Increased midgestational IFN-γ, IL-4 and IL-5 in women bearing a child with autism: A case-control study

The study sample was based on the Early Markers for Autism (EMA) Study. The EMA Study is a population-based, nested case-control study designed to evaluate biologic markers of susceptibility and exposure in archived maternal midpregnancy and neonatal blood specimens from the same mother-baby pairs. The study subjects are women residing in Orange County, California, who were pregnant in 2000 and 2001 and enrolled in the state's Prenatal Expanded AFP Screening Program [15]. Briefly, three groups were identified: mothers of children with autism spectrum disorder (ASD), mothers of children with DD but not ASD and mothers of GP children. Children with ASD or DD were ascertained from client records obtained from the Regional Center of Orange County. This is one of the 21 regional centers operated by the California Department of Developmental Services (DDS), which are designed to coordinate services for persons with autism and other developmental disabilities. Clients receiving DDS services for ASD or suspected ASD were ascertained as possible subjects for inclusion in this study. Other subjects with moderate to profound developmental disabilities but not ASD (specifically children with an IQ <70 based on standardized tests) were ascertained as other possible DD cases. Diagnoses were confirmed by expert review of all ASD and DD cases as described in the next subsection. GP controls were randomly sampled from the birth certificate files and frequency-matched to ASD cases by sex, birth month and birth year at a 2:1 ratio. All past or current DDS/regional center clients were excluded from the GP population. All study procedures were approved by the institutional review boards of the California Health and Human Services Agency and Kaiser Permanente Northern California.

After subjects were ascertained from the Regional Center Orange County, ASD and DD diagnoses were verified by trained medical record abstractors following a protocol initially developed by the Metropolitan Atlanta Developmental Disabilities Surveillance Program [23]. Medical record abstractors compiled detailed diagnostic and clinical data from the Regional Center Orange County records for all children initially ascertained as possibly having ASD or DD. Expert clinical review of abstracted data was then conducted by a developmental pediatrician to confirm the ASD or DD diagnoses for this study using Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria. Children with ASD were further categorized on the basis of disease onset type and cognitive status (presence or absence of intellectual disability (ID) in addition to ASD) using the DSM-IV criteria. Onset type was determined by parental report or clinical observations derived from chart reviews and categorized as "early" (no statement of loss of social and/or language skills, or skill plateau without actual loss), "regressive" (clear loss of previously acquired language and/or social skills) or unable to discern from the record review. The ID determination among ASD subjects was based on composite scores on standardized cognitive and functional tests (with ID: composite score <70; without ID: all scores ≥70 or some scores <70 and others ≥70; unknown: no standardized scores in chart). The final analytic sample consisted of 84 children with ASD (Table 1), 49 children with ID but not ASD (DD) (mild [DSM-IV Text Revision 317] (n = 20), moderate [DSM-IV Text Revision 318.0] (n = 12), severe [DSM-IV Text Revision 318.1] (n = 11), profound [DSM-IV Text Revision 318.2] (n = 3), unspecified [DSM-IV Text Revision 319] (n = 3), known etiology (n = 29) or unknown etiology (n = 20)) and 159 GP controls.

Maternal midpregnancy serum specimens were retrieved from the Project Baby's Breath prenatal screening specimen archive maintained by the California Genetic Disease Screening Program, at the California Department of Public Health. As part of the screening program, venous blood was collected at 15 to 19 weeks' gestation in serum separator tubes by obstetrical care service providers and underwent expanded α-fetoprotein screening at a single regional laboratory, typically within seven days of collection (median time = 3 days). During transit via the US Postal Service to the regional screening laboratory, no effort was made to control the temperature of the specimens. After testing, leftover specimens were kept under refrigeration for 1 to 2 days and then stored at -20°C. Aliquots of the samples used for this study were stored at -80°C until use with no freeze-thaws prior to testing. All samples were exposed to the same collection and storage protocols.

Serum concentrations of 17 cytokines and chemokines, including eotaxin, granulocyte macrophage colony-stimulating factor (GM-CSF), IFN-γ, IL-10, IL-12, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IFN-γ-induced protein 10, macrophage inflammatory protein (MIP)-1α, MIP-1β, RANTES and TNF-α were determined using a commercially available multiplex bead-based kit (BioSource Human Bead Kit; Invitrogen, Carlsbad, CA, USA). The assay was carried out in accordance with the protocols provided by the manufacturer. Briefly, 50 μL of serum was incubated with anti-cytokine-conjugated beads in a 96-well filter-bottomed plate on a plate shaker. After two hours, the beads were washed using a vacuum manifold, and biotin-conjugated detection antibodies were added for one-hour incubation. Following a repeat of the washing step, beads were incubated with streptavidin phycoerythrin for 30 minutes. The plates were then read on a Bio-Plex 100 system (Bio-Rad Laboratories, Hercules, CA, USA) and analyzed using Bio-Plex Manager software (Bio-Rad Laboratories) with a five-point standard curve. Reference samples were run on each plate to determine assay consistency.

The distribution of the cytokine concentration values was skewed, and natural log transformation was used to approximate normality. To examine the association of cytokine levels with developmental outcomes after adjustment for possible confounders, we fit separate logistic regression models for ASD vs. GP, ASD vs. DD and DD vs. GP. Case vs. control status was regressed on natural log-transformed cytokine levels with adjustment for several covariates related to the maternal blood draw (maternal weight and gestational age at time of draw) or associated with autism in previous epidemiologic studies (maternal age, race, ethnicity and country of origin). Separate models were run for each cytokine. For all cytokine values that were below the limit of detection (LOD), we assigned a value of LOD/2 prior to log transformation. Fisher's exact tests were used to determine whether there were differences between groups in the proportion of subjects falling within the LOD for each cytokine. Finally, the correlation of individual cytokine levels was tested separately for cases and controls on the basis of the Pearson correlation coefficient.

The maternal immune system is uniquely regulated during pregnancy to optimize the gestational environment of the fetus [24]. Primarily, it must be poised to protect the mother and fetus from pathogens and other potentially harmful environmental factors. Simultaneously, robust maternal cellular immune responses must be suppressed to avoid rejection of the fetus as a foreign allograft [25]. Evidence suggests that under normal circumstances, pregnancy shifts the maternal immune system toward a more tolerant, low inflammatory state that involves decreased production of cytokines such as IL-6 and IFN-γ and increased production of the more regulatory cytokines, including IL-4, IL-5 and IL-10 [26‐29]. Mothers of children with ASD and DD demonstrated increased levels of the inflammatory cytokines IFN-γ and IL-6, respectively, which may be indicative of an atypical immune state during gestation.

Dramatic changes in maternal immune homeostasis during pregnancy (in response to infection, disease or other environmental influences) are associated with complications such as miscarriage, preterm delivery and preeclampsia [30]. Maternal immune responses can also affect the development of the fetal nervous system [31, 32]. Epidemiological studies have suggested that prenatal infections may be related to neurological disorders such as schizophrenia and autism [33, 34]. Furthermore, animal models have repeatedly demonstrated that robust maternal immune responses during pregnancy can alter offspring behavior and brain histopathology [17, 19, 35‐37]. Various cytokines, including IL-6 and IL-2, have been shown to mediate some of these effects [18, 20, 21]. We suggest that atypical maternal immune function during pregnancy may be related to ASD or DD outcomes among children.

In the current study, mothers bearing a child with ASD had significantly increased levels of serum IFN-γ, IL-4 and IL-5. IFN-γ, the most dramatically elevated cytokine in this population, is involved in aspects of defense against intracellular pathogens, tumor surveillance, autoimmunity, allergy and pregnancy. Peripheral IFN-γ levels are low in healthy pregnancies, and increased production of peripheral IFN-γ is often related to complications such as preeclampsia [38]. IFN-γ is produced by a subset of activated T cells, though its primary source is natural killer (NK) cells. During pregnancy, a unique population of IFN-γ-producing NK cells accumulates in the uterus, where they have a vital role in placental development [39]. The increased levels of serum IFN-γ observed in mothers bearing a child with ASD may be indicative of imbalanced immune function at the maternal-fetal interface, which could lead to improper placental formation and thereby incomplete support of fetal development. Alternatively, increased serum IFN-γ may be due to peripheral immune activity, including a response to infection, or to an immune-mediated disorder. Interestingly, a recent examination of archived neonatal blood spots from children with ASD revealed no elevation in pathogen-specific immunoglobulin G (IgG) levels relative to controls, suggesting that prenatal infection may not be involved [40].

Epidemiological data previously reported by our group indicated a higher prevalence of allergy and asthma during pregnancy in mothers of children with ASD [11]. That 2005 study examined physician-diagnosed medical conditions in over 2,500 women enrolled in the Kaiser Permanente Medical Care Program. Interestingly, the midpregnancy cytokine profile we describe in the ASD group in the present study (increased IL-4, IL-5 and IFN-γ) may be consistent with an allergic asthma clinical phenotype [41‐44]. While IL-4 and IL-5 are known to be upregulated in allergic asthma, IFN-γ is generally thought to be downregulated [45]. However, several reports have shown increased production of IFN-γ in addition to IL-4 and IL-5 in allergic asthma [41‐44]. This has also been observed during pregnancy in women with asthma, when higher levels of IFN-γ correlated with worsening maternal and fetal health [43]. Future studies should address the impact of prenatal allergy and asthma on fetal neurodevelopment and further explore the possible connection to behavioral disorders.

We noted that mothers bearing a child with DD but not ASD demonstrated a different midgestational immune profile. The risk of DD was associated with higher levels of the cytokines IL-2, IL-4 and IL-6. Interestingly, IL-6 is part of a cytokine family with well-defined neurological impacts [46]. Extensive evidence links prenatal immune responses involving increased production of inflammatory cytokines such as IL-6 to pregnancy complications and neurological abnormalities among offspring [30, 47]. Mouse models have shown that prenatal exposure to IL-6 or mimics of infectious agents such as LPS or poly(I:C) can induce behavioral changes and brain abnormalities among offspring [18, 19, 21, 48]. Similarly, prenatal exposure to high levels of IL-2 has been shown to induce behavioral differences in murine models [20]. On the basis of these animal studies, it has been suggested that prenatal exposure to these inflammatory conditions may be relevant to the development of autism [20, 48]. However, our results showed elevated IL-6 and IL-2 in mothers bearing a child with DD but not autism. Therefore, we propose that elevated levels of these cytokines have a global effect on neurodevelopment, resulting in cognitive impairment but not necessarily autism.

The mechanism by which maternal cytokines affect fetal neurodevelopment is unclear, though the central nervous system (CNS) and immune system interact extensively during fetal development and throughout life. Neuroimmune cross-talk is facilitated by shared signaling pathways and commingling of cellular and soluble components from each system [49, 50]. Evidence suggests that immune components such as cytokines can affect aspects of neurogenesis, neuronal migration and synaptic plasticity, depending on the timing and level of exposure [46, 51, 52]. The developing CNS is especially vulnerable to immunological and environmental influences because the fetus has an immature blood-brain barrier and limited capacity for detoxification and excretion [53]. Under normal circumstances, the placenta forms a barrier between the maternal and fetal circulation, though some maternal immune factors, including IgG and IL-6, are permitted to cross the placenta [22, 54, 55]. When passage of maternal immune components is blocked, the placenta may respond to entities at the maternal-fetal interface and alter the fetal compartment [56]. For example, IFN-γ is not known to pass between the maternal and fetal circulation, though IFN-γ and its receptors are expressed by maternal and fetal cells at the maternal-fetal interface [57]. Therefore, maternal immune components can interact with fetal development both directly and indirectly. The specific neurodevelopmental impact of the different cytokine profiles observed in the present study remains to be determined.

Although our study provides valuable, temporally relevant information regarding prenatal immune status and the child's developmental outcome, a few primary limitations must be considered. First, immune activation in the peripheral blood is not necessarily representative of immune activity at the maternal-fetal interface. Examination of more spatially relevant immune parameters would require placental or amniotic specimens, which were not available in this study. Despite this limitation, the archived serum samples examined provide valuable insight into global maternal immune status during a developmentally relevant window. Second, this study is cross-sectional, as the serum specimens represent a single time point between 15 and 19 weeks of gestation. Maternal immune activity is likely to change throughout pregnancy, and the gestational immune environment outside 15 to 19 weeks' gestation is also developmentally relevant. Future longitudinal studies will provide a more complete picture of the relationship between maternal immune activation throughout pregnancy and fetal neurodevelopment. Third, data regarding the occurrence of infection, allergy and asthma were not available for the population included in this study, so the factors underlying the observed cytokine profiles are unknown. Replication studies are required to further verify the findings described herein. Fourth, it should be noted that the study groups were matched on the basis of child characteristics rather than maternal characteristics. However, two of the three offspring characteristics used for matching, birth month and birth year, relate directly to an important characteristic of the mother (that is, season during midpregnancy) that may influence cytokine levels through their association with seasonal illness. While the remaining covariates were adjusted for in multivariable logistic regression analysis, there is a possibility that our results could be biased because of residual confounding. Finally, diagnoses were made on the basis of medical record abstraction rather than via direct assessment. While we are confident in the consistency and accuracy of our expert medical record review, we recognize that this approach is likely to introduce some level of error, in part because of the differences in the amount and specificity of documentation in the medical records. Our future analyses will involve direct observation and diagnosis of subjects.