Folic Acid and Autism: A Systematic Review of the Current State of Knowledge

Folic acid has been identified to be integral in rapid tissue growth and cell division during fetal development. Different studies indicate folic acid’s importance in improving childhood behavioral outcomes and underline its role as a modifiable risk factor for autism spectrum disorders. The aim of this systematic review is to both elucidate the potential role of folic acid in autism spectrum disorders and to investigate the mechanisms involved. Studies have pointed out a potential beneficial effect of prenatal folic acid maternal supplementation (600 µg) on the risk of autism spectrum disorder onset, but opposite results have been reported as well. Folic acid and/or folinic acid supplementation in autism spectrum disorder diagnosed children has led to improvements, both in some neurologic and behavioral symptoms and in the concentration of one-carbon metabolites. Several authors report an increased frequency of serum auto-antibodies against folate receptor alpha (FRAA) in autism spectrum disorder children. Furthermore, methylene tetrahydrofolate reductase (MTHFR) polymorphisms showed a significant influence on ASD risk. More clinical trials, with a clear study design, with larger sample sizes and longer observation periods are necessary to be carried out to better evaluate the potential protective role of folic acid in autism spectrum disorder risk.


Introduction
Autism spectrum disorders (ASDs) are complex neurodevelopmental disorders, characterized by social and communication impairments, sensory hyper-sensitivity, and difficulties adjusting to unexpected change, as well as restricted interests and repetitive behaviors. ASDs are estimated to affect up to 3% of children in the United States with an overall prevalence of 13.4 per 1000 children aged 4 years in 2010, 15.3 in 2012, and 17.0 in 2014 [1].
Many etiologic and risk factors, including genetic and environmental influences, have been proposed over time for autism and ASDs. Genetic studies have focused, for several years, on single gene mutations or small groups of genes but, despite about 600 genes considered, only a small number of genes, such as Fragile X, SHANK3, and CASPR2, have shown a relevant association with ASDs. More recently, ongoing studies are investigating the role of different genetic mechanisms and on the whole exome sequencing [2], and overall, the multifactorial theory of ASDs risk has been broadly accepted, as well of other During neurogenesis and cell migration in most cortical and subcortical structures, high concentrations of methyl donors are required [23]. Several studies have been performed about the possible causative relationship between folate intake, pre or during pregnancy, its metabolism, and the onset of ASDs because of the modulation that cell folate exerts on the developing brain through the synthesis of DNA, neurotransmitters, and myelination. The results, however, are unfortunately still conflicting.
Two main MTHFR single nucleotide polymorphisms (SNPs), cytosine to thymidine switch at nucleotide position 677 and adenosine to cytosine switch at nucleotide position 1298, have been linked to functional alterations of folate levels [24]. The attenuation of this enzyme activity, prevalently due to the 677C → T mutation, leads to impaired methylation reactions and nucleotide synthesis. The abnormal one-carbon metabolism related to the MTHFR polymorphism might result in a hyperhomocysteinemia and hypomethylation function related to the increased synthesis of S-adenosyl homocysteine (AdoHcy) [25]. The influence of MTHFR genotype in serum folate and homocysteine levels has been associated to several neurologic and psychiatric traits [26]. Furthermore, the study of tissue concentrations of one-carbon metabolites in the liver, cerebral cortex, basal forebrain, and of MTHFR genotyping in mice shows that MTHFR deficiency could increase the risk of ASD-like behavior, underlining the relevance of the prenatal dietary intervention focused on MTHFR genotypes [27]. In addition, a diminished red blood cell (RBC) folate uptake and a decreased serum folate level resulting in an irregular one-carbon metabolism has been associated with the AG and GG genotypes of the reduced folate carrier (RFC1) [28,29].
Modulation of folate uptake at the blood-brain barrier (BBB) through RFC1, folate receptor alpha (FRα), or proton-coupled folate transporter (PCFT) may have clinical importance in establishing optimal therapies for childhood neurodegenerative disorders caused as a result of the inactivation or mutations of these folate transport systems [30].
Human and animal experimental studies have also been conducted to evaluate the role of auto-antibodies against FRα, blocking at the choroid plexus 5-methyltetrahydrofolate (5-MTHF) transfer to the brain, and their association with pregnancy-related complications, as well as neurodevelopmental disorders. Ramaekers V. et al., showed that reduced cerebrospinal fluid (CSF) folate could be explained by serum FR autoantibodies blocking the folate binding site of the membrane-attached FR on the choroid epithelial cells, while oral folinic acid supplements (starting dose of 1.0 mg/kg/day), as a 5-formyl derivative that does not require the action of dihydrofolate reductase for its conversion, can lead to normal CSF 5-MTHF and partial or complete clinical recovery after 12 months [31].
The aim of this systematic review is to evaluate the current state of knowledge on the potential role of FA in ASDs, both in animal and human studies.

Materials and Methods
This systematic review was conducted rigorously following the preferred reporting items for systematic reviews (PRISMA) guidelines [32]. We collected relevant data conformed to the eligibility criteria of our study.

Study Design
According to previous published evidence of FA deficiency involvement in ASDs events [33], we conducted a systematic review to assess the association of FA with ASDs in both animal and human clinical studies.

Eligibility Criteria
Predefined eligibility criteria for inclusion of the studies were as follows: all published randomized controlled trials (RCTs), observational studies (cohort or case-control design), and reviews (Table 6) dealing with the association between FA and ASDs. The following events were considered as primary outcomes: maternal exposure to FA and/or multivitamin supplements and ASD risk in offspring; improvement of autism symptoms towards sociability, cognitive verbal/preverbal, receptive language, and affective expression and communication; improvement of verbal communication; development of language and communication skills; MTHFR genetic variants as a risk factor for ASDs; and the effect of exposure to FRα auto-antibodies. Secondary outcomes included: serum homocysteine levels; serum folate and vitamin B12 levels; glutathione metabolism before and after treatment; CSF and serum 5-MTHF levels; plasma levels of: methionine, S-adenosyl-methionine (SAM), S-adenosyl-homocysteine (SAH), SAM:SAH ratio, cysteine, and cysteinylglycine.
Our search without language limitation reviewed articles with no restriction by year of publication or age of the patients. All published data until 31 May 2021 were included. Abstracts, conference papers, posters, and in vitro studies have not been considered.

Literature Search and Selection of Articles
We searched in PubMed, Scopus, Medline, and Embase databases using different key words to identify all studies that indicated the association between FA and ASDs. Specific text words were used: "folic acid and autism", "folic acid" AND "autism", "polymorphism and autism", "MTHFR polymorphism and autism", "antibodies FRA and autism", "autism and folic acid absorption", "autism and folic acid metabolism", "cholecalciferol and folic acid", "folic acid deficiency and autism", "FRAs and autism", "smoking and folic acid and autism", "western people and autism", "women in reproductive age and autism", "folate and autism", "folate" AND "autism", "autism and folate absorption", "autism and folate metabolism", "cholecalciferol and folate", "folate deficiency and autism", "smoking and folate and autism". After data extraction, we reviewed the titles and the respective abstracts for all records. Two authors independently reviewed the full texts to further assess if the selected studies fulfilled the eligibility criteria, verifying the results and removing duplicates. No disagreements were noted between the two authors. In the last phase, two authors independently evaluated the studies that met the criteria. Figure 2 reports a schematic diagram of the literature search procedure.

Data Extraction
Fifty six articles met the criteria and were selected for inclusion in our systematic review. Data extracted from each eligible article included the study name, publication year, main outcome, study outcome parameters, sample size, type of study, age of children, effect of folic acid, supplementation used, period of folic acid intake, and follow up period. In regard to animal studies, the type of animal and treatment used were reported.

Risk of Bias across Studies
In the majority of studies reviewed, there was the possibility of bias due to the small sample sizes and different case-control study designs that could potentially limit the sensitivity of the analyses to spot the treatment effects. The lack of reliable biomarkers that could specifically estimate who may benefit from a certain intervention is another factor that is needed to better combine treatment protocols. In studies investigating the prenatal use of FA, different unmeasured factors that could impact the background of ASD risk in offspring were not estimated, such as medication use during pregnancy, parental age, parental education, maternal exercise, maternal body mass index, maternal toxicant exposures, whether the pregnancy was planned, maternal smoking during pregnancy, weight gain during pregnancy, and year of birth. A possible misclassification due to insufficient information on gestational age may actually decrease the accuracy of exposure classifications. Also, a misclassifications of FA exposure as a result of not recorded maternal supplementation use might confound the evaluation of the risk reduction. Another limitation was the unavailable information of mother's whole blood and serum folate levels in several studies.
The phenotypic expression of MTHFR genotype might change during FA supplementation. In addition, in the genetic studies, there was a lack in the evaluation of serum folate levels or folate intake in cases and controls. Also, a further limitation was the absence of estimation in most studied cases and controls of vitamin B12, folate, and their metabolites serum levels.
In the animal studies, the limited sample sizes and the use of different mouse strains or pups from a single dam could possibly influence the tested outcomes.

Overview of Literature Search Results
A total number of 3609 articles were identified with our search strategy as shown in Figure 2. We eliminated duplicates and other studies not eligible for different reasons, such as in vitro studies, reviews and/or articles not correlated with the main argument (FA and ASDs). Finally, we systematically reviewed 56 studies.

Summary of the Results Reported by Human Clinical Trials Included in the Systematic Review
Studies regarding the association between the use of FA and ASDs lacked homogeneity due to a number of factors, including FA intake (mainly assessed by telephone interviews and self-reported questionnaire/form), period of intake, variability in the measured outcome parameters, and the specification of FA supplement used.
The essential characteristics of all human clinical studies systematically reviewed are summarized in Tables 1-4 according to specific primary outcome effects, such as the association between maternal FA supplementation and the risk of ASDs in offspring (10 studies); folate status or clinical benefits observed after folate supplementation in ASD diagnosed children (nine studies); role of MTHFR Gene C677T polymorphism in ASDs risk (10 studies); and frequency of serum FRAA in ASDs children (three studies).   The study showed a significant improvement in sleep and gastrointestinal problems compared with the placebo group. The intake of vitamins B 6 and B 12 , together with folic acid, was found to be more effective in lowering the levels of urinary homocysteine than the intake of vitamins B 6 and B 12 alone [46]. The results indicated that mean serum Hcy levels were significantly higher in autistic children as compared to controls. Significantly lower serum folate and vitamin B 12 levels were observed in autistic children as compared to controls. The levels of homocysteine in autistic children were also much higher as compared to normal reference values (5-15 µmol/L) [47]. 6 Guo M et al.,  The results showed that serum folate levels were lower in ASDs children comparing to the levels found in typically developing children. Moreover, the author underlined the necessity to evaluate folate status in children with ASDs aged three and under [51]. The results showed a significant increase in the reduced folate carrier (RFC1) G allele frequency among case mothers but not among fathers or affected children. Subsequent log linear analysis of the RFC1 A80G genotype within family trios revealed that the maternal G allele was associated with a significant increase in risk of autism, whereas the inherited genotype of the child was not.Results suggest that the maternal genetics/epigenetics may influence fetal predisposition to autism [61].

Studies on the Association between Maternal FA Supplementation in Reducing the Risk of Asds in Offspring
In Table 1, the results of Roth et al. [43], in the Norwegian mother and child cohort study are reported: they observed a lower risk of language disability following preconception maternal FA supplementation in 3 year old children; in 2017, in the randomized controlled trial by Christian et al. [42] a positive effect between maternal FA supplementation and neurological aspects was found. The Stockholm study [37] indicates that maternal multivitamin supplementation during pregnancy may be associated to a reduced risk of ASDs with intellectual disability by 0.26% (158 cases out of 61,934), with respect to 0.48% (430 cases out of 90,480) in the no nutritional supplementation use group. Nilsen et al., further observed that maternal prenatal FA supplement use was associated with a 14-17% adjusted risk reduction for ASDs [41]. The FA supplement dose was not specified in all studies, and the intake period was also variable. No correlation between early folate or multivitamin intake for ASDs was found in respect to women who did not have a supplement use in the same period [34]. This study has several limitations, such as the inability to assess FA use in the absence of multivitamin or other supplement use reflecting insufficient folate intake to accomplish levels necessary to obtain the desired protective effect. In addition, it suggested no ASDs risk reduction in one population at the time period this study was conducted (folic acid-adjusted risk ratio: 1.06, 95% confidence interval: 0.82-1.36; multivitamin-adjusted risk ratio: 1.00, 95% confidence interval: 0.82-1.22) [34].
In a case-control cohort study of 45,300 Israeli children, with 572 (1.3%) ASDs diagnosed cases, Levine et al., observed a statistically significant association between maternal FA or multivitamin supplement use before pregnancy (RR, 0.39; 95% CI, 0.30-0.50; p < 0.001) and/or during pregnancy (RR, 0.27; 95% CI, 0.22-0.33; p < 0.001) and reduced risk of ASDs in offspring [35]. In a group of 85,176 children from the Norwegian mother and child cohort study (MoBa) [36], Surén et al., found that the use of prenatal FA supplements around the time of conception was associated with an approximately 45% lower risk of autistic disorder. In a recent paper, Raghavan et al., analyzed the maternal plasma levels of folate, vitamin B12, and homocysteine from samples taken at birth in a cohort study of 1257 mother-child pairs [38]. Furthermore, the authors correlated the risk of ASDs with the frequency of prenatal multivitamin supplementation (low: ≤2 times/week; high: >5 times/week). Interestingly, a "U" shaped relationship was found between the frequency of maternal supplementation and risk of ASDs: this was reduced following moderate intake (3-5 times/week), while increased in case of low or high frequency. Likewise, the risk of ASDs was higher in children from mothers whose plasma levels of folate and B12 were elevated (>90th percentile) [38].
However, another study found that prenatal FA use was associated with less child autistic traits, while no association of autistic traits in the offspring was found with maternal plasma folate concentration measured in early pregnancy [40]. The risk of ASDs in children whose mothers had, at delivery, very high concentrations of plasma folate and B12 (≥90th percentile) could find an explanation in the sensitivity of the fetus brain exposed to higher levels of micronutrients, especially in the third trimester when some relevant neurological processes are ongoing.
In the recent MARBLES study (markers of autism risk in babies: learning early signs), Schmidt analyzed data from children (n = 332) and their mothers (n = 305) to evaluate the association between risk of ASDs recurrence and maternal prenatal multivitamin use [39]. Women that took prenatal FA (~600 µg) in the first month of pregnancy had children that were half as likely to receive an ASDs diagnosis (adjusted RR, 0.50; 95% CI, 0.30-0.81), with significant lower autism symptom severity (adjusted estimated difference, −0.60; 95% CI, −0.97 to −0.23) compared to children whose mothers did not take prenatal vitamins in the first month.

Studies of Folate Supplementation or Folate Levels in ASD Diagnosed Children
In a recent randomized controlled single-blind study of 67 children and adults with ASDs from Arizona vs. 50 non-sibling neurotypical controls of similar age and gender, it was confirmed a significant improvement in nonverbal intellectual ability in the treatment group (~600 µg FA) compared to the non-treatment group by using different tests, such as IQ (+6.7 ± 11 IQ points vs. −0.6 ± 11 IQ points, p = 0.009) and non-verbal intelligence index (+10% in treatment group vs.−1%, p value 0.01) based on a blinded clinical assessment ( Table 2) [49]. The treatment group had significantly higher improvement in autism symptoms and developmental age with an increased level of docosahexaenoic acid (DHA); eicosapentaenoic acid (EPA); carnitine; Coenzyme Q10; and vitamins A, B2, B5, B6, B12, FA, observed as compared to the non-treatment group. Nutritional status, non-verbal IQ, autism symptoms, and other symptoms in most ASDs patients were improved, based on the semi-blinded assessment, suggesting the efficacy of a comprehensive nutritional and dietary intervention. Improvement in verbal communication, motor skills, and plasma levels of homocysteine, FA, B12, glutathione following FA (400-600 µg) or folinic acid (0.4-2 mg/kg) intake are reported also in several studies [44][45][46]48]. ASD Omani children had statistically (p < 0.05) higher homocysteine levels (20.1±3.3 µmol/L) in respect to controls (9.64 ± 2.1 µmol/L) [47]. Additionally, the homocysteine levels in ASD children were considerably higher compared to normal reference values (5-15 µmol/L), whereas serum concentrations of folate and Vitamin B12 were much below the values defined as deficiency levels (3.0 µg/L and < 250 pg/mL, respectively) for these nutrients [47].
Regarding the studies that have investigated the folate levels in ASDs children, Shoffner in his cohort study examined CSF 5-MTHF concentrations in 67 ASD diagnosed children and did not observe any significant correlation between CSF 5-MTHF and autism symptoms. CSF 5-MTHF levels less than 40 nmol/L were observed in 11 of 67 children but only in one of two repeated CSF evaluations. Findings showed that CSF 5-MTHF levels differ significantly over time in an unpredictable way, indicating a biological variability, and have no relationship with typical clinical features of autism [50].
In a recent multi-center study performed in China and involving 1300 ASDs diagnosed children, Li et al., correlated the serum folate levels with the clinical aspects and the neurodevelopmental levels of different age groups; the authors found serum folate levels lower than in normally developing children. Furthermore, the folate serum concentration in the ASD children is interestingly associated with the neurodevelopmental level [51].

Studies on MTHFR Polymorphisms and ASDs
On the other hand, an important consideration clarified the role of the MTHFR C677T polymorphism and ASDs (Table 3) [53,55,56,59]. A population-based case-control study in Chinese Han assessed the frequency of genotype MTHFR 677TT in 372 children. The frequency of these genotype in children with autism (16.1%) was significantly higher (odds ratio [OR] = 2.04; 95% confidence interval [CI] = 1.07, 3.89; p = 0.03) than those in controls (8.6%), suggesting that MTHFR C677T is a risk factor for autism in children of this population [52].
A Brazilian case-control study showed no significant changes between cases and controls (p = 0.72) in terms of frequency of T allele (0.38 for the case group and 0.35 for the control group) (p = 0.77) [54]. Concordant observations were made in addition by Sener and Zhang [57,58]. Moreover, the authors suggested a replication of the studies with a larger well-characterized scale. The study of 529 case-parent trios vs. 566 neurotypical controls showed that maternal genetics/ epigenetics may affect fetal predisposition to autism [61]. Plasma homocysteine, adenosine, and AdoHcy resulted as significantly elevated between autism mothers, corresponding to reduced methylation capacity and significant DNA hypomethylation (p < 0.001). Further analysis revealed a significant increase in the reduced folate carrier (RFC1) G allele frequency among case mothers, but not among fathers or affected children, determining a significant increase of the risk of autism. Table 4 contains the results obtained in three human studies investigating the frequency of FRAA in ASD children [31,62,63]. The first cohort study includes 40 ASDs children and 42 gender-age matched typical development (TD) children, age range of 2 to 6 years. The author found that serum FRAA concentration in ASD children was higher than in TD children (138.61 ±373.27 ng/mL vs. 37.68 ±71.54 ng/mL, p = 0.09829), where 77.5% (31/40) of children with ASDs and 54.8% (23/42) of TD children were positive for serum FRAA (p = 0.03746) [62]. In the second study, Ramaekers [31] et al., assessed additional parameters analyzing serum and CSF folate, serum vitamin B12, homocysteine and amino acids concentration, serum, and CSF FRAA and FR1 and FR2 genes. Results highlighted that all 25 age-matched controls were negative for FRAA, while 19 out of the 23 patients with low CSF 5MTHF had serum FRAA with a mean value of 1.05 pmol FR blocked/mL serum (range: 0. 1-4.19). Furthermore, treatment with folinic acid (1-3 mg/kg/day) in these patients led to partial or complete recovery from ASDs in early diagnosed cases (before the age of three years old). An improvement in verbal communica-tion was found by Frye et al. [63] in his cohort of non-syndromic ASD children through folinic acid treatment.

Summary of the Results Reported by Animal Clinical Studies Included in the Systematic Review
In Table 5, the principal findings of the nine animal clinical studies are reported. Animal studies showed how maternal periconceptional deficit of folate in rats provokes behavior alterations in the offspring relevant to the autistic-like phenotype [64][65][66]. Analysis of gene expression in the cerebellum of offspring mice from 8 to 10-week-old pups revealed that the expression pattern of a significant number of genes were found to be altered by ≥2.5 fold at a significance of p > 0.05 after exposure to 20 mg/kg high maternal FA (HMFA) diet during gestation, suggesting a dysregulated expression of several genes in the cerebellum of both male and female pups [67]. The results showed that HMFA supplementation alters offsprings' CH gene expression in a sex-specific manner. These changes may influence infants' brain development. In addition, it was also found that HMFA had no impact on global DNA methylation levels of the offspring epigenome [65]. The study suggests that MTHFR deficiency can increase the risk of ASD-like behavior in mice and that prenatal dietary intervention focused on MTHFR genotypes can reduce the risk of ASDs-like behavior. Findings emphasize the critical role of in utero C1 metabolism in developmental trajectories that lead to the presentation of autistic behavior. Aberrations in both the GABAergic and glutamatergic pathways suggest that Mthfr deficiency is linked to deleterious alterations in the basal cortical circuit activities in the affected mice [27]. 6 Sadigurschi N et al., (2019) [68] Mice on a Balb/cAnNCrlBR backgroundheterozygote Mthfr-KO mice

No treatment
Genotyping of DNA isolated from toe clips. Immuno-fluorescence analysis of brain tissue. Morphogenic and behavioral assessments N.S (not specified) The study provides evidence for the profound impact of a genetic deficiency in the MTHFR gene on the induction of autistic features. In the mouse model of ASDs, this deficiency directly regulates metabolite availability and indirectly controls the environment of the developing embryonic brain [68]. Deficits in rats exposed to Ab during gestation and pre-weaning (GST+PRW) included indications of increased levels of anxiety. None of these rats learned the active place avoidance task, indicating severe learning deficits and cognitive impairment.
Similar but less severe deficits were observed in rats exposed to Ab during GST alone or only during the PRW period, suggesting the extreme sensitivity of the fetal as well as the neonatal rat brain to the deleterious effects of exposure to Ab during this period [69]. Findings suggest severe behavioral and cognitive changes mirroring ASD symptoms following gestational Ab exposure in a rat model and protection afforded by folinic acid and dexamethasone treatment [71]. Table 6. An overview of all reviews, systematic reviews, and metanalysis included in the study. Maternal folate and autism spectrum disorders and related traits. Self-reported maternal folate and autism spectrum disorder traits. Maternal folate biomarker and autism spectrum disorder traits.

Folate supplementation and ASDs
Inconclusive evidences underline the need for future studies of maternal folate status during the pre-and peri-conceptional periods. In addition, an incorporation of genetic data could complete better these assessments [14]. Elucidate the association of maternal FA intake during the prenatal period and ASD risk in offspring FA intake.Period of FA intake. FA intake and risk of ASDs subtypes. FA supplementation (excluding diet consumption) and risk of ASDs.

Geographical area and risk of ASDs
Findings do not support the link between FA supplementation during prenatal period and ASD reduced risk in offspring. In addition, more investigation is needed because of many study limitations [74]. This meta-analysis found that periconceptional FA supplementation may reduce ASD risk in those with MTHFR 677C>T polymorphisms where an increased risk of ASDs was indicated. The C677T polymorphism was found to be associated with ASDSs only in children from countries without food fortification [75]. In genetically susceptible individuals with altered DNA-methylation patterns, a potential protective effect of supplementation taken before conception was suggested [16]. 7 Modabbernia Evidence underlined the need of further studies for specific biomarkers of the folate pathway that might help to detect ASDs early and diagnose ASDs, as the abnormality of FA metabolism has a potential impact in ASD offspring. Thus, specific type and dose of folate and other cofactors could be used for treating or preventing ASD traits [80].
Likewise, studies on mice with a Balb/cAnNCrlBR background and heterozygous MTHFR-knockout indicated that maternal and offspring MTHFR deficiency increased the risk for an ASD-like phenotype in the offspring [27,68]. Orenbunch observed a reduced risk of ASD-like behavior in MTHFR-deficient mice supplemented with one-carbon nutrients prenatally (FA 9 mg/mL, betaine 2%, choline 2%) [27]. Specifically, among offspring of MTHFR+/− dams, prenatal diet supplementation was protective against ASD-like symptomatic behavior compared to the control diet with an odds ratio of 0.18 (CI:0.035, 0.970). Additionally, a change in the cerebralcortex of the proportions between betaine/choline and SAM/SAH were correlated with behavior similar to ASDs. Males had an altered ratio of the glutamate receptor subunits GluR1/GluR2 with respect to NR2A/NR2B. Moreover, symptomatic mice with ASD-like behavior had lower levels of GABA pathway proteins (GAD65/67 and VGAT) [27].
Regarding the frequency of FRAA on ASDs only two studies, conducted in 2016 in rat models, suggested severe behavioral and cognitive changes mirroring ASD symptoms when exposed to FRα antibodies during gestation [69,71].
In a recent paper, Chu et al., supplemented the diet of male and female mice prior to mating, during pregnancy and lactation with moderate and high FA doses (2.5 and 10 times the normal dietary intake respectively) and evaluated the effect to offspring at weaning. Interestingly, the authors found that behavioral abnormalities were more evident in mice fed with moderate FA doses than in those treated with higher FA doses. However, both moderate and high FA supplementation modified the cerebral gene expression in offspring at weaning, but these FA doses were not sufficient to induce autism-like behavior [70].
The relationship between folate and ASDs is extremely complex overall because the low folate-dependent DNA hypomethylation can be due not only to a reduced intake but also to an altered folate metabolism [14], an example of which is the debated epigenetic role of MTHFR polymorphisms (C677T and A1298C) in the DNA methylation. MTHFR C667T polymorphism shows regional and ethnic variations; additionally, the frequency of the T allele was higher in ASD offspring consequently identified as a risk factor. Cytosolic serine hydroxyl methyl transferase (SHMT1 C1420T), MTHFR A1298C, methionine synthase reductase (MTRR A66G), and methionine synthase (MS A2756G) were assessed as well. This issue has been investigated in a meta-analysis by Pu et al. [75], who interestingly found that MTHFR C677T can be associated with an increased risk of ASDs, but only in countries without mandatory FA fortification; these results emphasize the potential relevance of the FA prenatal supplementation in modulating the ASDs risk in the presence of MTHFR C677T polymorphism [81].
Another principal observation of this review was the frequency of FRAA in serum or CSF and their association with ASDs. Serum FR autoimmunity appears to represent an important factor in the pathogenesis of reduced folate transport to the nervous system among children with early-onset low-functioning autism, associated with or without neurological deficits. The elevated presence of FRAA in ASD offspring could reduce folic acid uptake in the choroid plexus, explaining the reduced CSF folate levels, representing a useful screening biomarker for ASDs [31]. Early detection of FRAA may be a key factor in the prevention and therapeutic intervention among this subgroup of patients with autism. We further observed that clinical evaluation and confirmation in large sample sizes of the results obtained is necessary, considering additionally to investigate whether FA intake can improve the syndrome of FRAA positive children with ASDs. In this subgroup of ASDs children, in some studies [63,89], the treatment with folinic acid has been successfully used; folinic acid, as a reduced form of FA, can enter the nervous central system through the blood-brain barrier by using the reduced folate carrier when the FR is unavailable because of the presence of FRAA.
According to an interesting recent study [80], also the maternal post-delivery high concentrations of folate could be the consequence of an altered mechanism of transport of the micronutrient across the placenta, depending on a higher prevalence of FRAAs or of a mutation in RFC, as shown in mothers of ASD offspring. Then, according to this author, FRAAs could induce a condition of folate deficiency in the fetus, even with a normal maternal folate status.
Only a few studies measured vitamin B12, homocysteine levels, and glutathione levels, while none of them analyzed red blood cell folate. The relevance of vitamin B12 in the onset of autism and ASDs can be strictly connected to its role in DNA methylation, the epigenetic regulatory process known to be relevant to brain development [90]. Amanat indicated that serum homocysteine levels were significantly higher in autistic children compared to controls, and significantly lower serum folate and vitamin B12 levels were observed in autistic children compared to controls [47]. Intervention treatment with FA resulted in improvement of ASD-associated behaviors and metabolic profile in autistic children, especially in those with early established diagnosis [44]. These findings suggest a recovery of the methyl groups transfer via the methionine cycle lowering homocysteine levels, redirecting it to the transsulfuration pathway as well to avoid brain dysfunction via oxidative damage and abnormal DNA methylation. Clearly, methyl donors are functionally dependent of one another; therefore, further studies should incorporate homocysteine and vitamin B12 concentration measurements into their exposure evaluations during the periconceptional period and early pregnancy to better explicate the association of folate with ASDs.
In studies that have evaluated the risk of ASD, different doses of FA maternal supplementation have been considered, in addition to different windows of exposure to the vitamin supplementation.
This lack of homogeneity gave rise to an interesting debate about the potential enhanced risk of ASDs following high dosage of FA intake pre or during pregnancy; most of the studies, performed both in humans and in animals, emphasizing the warning in the use of high FA dosages, focused on the presence of circulating, and very likely detrimental, unmetabolized FA [14,15,72,79,91], usually associated with clinical conditions such as the reduced cytotoxicity of natural killer cells in animal studies [92], neurological and cognitive disorders [93], or cancer.
Several neurological disorders in brain development have been observed in some animal studies following maternal supplementation with large doses of FA: synaptic defects [94], dysregulation of gene expression [67], and effects on brain DNA methylation [84]. Synthetic FA is reduced to tetrahydrofolate by dihydrofolate reductase (DHFR), whose activity in humans is slow and can be inhibited by higher concentrations of the same FA [95]; however, unmetabolized FA, usually is not present in serum in the case of FA supplementation not exceeding 400 ug/day [96].
A lower concentration of 5-MTHF in the cell, a reduced transformation to methionine, and an altered methylation process can be the consequences of high concentrations of unmetabolized FA; in this regard, several animal studies have interestingly shown both gene-specific hypermethylation and DNA hypomethylation [77,97,98] in animals fed with higher dosages of FA, and a different grade of modulation of the global DNA methylation has been recently proposed also by Chu et al. [70]. Further investigations are, however, requested in humans.
A detrimental effect of unmetabolized FA on neural districts during neurological development might be then hypothesized [86,99,100].
Unmetabolized FA can arise not only from FA intake, pre or during pregnancy, at higher dosages than 1 mg/d but, according to a recent study [85], it could depend also on an impaired function of the one-carbon metabolism with or without vitamin B12 involvement [78]; another consideration can be made in this regard: the exact critical time windows during which the neurological development can be influenced by the one-carbon metabolism are not yet fully defined.
As previously underlined, folate is currently and successfully used in early pregnancy to reduce the risk of NTDs at a dosage, usually of 400 ug/day or up, at least to the end of the first trimester. However, many women at increased risk for NTDs, or at risk for recurrence of NTDs, are advised to intake FA dosage higher than 1 mg/day.
Further studies then aimed to investigate the real responsibility, if any, of unmetabolized FA in the onset of ASDs are absolutely necessary, especially because the presence of high levels of FA in the maternal blood is not an extraordinary finding, particularly in countries where food fortification with FA is mandatory.

Limitations
Some of the limitations of the studies are the heterogeneity of FA supplementation, the duration of use of these supplements, and the residence of the study participants in regard to the approved nutritional fortification. Moreover, the small sample size relative in some studies can be an additional limitation.In order to evaluate the genetic influence on ASDs in offspring only a few studies reported data from the potential presence of paternal polymorphisms in one-carbon metabolism.

Conclusions
In summary, this systematic review aimed to clarify the association between FA and ASDs, taking into consideration biological, genetic, and epidemiological evidence as important complex mechanisms required for maintaining optimal folate levels. One of the main concerns emerging from the studies is that nutritional fortification with folic acid present in some countries can cause higher maternal blood folic acid levels, potentially leading to detrimental circulating of unmetabolized FA, especially if associated with FA supplements pre or during pregnancy. Another concern not fully defined is the exact critical time window during which the neurological development can be influenced by the one-carbon metabolism. Further research taking in consideration the limitations of the current studies reported in this systematic review should be carried out to give a clear overview of the correlation between FA and ASDs.

Data Availability Statement:
No new data were created in this study. Data sharing is not applicable to this article.

Conflicts of Interest:
The authors declare no conflict of interest.