Elevated plasma cytokines in autism spectrum disorders provide evidence of immune dysfunction and are associated with impaired behavioral outcome

Brain, Behavior, and Immunity xxx (2010) xxx–xxx

Contents lists available at ScienceDirect

Brain, Behavior, and Immunity

journal homepage: www.elsevier .com/locate /ybrbi

Short Communication

Elevated plasma cytokines in autism spectrum disorders provide evidenceof immune dysfunction and are associated with impaired behavioral outcome

Paul Ashwood a,f,⇑, Paula Krakowiak b, Irva Hertz-Picciotto b,f, Robin Hansen c,f, Isaac Pessah d,f,Judy Van de Water e,f

a Department of Medical Microbiology and Immunology, University of California, Davis, CA, USAb Department of Public Health Sciences, Division of Epidemiology, University of California, Davis, CA, USAc Department of Pediatrics, School of Medicine, University of California, Davis, CA, USAd Department of Molecular Biosciences, School of Veterinary Medicine, University of California, Davis, CA, USAe Division of Rheumatology, Allergy and Clinical Immunology, University of California, Davis, CA, USAf The Medical Investigation of Neuodevelopmental Disorders (M.I.N.D) Institute, UC Davis Health System, Sacramento, CA, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 6 May 2010Received in revised form 24 July 2010Accepted 6 August 2010Available online xxxx

Keywords:AutismCytokinesBehaviorImmunologyRegression

0889-1591/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.bbi.2010.08.003

⇑ Corresponding author at: Department of MicrobMedical Investigation of Neuodevelopmental Disord50th Street, Sacramento, CA 95817, USA. Fax: +1 916

E-mail address: [email protected] (P. Ashwo

Please cite this article in press as: Ashwood, P., eare associated with impaired behavioral outcom

Autism spectrum disorders (ASD) are characterized by impairment in social interactions, communicationdeficits, and restricted repetitive interests and behaviors. A potential role for immune dysfunction hasbeen suggested in ASD. To test this hypothesis, we investigated evidence of differential cytokine releasein plasma samples obtained from 2 to 5 year-old children with ASD compared with age-matched typicallydeveloping (TD) children and children with developmental disabilities other than autism (DD). Partici-pants were recruited as part of the population based case-control CHARGE (Childhood Autism Risks fromGenetics and Environment) study and included: 97 participants with a confirmed diagnosis of ASD usingstandard assessments (DSM IV criteria and ADOS, ADI-R), 87 confirmed TD controls, and 39 confirmed DDcontrols. Plasma was isolated and cytokine production was assessed by multiplex Luminex™ analysis.Observations indicate significant increases in plasma levels of a number of cytokines, including IL-1b,IL-6, IL-8 and IL-12p40 in the ASD group compared with TD controls (p < 0.04). Moreover, when theASD group was separated based on the onset of symptoms, it was noted that the increased cytokine levelswere predominantly in children who had a regressive form of ASD. In addition, increasing cytokine levelswere associated with more impaired communication and aberrant behaviors. In conclusion, using largernumber of participants than previous studies, we report significantly shifted cytokine profiles in ASD.These findings suggest that ongoing inflammatory responses may be linked to disturbances in behaviorand require confirmation in larger replication studies. The characterization of immunological parametersin ASD has important implications for diagnosis, and should be considered when designing therapeuticstrategies to treat core symptoms and behavioral impairments of ASD.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

Autism Spectrum Disorders (ASD) are a heterogeneous group ofneurodevelopmental disorders characterized by severe impair-ments in social interaction and communication, and restricted, ste-reotyped interests that manifest in early childhood (AmericanPsychiatric Association, 2000). Recent epidemiologic data suggestthat approximately 1% of children are diagnosed with an ASD(MMWR, 2009). In the majority of cases, the etiology of ASD isnot known and likely involves complex interactions between ge-netic, epigenetic and environmental factors. Recent papers have

ll rights reserved.

iology and Immunology, Theers (M.I.N.D) Institute, 2805,703 0367.od).

t al. Elevated plasma cytokinese. Brain Behav. Immun. (2010

described links between genes that encode for immune-relatedproteins and ASD, suggesting that abnormalities in the immunesystem may influence aspects of brain development and synapticfunctions that negatively impact clinical outcomes relevant toASD (reviewed in Enstrom et al., 2009a). Taken together withwell-established reports of cytokine mediated influences on neuro-nal function, differentiation, migration, proliferation, and behav-ioral impairments in animal models, there is an emerging view ofsynergistic relationships between immune dysfunction and geneticpredisposition that contribute to a subset of ASD cases.

Altered immune responses in individuals with ASD have beenreported for nearly 40 years including the presence of self-reactiveantibodies to brain and CNS proteins (Cabanlit et al., 2007; Con-nolly et al., 2006; Silva et al., 2004; Todd et al., 1988; Vojdaniet al., 2002; Wills et al., 2009) and evidence for increased neuroin-flammation in brain and CNS specimens obtained from individuals

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003

2 P. Ashwood et al. / Brain, Behavior, and Immunity xxx (2010) xxx–xxx

with ASD (Vargas et al., 2005; Li et al., 2009; Garbett et al., 2008).Several studies have shown peripheral immune abnormalities inchildren with ASD including abnormal or skewed T helper cellcytokine profiles (Ashwood and Wakefield, 2006; Gupta et al.,1998), decreased lymphocyte numbers (Ashwood et al., 2003), al-tered T cell responses to mitogens and recall responses (Molloyet al., 2006), an imbalance of serum immunoglobulin levels (En-strom et al., 2009b), NK cell activation (Enstrom et al., 2009c), in-creased monocyte responses (Enstrom et al., 2010) and increasedlevels of complement components (Corbett et al., 2007). Taken to-gether, these findings support alterations in the immune responsesin a significant proportion of children with ASD.

Few studies have addressed if cytokine patterns differ in the seraor plasma in ASD (Ashwood et al., 2008a,b; Croonenberghs et al.,2002; Enstrom et al., 2008a; Okada et al., 2007; Singh et al. 1991;Singh, 1996; Sweeten et al., 2004; Zimmerman et al., 2005) and ofthese only three have attempted to evaluate whether cytokine lev-els are associated with core deficits of ASD or impairments in asso-ciated behaviors and/or onset patterns of ASD (Ashwood et al.,2008a,b; Enstrom et al., 2008a; Okada et al., 2007). A subset of thesestudies have demonstrated increased levels of cytokines that can in-duce inflammation in ASD, such as IFNc or IL-12, (Sweeten et al.,2004; Singh, 1996), or a decreased production of cytokines that neg-atively regulate inflammation, such as TGFb1 (Ashwood et al.,2008a,b; Okada et al., 2007). However, varied experimental designs,diagnostic criteria, ages of the probands and control populations(including the comparison of children with ASD with adult controls)have confounded interpretation of the results. Moreover, smallsample sizes and the use of non-clinically assessed siblings of chil-dren with ASD as controls have dampened the ability to detect dif-ferences in cytokine patterns, thwarting scientific progress towardsa consensus regarding whether particular cytokine patterns reflectan inherent immune dysfunction in ASD.

Based on the abnormal immune dysfunction observed in ASD,we herein evaluated a comprehensive panel of inflammatorycytokines associated with general immune activation in plasmasamples from a large series of well-characterized participantsenrolled in a population based case-control study. To better definethe immune status of children with ASD, cytokine levels wereassessed in ASD children 24–60 months and compared withtypically developing children and children with developmentaldisabilities other than ASD who were of the same age. In addition,cytokine profiles in children with ASD were investigated for anyassociations with clinical behavioral outcomes.

2. Methods

2.1. Subjects

Participants in the study were enrolled in the CHARGE (Child-hood Autism Risks from Genetics and Environment) study con-ducted at the UC Davis M.I.N.D. Institute (Hertz-Picciotto et al.,2006). The study protocols including recruitment and behavioralassessments have been described in detail (Ashwood et al.,2008a,b; Hertz-Picciotto et al., 2006; Enstrom et al., 2009b). In brief,after clinical evaluations, participants were placed in one of threegroups: (1) diagnosed with autism spectrum disorders; (2) diag-nosed with developmental delay but not ASD; or (3) confirmed astypically developing controls. Two hundred and twenty-three(223) children were selected based on available volumes of plasmafrom consecutively recruited participants. Participants were not dif-ferent in age or sex ratios. All children were medication free and ingood health at time of blood draw. Participants included 97 childrenwith ASD (median age 3.4 (interquartile range 2.9–4.3), 84 males),87 TD controls (3.4 (2.8–4.1), 71 males) and 39 DD controls (3.5(3.0–4.1), 28 males). This study was approved by the UC Davis insti-

Please cite this article in press as: Ashwood, P., et al. Elevated plasma cytokinesare associated with impaired behavioral outcome. Brain Behav. Immun. (2010

tutional review board and complied with all requirements regard-ing human subjects. Parents gave informed consent. Autismspectrum disorder diagnosis was performed using gold-standardassessments based on Diagnostic and Statistical Manual of MentalDisorders, Fourth Edition (DSM-IV) (American Psychiatric Associa-tion, 2000) criteria by qualified trained clinicians at the M.I.N.D.Institute. To confirm and further evaluate the diagnosis, all ASDparticipants completed the Autism Diagnostic Interview-Revised(ADI-R) and the Autism Diagnostic Observation Schedule (ADOS).Children confirmed as ASD were divided into subgroups of non-regressive or regressive autism based on the stage of developmentthe child began to display autistic behavior (Hansen et al., 2008).Children with ASD were classified as having no-regression if thechild exhibited traits of autism from infancy, and regressive autismif they had typical early development and later lost function in lan-guage and/or social interactions based on ADI-R scores (languageloss (Q11) or any social skills loss (Q25)). There were 40 childrenin the ASD who had regression and 53 without regression, it wasnot possible to confirm onset in the remaining four ASD partici-pants. Children from the TD and DD groups were screened forautism traits using the Social Communication Questionnaire(SCQ). For all children, adaptive function was assessed by parentalinterview using the Vineland Adaptive Behavior Scales (VABS).Additional measures of cognitive ability were determined usingthe Mullen Scales of Early Learning (MSEL) and abnormal behaviorprofiles using the Aberrant Behavior Checklist (ABC).

2.2. Cytokine analysis

For each subject peripheral blood was collected in acid-citrate-dextrose Vacutainers (BD Biosciences; San Jose, CA), centrifuged at2300 rpm for 10 min, and the plasma harvested. Plasma was ali-quoted and stored at �70 �C until cytokine analysis. The quantifi-cation of the cytokines GM-CSF, IFN-c, IL-1b, IL-2, IL-4, IL-5, IL-6,IL-8, IL-10, IL-12(p40), IL-13, and TNF-a in the plasma was deter-mined using human multiplexing bead immunoassays (Biosource,Camarillo, CA) that are based on a sandwich immunoassay that uti-lize the Luminex� fluorescent-bead-based technology. Plasmasamples were run in concordance with the instructions of the kitprotocol. Briefly, 50 lL of plasma sample were incubated with anti-body-coupled beads. After a series of washes, a biotinylated detec-tion antibody was added to the beads, and the reaction mixturewas detected by the addition of streptavidin–phycoerythrin. Thebead sets were analyzed using a flow-based Luminex™ 100 sus-pension array system (Bio-Plex 200; Bio-Rad Laboratories, Inc.).Unknown sample cytokine concentrations were calculated byBio-Plex Manager software using a standard curve derived fromthe known reference cytokine concentrations supplied by the man-ufacturer. A five-parameter model was used to calculate final con-centrations and values are expressed in pg/ml. The sensitivity ofthis assay allowed the detection of cytokine concentrations withinthe following ranges: GM-CSF 15–9940 pg/ml; IFN-c 5–4740 pg/ml; IL-1b 15–6810 pg/ml; IL-2 6–7160 pg/ml; IL-4 5–4910 pg/ml;IL-5 3–5390 pg/ml; IL-6 3–5200 pg/ml; IL-8 3–6330 pg/ml; IL-105–4550 pg-ml; TNF-a 10–11,390 pg/ml. Concentrations obtainedbelow the sensitivity limit of detection (LOD) of the method werecalculated as LOD/2 for statistical comparisons. Values obtainedfrom the reading of samples that exceeded the upper limit of thesensitivity method were further diluted and cytokine concentra-tions calculated accordingly. Plasma aliquots had not undergoneany previous freeze/thaws cycle.

2.3. Statistical analysis

Descriptive statistics were computed for selected demographicvariables across diagnostic groups. Covariates of interest as possi-

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003

P. Ashwood et al. / Brain, Behavior, and Immunity xxx (2010) xxx–xxx 3

ble confounders included the child’s age at blood draw and gender.In analyses involving questionnaire and behavioral assessmentscore data, diagnostic group was also evaluated as a covariate. Inprimary analyses, natural log-transformed cytokine levels (out-come) were compared by group (predictor) in multiple linearregression models adjusted for child’s age and gender. P-valueswere corrected for multiple comparisons using the Benjamini–Hochberg False Discovery Rate. Secondary analyses examined theassociation between cytokine levels (predictor) and behavioralassessment scores. These multiple linear regression models wereadjusted for diagnostic group, as well as child’s age, gender andmultiple comparisons (as above). Findings with P-value < 0.05 aftercorrection were considered significant. All analyses were carriedout using SAS version 9.1 (SAS Inc., Cary, NC) and Prism 5 Software(GraphPad Software; San Diego, CA).

3. Results

Levels of IL-6 and IL-12p40 were significantly higher in childrenwith ASD compared with TD and DD controls (Table 1). Afteradjusting for child’s age and gender, levels of IL-6 were approxi-mately twofold higher in plasma collected from ASD children(median 20.3; interquartile range 9.5–43.5 pg/mL) when comparedwith TD (11.8; 3–29.2 pg/mL; p = 0.01), and elevated nearly sevenfold compared with DD controls (3; 3–18.8 pg/mL; p = 0.03). Levelsof IL-12p40 were significantly higher in plasma sampled from chil-dren with ASD (199.2; 162.4–277.9 pg/mL) compared with TD(171.7; 125.6–241.7 pg/mL; p = 0.04) and DD controls (169.7;119.9–221.9 pg/mL; p = 0.03). In addition, levels of IL-8 sampledfrom children with ASD were significantly higher (14.1; 3.3–22.1 pg/mL) compared with those from TD controls (3.9; 3–10.6 pg/mL; p = 0.002) and were elevated compared with DD con-trols but the difference did not reach statistical significance (4.9;3–19.6 pg/mL; p = 0.14). Similarly, IL-1b was significantly raisedin plasma samples from children with ASD (110.6; 25.7–

Table 1Comparison of plasma cytokine levels in children with autism spectrum disorders(n = 97), typically developing (n = 87) controls and children with developmentaldisabilities other than autism (n = 39). Data are presented as median and interquartileranges.

Typicallydeveloping

Developmentaldelay

Autism spectrumdisorders

IL-1b 62.8 46.1 110.6*

(15–148.8) (15–153.8) (25.7–245.8)IL-2 8 7.3 22.9

(6–83.9) (6–97.8) (6–89.3)IL-4 36.7 39.9 35.5

(16.8–83.3) (14.1–58.2) (18.2–66.2)IL-5 9.8 8.5 9.9

(3–17.1) (3–14.5) (3–14.9)IL-6 11.8 3 20.3*,#

(3–29.2) (3–18.8) (9.5–43.5)IL-8 3.9 4.9 14.1*

(3–10.6) (3–19.6) (3.3–22.1)IL-10 16.4 12.4 11.9

(5–38.8) (5–24.1) (5–33.8)IL-12 p40 171.7 169.7 196.2*,#

(125.6–241.7) (119.9–221.9) (162.4–277.9)IL-13 20.9 18 20

(11.5–36.1) (10–36.4) (10–38.7)GM-CSF 54.2 52.3 63

(15–92.8) (15–123.8) (32.7–118.6)IFNc 62.8 84.3 70.2

(23.4–172.5) (16.8–142.6) (25.9–135.8)TNFa 63.9 51.6 81.8

(20.7–128.1) (29.1–144.4) (34.5–204.7)

* p < 0.05 compared with typically developing controls.# p < 0.05 compared with developmentally delay controls.

Please cite this article in press as: Ashwood, P., et al. Elevated plasma cytokinesare associated with impaired behavioral outcome. Brain Behav. Immun. (2010

245.8 pg/mL) compared with TD controls (62.8; 15–148.8 pg/mL;p = 0.04), but after correcting for multiple comparisons, differencesdid not reach statistical significance when compared with DD con-trols (46.1; 15–153.8 pg/mL; p = 0.1), and may be due to the rela-tively fewer numbers of DD subjects in our study. Finally, therewas a trend showing higher, GM-CSF in children with ASD (63;32.7–118.6 pg/mL) compared with TD (54.2; 15–92.8 pg/mL) andDD controls (52.3; 15–123.8 pg/mL) but this was not significantafter correcting for multiple comparisons. There were no differ-ences in any cytokine measure between the TD and DD groupsand cytokine levels were not related to IQ (data not shown).

The children with ASD were further subdivided into those whoexhibited regression and those with early onset ASD (Table 2).There were no significant differences in psychological measuresbased on the ABC, ADI-R, ADOS, MSEL and VABS scores betweenASD children who had regressed and those that had not (data notshown). In general, cytokine levels were higher in children withregression compared with those that had early onset ASD (Table 2).Higher levels of IL-1b were measured in plasma of ASD childrenwith regression (median 144.3; range 64.6–353.3 pg/mL) com-pared with ASD children without regression (61; 15–241.8 pg/mL: p = 0.04). Observed levels of GM-CSF in children with ASDregression were also elevated (101.2; 45.8–133.4 pg/mL) comparedwith ASD children without regression (51.3; 15.4–81.2 pg/mL,p = 0.01). When compared with TD controls, the levels of IL-8 weresignificantly increased in the plasma of both regressive ASD children(p = 0.001) and non-regressive ASD children (p = 0.05). However,as compared with TD controls, levels of IL-1b (p = 0.002, IL-6(p = 0.008), GM-CSF (p = 0.05) and IL-8 (p = 0.001) were only signifi-cantly raised in plasma samples obtained from ASD children withregression (Table 2) and not in non-regressive ASD children.

We then examined whether there were associations betweencytokine levels and the severity of clinical behavioral outcomes.Based on ADI-R assessment scores in children with ASD, increased

Table 2Comparison of plasma cytokine levels in children with autism spectrum disorders(ASD) based on onset of symptoms. Children who were ASD and regressed (n = 40) arecompared with ASD children with no regression (n = 53) as based on definitionsdetermined using ADI-R. (Note in n = 4 subjects clinical regression could not bedetermined). Data are presented as median and interquartile ranges.

Typically developing Autism spectrum disorders

No-regression Regression

IL-1b 62.8 61 144.3*,�

(15–148.8) (15–241.8) (64.6–353.3)IL-2 8 17.7 19.3

(6–83.9) (6–63.7) (6–114.4)IL-4 36.7 28.8 39.4

(16.8–83.3) (14.9–65.3) (27.3–79.3)IL-5 9.8 9.2 11.5

(3–17.1) (3–12.5) (5.1–19.4)IL-6 11.8 15.1 32.6*

(3–29.2) (3–34.7) (14.2–62.5)IL-8 3.9 6.8* 14.5*

(3–10.6) (3–23.7) (3.7–21.7)IL-10 16.4 7.5 15.6

(5–38.8) (5–25.9) (5–34.66)IL-12 p40 171.7 192.3* 198.6*

(125.6–241.7) (160.4–272.7) (155–255.7)IL-13 20.9 14.1 29.4

(11.5–36.1) (10–40.1) (13.4–40.1)GM-CSF 54.2 51.3 101.2*,�

(15–92.8) (15.4–81.2) (45.8–133.4)IFNc 62.8 51.2 94.1

(23.4–172.5) (16.8–135.9) (45.4–194.6)TNFa 63.9 56.2 111.1

(20.7–128.1) (26.3–136.8) (43.2–208.8)

* p < 0.05 compared with typically developing controls.� p < 0.05 compared with no-regression autism spectrum disorder children.

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003

4 P. Ashwood et al. / Brain, Behavior, and Immunity xxx (2010) xxx–xxx

IL-4 was associated with greater impairments in non-verbal com-munication (t = 2.61, p = 0.03). There was also trends for associa-tions between more impaired social interactions measured byADI-R and plasma IL-b levels. (t = 2.15, p = 0.09) and IL-10(t = 1.82, p = 0.07) but these did not reach statistical significanceafter correction for multiple comparisons. There was a trend foran association between elevated IL-13 levels and the degree of im-paired social interactions assessed by ADOS module 1 but this didnot reach statistical significance after correction for multiple com-parisons (t = 2.15, p = 0.06). In addition, significant associationswere observed between cytokine levels and aberrant behaviorsmeasured in the ABC such that increased lethargy (t = 2.58,p = 0.03), stereotypy (t = 2.56, p = 0.02), hyperactivity (t = 2.89,p = 0.02) as assessed by ABC were associated with increased IL-8levels. Increased lethargy (t = 2.41, p = 0.04) and stereotypy(t = 2.55, p = 0.05) were associated with increased IL-12p40 levels.In addition, increased stereotypy (t = 2.21, p = 0.05) was associatedwith increased IL-6 levels and also with increased IL-1b levels(t = 3.07, p = 0.01). Finally, IL-8 was associated with MSEL scoresof visual reception (t = �2.65, p = 0.04), receptive language(t = �2.35, p = 0.04) and expressive language (t = �2.27, p = 0.04)as well as VABS scores of daily living (t = �2.32, p = 0.04) such thatas IL-8 decreased cognitive and adaptive ability improved.

4. Discussion

Although the etiology and pathogenesis of ASD remain unclear,it is suggested that there may be an association with immunedysfunction (Ashwood et al., in press; Enstrom et al., 2009a).Marked neuroinflammation and microglia cell activation is oneof the most prominent ASD characteristics (Vargas et al., 2005),along with increased circulating autoantibodies to brain or CNStissue (Wills et al., 2009). Further, altered T cell activation (Ash-wood et al., 2004; Ashwood and Wakefield, 2006; Molloy et al.,2006; Onore et al., 2009; Saresella et al., 2009), increased mono-cyte cell activation (Enstrom et al., 2010; Jyonouchi et al., 2005;Sweeten et al., 2004), increased basal levels of NK cell activationbut decreased response to stimulation (Enstrom et al., 2009c),skewed immunoglobulin profiles (Croonenberghs et al., 2002; En-strom et al., 2009b) and alterations in the levels of complementcomponents (Corbett et al., 2007) have all been described inASD and may provide pathogenic clues. The underlying mecha-nisms linking immune dysregulation and neuronal dysfunctionis not clear, but there is evidence indicating that certain cytokinescan impair neurodevelopment and behavior. Cytokines and theirreceptors are found in the healthy CNS during neural differentia-tion and plasticity and play important modulatory roles in theseprocesses. For example, neuropoietic cytokines, such as IL-6, candirectly alter neuron proliferation, survival, death, cortical neurondendrite development, neural activity, long-term potentiation andneurodevelopment that may impact behavior (Gadient and Patt-erson 1999; Gilmore et al., 2005; Juttler et al., 2002; Mehlerand Kessler 1998; Shi et al., 2003) while IL-1b and TNFa havebeen linked with oligodendrocyte toxicity, neurite growth andthe regulation of homeostatic synaptic plasticity in the hippocam-pus (Barker et al., 2001; Cacci et al., 2008; Munoz-Fernandez andFresno, 1998). Thus cytokine dysregulation may have importantbiological effects on neuronal development and activity that ad-versely affect behavior.

In the present study, we report statistically significant elevationof IL-1b, IL-6, IL-8 and IL-12p40 cytokine levels in the plasma ofASD children compared with TD controls. The elevated cytokinelevels seemed to be predominantly driven by the children with aregressive form of ASD who exhibited higher cytokine levels thanthose ASD children with no regression. These findings are compat-

Please cite this article in press as: Ashwood, P., et al. Elevated plasma cytokinesare associated with impaired behavioral outcome. Brain Behav. Immun. (2010

ible with altered levels of immunomodulatory factors in childrenwith regressive ASD compared to non-regressive ASD children(Ashwood et al., 2008b; Bu et al., 2006; Enstrom et al., 2008a; En-strom et al., 2008b). In addition, we found associations betweenthe plasma levels of several cytokines and severity of certain coremeasures of ASD as assessed by ADI-R (IL-4), as well as aberrantbehaviors as measured by ABC (IL-1b, IL-6, IL-8, IL-12p40), suchthat impairments in behavior were more pronounced as certaincytokine levels increased. In particular, several cytokines were ob-served to be associated with more impaired stereotypical behav-iors (IL-1, IL-6, IL-8 and IL-12p40). It is currently unclear howcytokines could affect stereotypical behaviors during childhoodin ASD and these data should be treated with caution until furthervalidation can be performed. Notably several other studies haveshown increased severity of impairments in social interactions,communication and stereotypical behaviors, as well as aberrantbehaviors, to be associated with altered levels of immune factorsincluding: TGF 1 (Ashwood et al., 2008a), macrophage inhibitoryfactor (Grigorenko et al., 2008), platelet–endothelial adhesion mol-ecule (Tsuchiya et al., 2007), immunoglobulin-G4 (Enstrom et al.,2009b), monocyte activation (Enstrom et al., 2010) and IL-23(Onore et al., 2009). These observations and our own results maysuggest that dysfunctional immune responses and/or activitymay affect core features of ASD as well as associated behaviors inchildren with ASD. Another possibility is that dysfunction in com-mon basic cellular processes, such as those involved in signaling,may be manifest as aberrations in both the immune and neuronalsystems. As ASD may encompass several distinct phenotypes, thepotential separation of ASD subgroups based on immunologicalparameters and/or associations with worsening behavior may haveimportant implications for diagnosis, as well as the design andmonitoring of therapeutic treatments of ASD. Further validationof the associations observed here between cytokine levels andbehaviors is necessary, including analysis of other features oftenassociated with ASD such as gastrointestinal problems, seizures,macrocephaly, cognitive impairments, and sleep disorders. Mostcritically, longitudinal studies beginning prior to diagnosis wouldprovide the best means for understanding whether these changesplay an etiologic role, have prognostic potential, or are purelyphenomenological.

The pattern in ASD of increased IL-6 and IL-1b cytokine levels,which are often described as archetypal pro-inflammatory cyto-kines, is consistent with in vitro evidence for higher productionof these same cytokines by cultured monocytes from children withASD when challenged with ligands from infectious agents (En-strom et al., 2010) and may thus represent an imbalance in the in-nate immune compartment favoring a breakdown of self-tolerance(Enstrom et al., 2009a). It should be noted that due to well docu-mented sensitivity issues with the Luminex assay, baseline levelsand detected levels of some cytokines are not the same as thoseobserved by ELISA techniques, in particular this is true of the IL-1b and TNFa results; however, additional experiments were ableto confirm between groups differences (data not shown). More-over, increased expression of gene transcripts for IL-6 and IL-1pathways are noted in postmortem specimens of the temporal cor-tex of the brain of individuals with ASD (Garbett et al., 2008) andincreased protein levels of IL-6, and IL-1b have also been foundin the brain and CSF of individuals with ASD (Vargas et al., 2005;Li et al., 2009). These cytokines appear to have been derived frommicroglial and astroglial cells and also implicate that disturbancesof the innate immune system are relevant in ASD (Vargas et al.,2005). Few studies have analyzed plasma cytokines levels in chil-dren with ASD; however, consistent with our results is a previouslyreported finding of increased IL-12 in plasma from ASD individualscompared with that from controls (Singh, 1996) which may sug-gest a TH1 plasma cytokine profile. In this study we also found in-

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003

P. Ashwood et al. / Brain, Behavior, and Immunity xxx (2010) xxx–xxx 5

creased IL-8 levels, a chemoattractant cytokine important ininflammatory process, a finding that corresponds to similar in-creases of IL-8 seen in the brain and CSF of individuals with ASD(Vargas et al., 2005; Li et al., 2009). Other studies have reportedcytokine dysregulation in ASD including increased IFNc, IL-2,TNFa, GM-CSF, IL-4, IL-5 in either the brain tissue, CSF, plasma, ser-um, cultured PBMC or cultured NK-cells from individuals with ASD(Croonenberghs et al., 2002; Enstrom et al., 2009c; Jyonouchi et al.,2005; Li et al., 2009; Molloy et al., 2006; Vargas et al., 2005). How-ever, other studies have failed to demonstrate any differences inplasma IFNc or TNFa levels (Singh, 1996; Sweeten et al., 2004),or have demonstrated decreased IFNc or IL-2 responses to stimula-tion in mononuclear cell cultures in individuals with ASD (Guptaet al., 1998). These discrepancies are most likely due to character-istics of the patient populations under investigation and/or mis-matching of the age of the cases and controls – with severalstudies comparing adult controls with ASD children – as well asthe analytical techniques, power of the statistical analysis and tis-sue/specimens used. Often further complicating these studies hasbeen the use of siblings of ASD children as controls. As ASD is highlyheritable, similar findings in analyses of the immune system in ASDchildren and their siblings most likely reflects shared immune sus-ceptibilities that are upstream of the CNS pathology (Ashwoodet al., 2008b; Saresella et al., 2009). A further issue highlighted bythe current study is the difference in the cytokine profile betweenthose children with no regression and those with regression.Although it is hard to ascertain from previous studies how manystudy subjects with regression were included, it may be differencesin the subgroups of ASD that lead to the variable reports forplasma cytokine levels reported in the literature thus far. Whetherincreased cytokine production occurs in all children with regressionor that the regressive group has a higher number of individuals withhigh cytokine production – a putative ‘‘high-immune responder”group – needs further investigation, as do the apparent associationsbetween cytokines and more impaired behaviors.

In summary, using a large number of participants from a popu-lation based case-control study, we demonstrate that there are sig-nificant increases in plasma IL-1b, IL-6, IL-8 and IL-12p40 levels inyoung children who have confirmed ASD as compared with con-firmed and age-matched typically developing children. These lev-els were predominantly driven by increases in cytokines inchildren who experienced loss of language or social skills. Further-more, elevations in these cytokine levels were associated withgreater severity in the core domains of ASD and with greaterimpairments in aberrant behaviors. These findings suggest thatongoing inflammatory responses may be linked to disturbancesin behavior, a conjecture that requires confirmation in a largerstudy, and most critically, in longitudinal investigations of thechanges in cytokine levels prior to the diagnosis or developmentof ASD symptoms and over the natural history of the disorder.The biological impact of increased cytokines in ASD children andtheir association with more impaired behaviors is intriguing andwarrants further consideration.

Acknowledgments

This study was funded by the NIEHS Children’s Center Grant (2P01 ES011269), US EPA STAR program Grant (R833292 andR829388), NIEHS CHARGE study (R01ES015359) Autism SpeaksFoundation, Peter Emch Foundation, The Boler Company Founda-tion and the Johnson Foundation. We would like to thank andthe staff of both the UC Davis M.I.N.D Institute and the CHARGEstudy, and Christina Kwong and Angela Tarver for their technicalsupport. The commitment of the families who took part in thesestudies is gratefully acknowledged.

Please cite this article in press as: Ashwood, P., et al. Elevated plasma cytokinesare associated with impaired behavioral outcome. Brain Behav. Immun. (2010

References

American Psychiatric Association, 2000. Diagnostic and Statistical Manual of MentalDisorders DSM-IV-TR, fourth edition. American Psychiatric AssociationPublishing Inc., Washington DC, USA.

Ashwood, P., Anthony, A., Pellicer, A.A., Torrente, F., Walker-Smith, J.A., Wakefield,A.J., 2003. Intestinal lymphocyte populations in children with regressiveautism: evidence for extensive mucosal immunopathology. J. Clin. Immunol.23, 504–517.

Ashwood, P., Anthony, A., Torrente, F., Wakefield, A.J., 2004. Spontaneous mucosallymphocyte cytokine profiles in children with autism and gastrointestinalsymptoms: mucosal immune activation and reduced counter regulatoryinterleukin-10. J. Clin. Immunol. 24 (6), 664–673.

Ashwood, P., Wakefield, A.J., 2006. Immune activation of peripheral blood andmucosal CD3+ lymphocyte cytokine profiles in children with autism andgastrointestinal symptoms. J. Neuroimmunol. 173 (1–2), 126–134.

Ashwood, P., Enstrom, A., Krakowiak, P., Hertz-Picciotto, I., Hansen, R.L., Croen, L.A.,Ozonoff, S., Pessah, I.N., Van de Water, J., 2008a. Decreased transforming growthfactor b1 in autism: a potential link between immune dysregulation andimpairment in clinical behavioral outcomes. J. Neuroimmunol. 204 (1–2), 149–153.

Ashwood, P., Kwong, C., Hansen, R., Hertz-Picciotto, I., Croen, L., Krakowiak, P.,Walker, W., Pessah, I.N., Van de Water, J., 2008b. Plasma leptin levels areelevated in autism: association with early onset phenotype? J. Autism Dev.Disord. 38 (1), 169–175.

Ashwood, P., Nguyen, D.V., Hessl, D., Hagerman, R.J., Tassone, F. in press. Plasmacytokine profiles in Fragile X subjects: Is there a role for cytokines in thepathogenesis? Brain Behav. Immun.

Barker, V., Middleton, G., Davey, F., Davies, A.M., 2001. TNFa contributes to thedeath of NGF-dependent neurons during development. Nat. Neurosci. 4 (12),1194–1198.

Bu, B., Ashwood, P., Harvey, D., King, I.B., Van de Water, J., Jin, L.W., 2006. Fatty acidcomposition of red blood cell phospholipids in children with autism.Prostaglandins Leukot. Essent. Fatty Acids 74 (4), 215–221.

Cacci, E., Ajmone-Cat, M.A., Anelli, T., Biagioni, S., Minghetti, L., 2008. In vitroneuronal and glial differentiation from embryonic or adult neural precursorcells are differently affected by chronic or acute activation of microglia. Glia 56(4), 412–425.

Cabanlit, M., Wills, S., Goines, P., Ashwood, P., Van de Water, J., 2007. Brain-specificautoantibodies in the plasma of subjects with autistic spectrum disorder. Ann.N. Y. Acad. Sci. 1107, 92–103.

Connolly, A.M., Chez, M., Streif, E.M., Keeling, R.M., Golumbek, P.T., Kwon, J.M.,Riviello, J.J., Robinson, R.G., Neuman, R.J., Deuel, R.M., 2006. Brain-derivedneurotrophic factor and autoantibodies to neural antigens in sera of childrenwith autistic spectrum disorders, Landau–Kleffner syndrome, and epilepsy. Biol.Psychiatry 59 (4), 354–363.

Corbett, B.A., Kantor, A.B., Schulman, H., Walker, W.L., Lit, L., Ashwood, P., Rocke,D.M., Sharp, F.R., 2007. A proteomic study of serum from children with autismshowing differential expression of apolipoproteins and complement proteins.Molecular Psychiatry 12 (3), 292–306.

Croonenberghs, J., Bosmans, E., Deboutte, D., Kenis, G., Maes, M., 2002. Activationof the inflammatory response system in autism. Neuropsychobiology 45 (1),1–6.

Enstrom, A., Onore, C., Hertz-Picciotto, I., Hansen, R., Croen, L., Van de Water, J.,Ashwood, P., 2008a. Detection of IL-17 and IL-23 in plasma samples of childrenwith autism. Am. J. Biochem. Biotechnol. 4, 114–120.

Enstrom, A., Onore, C., Tarver, A., Hertz-Picciotto, I., Hansen, R., Croen, L., Van deWater, J., Ashwood, P., 2008b. Peripheral blood leukocyte production of BDNFfollowing mitogen stimulation in early onset and regressive autism. Am. J.Biochem. Biotechnol. 4 (2), 121–129.

Enstrom, A.M., Van de Water, J.A., Ashwood, P., 2009a. Autoimmunity in autism.Curr Opin Investig Drugs 10, 463–473.

Enstrom, A., Krakowiak, P., Onore, C., Pessah, I.N., Hertz-Picciotto, I., Hansen, R.L.,Van de Water, J.A., Ashwood, P., 2009b. Increased IgG4 levels in children withautism disorder. Brain Behav. Immun. 23, 389–395.

Enstrom, A.M., Lit, L., Onore, C.E., Gregg, J.P., Hansen, R.L., Pessah, I.N., Hertz-Picciotto, I., Van de Water, J.A., Sharp, F.R., Ashwood, P., 2009c. Altered geneexpression and function of peripheral blood natural killer cells in children withautism. Brain Behav. Immun. 23 (1), 124–133.

Enstrom, A., Onore, C., Van de Water, J., Ashwood, P., 2010. Differential monocyteresponses to TLR ligands in children with autism spectrum disorders. BrainBehav. Immun. 24 (1), 64–71.

Gadient, R.A., Patterson, P.H., 1999. Leukemia inhibitory factor, interleukin 6, andother cytokines using the GP130 transducing receptor: roles in inflammationand injury. Stem Cells 17 (3), 127–137.

Garbett, K., Ebert, P.J., Mitchell, A., Lintas, C., Manzi, B., Mirnics, K., Persico, A.M.,2008. Immune transcriptome alterations in the temporal cortex of subjects withautism. Neurobiol Dis 30 (3), 303–311.

Gilmore, J.H., Jarskog, L.F., Vadlamudi, S., 2005. Maternal poly I:C exposure duringpregnancy regulates TNFa, BDNF, and NGF expression in neonatal brain and thematernal-fetal unit of the rat. J. Neuroimmunol. 159 (1–2), 106–112.

Grigorenko, E.L., Han, S.S., Yrigollen, C.M., Leng, L., Mizue, Y., Anderson, G.M.,Mulder, E.J., de Bildt, A., Minderaa, R.B., Volkmar, F.R., Chang, J.T., Bucala, R.,2008. Macrophage migration inhibitory factor and autism spectrum disorders.Pediatrics 122 (2), 438–445.

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003

6 P. Ashwood et al. / Brain, Behavior, and Immunity xxx (2010) xxx–xxx

Gupta, S., Aggarwal, S., Rashanravan, B., Lee, T., 1998. Th1- and Th2-like cytokines inCD4+ and CD8+ T cells in autism. J. Neuroimmunol. 85 (1), 106–109.

Hansen, R.L., Ozonoff, S., Krakowiak, P., Angkustsiri, K., Jones, C., Deprey, L.J., Le, D.N.,Croen, L.A., Hertz-Picciotto, I., 2008. Regression in autism: prevalence andassociated factors in the CHARGE Study. Ambul Pediatr (1), 25–31.

Hertz-Picciotto, I., Croen, L.A., Hansen, R., Jones, C.R., Van de Water, J., Pessah, I.N.,2006. The CHARGE study: an epidemiologic investigation of genetic andenvironmental factors contributing to autism. Environ. Health Perspect. 114,1119–1125.

Juttler, E., Tarabin, V., Schwaninger, M., 2002. Interleukin-6 (IL-6): a possibleneuromodulator induced by neuronal activity. Neuroscientist 8, 268–275.

Jyonouchi, H., Geng, L., Ruby, A., Zimmerman-Bier, B., 2005. Dysregulated innateimmune responses in young children with autism spectrum disorders: theirrelationship to gastrointestinal symptoms and dietary intervention.Neuropsychobiology 51 (2), 77–85.

Li, X., Chauhan, A., Sheikh, A.M., Patil, S., Chauhan, V., Li, X.-M., Ji, L., Brown, T., Malik,M., 2009. Elevated immune response in the brain of autistic patients. J.Neuroimmunol. 207 (1–2), 111–116.

Mehler, M.F., Kessler, J.A., 1998. Cytokines in brain development and function. Adv.Protein Chem. 52, 223–251.

MMWR Surveill. Summ, 2009. Prevalence of autism spectrum disorders – Autismand Developmental Disabilities Monitoring Network, United States, 2006.Autism and Developmental Disabilities Monitoring Network Surveillance Year2006 Principal Investigators; Centers for Disease Control and Prevention (CDC).58 (10): 1–20.

Molloy, C.A., Morrow, A.L., Meinzen-Derr, J., Schleifer, K., Dienger, K., Manning-Courtney, P., Altaye, M., Wills-Karp, M., 2006. Elevated cytokine levels inchildren with autism spectrum disorder. J. Neuroimmunol. 172 (1–2), 198–205.

Munoz-Fernandez, M.A., Fresno, M., 1998. The role of tumour necrosis factor,interleukin 6, interferon-c and inducible nitric oxide synthase in thedevelopment and pathology of the nervous system. Prog. Neurobiol. 56 (3),307–340.

Okada, K., Hashimoto, K., Iwata, Y., Nakamura, K., Tsujii, M., Tsuchiya, K.J., Sekine, Y.,Suda, S., Suzuki, K., Sugihara, G., Matsuzaki, H., Mori, N., 2007. Decreased serumlevels of transforming growth factor-b1 in patients with autism. Prog.Neuropsychopharmacol. Biol. Psychiatry 31 (1), 187–190.

Onore, C., Krakowiak, P., Enstrom, A., Hansen, R., Hertz-Picciotto, I., Van de Water, J.,Ashwood, P., 2009. Decreased cellular IL-23 production in children with autismspectrum disorders. J. Neuroimmunol. 216 (1–2), 126–129.

Please cite this article in press as: Ashwood, P., et al. Elevated plasma cytokinesare associated with impaired behavioral outcome. Brain Behav. Immun. (2010

Saresella, M., Marventano, I., Guerini, F.R., Mancuso, R., Ceresa, L., Zanzottera, M.,Rusconi, B., Maggioni, E., Tinelli, C., Clerici, M., 2009. An autistic endophenotyperesults in complex immune dysfunction in healthy siblings of autistic children.Biol. Psychiatry 66 (10), 978–984.

Shi, L., Fatemi, S.H., Sidwell, R.W., Patterson, P.H., 2003. Maternal influenza infectioncauses marked behavioral and pharmacological changes in the offspring. J.Neurosci. 23 (1), 297–302.

Singh, V.K., Warren, R.P., Odell, J.D., Cole, P., 1991. Changes of soluble interleukin-2,interleukin-2 receptor, T 8 antigen, and interleukin-1 in the serum of autisticchildren. Clin. Immunol. Immunopathol. 61 (3), 448–455.

Singh, V.K., 1996. Plasma increase of interleukin-12 and interferon-c. Pathologicalsignificance in autism. J. Neuroimmunol. 66 (1–2), 143–145.

Silva, S.C., Correia, C., Fesel, C., Barreto, M., Coutinho, A.M., Marques, C., Miguel, T.S.,Ataide, A., Bento, C., Borges, L., Oliveira, G., Vicente, A.M., 2004. Autoantibodyrepertoires to brain tissue in autism nuclear families. J. Neuroimmunol. 152 (1–2), 176–182.

Sweeten, T.L., Posey, D.J., Shankar, S., McDougle, C.J., 2004. High nitric oxideproduction in autistic disorder: a possible role for interferon-c. Biol. Psychiatry55 (4), 434–437.

Todd, R.D., Hickok, J.M., Anderson, G.M., Cohen, D.J., 1988. Antibrain antibodies ininfantile autism. Biol. Psychiatry 23 (6), 644–647.

Tsuchiya, K.J., Hashimoto, K., Iwata, Y., Tsujii, M., Sekine, Y., Sugihara, G., Matsuzaki,H., Suda, S., Kawai, M., Nakamura, K., Minabe, Y., Yagi, A., Iyo, M., Takei, N., Mori,N., 2007. Decreased serum levels of platelet-endothelial adhesion molecule(PECAM-1) in subjects with high-functioning autism: a negative correlationwith head circumference at birth. Biol. Psychiatry 62, 1056–1058.

Vargas, D.L., Nascimbene, C., Krishnan, C., Zimmerman, A.W., Pardo, C.A., 2005.Neuroglial activation and neuroinflammation in the brain of patients withautism. Ann. Neurol. 57 (1), 67–81.

Vojdani, A., Campbell, A.W., Anyanwu, E., Kashanian, A., Bock, K., Vojdani, E., 2002.Antibodies to neuron-specific antigens in children with autism: Possible cross-reaction with encephalitogenic proteins from milk, Chlamydia pneumoniae andStreptococcus group A. J. Neuroimmunol. 129 (1–2), 168–177.

Wills, S., Cabanlit, M., Bennett, J., Ashwood, P., Amaral, D.G., Van de Water, J., 2009.Detection of autoantibodies to neural cells of the cerebellum in the plasma ofsubjects with autism spectrum disorders. Brain Behav. Immun. 23 (1), 64–74.

Zimmerman, A.W., Jyonouchi, H., Comi, A.M., Connors, S.L., Milstien, S., Varsou, A.,Heyes, M.P., 2005. Cerebrospinal fluid and serum markers of inflammation inautism. Pediatr. Neurol. 33 (3), 195–201.

in autism spectrum disorders provide evidence of immune dysfunction and), doi:10.1016/j.bbi.2010.08.003