NEW RESEARCH

Randomized, Controlled Trial of Intravenous

Immunoglobulin for Pediatric Autoimmune
Neuropsychiatric Disorders Associated With
Streptococcal Infections
Kyle A. Williams, MD, PhD, Susan E. Swedo, MD, Cristan A. Farmer, PhD, Heidi Grantz, LCSW,
Paul J. Grant, MD, Precilla D’Souza, CRNP, Rebecca Hommer, MD, Liliya Katsovich, MA,
Robert A. King, MD, James F. Leckman, MD, PhD

Objective: Pediatric autoimmune neuropsychiatric disor- were not statistically significant. Twenty-four participants
ders associated with streptococcal infections (PANDAS) met criteria for nonresponse to double-blind infusion
are hypothesized to occur as a result of cross-reactive anti- and received open-label IVIG at week 6. Among all
bodies produced in response to group A streptococcal participants, the mean CY-BOCS improvement from
infections. Previous research suggests that immunomodu- baseline was 55%  33% at week 12 and 62%  33% at
latory therapies, such as intravenous immunoglobulin week 24.
(IVIG), may lead to rapid and sustained symptom
improvement in patients with PANDAS. Conclusion: IVIG was safe and well tolerated. Between-
group differences were smaller than anticipated, and the
Method: A total of 35 children meeting criteria for
double-blind comparison failed to demonstrate superior-
PANDAS and moderate to severe obsessive-compulsive
ity of IVIG over placebo. The observed open-label im-
disorder (OCD) were enrolled in a randomized-entry,
provements indicate that future trials would benefit from
double-blind, placebo-controlled, 6-week trial of IVIG (1
larger sample sizes designed in part to aid in the identi-
g/kg/day on 2 consecutive days), followed by optional
fication of biomarkers predictive of a positive response to
open-label treatment for nonresponders, with follow-up at
immunotherapy. Future investigations focused on the
12 and 24 weeks. Primary outcome measures were the
natural history of PANDAS are also warranted.
Children’s Yale
Brown Obsessive Compulsive Scale (CY-
BOCS) and the Clinical Global Impressions
Improvement
Clinical trial registration information—Intravenous Immu-
(CGI-I) rating. “Responders” were defined, a priori, by
noglobulin for PANDAS (Pediatric Autoimmune Neuropsy-
a  30% decrease in CY-BOCS total score, and a “much”
chiatric Disorders Associated With Streptococcal Infections);
or “very much” improved rating on CGI-I.
http://clinicaltrials.gov/; NCT01281969 ZIAMH002666.
Results: During the double-blind phase, the mean
decrease in CY-BOCS score was 24%  31% in the IVIG Key words: PANDAS, obsessive-compulsive disorder,
group (n ¼ 17) and 12%  27% in the placebo group IVIG, group A streptococcus pyogenes
(n ¼ 18), with six responders in the IVIG group (35%)
versus four (22%) in the placebo group; these differences J Am Acad Child Adolesc Psychiatry 2016;-(-):-–-.

A ffecting 1% to 2% of children and adolescents,

obsessive-compulsive disorder (OCD) is charac-
terized by the presence of recurrent, unwanted
thoughts (obsessions) and repetitive behaviors (compul-
sions) that create psychological distress and significant
impairment.1 Although most patients experience a
gradual onset of symptoms, some children have a sudden
onset of severe OCD and neuropsychiatric symptoms
This article is discussed in an editorial by Dr. Jonathan Mink on
page xx.
Supplemental material cited in this article is available online.
following infections with group A streptococci (GAS).2-4
Such cases are identified by the acronym PANDAS
(pediatric autoimmune neuropsychiatric disorders associ-
ated with streptococcal infections). The etiology and
pathogenesis of PANDAS is hypothesized to be similar
to that of rheumatic fever and Sydenham chorea, in
which susceptible individuals produce cross-reactive an-
tibodies in response to a GAS infection. These antibodies
react both to components of the GAS cell wall and
neuronal proteins in the basal ganglia.5 Among patients
with PANDAS, high concentrations of cross-reactive an-
tibodies are found in acute serum samples, and these
antibody titers are decreased during periods of symptom
remission.5-7 Support for the pathogenic role of cross-
reactive antibodies has been provided through rodent
models in which immunization with GAS proteins elicits

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME - NUMBER - - 2016 www.jaacap.org 1
WILLIAMS et al.

behavioral abnormalities, production of cross-reactive double-blind, placebo-controlled trial demonstrated that

antibodies, and antibody deposition in the basal ganglia
and cerebellum.8-14
Given these findings, it is hypothesized that treatments
that remove or inactivate pathologic, cross-reactive anti-
bodies should produce therapeutic benefits in PANDAS
similar to those reported for Sydenham chorea, Guillain-
Barre syndrome, and antibody-mediated autoimmune
encephalitis.15-17 Intravenous immunoglobulin (IVIG), an
immunomodulatory therapy, has been shown to have ben-
efits for each of these disorders, although the therapeutic
mechanisms are unknown.18,19 In PANDAS, an earlier
IVIG and therapeutic plasma apheresis (apheresis) were
both effective in reducing obsessive-compulsive symptoms
in patients with PANDAS (by 45% and 58%, respectively),
whereas a placebo infusion had no discernible effect.20
Subsequent to this controlled trial, several case reports and
two case series have provided additional support for the
therapeutic benefits of apheresis and IVIG.21-27 These
improvements are hypothesized to be related to the immu-
nomodulatory effects of IVIG, as apheresis and IVIG have
shown no benefit in the treatment of non-PANDAS OCD28
and tic disorders.29

TABLE 1 Sociodemographic and Clinical Characteristics

Placebo IVIG

Characteristic n (%) Mean  SD n (%) Mean  SD

Age, y 18 (100) 9.61  2.32 17 (100) 8.99  2.37
Male sex 11 (61) 12 (70)
Race/ethnicity
White 16 (89) 14 (82)
African American 0 1 (6)
Asian American 0 1 (6)
American Indian 0 1 (6)
Multiple 2 (11) 0
Not Hispanic/Latino 17 (94) 16 (94)
Antibiotic
Amoxicillin/penicillin 14 (78) 13 (76)
Other 4 (22) 4 (24)
Days of prophylactic antibiotic 12 26.91  26.35 16 29.38  24.71
Other baseline medications
SSRI 4 (22) 2 (12)
a-Adrenergic 0 (0) 2 (12)
Serotonin/dopamine agonist (antipsychotic) 0 (0) 1 (1)
Antihistamine 1 (6) 7 (41)
Bronchodilator/respiratory steroid 1 (6) 3 (18)
Hormonal supplement (sleep) 6 (33) 4 (24)
First episode 12 (67) 12 (71)
Mean duration of illness at baseline (days) 115.00  51.83 92.12  60.25
CGI Severity 5.27  0.89 5.29  0.69
Moderate (4) 3 (17) 2 (12)
Marked (5) 9 (50) 8 (47)
Severe (6) 4 (22) 7 (41)
Extreme (7) 2 (11) 0 (0)
CY-BOCS total 18 (100) 28.78  3.98 17 (100) 26.47  5.14
Obsessions total 14.5  2.18 13.65  2.62
Compulsions total 14.28  2.63 12.82  3.24
CaM kinase II, baseline
Serum 18 (100) 155.23  32.06 17 (100) 154.55  33.37
Elevated (>130) serum 13 (72) 12 (71)
CSF 17 (94) 116.80  24.91 17 (100) 114.87  27.76
Elevated (>130) CSF 5 (28) 3 (18)
Note: Randomization groups do not differ significantly on any baseline characteristic (all p > .10), with the exception of antihistamine medications (c2¼5.89, p ¼ .02).
Antibiotics are mutually exclusive. Other antibiotics were azithromycin, clarithromycin, clindamycin, or Omnicef. Other medications are not mutually exclusive.
CGI ¼ Clinical Global Impressions scale; CSF ¼ cerebrospinal fluid; CY-BOCS ¼ Children’s Yale
Brown Obsessive Compulsive Scale; IVIG ¼ intravenous
immunoglobulin; SSRI ¼ selective serotonin reuptake inhibitor.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

2 www.jaacap.org VOLUME - NUMBER - - 2016

Despite these findings, IVIG has not been widely adopted
in clinical practice for patients with PANDAS. Furthermore,
biological markers of treatment response have not been
systematically evaluated in children with PANDAS treated
with IVIG therapy. To address these issues, we conducted a
randomized, placebo-controlled trial of IVIG to evaluate the
safety and efficacy of its use in children with PANDAS.
METHOD
Participants and Study Design
This randomized clinical study was conducted at two academic
centers: the National Institute of Mental Health (NIMH), Bethesda,
Maryland, and the Yale Child Study Center, New Haven, Con-
necticut. Institutional review boards at both sites approved the
study. Participants were recruited nationwide through postings on
the NIMH website and ClinicalTrials.gov (NCT01281969), as well as
direct referrals from clinicians. Participants’ parents provided
informed consent, and study participants ( age 7 years) provided
written assent.
Eligible participants were required to meet all diagnostic criteria
for PANDAS, to have moderate to severe OCD symptoms (Chil-
dren’s Yale
Brown Obsessive Compulsive Scale [CY-BOCS]30 total
score 20), and not to initiate or modify psychoactive medications
or behavioral therapy for 6 weeks before study entry. Key inclusion
criteria were as follows: age 4 to 13 years; first episode of PANDAS
symptoms or first recurrence of symptoms; onset of current symp-
toms within 6 months of entrance into study; documentation that
symptom onset occurred within 6 to 8 weeks of a GAS infection or
exposure; and sudden onset or exacerbation of OCD (reaching peak
severity and impairment within 24 to 48 hours). Participants also
had to have concomitant onset of at least three comorbid neuro-
psychiatric symptoms, and thus met criteria for pediatric acute-onset
neuropsychiatric syndrome (PANS).28,29,31 Exclusion criteria were as
follows: a history of Sydenham chorea or acute rheumatic fever;
symptoms consistent with autism spectrum disorder or schizo-
phrenia; severe physical, behavioral, or psychiatric symptoms that
would prevent study participation; or prior corticosteroid or
immunomodulatory therapy for PANDAS.
Eligibility was established by trained interviewers at Yale and
NIMH during a multi-stage screening process (see Consolidated
Standards of Reporting Trials [CONSORT] diagram, Figure S1,
available online). After a focused telephone interview, parents
completed questionnaires and provided the child’s medical records,
which were reviewed by study investigators. Children with docu-
mented GAS infections preceding symptom onset and/or exacer-
bation, and symptoms considered “very likely” to fulfill study
criteria were invited for an interview at the National Institutes of
Health (NIH) Clinical Center. This interview was conducted by
Yale investigators through a video-conferencing link, with NIMH
investigators in attendance. If all interviewers agreed on a partici-
pant’s eligibility, the participant underwent a baseline evaluation,
consisting of physical and neurological examinations, phlebotomy,
and assessments to rule out other organic causes of neuropsychiatric
disease, including electroencephalography (EEG), lumbar puncture,
and a structural magnetic resonance imaging (MRI) scan. Electro-
cardiography and echocardiography were performed to exclude
occult rheumatic carditis. Participants who successfully completed
baseline assessments were randomized.
Prophylactic antibiotics were prescribed for all study participants
to prevent GAS infections during the study; most were prescribed
penicillin (n ¼ 29), but children with a penicillin allergy or who had
been started on a macrolide antibiotic by their primary care pedia-
trician (n ¼ 6) were maintained on macrolide therapy.
JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY
VOLUME - NUMBER - - 2016
RANDOMIZED CONTROLLED TRIAL OF IVIG IN PANDAS
Behavioral Assessments
Trained masters- or doctoral-level clinicians at Yale administered all
psychometric rating scales through a video-conferencing link. The
rating scales included the CY-BOCS symptom list and severity scale,
the Yale Global Tic Severity Scale (YGTSS), the Clinical Global Im-
pressions Severity and Improvement (CGI-S and CGI-I) scales, and
the Children’s Depression Rating Scale (CDRS-R).30,32-35 Parents and
children provided information for these clinician-rated measures.
CY-BOCS Total score and CGI-I ratings were the primary outcome
measures.
Laboratory Assessments
Serum and cerebrospinal fluid (CSF) samples were de-identified, masked
to treatment assignment, and sent to Dr. Madeleine Cunningham
at the University of Oklahoma (M.W.C.) for blinded analysis of
antibody-mediated activation of calcium calmodulin-dependent
protein kinase II (CaMKII), using previously described methods.36
The NIH Clinical Center Laboratory performed all clinical labora-
tory studies, including the anti-nuclear antibody (ANA) assessment,
which was performed by enzyme immunoassay on the Dynex
DSX instrument.
Randomization and Study Drugs
Participants were randomized to receive either IVIG (1 g/kg/day on
2 consecutive days; total dose 2 g/kg) or IV placebo (saline solution).
This IVIG dose was used in the previous PANDAS treatment trial
and is recommended for treatment of other post-infectious autoim-
mune diseases of childhood.20,21 Randomization was stratified
by sex with 1:1 allocation, using random block sizes of 4 and 6.
Participants, parents, and study staff were blinded to randomization
status, which was maintained by the Pharmaceutical Development
Service of the NIH Clinical Center. The NIH pharmacy prepared the
IVIG and placebo for infusion and supplied it in wrapped infusion
kits. Grifols (formerly Talecris Laboratories) supplied the IVIG for
the trial but did not have access to the data or any role in study
design, analysis, or manuscript preparation. Multiple lots of IVIG
were used by the NIH pharmacy, and individual infusions often
contained multiple lots of IVIG.
FIGURE 1 Unadjusted mean (SD) Children’s Yale
Brown
Obsessive Compulsive Scale (CY-BOCS) total scores during the
double-blind phase. Note: Accounting for baseline scores, the
randomization groups did not differ significantly at week 6.
IVIG ¼ intravenous immunoglobulin.
www.jaacap.org 3
WILLIAMS et al.

TABLE 2 Acute Phase Outcomes

Baseline Observed Week 6 Observed Difference in Estimated
Marginal Means at
Placebo IVIG Placebo IVIG Week 6, 95% CI Comparison
CY-BOCS, mean  SD
Total score 28.78  3.98 26.47  5.14 25.67  8.65 20.59  10.12
1.97 (
7.1, 3.15) F1,34 ¼ 0.62, p ¼ .44
Obsessions 14.5  2.18 13.65  2.62 12.39  4.62 10.94  4.8
0.53 (
3.32, 2.27) F1,34 ¼ 0.15, p ¼ .70
Compulsions 14.28  2.63 12.82  3.24 13.28  4.27 9.65  5.85
1.88 (
4.53, 0.77) F1,34 ¼ 2.10, p ¼ .16
CGI-I, mean  SD 3.53  1.62 2.88  1.20 Z ¼
1.18, p ¼ .12a
Responder, n (%) 4 (22) 6 (35) c2 ¼ 0.72, p ¼ .40
Note: Observed means were the actual mean scores in the treatment groups at baseline and week 6. Estimated marginal mean corresponds to the F value in the
Comparison column, which was obtained from an analysis of covariance, controlling for baseline score. The a priori definition of Responder was a Clinical Global
Impressions
Improvement (CGI-I) of 1 (very much improved) or 2 (much improved) and at least 30% improvement in Children’s Yale
Brown Obsessive Compulsive
Scale (CY-BOCS) total score. IVIG ¼ intravenous immunoglobulin.
a
Wilcoxon two-sample test. Placebo mean rank sum ¼ 19.92; IVIG mean rank sum ¼ 15.97.

Procedure failure to meet all PANDAS diagnostic criteria or a lack of

The study consisted of four visits: baseline, week 6 (end of the
blinded phase), week 12 (end of the open-label phase), and week 24
(follow-up). At baseline, participants were admitted to the NIH
Clinical Center, infused with randomized study drug, and observed
for adverse effects for 24 to 48 hours after infusion. Adverse events
were assessed by the treating NIH clinician after each infusion and
before discharge; a nurse conducted post-discharge status checks at
3 and 7 days post-hospitalization.
Six weeks following baseline, participants returned to NIH.
Psychometric rating scales, phlebotomy, and lumbar puncture were
repeated. “Responder” status was defined as a decrease in CY-BOCS
score of 30%, and “Much” or “Very Much” improved rating on
CGI-I. Nonresponders to the blinded infusion were offered open-
label IVIG infusion of 2 g/kg (administered as 1 g/kg/day on 2
consecutive days), and observed for 24 to 48 hours. All participants
returned for follow-up 12 and 24 weeks post-randomization to
complete a clinical interview, psychometric rating scales, and phle-
botomy; structural MRI was also performed at week 12.
Statistical Analysis
Data were analyzed using SAS version 9.3 software (SAS Institute,
Cary, NC). Continuous outcomes were assessed at the 6-week time
point using an analysis of covariance (ANCOVA) model controlling
for baseline scores and categorical outcomes, or the Wilcoxon
two-sample test (CGI Improvement ratings) or c2 (responder status).
Data were analyzed following the intent-to-treat model. Effect sizes
were Cohen’s d with pooled standard deviations, using adjusted
means from the ANCOVAs. No corrections were made for multiple
comparisons. A priori power calculations based on the Perlmutter
et al.20 study (IVIG effect size 1.2) suggested that a sample size of 16
per group would be sufficient to detect an effect size of 1.0 with 80%
power.
RESULTS
Study Population
From 1,183 inquiries, 214 potential participants were
screened between February 2011 and May 2013. Of these,
48 traveled to NIMH for an in-person assessment (Figure S1,
available online), and 36 met criteria for study enrollment.
The two most frequent reasons for study exclusion were
sufficient severity of current symptoms.
No significant differences were noted in demographic
characteristics, symptom severity, or medication status be-
tween the participants randomized to the two study groups,
with the exception of anti-histamine medications (Table 1).
One randomized participant (IVIG) withdrew from the
study before study drug infusion because of post
lumbar
puncture anxiety. As this participant was removed from the
study before initiation of treatment, the participant was
excluded from the data analysis.
Efficacy
Double-Blind Phase. Improvement on the primary outcome
measure, CY-BOCS Total score, was observed in both
groups at week 6 (Figure 1). A mean decrease of 23.9% 
31.0% in CY-BOCS Total score was observed in the IVIG
group, compared to a mean decrease of 11.7%  26.6% in
the placebo group. The mean difference between groups
did not reach statistical significance, with an effect size
of 0.28 (95% CI ¼
0.39 to 0.95) (Table 2).
Week 6 mean CGI-Improvement scores also did not differ
significantly between groups (Table 2), with individual
responses varying widely. CGI-Improvement scores were 4
(No Change) or better for all members of the IVIG group,
with n ¼ 3 (18%) rated as Very Much Improved, n ¼ 3 (18%)
as Much Improved, and n ¼ 3 (18%) as Mildly Improved.
Fourteen participants (78%) in the placebo group were rated
as No Change or better: Very Much Improved: n ¼ 3 (17%),
Much Improved: n ¼ 2 (11%), Mildly Improved: n ¼ 2 (11%),
No Change: n ¼ 7 (39%); two (11%) were considered Mildly
Worse, and two (11%) were rated Very Much Worse. Six
participants (35%) in the IVIG group were characterized
as responders, compared to four participants (22%) in the
placebo group (not significant) (Table 2).
Open-Label Phase. Children who met criteria for response
during the double-blind phase of the trial were not eligible
to receive open-label IVIG. However, this criterion was
violated for one participant in the placebo group who

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

4 www.jaacap.org VOLUME - NUMBER - - 2016
 RANDOMIZED CONTROLLED TRIAL OF IVIG IN PANDAS

TABLE 3 Open-Label Primary Outcomes

Randomized to Randomized to Randomized to Randomized to
IVIG, Received Placebo, Received IVIG or Placebo, IVIG or Placebo, All
Open-Label IVIG Open-Label IVIGa Received Open-Label IVIGa No Open-Label IVIGa Participantsa
n 10 14 24 10 34
Baseline CY-BOCS 26.80  5.920 29.07  3.81 28.25  4.69 25.80  3.49 27.44  4.55
Week 6 CY-BOCS 23.50  10.04 29.07  5.12 27.67  5.58 14.60  8.60 23.18  9.75
Week 12 CY-BOCS 12.90  7.84 14.36  10.86 13.96  9.29 9.50  10.63 12.50  9.91
Week 24 CY-BOCS 10.10  10.31 13.50  8.35 11.38  9.42 7.00  8.62 10.59  9.18
Week 6 CGI-I 3.40  0.97 4.23  1.09 4.00  0.93 1.90  1.20 3.25  1.46
Week 12 CGI-I 1.70  0.67 2.14  1.56 1.96  1.27 1.80  1.23 1.91  1.24
Week 24 CGI-I 1.70  1.06 1.83  0.94 1.76  1.00 1.60  0.84 1.74  0.93
Responder at week 12, n (%) 7 (70) 10 (67) 16 (67) 8 (80) 24 (71)
Responder at week 24, n (%) 8 (80) 10 (67) 17 (71) 8 (80) 25 (74)
Note: With one exception, participants who did not receive open label intravenous immunoglobulin (IVIG) were responders to IVIG (n ¼ 6) or placebo (n ¼ 3) during the
double-blind phase. One participant in the IVIG group was not a responder during the double-blind phase, but refused open-label treatment. CGI-I ¼ Clinical Global
Impressions
Improvement; CY-BOCS ¼ Children’s Yale
Brown Obsessive Compulsive Scale.
a
Contrary to protocol, one participant randomized to placebo responded during the double-blind phase and received open-label IVIG. This participant is excluded
from open-label outcomes.

displayed marked improvement in OCD severity but
remained impaired by comorbid symptoms. This partici-
pant received open-label IVIG and was excluded from
12- and 24-week analyses. One nonresponder to blinded
infusion declined open-label IVIG. Twenty-four partici-
pants (double-blind randomization status: n ¼ 10 IVIG,
n ¼ 14 placebo) met criteria for nonresponse, received
open-label IVIG at week 6, and were re-evaluated at week
12 (Table 3). The overall effect size for improvement after
open-label administration of IVIG was 1.79 (95%
CI ¼ 1.07
2.51; paired t23 ¼ 6.93, p < .0001) (Figure 2).
Participants who were randomized to placebo and received
IVIG at week 6 (n ¼ 14) displayed a 50% mean reduction
in CY-BOCS scores at week 12 (Cohen’s d ¼ 1.73, 95%
CI ¼ 0.76
2.71); participants who received a second IVIG
dose at week 6 (n ¼ 10) displayed a 55% reduction in
CY-BOCS score (Cohen’s d ¼ 1.18, 95% CI ¼ 0.45
1.90).
Among those participants who did not receive an open-
label IVIG dose (n ¼ 10), additional mean improvement
in CY-BOCS Total Score of 46.3%  47.8% was observed
between week 6 and week 12, with a total improvement of
64.9%  34.9% from baseline (Table 3).
On average, improvements observed at week 12 were
maintained at week 24. Compared to baseline, CY-BOCS
Total scores were improved by 62%  33%, and 25 partici-
pants (74%) met response criteria at week 24.
Several baseline variables were explored as possible
predictors or moderators of response, including duration of
illness, antibiotic class and duration of prophylaxis, serum
and CSF CaMKII activity, and ANA positivity. Of these,
only baseline serum CaMKII and ANA positivity were
found to be potentially related to response. Neither
elevated CaMKII (defined as 130; p ¼ .06) nor positive
ANA (defined as 1; p ¼ .07) showed a statistically sig-
nificant main effect in the ANCOVA model, but their
combination (positive CaMKII and positive ANA, n ¼ 4 in
the placebo group and n ¼ 7 in the IVIG group) was pre-
dictive of outcome regardless of randomization (F ¼ 7.68,
p ¼ .0095). The interaction between randomization and
positive ANA, positive CaMKII, or their combination was
not significant in any case (CaMKII, p ¼ .31; ANA, p ¼ .51;
combination, p ¼ .44), indicating that these variables were
not moderators of treatment. Raw CY-BOCS Total score
data for participants grouped by ANA and CaMKII status
(negative/negative, mixed, and positive/positive) are
shown in Figure 3.
Adverse Events
Adverse effects of IVIG infusion included one possible
allergic reaction, which resolved without complication, and
a number of minor discomforts (Tables S1 and S2, available
online).

Exploratory Analyses
Five participants (3 IVIG, 2 placebo) developed culture-
positive GAS infection during the study period despite
prophylactic antibiotics. As GAS infections are reported
to affect OCD symptoms in patients with PANDAS, a sec-
ondary ANCOVA was conducted with these participants
removed; primary outcomes remained nonsignificant
(F ¼ 1.36, p ¼ .25), although the effect size increased to 0.45
(95% CI ¼
0.28 to 1.18).
DISCUSSION
We report here the results of an investigation of IVIG
administration for the treatment of OCD symptoms of
PANDAS in a double-blind, placebo-controlled trial. No
significant difference in primary outcome measures was
observed in the blinded phase between participants
receiving IVIG or placebo. A significant mean reduction in
OCD symptoms was observed in participants who received

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME - NUMBER - - 2016 www.jaacap.org 5
WILLIAMS et al.

FIGURE 2 Unadjusted mean (SD) Children’s Yale
Brown
Obsessive Compulsive Scale (CY-BOCS) total scores
during the open-label phase. Note: DB ¼ double blind;
IVIG ¼ intravenous immunoglobulin; OL ¼ open label.
an open-label infusion of IVIG, regardless of the type of
infusion (placebo or IVIG) that they received in the blinded
phase of the trial. Adverse events in the trial were few,
and IVIG was generally well tolerated by the study
participants.
These results stand in contrast to the results of the
previous placebo-controlled trial, which demonstrated a
significant effect of IVIG therapy on OCD symptoms in
individuals with PANDAS.20 A number of factors may
explain this discrepancy. First, the placebo response
rate was higher, and the degree of improvement
from IVIG smaller, in the present trial compared to the
previous study, leading to a smaller than anticipated
effect size and insufficient power to detect the more
moderate observed effect. Although the double-blind
phase failed to show superiority of IVIG over placebo,
open-label administration of IVIG produced a 49%
reduction in OCD symptom severity. Further research is
needed to determine whether these benefits were related
to expectation bias, the natural course of the disorder, or
the immunomodulatory properties of IVIG. Analysis of
the study design of the present trial suggests that partic-
ipants may have been inadvertently biased toward
nonresponse during the double-blind portion, as only
those who had failed to respond to double-blind treat-
ment were eligible to receive IVIG at 6 weeks.
The initiation of prophylactic antibiotics during partici-
pants’ baseline evaluations may have further compromised
the between-group comparisons, as a recent case series re-
ported that antibiotic therapy alone produced symptomatic
improvements in children with PANDAS.37 In addition, we
observed a high rate of newly acquired GAS infections in
study participants despite reported adherence to antibiotic
prophylaxis, which may have further affected the efficacy of
the IVIG infusions.
The study’s power to detect between-group differences
was tempered by the high variability in individual
improvement after double-blind administration of IVIG.
Given that PANDAS is a clinical diagnosis, it is possible
that even in a rigorously screened cohort, patients who
meet criteria for PANDAS represent a heterogeneous
population with a variable response to immunomodulatory
interventions, including IVIG. Varying responses to IVIG
have been reported in previous trials of IVIG for autoim-
mune disorders such as Kawasaki disease, with response

FIGURE 3 Effect of randomization in subgroups based on the anti-nuclear antibody (ANA) and serum calcium calmodulin-
dependent protein kinase II (CaMKII). Note: Participants were grouped by baseline ANA and CaMKII results; those who were
positive on one and negative on the other are labeled Mixed (intravenous immunoglobulin [IVIG], n ¼ 7; Placebo, n ¼ 10); negative
on both are labeled ANA
, CaM
(IVIG, n ¼ 3; placebo, n ¼ 4); positive on both are labeled ANAþ, CaMþ (IVIG, n ¼ 7;
placebo, n ¼ 4). Figure reflects that the combination of ANA and CaMKII positivity was a significant exploratory predictor of
response, regardless of randomization (p ¼ .0095). CY-BOCS ¼ Children’s Yale-Brown Obsessive Compulsive Scale.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

6 www.jaacap.org VOLUME - NUMBER - - 2016

correlating with host inflammatory factors.38 In the present
trial, exploratory analyses revealed that improvement ten-
ded to be greater for children with elevations of both ANA
and CaMKII titers, suggesting that these serum biomarkers
might be useful in predicting clinical course and should be
considered in future study design. However, given that we
found no evidence of moderation of treatment effect, further
research is required before the titers can be used to guide
therapy, as has occurred in anti-N-methyl-D-aspartate
(NMDA) receptor autoimmune encephalitis.17
In addition to CaMKII and ANA titers, a variety of clin-
ical and demographic variables require systematic explora-
tion in PANDAS. For example, volumetric MRI scans have
revealed enlargements of the caudate, putamen, and globus
pallidus among patients with PANDAS,32 with normali-
zation after successful treatment with therapeutic plasma-
pheresis.29 Positron emission tomography (PET) and
[18F]-peripheral benzodiazepine receptor, a ligand that rec-
ognizes activated microglia, demonstrated basal ganglia
inflammation at baseline in patients with PANDAS and tics,
and normalization after IVIG treatment in one case.39 In
addition, abnormalities of sleep architecture, particularly
during rapid-eye movement (REM) sleep stage, have been
reported to occur in more than 80% of patients with this
disorder.40
These findings suggest that future investigations may
identify diagnostic and therapeutic biomarkers to define the
subgroup of individuals with PANDAS who are most likely
to benefit from immunomodulatory interventions. In addi-
tion, future studies of PANDAS will need to carefully
consider the role of immunomodulatory therapy in the
context of the natural history of the disorder, to more
definitively separate response to treatment from the episodic
nature of the disorder. Future research would also benefit
from considering the potential impacts of antibiotic treat-
ment and standard psychotherapeutic interventions (e.g.,
cognitive-behavioral therapy/exposure response preven-
tion, selective serotonin reuptake inhibitors). Future studies
of immunomodulatory treatments for PANDAS may require
enrolling only patients who do not respond to antibiotic
therapy, and should be powered for a more modest thera-
peutic effect from IVIG than the current trial.
Multiple independent investigations have provided
support for a subtype of OCD originating from inflamma-
tory or autoimmune mechanisms.41 However, this ran-
domized controlled trial failed to provide definitive support
for clinical use of IVIG in PANDAS. Despite these negative
REFERENCES
1. Kalra SK, Swedo SE. Children with obsessive-compulsive disorder: are
they just “little adults”? J Clin Invest. 2009;119:737-746.
2. Garvey MA, Giedd J, Swedo SE. Topical review: PANDAS: The search for
environmental triggers of pediatric neuropsychiatric disorders. Lessons
from rheumatic fever. J Child Neurol. 1998;13:413-423.
3. Murphy TK, Snider LA, Mutch PJ, et al. Relationship of movements and
behaviors to Group A Streptococcus infections in elementary school
children. Biol Psychiatry. 2007;61:279-284.
4. Williams KA, Swedo SE. Post-infectious autoimmune disorders:
Sydenham’s chorea, PANDAS and beyond. Brain Res. 2014;1617:
144-154.
JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY
VOLUME - NUMBER - - 2016
RANDOMIZED CONTROLLED TRIAL OF IVIG IN PANDAS
results during the double-blind phase, the benefit witnessed
from open-label IVIG administration, combined with pre-
vious reports of the therapeutic effect of IVIG therapy
in patients with PANDAS, suggest that future controlled
trials that address the limitations of the current study are
warranted. &
Accepted July 28, 2016.
This article was reviewed under and accepted by deputy editor John T.
Walkup, MD.
Drs. Williams, Leckman, King, and Mss. Grantz and Katsovich are with the
Child Study Center at Yale University School of Medicine, New Haven, CT.
Dr. Williams is also with the Pediatric Neuropsychiatry and Immunology Pro-
gram at Massachusetts General Hospital, Boston, MA. Drs. Swedo, Farmer,
Grant, Hommer, and Ms. D’Souza are with the Pediatrics and Developmental
Neuroscience Program of the National Institute of Mental Health, Bethesda,
MD.
Grifols Laboratories supplied intravenous immunoglobulin for the trial, as well
as financial support to the Yale investigators. Additional support was provided
by the National Institutes of Health (NIH) Bench-to-Bedside Program, and by
the Intramural Research Program of the National Institute of Mental Health
(NIMH) (NCT01281969; protocol 11-M-0058). Dr. Williams had full ac-
cess to all the data in the study and takes responsibility for the integrity of the
data analysis.
This work was prepared as part of some authors’ (Swedo, Farmer, Grant,
D’Souza, Hommer) official duties as federal government employees. The views
expressed in this manuscript do not necessarily represent the views of the
NIMH, NIH, the US Department of Health and Human Services (HHS), or the
United States Government.
The authors acknowledge and thank the patients, their families, and the nurses
and staff of the NIH Clinical Center.
Disclosure: Dr. Leckman has received grant or research support from the National
Institutes of Health, the UBS Optimus Foundation, and the Open Road Alliance.
He has served on the advisory boards/DSMB of the Brain and Behavior Research
Foundation, the National Organization for Rare Disorders, Fondazione Child, the
European Muilticentre Tics in Children Studies, and How I Decide. He has
authored the Yale Global Tic Severity Scale (YGTSS) assessment tool, which is
open access. He has received honoraria from the European Society for the Study
of Tourette Syndrome and the Brazilian Psychiatric Association. He has received
royalties from John Wiley and Sons, McGraw Hill, and Oxford University Press.
He has received travel expenses from the University of Illinois e Chicago, Cornell
Weill Medical College, the Medical University of South Carolina, Rutgers Uni-
versity, the British Academy, and the Brazilian Psychiatric Association. He has
received additional support from Anne Çocuk E gitim Vakfi (AÇEV; Mother Child
Education Foundation) and private donors. Drs. Williams, Swedo, Farmer, Grant,
Hommer, King, and Mss. Grantz, D’Souza, and Katsovich report no biomedical
financial interests or potential conflicts of interest.
Correspondence to Susan E. Swedo, MD, 10 Center Drive, MSC 1255,
Bethesda, MD 20892; e-mail: swedos@mail.nih.gov
0890-8567/$36.00/ª2016 Published by Elsevier Inc. on behalf of the
American Academy of Child and Adolescent Psychiatry.
http://dx.doi.org/10.1016/j.jaac.2016.06.017
5. Kirvan CA, Swedo SE, Snider LA, Cunningham MW. Antibody-mediated
neuronal cell signaling in behavior and movement disorders.
J Neuroimmunol. 2006;179:173-179.
6. Kirvan CA, Swedo SE, Heuser JS, Cunningham MW. Mimicry and
autoantibody-mediated neuronal cell signaling in Sydenham chorea.
Nature Med. 2003;9:914-920.
7. Kirvan CA, Cox CJ, Swedo SE, Cunningham MW. Tubulin is a neuronal target
of autoantibodies in Sydenham’s chorea. J Immunol. 2007;178:7412-7421.
8. Hallett JJ, Harling-Berg CJ, Knopf PM, Stopa EG, Kiessling LS. Anti-
striatal antibodies in Tourette syndrome cause neuronal dysfunction.
J Neuroimmunol. 2000;111:195-202.
www.jaacap.org 7
 WILLIAMS et al.
9. Taylor JR, Morshed SA, Parveen S, et al. An animal model of Tourette’s
syndrome. Am J Psychiatry. 2002;159:657-660.
10. Lotan D, Benhar I, Alvarez K, et al. Behavioral and neural effects of intra-
striatal infusion of anti-streptococcal antibodies in rats. Brain Behav
Immun. 2014;38:249-262.
11. Brimberg L, Benhar I, Mascaro-Blanco A, et al. Behavioral, pharmaco-
logical, and immunological abnormalities after streptococcal exposure:
a novel rat model of Sydenham chorea and related neuropsychiatric
disorders. Neuropsychopharmacology. 2012;37:2076-2087.
12. Yaddanapudi K, Hornig M, Serge R, et al. Passive transfer of
streptococcus-induced antibodies reproduces behavioral disturbances
in a mouse model of pediatric autoimmune neuropsychiatric disor-
ders associated with streptococcal infection. Mol Psychiatry. 2010;15:
712-726.
13. Hoffman KL, Hornig M, Yaddanapudi K, Jabado O, Lipkin WI. A murine
model for neuropsychiatric disorders associated with group A beta-
hemolytic streptococcal infection. J Neurosci. 2004;24:1780-1791.
14. Dileepan T, Smith ED, Knowland D, et al. Group A streptococcus
intranasal infection promotes CNS infiltration by streptococcal-specific
Th17 cells. J Clin Invest. 2016;126:303-317.
15. Garvey MA, Snider LA, Leitman SF, Werden R, Swedo SE. Treatment
of Sydenham’s chorea with intravenous immunoglobulin, plasma
exchange, or prednisone. J Child Neurol. 2005;20:424-429.
16. Hughes RA, Swan AV, van Doorn PA. Intravenous immunoglobulin for
Guillain-Barre syndrome. Cochrane Database Syst Rev. 2014;9:Cd002063.
17. Dalmau J, Lancaster E, Martinez-Hernandez E, Rosenfeld MR, Balice-
Gordon R. Clinical experience and laboratory investigations in patients
with anti-NMDAR encephalitis. Lancet Neurol. 2011;10:63-74.
18. Heitink-Polle KM, Nijsten J, Boonacker CW, de Haas M, Bruin MC.
Clinical and laboratory predictors of chronic immune thrombocytopenia
in children: a systematic review and meta-analysis. Blood. 2014;124:
3295-3307.
19. Wong PH, White KM. Impact of immunoglobulin therapy in pediatric
disease: a review of immune mechanisms. Clin Rev Allergy Immunol.
2015 Jul 4 [Epub ahead of print].
20. Perlmutter SJ, Leitman SF, Garvey MA, et al. Therapeutic plasma
exchange and intravenous immunoglobulin for obsessive-compulsive
disorder and tic disorders in childhood. Lancet. 1999;354:1153-1158.
21. Latimer ME, L’Etoile N, Seidlitz J, Swedo SE. Therapeutic plasma
apheresis as a treatment for 35 severely ill children and adolescents with
pediatric autoimmune neuropsychiatric disorders associated with strep-
tococcal infections. J Child Adolesc Psychopharmacol. 2015;25:70-75.
22. Kovacevic M, Grant P, Swedo SE. Use of intravenous immunoglobulin in
the treatment of twelve youths with pediatric autoimmune neuropsy-
chiatric disorders associated with streptococcal infections. J Child Ado-
lesc Psychopharmacol. 2015;25:65-69.
23. Elia J, Dell ML, Friedman DF, et al. PANDAS with catatonia: a case
report. Therapeutic response to lorazepam and plasmapheresis. J Am
Acad Child Adolesc Psychiatry. 2005;44:1145-1150.
24. Tucker DM, Leckman JF, Scahill L, et al. A putative poststreptococcal case
of OCD with chronic tic disorder, not otherwise specified. J Am Acad
Child Adolesc Psychiatry. 1996;35:1684-1691.
25. Hachiya Y, Miyata R, Tanuma N, et al. Autoimmune neurological dis-
orders associated with group-A beta-hemolytic streptococcal infection.
Brain Dev. 2013;35:670-674.
8 www.jaacap.org
26. Frankovich J, Thienemann M, Rana S, Chang K. Five youth with pediatric
acute-onset neuropsychiatric syndrome of differing etiologies. J Child
Adolesc Psychopharmacol. 2015;25:31-37.
27. Gerardi DM, Casadonte J, Patel P, Murphy TK. PANDAS and comorbid
Kleine-Levin syndrome. J Child Adolesc Psychopharmacol. 2015;
25:93-98.
28. Murphy TK, Storch EA, Lewin AB, Edge PJ, Goodman WK. Clinical
factors associated with pediatric autoimmune neuropsychiatric disorders
associated with streptococcal infections. J Pediatr. 2012;160:314-319.
29. Swedo SE, Leckman J, Rose NR. From research subgroup to clinical
syndrome: modifying the PANDAS criteria to describe PANS (pedi-
atric acute-onset neuropsychiatric syndrome). Pediatr Therapeut. 2012;
2:113.
30. Scahill L, Riddle MA, McSwiggin-Hardin M, et al. Children’s Yale-Brown
Obsessive Compulsive Scale: reliability and validity. J Am Acad Child
Adolesc Psychiatry. 1997;36:844-852.
31. Bernstein GA, Victor AM, Pipal AJ, Williams KA. Comparison of clin-
ical characteristics of pediatric autoimmune neuropsychiatric disorders
associated with streptococcal infections and childhood obsessive-
compulsive disorder. J Child Adolesc Psychopharmacol. 2010;20:
333-340.
32. Leckman JF, Riddle MA, Hardin MT, et al. The Yale Global Tic Severity
Scale: initial testing of a clinician-rated scale of tic severity. J Am Acad
Child Adolesc Psychiatry. 1989;28:566-573.
33. Storch EA, Murphy TK, Geffken GR, et al. Reliability and validity of the
Yale Global Tic Severity Scale. Psychol Assess. 2005;17:486-491.
34. Guy W. Clinical Global Impressions. In: ECDEU Assessment Manual For
Psychopharmacology. Rockville, MD: National Institute of Mental
Health; 1976:218-222.
35. Poznanski EO, Grossman JA, Buchsbaum Y, Banegas M, Freeman L,
Gibbons R. Preliminary studies of the reliability and validity of the
Children’s Depression Rating Scale. J Am Acad Child Psychiatry. 1984;23:
191-197.
36. Singer HS, Mascaro-Blanco A, Alvarez K, et al. Neuronal antibody bio-
markers for Sydenham’s chorea identify a new group of children with
chronic recurrent episodic acute exacerbations of tic and obsessive
compulsive symptoms following a streptococcal infection. PLoS One.
2015;10:e0120499.
37. Snider LA, Lougee L, Slattery M, Grant P, Swedo SE. Antibiotic pro-
phylaxis with azithromycin or penicillin for childhood-onset neuropsy-
chiatric disorders. Biol Psychiatry. 2005;57:788-792.
38. Sato S, Kawashima H, Kashiwagi Y, Hoshika A. Inflammatory cytokines
as predictors of resistance to intravenous immunoglobulin therapy in
Kawasaki disease patients. Int J Rheum Dis. 2013;16:168-172.
39. Kumar A, Williams MT, Chugani HT. Evaluation of basal ganglia and
thalamic inflammation in children with pediatric autoimmune neuro-
psychiatric disorders associated with streptococcal infection and Tourette
syndrome: a positron emission tomographic (PET) study using 11C-[R]-
PK11195. J Child Neurol. 2015;30:749-756.
40. Gaughan T, Buckley A, Hommer R, et al. Rapid eye movement sleep
abnormalities in children with pediatric acute-onset neuropsychiatric
syndrome (PANS). J Clin Sleep Med. 2016;12:1027-1032.
41. Pearlman DM, Vora HS, Marquis BG, et al. Anti-basal ganglia antibodies
in primary obsessive-compulsive disorder: systematic review and meta-
analysis. Br J Psychiatry. 2014;205:8-16.
JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY
VOLUME - NUMBER - - 2016
 RANDOMIZED CONTROLLED TRIAL OF IVIG IN PANDAS

FIGURE S1 Consolidated Standards of Reporting Trials (CONSORT) diagram. Note: CY-BOCS ¼ Children’s Yale
Brown
Obsessive Compulsive Scale; NIMH ¼ National Institute of Mental Health; IVIG ¼ intravenous immunoglobulin; OCD ¼ obsessive-
compulsive disorder; PANDAS ¼ pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME - NUMBER - - 2016 www.jaacap.org 8.e1
WILLIAMS et al.

TABLE S1 Adverse Events, n (%)

Acute Phase
Week 1 (Week 1
6)

IVIG PLA IVIG PLA

n ¼ 17 n ¼ 18 n ¼ 17 n ¼ 18
Headache 8 (47) 3 (17) 2 (12) 1 (6)
Sore throat 1 (6) 2 (11) 3 (18) 1 (6)
Stomach or abdominal 3 (18) 1 (6) 2 (12) 1 (6)
discomfort
Nausea (vomiting)a 4 (24) 1 (6) 5 (29) 1 (6)
Muscle/bone/joint pain 3 (18) 2 (11) 1 (6) 1 (6)
Tiredness/fatigue 2 (12) 1 (6) 4 (24) 2 (11)
Anxiety 2 (12) 2 (11) 1 (6) 2 (11)
Note: Groups did not differ on rate of any adverse effect. Rate of headache at
1 week approached significance (c2 ¼ 3.75, p ¼ .053). IVIG ¼ intrave-
nous immunoglobulin; PLA ¼ placebo.
a
Three participants in the IVIG group at week 1 and two participants in the
IVIG group during the acute phase reported nausea and vomiting; the
remainder of the reported events in both groups was nausea only.

TABLE S2 Open-Label Adverse Effects, n (%)

Open Label
Week 7 (Week 7
12)

IVIG None IVIG None

n ¼ 25 n ¼ 10 n ¼ 25 n ¼ 10
Muscle/bone/joint pain 0 (0) 1 (10)a 3 (12) 0 (0)
Nausea (vomiting)b 3 (12) 0 (0) 2 (8) 1 (10)
Tiredness/fatigue 3 (12) 0 (0) – –
Stomach or abdominal 4 (16) 0 (0) 2 (8) 0 (0)
discomfort
Appetite decrease 5 (20) 0 (0) – –
Headache 7 (28) 1 (10) 4 (16) 0 (0)
Note: No significant differences between groups. IVIG ¼ intravenous
immunoglobulin.
a
Event received a rating of 3 (“required consultation by medical
professional”).
b
Two participants in the IVIG group at week 7 reported nausea and vom-
iting; the remainder of the reported events in both groups was
nausea only.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

8.e2 www.jaacap.org VOLUME - NUMBER - - 2016