Table 4 Summary of EMA studies published as of October 2020

Immune-related factors

Croen et al. [50]

1

84 ASD, 49 ID, 160 GP

Antibody reactivity in maternal samples

Maternal mid-gestation antibody reactivity to human fetal brain protein differed by study group and by autism onset type, though most differences did not reach statistical significance. Reactivity to a band at 39 kDa was more common in mothers of children with autism (7%) compared with mothers of ID and GP controls (0% and 2%, p = 0.09 and 0.07, respectively). Reactivity to bands at 39 kDa and 73 kDa was found only in mothers of children with early onset autism (n = 3)

Goines et al. [31]

1

84 ASD, 49 ID, 159 GP

Cytokine and chemokines in maternal samples

Differing cytokine profiles observed in ASD and ID groups relative to GP controls: Elevated levels of IFN-γ, IL-4, IL-5 in midgestation maternal serum were significantly associated with a 50% increased risk of ASD (AOR = 1.51, 95% CI 1.12, 2.03, for IL-4 and similar for IL-5) while elevated IL-2, IL-4 and IL-6, and GM-CSF and chemokine MIP-1α, were significantly associated with an increased risk of ID without autism (AOR = 2.18, 95% CI 1.24, 3.85 for IL-4 and AORs ranging from 1.22 to 1.72 for others). In both ASD and ID IL-10 was elevated relative to GP controls

Grether et al. [54]

1

84 ASD, 49 ID, 159 GP

Immunoglobulin levels in maternal and newborn samples

Inverse associations between immunoglobulin levels and ASD: Higher Toxo IgG associated with lower risk of ASD in both maternal and newborn specimens (AOR for newborn = 0.25, 95% CI 0.08, 0.78, similar for maternal though attenuated). Overall lower immunoglobulin levels associated with higher ASD risk, but most did not reach statistical significance. No associations seen for ID

Zerbo et al. [32]

1

84 ASD, 49 ID, 159 GP

Cytokine and chemokines in newborn samples

Cytokines were not detected in most newborn samples, regardless of case or control status. However, certain chemokines were associated with ASD and ID: MCP-1 was elevated and RANTES was decreased in ASD cases compared to GP controls; MIP-1α and RANTES were both decreased in children with ID compared to GP controls

Zerbo et al. [49]

1, 2

500 ASD, 235 ID, 580 GP

CRP in maternal samples

Maternal CRP levels were lower in mothers of ASD cases compared with GP controls. Maternal CRP levels in the upper quartile were associated with a decreased risk of ASD (AOR = 0.58, 95% CI 0.38, 0.89). No difference was found between maternal CRP levels of ID and GP controls

Jones et al. [55]

2

415 ASD, 188 ID, 428 GP

Cytokines and chemokines in maternal samples

Mothers of children with ASD + ID had significantly elevated levels of a number of cytokines and chemokines, including GM-CSF, IFN-γ, IL-1α, and IL-6, compared to mothers of children with either ASD without ID, those with ID, or GP controls. Mothers of children with either ASD-without ID or with ID had significantly lower levels of the chemokines IL-8 and MCP-1 compared to mothers of GP controls

Heuer et al. [56]

2

370 ASD, 140 ID, 378 GP

Cytokines and chemokines in newborn samples

Children with ASD had significantly increased neonatal levels of interleukin-6 (IL-6) and IL-8 compared with GP controls. In particular, higher IL-8 was associated with ASD with early onset. Children with ASD also had significantly higher levels of eotaxin-1, interferon-γ, and IL-12p70 compared to children with ID. No significant differences were noted between the ID and GP groups

Additional endogenous factors

Croen et al. [50]

1

84 ASD, 49 ID, 159 GP

BDNF in maternal and newborn samples

Prenatal and neonatal BDNF levels were not significantly associated with ASD or ID

Yau et al. [34]

1

84 ASD, 49 ID, 159 GP

Thyroid stimulating hormone levels in maternal and newborn samples

Inverse associations between ASD and TSH levels in maternal serum samples (ASD vs. GP: AOR 0.33 95% CI 0.12–0.91, Early Onset ASD vs. GP: 0.31 95% CI 0.10–0.98). Results for thyroid levels in newborn blood samples were similar though not significant (ASD vs. GP: AOR = 0.61, 95% CI 0.18–2.04). Among children with ID vs. GP, relationships between maternal TSH (AOR = 0.9, 95% CI 0.02–0.42) and neonatal TSH (AOR = 0.36, 95% CI 0.08, 1.61) were similar to ASD vs. GP findings

Ames et al. [33]

2

518 ASD, 145 ID, 399 GP

Thyroid stimulating hormone in newborn samples

No association between neonatal TSH levels and ASD vs. GP nor ID vs GP. Among ASD sub-phenotypes, a suggestive inverse trend was observed between ASD with regression and TSH, though the association was only statistically significant in the highest TSH quartile (AOR: 0.50, 95% CI: 0.26–0.98)

Windham et al. [37]

2

534 ASD, 181 ID, 421 GP

Neonatal vitamin D levels

No association between newborn 25OHD levels and ASD nor ID, but potential differences by sex and ethnicity observed. In non-Hispanic whites, as well as in male children, a protective effect of maternal 25OHD was observed (AOR for non-Hispanic whites = 0.82, 95%CI 0.69–0.98 per 25 nmol/L; similar in males), and suggestion of non-linear effects (inverted j-shaped) for the latter. In females, an increased risk with higher 25OHD was observed

Windham et al. [71]

2

534 ASD, 181 ID, 421 GP

Maternal vitamin D levels

No overall associations between maternal 25OHD levels and ASD nor ID. Non-linear, inverted j-shape pattern between maternal 25OHD and ASD (p = 0.009 for linearity), and suggestion of interactions with sex and ethnicity (AOR AOR = 0.82, 95%CI 0.69–0.9 for non-Hispanic Whites, similar for males), were observed, as for neonatal vitamin D

Environmental chemicals

Yau et al. [83]

1

84 ASD, 49 ID, 159 GP

Mercury levels in maternal and newborn samples

No statistically significant associations found with either maternal or newborn levels, though there were non-significant elevations in odds of ASD with higher levels of mercury, and odds of ASD relative to ID were higher for higher quartiles of mercury

Lyall et al. [3]

2

545 ASD, 181 ID, 418 GP

Organochlorine chemicals (OC pesticides and PCBs) in maternal samples

General pattern of increased odds of ASD with higher maternal PCB levels, particularly for certain PCB congeners (138/158, 153, 170, and 180; AOR for Q4 v Q1 = 1.82, 95% CI 1.10, 3.02 for PCB153, similar though attenuated for most others). No significant associations were seen for ASD and the two pesticides examined (p,p’-DDE and trans-nonachlor). Results were similar for ID, though the 3rd quartile of p,p’-DDE showed a significantly increased risk of ID relative to the first (1.89, 95% CI 1.03, 3.44), though no trend was suggested

Lyall et al. [39]

2

545 ASD, 181 ID, 418 GP

PBDEs in maternal samples

General pattern of reduced odds of ASD with higher maternal brominated flame retardant levels; significant associations were observed for certain congeners (BDE-153, as well as the sum across BFR congeners, AOR 0.54, 95% CI 0.36, 0.80). In analyses of ASD stratified by sex, estimates for all BFR congeners except BB153 suggested positive associations for female children, but inverse associations for male children. Results were similar for ID

Lyall et al. [38]

2

553 ASD, 189 ID, 433 GP

PFASs in maternal samples

Associations with maternal PFASs were null to inverse; those with the highest levels of PFOA had reduced odds of ASD (Q4 vs. Q1 AOR = 0.62; 95% CI: 0.41, 0.93). Associations were similar for ID

Hamra et al. [79]

2

491 ASD, 155 ID, 373 GP

EDCs listed above

Applying a Bayesian mixture approach, no significant associations with EDCs and ASD nor ID were observed, even for chemicals identified in prior analyses of individual chemicals. AORs were all close to the null

Volk et al. [82]

2

379 ASD, 164 ID, 414 GP

Air pollution and cytokines

Observed several relationships between air pollution exposure (PM_2.5, PM₁₀, O₃, NO₂,) during the prior month and maternal serum cytokine and chemokine levels measured mid-gestation. However, no direct relationships of air pollutants and neurodevelopmental outcomes (ASD+/I ID, ID only), nor was there evidence of mediation by maternal mid-pregnancy immune response

Genetic factors^b

Tsang et al. [41]

2

390 ASD, 400 GP for maternal and 385 ASD, 379 GP for neonatal

Transgenerational genetic effects

Associations with SNPs in or near several genes previously implicated in autism (including NLGN4X, RAPGEF4, RORA, FAM135B and CNTNAP2) were identified in mothers of ASD cases, though no results reached genome-wide significance. Transgenerational epistasis analysis found several suggestive associations (at the p < 10⁻⁴ level)

Desachy et al. [45]

2

349 ASD and 351 GP

Maternal CNVs

Higher autosomal burden of large, rare CNVs in females in ASD families (though not unique to ASD). Case mothers had more CNVs (both deletions and duplications) than control mothers, although this difference was not significant for duplications

Traglia et al. [42]

2

390 ASD, 400 GP for maternal and 385 ASD, 379 GP for neonatal

Genes and EDC levels

Maternal circulating levels of BB-153, BDE-47, -100, -153 and their sum were significantly controlled by common genetic factors, and maternal genome-wide significant associations were observed between p,p’-DDE and PBDE levels and a SNP (rs7259965 which maps within the CYP2B6 locus). In addition, mid-gestational levels of BB-153, BDE-47, -99, -100, -153 and the sum of PBDE congeners were more strongly regulated by fetal genetic factors than maternal genetics

Traglia et al. [43]

2

390 ASD, 400 GP for maternal and 385 ASD, 379 GP for neonatal

Genes and immune markers

Results supported a large contribution of genetic variation in regulating cytokine/chemokine status during pregnancy and at birth. Neonatal levels of immune molecules were not correlated with mid-gestational maternal levels. 17 specific maternal and fetal loci were identified as contributing to cytokine and chemokine status at the two different time points. Distinct maternal loci in novel regions were independently associated with cytokine/chemokine levels in newborns, and fetal loci in novel regions were independently associated with maternal cytokine/chemokine levels. Increased IL-8 was associated with ASD status only in the presence of a specific maternal SNP genotype

Traglia et al. [44]

2

507 ASD, 179 ID, 400 GP for maternal and 385 ASD, 379 GP for neonatal

Genes and vitamin D

Evidence for a genome-wide significant missense variant located in the GC gene (previously identified as part of the vitamin D pathway), influencing neonatal vitamin D levels. Novel loci from maternal and fetal genetics emerged at suggestive levels near genes important for immune function. For ID, an association between decreased neonatal vitamin D levels and mothers with a specific genotype was observed