member for arthritis, ankylosing spondylitis, psoriatic arthritis, psoriasis and
AbbVie, Allergan, Bayer, Clearside, Regeneron, and Xoma. JS has been a Crohn’s disease. Ann Rheum Dis 2013; 72: 517–24.
consultant for AbbVie, Alcon, Allergan, Aldeyra, Bausch & Lomb, 14 Diaz-Llopis M, Salom D, Garcia-de-Vicuna C, et al. Treatment of
Clearside, EyeGate, Ocular Therapeutix, Tear Lab, Tear Science, Santen, refractory uveitis with adalimumab: a prospective multicenter study
of 131 patients. Ophthalmology 2012; 119: 1575–81.
Shire; an investigator for Xoma, Bausch & Lomb, Lux Biosciences,
Eyegate, Alcon, Allergan, Clearside, Alimera, pSivida, Novartis, Shire, 15 Levy-Clarke G, Jabs DA, Read RW, Rosenbaum JT, Vitale A,
Van Gelder RN. Expert panel recommendations for the use of
Aldeyra; and has served on the steering committee for the VISUAL
anti-tumor necrosis factor biologic agents in patients with ocular
studies. LC has been on advisory boards and served as a consultant for
inflammatory disorders. Ophthalmology 2014; 121: 785–96 e3.
AbbVie. CP has received a research grant from Alcon; has served as a

www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3 9

 Articles

16 Suhler EB, Smith JR, Giles TR, et al. Infliximab therapy for 25 Suhler EB, Lowder CY, Goldstein DA, et al. Adalimumab therapy
refractory uveitis: 2-year results of a prospective trial. for refractory uveitis: results of a multicentre, open-label,
Arch Ophthalmol 2009; 127: 819–22. prospective trial. Br J Ophthalmol 2013; 97: 481–86.
17 Foster CS, Tufail F, Waheed NK, et al. Efficacy of etanercept in 26 Vallet H, Seve P, Biard L, et al. Infliximab versus adalimumab in
preventing relapse of uveitis controlled by methotrexate. the treatment of refractory inflammatory uveitis: multicenter study
Arch Ophthalmol 2003; 121: 437–40. from the french uveitis network. Arthritis Rheumatol 2016;
18 Schmeling H, Horneff G. Etanercept and uveitis in patients with 68: 1522–30.
juvenile idiopathic arthritis. Rheumatology 2005; 44: 1008–11. 27 Bidaut Garnier M, Vallet H, Seve P, et al. Efficacy and safety of anti
19 Jaffe GJ, Thorne JE, Scales D, et al. Adalimumab in patients with TNFα in non-infectious uveitis: a multi-center retrospective study.
active, non-infectious uveitis requiring high-dose corticosteroids: Acta Ophthalmol 2014; 92: 0.
The VISUAL-1 trial. Ann Rheum Dis 2015; 74 (suppl 2): 849 (abstr). 28 Gomez-Reino JJ, Carmona L, Valverde VR, Mola EM, Montero MD,
20 Jabs DA, Nussenblatt RB, Rosenbaum JT; Standardization of Group B. Treatment of rheumatoid arthritis with tumor necrosis
Uveitis Nomenclature (SUN) Working Group. Standardization of factor inhibitors may predispose to significant increase in
uveitis nomenclature for reporting clinical data. Results of the first tuberculosis risk: a multicenter active-surveillance report.
international workshop. Am J Opthamol 2005; 140: 509–16. Arthritis Rheum 2003; 48: 2122–27.
21 Nussenblatt RB, Palestine AG, Chan CC, Roberge F. 29 Medscape. Voclosporin withdrawn from approval for uveitis in EU.
Standardization of vitreal inflammatory activity in intermediate and Oct 18, 2011. http://www.medscape.com/viewarticle/751710
posterior uveitis. Ophthalmology 1985; 92: 467–71. (accessed April 20, 2016).
22 Chew EY, Kim J, Sperduto RD, et al. Evaluation of the age-related 30 Dick AD, Tugal-Tutkun I, Foster S, et al. Secukinumab in the
eye disease study clinical lens grading system AREDS report No. 31. treatment of noninfectious uveitis: results of three randomized,
Ophthalmology 2010; 117: 2112–19 e3. controlled clinical trials. Ophthalmology 2013; 120: 777–87.
23 Tynjala P, Kotaniemi K, Lindahl P, et al. Adalimumab in juvenile 31 Pasadhika S, Rosenbaum JT. Update on the use of systemic biologic
idiopathic arthritis-associated chronic anterior uveitis. Rheumatology agents in the treatment of noninfectious uveitis. Biologics 2014;
2008; 47: 339–44. 8: 67–81.
24 Diaz-Llopis M, Garcia-Delpech S, Salom D, et al. 32 Nussenblatt RB, Palestine AG, Chan CC, Roberge F.
Adalimumab therapy for refractory uveitis: a pilot study. Syandardization of vitreal inflammatory activity in intermediate and
J Ocul Pharmacol Ther 2008; 24: 351–61. posterior uveitis. Ophthalmology 1985: 92: 467–71.

10 www.thelancet.com Published online August 16, 2016 http://dx.doi.org/10.1016/S0140-6736(16)31339-3