Cytokine levels and associations with symptom severity in male and female children with autism spectrum disorder

Autism spectrum disorders (ASDs) are complex, pervasive, and heterogeneous neurodevelopmental disorders of primarily unknown etiology and pathogenesis. Recent epidemiological studies estimate prevalence to be as high as 1 in 68 [1]. Diagnosis is currently based on clinical observation of behavioral features, defined as persistent deficits in social communication and interaction, and restricted, repetitive patterns of behavior, interests or activities; severity classifications are based on required levels of support [2]. While ASD is considered a lifelong condition [3], developmental trajectories vary significantly [4] with a range of prognoses [5]. Sex is believed to have an important role in development, with significantly more males diagnosed with ASD (4:1) [6] and different characteristic phenotypes presenting for females [7]. Many genes have been proposed to converge on common pathways affecting neuronal and synaptic homeostasis [8]. Environmental factors may also play a role in etiology through, for example, the impact of maternal immune activation or antibodies on a developing brain or by impacting a susceptible physiology [9‐11]. This complex etiology, manifesting as vast clinical heterogeneity, has been a catalyst for investigations aiming to characterize potential biological subtypes of ASD.

Immune system alterations related to immunogenetics, maternal immune activation, and a family history of autoimmune disorders, suggests a relationship between ASD and the immune system [12]. Accumulating evidence of alterations in central and peripheral immune system functioning supports the proposal that there is a subgroup of individuals with ASD who have some form of immune system dysregulation [11]. Previous work has highlighted significantly altered peripheral levels of primarily pro-inflammatory cytokines in ASD compared to healthy controls [13]. Levels of both pro-inflammatory and anti-inflammatory cytokines and chemokines have been associated with the severity of aberrant behavior and more impaired developmental and adaptive function [14‐16]. However, numerous differences in immune function have been identified in ASD, suggesting that these aberrations may be due to reduced immune system regulation rather than a dominance of one type of inflammatory response signal [11].

Despite extensive immunological evidence suggesting immune system aberrations in ASD, further research is required to clarify the relationship between immune profiles and ASD symptoms [12]. Such investigations may further knowledge about potentially useful biomarkers for ASD and the severity of symptoms, and provide clues for potential causative relationships and factors that might characterize subgroups of ASD. The pathophysiological significance of commonly identified comorbidities in ASD, such as gastrointestinal dysfunction and sleep disorder, with respect to the identified immune system aberrations, is also unknown [17‐19]. While understanding the influence of sex has been a recent research priority [20], partially because of the higher ratio of ASD diagnosis of males [21], to date, there has been limited investigation of potential sex differences in immunological profiles [22]. Sex-specific immunological aberrations have been described in a rodent model, with females exhibiting attenuation of induced behavioral and immunological alterations compared to males [23]. Therefore, there is a need to characterize the relationship between immune system dysfunction, specifically cytokine alterations, and severity of ASD symptoms, including disordered sleep and gastrointestinal dysfunction, between males and females.

The present study utilized a large, well-characterized repository of biological samples and multiplex assay techniques to investigate the relationship between cytokine levels and symptom severity of ASD. Cytokine profiles were investigated in relation to clinical traits, including variability in social functioning, sleep problems, and gastrointestinal dysfunction, along with the role of sex.

Samples for this analysis were collected from 144 participants, aged between 2 and 18 (mean 8.2 SD ± 4.2). Sex distribution was 78% male and 22% female. Eighty-three percent of participants identified as Caucasian. Additional demographic and clinical characteristics of participants are presented in Table 1. The variable clustering procedure resulted in the inclusion of six variables in the reduced set: age, sex, gastrointestinal symptoms, severity of ASD, sleep score, and SRS total score (Table 1). Assessment of covariate balance with respect to the two different plates or runs identified a run difference only due to sex (p = 0.042) (Additional file 1: Table S3).

Descriptive statistics for cytokine levels in plasma samples are shown in Table 2. Significance analysis of mixed effect modeling identified an interaction in cytokine levels with respect to two clinical covariates: age (p < 0.001) and the severity of ASD (p = 0.02; Table 3, Additional file 2: Table S4). There was no interaction between cytokine levels and remaining clinical variables (SRS, sleep, and gastrointestinal symptoms) or a significant main effect. The effect of sex was also investigated to determine if the interaction between cytokine expression and covariates were due to sex differences. Statistically significant effects in females were identified for interactions between cytokine levels and age (p = 0.03), severity of presentation (p < 0.001), and SRS score (p = 0.04), while in males there was a statistically significant effect between cytokine levels and age (p < 0.001; Table 3). Significance for the interaction effects with respect to age, severity, and SRS were then investigated for males and females. Reduced levels of IL-1β (p < 0.001), IL-8 (p < 0.001), MIP-1β (p = 0.006), PDGF-BB (p = 0.005), and VEGF (p = 0.03) were associated with increased severity of autism in females (Table 4). Reduced levels of PDGF-BB (p = 0.03) were associated with increased severity of autistic symptomatology in males (Table 4). With the exception of IL-15, cytokine levels appear to decrease with respect to increasing severity (Table 4). Higher levels of IL-8 (p = 0.03) were associated with higher scores on the SRS in females. Increased levels of IL-15 (p = 0.04) and reduced levels of IL-8 (p = 0.005) and MIP-1β (p = 0.009) were associated with increasing age in males, while reduced levels of IP-10 (p = 0.001) was associated with increasing age in females (Table 4).

Correlations between the levels of each of the 27 cytokines were analyzed (Additional file 1: Figures S3 and S4), with Spearman’s (ρ) correlations above 0.80 highlighted (Additional file 1: Tables S5-S7). A network visualization of the Spearman’s rank correlation matrix between cytokine expression values was created and also optimized for networks separated for sex (Fig. 1, Additional file 1: Figure S5). In the optimal graph networks, visual inspection suggests different network topologies for males and females.

To our knowledge, this study consists of the largest cohort of male and female ASD children to date investigating the relationship between clinical traits of ASD and cytokine levels. Overall, results show evidence of a significant association between decreased cytokine levels and increasing severity of ASD-related symptoms. In particular, decreased levels of the growth factor PDGF-BB were associated with increased severity of ASD-related symptoms. Results also suggested cytokine expression was moderated by the sex of participants. Exploratory analysis showed that decreased levels of an additional four cytokines (IL-1β, IL-8, MIP-1β, and VEGF) were associated with increased severity of presentation in females only. Additionally, increased levels of IL-8 were associated with greater caregiver reported social difficulties. An interaction effect for age was identified across both males and females, although there were different cytokines interacting with age in males and females. Increasing age in males was associated with higher levels of IL-15, and reduced levels of IL-8 and MIP-1β, while increasing age in females was associated with lower levels of IP-10. There were no associations between the other quantitative clinical traits, sleep behavior or gastrointestinal symptoms, and cytokine levels.

Previous studies have reported associations between levels of various cytokines and aberrant behaviors. For example, higher expression of cytokines, including IL-1β, IL-6, IL-8, and IL-12p40 were associated with more impaired stereotyped patterns of behavior [15] while increased levels of the chemokines MCP-1, RANTES, and eotaxin were associated with worsening behavioral symptoms and cognitive and adaptive ability [14]. Further, lower concentration levels of transforming growth factor beta1 (TGFβ1), a regulatory cytokine, has previously been associated with worsening behavior symptoms and lower adaptive behaviors [52]. These findings are in line with a heightened inflammatory state associated with the severity of autism-related symptoms, a cytokine profile similar to that previously identified in participants with ASD compared to healthy controls [13]. In the present study, higher levels of IL-8, which is involved in the recruitment of predominantly neutrophils to settings of tissue inflammation, were associated with higher severity of social impairments in females only. Our remaining findings revealed primarily decreased levels of significant cytokines associated with increased severity of autism-related symptoms, suggesting an altered cytokine profile may be associated with severity of autism-related symptoms. Unlike the possibility of a negative correlation of levels of PDGF-BB with severity of ASD in males and females identified in this study, a positive association between serum levels of PDGF-BB and worsening stereotyped patterns of behavior has previously been identified in a smaller male only cohort [53]. PDGF-BB has been found to be upregulated in healthy children between ages 7 to 17 compared to younger children and adults [54]. PDGF-BB has a complex profile, with PDGF-BB signaling implicated in development and a range of diseases through numerous roles and pathways [55]. Overall, with the exception of IL-15, a pro-inflammatory cytokine considered to be a central factor in the pathogenesis of celiac disease [56], cytokine levels appeared to decrease with respect to increasing symptom severity. However, for most of these cytokines, these differences did not reach significance at the individual cytokine level.

The association of reduced levels of IL-1β, IL-8, MIP-1β, and VEGF, with higher severity levels of ASD in females suggests sex differences in cytokine expression in children with ASD. This was also supported by the exploratory optimal network visualization graphs of cytokine correlations for males and females. Results from recent DNA analysis suggests that females could be more resilient to genetic insults due to carrying more extreme neurodevelopmental-related genetic mutations than males with the same symptoms [57]. Further investigation, particularly with larger female cohorts of ASD that use repeated testing over time, are now required to further support a potential sex-specific biology underlying different cytokine profiles between males and females with ASD. Our results also suggest that age appears to play an important moderating role in these effects; in males, an overall decline in cytokine levels was associated with increasing age. However, this effect was more mixed in females with only about half of the cytokine levels declining with older age. The physiological changes occurring during the onset of puberty and beyond may contribute to the variations of cytokine levels with age and sex. Age-related changes in cytokine expression in healthy children have been described elsewhere [54, 58, 59], although there is limited age-matched reference data of cytokine expression that can assist with more specific interpretation of these findings.

The results of the present study suggest that an altered cytokine response could be associated with severity of presentation, particularly in females. These results do not corroborate with some previous work that has identified associations between increased levels of pro-inflammatory cytokines and behavioral impairments in ASD, and are also not consistent with peripheral immune system abnormalities in ASD manifesting only as a pro-inflammatory bias [60]. The elucidation of possible neural circuitry mechanisms that are impacted by an aberrant peripheral immune system resulting in downstream changes in behavior are imperative. Evidence from investigations involving a rodent model deficient in adaptive immunity suggest that social deficits may result from impaired circuitry homeostasis derived from dysfunctional immunity [61], supporting the importance of this type of research in ASDs. The heterogeneity in ASD substantially increases the complexity associated with the identification of either markers of a pathological state that manifest in a subgroup of ASD or biomarkers of response to treatment that are specific and sensitive enough to be consistent measures of change.

The objective of this study was to utilize a well characterized cohort to establish if there is a relationship between cytokine levels and symptom severity of ASD and other clinical traits, including variability in social functioning, sleep problems, and gastrointestinal dysfunction. The levels of cytokines were analyzed using fluorescent data. There is a need for consensus in the literature about the use of absolute or fluorescent levels, and the treatment of concentrations obtained below the sensitivity LOD when presenting cytokine data. The use of absolute values has been more common in past autism research [14, 15]; however, the use of fluorescent levels is also an acceptable method [36]. While the concurrent analysis of a control group would have highlighted any cytokine differences between the ASD cohort and healthy controls, these differences have previously been well established [13] and would not facilitate the identification of potential relationships between the immune system aberrations and severity in ASD. The elucidation of these latter relationships have greater potential to contribute to knowledge about the relationship between symptoms and immune function in autism. Control groups typically vary widely on many characteristics beyond autism symptoms and their use in understanding the immune circuitry underpinning autism profiles may be debated. This cohort reflects the ratio of approximately four affected males to one affected female in the general population; however, larger cohorts of females are required to replicate and extend the sex-related findings. Furthermore, the duplication of samples across plates would have clarified the run difference that was linked to sex. The assessment of the severity of ASD-related symptoms undertaken using the ADOS was conducted by one assessor, strengthening the reliability of diagnoses and evaluations of severity. Additionally, ADOS assessments were conducted on the day of blood draw, therefore cytokine expression levels were relevant to that time point. While the ADOS-2 calibrated severity metric was approximated using a large sample [27], it is not representative of a population distribution of ADOS scores. Diagnostic and assessment measures, such as the ADOS, used in the ASD population are primarily behaviorally defined. There is a growing body of evidence implicating comorbid psychopathology and medical conditions in ASD [62‐64], highlighting the importance of comprehensive characterization of participants, particularly with respect to the assessment and reporting of all medical and mental health comorbidities. It is possible that there were participants with an undiagnosed comorbidity, allergies or concurrent illness at the time of blood draw that may elevate risk of immune system dysfunction and peripheral upregulation or downregulation of cytokines. Additionally, describing the onset of symptoms is important as previous research has suggested that regression may be associated with elevated cytokine expression [15].

The present study highlights the importance of an interdisciplinary approach to address the complexity of an autism presentation [65], particularly with respect to the pathological significance of immune system abnormalities. Longitudinal investigations in large cohorts with comprehensive, clinically relevant comprehensive biological data relating to medical comorbidities found in ASD, such as sleep disorders, and gastrointestinal and immune dysfunction, and controlling for variables that continually interact with the immune system, such as age, pubertal status, infections, and allergies, are required in order to thoroughly characterize the mechanisms of action behind these types of results. Further work is required to extend these findings to determine whether cytokines can be employed as either a sensitive marker of a pathological state-related to a subgroup in ASD or as a biomarker of response to treatment.