A Scoping Literature Review of the Relation between Nutrition and ASD Symptoms in Children

Autism spectrum disorder (ASD) is characterized by impairments in social interaction, communication skills, and repetitive and restrictive behaviors and interests. Even though there is a biological basis for an effect of specific nutrition factors on ASD symptoms and there is scientific literature available on this relationship, whether nutrition factors could play a role in ASD treatment is unclear. The goal of the current literature review was to summarize the available scientific literature on the relation between nutrition and autism spectrum disorder (ASD) symptoms in childhood, and to formulate practical dietary guidelines. A comprehensive search strategy including terms for ASD, nutrition factors (therapeutic diets, dietary patterns, specific food products, fatty acids and micronutrients) and childhood was developed and executed in six literature databases (Cinahl, Cochrane, Ovid Embase, PsycInfo, PubMed and Web of Science). Data from meta-analyses, systematic reviews and original studies were qualitatively summarized. A total of 5 meta-analyses, 29 systematic reviews and 27 original studies were retrieved that focused on therapeutic diets, specific food products, fatty acids and micronutrients and ASD symptoms during childhood. Results of the available studies were sparse and inconclusive, and hence, no firm conclusions could be drawn. There is currently insufficient evidence for a relation between nutrition and ASD symptoms in childhood, making it impossible to provide practical nutrition guidelines; more methodological sound research is needed.


Introduction
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by three core behavioral symptoms: impairment in social interaction skills, impairment in communication, and repetitive and restrictive behaviors and interests [1]. Estimations of the prevalence of ASD vary and is nowadays estimated to be globally around 0.06-1.4% [2]. during childhood and ASD symptoms during childhood (here defined as children aged 0 to 12 years). Dietary pattern was defined as the habitual total food intake, while a therapeutic diet was defined as a specific diet prescribed for the treatment of ASD, e.g., the gluten and casein free diet.
To execute the knowledge synthesis, a comprehensive search strategy was developed which included search terms related to (1) nutritional factors (e.g., dietary patterns, therapeutic diet, and nutrients), (2) mental health problems specific for the life phase (i.e., for the current review: ASD), (3) the specific age group (i.e., children). The overall childhood search strategy can be found in the File S1. The search strategy was executed in the following databases: Cinahl, Cochrane, Ovid Embase, PsycInfo, PubMed, and Web of Science. No restrictions on publication year were entered in any of the databases. The overall search strategy was adapted for each of the databases searched.
The search strategy was executed in July 2018 and updated in April 2020. All found articles were first title/abstract screened, and then full text screened by the first author. Title/abstract screening was checked by a second screener and discrepancies were solved by discussion.
If multiple meta-analyses or systematic reviews were available, a review of reviews was executed. This review of reviews was supplemented with original articles not included in any of the meta-analyses or systematic reviews. If no meta-analyses or systematic reviews were available, the results of the original studies were summarized.

Inclusion and Exclusion Criteria
Studies (i.e., studies in meta-analyses, studies in systematic reviews and original studies) were included in this review if (1) they reported on ASD or ASD symptoms or characteristics; (2) the ASD outcome measure used was either an ASD specific clinicaldiagnostic instrument or a parent or teacher-rated instrument related to ASD; (3) dietary patterns, therapeutic diet, specific food product, or nutrient intake were measured with a quantitative instrument OR the dietary pattern, diet, or nutrient intake was manipulated by diet intervention, dietary advise or supplementation; (4) the study included children aged 0-12 years, or the average age of the included participants was 12 years or younger; (5) the study was reported in English, Dutch or German; and (6) the manuscript was a published meta-analysis, systematic review, or an original study not included in the meta-analyses or systematic reviews included in the current review.
Studies (i.e., studies in meta-analyses, studies in systematic reviews and original studies) were excluded if the study (1) was a single-case study; (2) a cross-sectional study reporting on nutrition intake (e.g., 24 h diet recall, food frequency questionnaires or diet diary) or nutritional status (e.g., blood values) only of children with ASD (3) a case-control study in which the nutrition intake (e.g., 24 h diet recall, food frequency questionnaires or diet diary) or nutritional status (e.g., blood values) only of children with ASD was compared to children without ASD; (4) reported on pre-or probiotics, food coloring, preservatives, alcohol consumption, caffeine consumption, or herbal supplements, as these are not considered as normal elements of a regular dietary pattern for children.
Note that in some meta-analyses and systematic reviews article were included that did not meet in-and exclusion criteria of this review, these were not mentioned in the study design(s) column of the meta-analyses/systematic review result tables nor counted in the total of included studies in the text. However, they might be present in meta-analyses results, if this is the case this is noted via a superscript next to the study design cell in the meta-analyses/systematic review result tables.

Data Extraction
For the review of reviews, the following data were extracted from each included metaanalysis/systematic review: type of study (meta-analysis, systematic review, combination), design of studies included, number of studies included relevant for our review, characteristics of included participants (i.e., age range and specifics regarding ASD diagnosis), nutrition component or described intervention, name of the first author, publication year and results. Regarding the results for meta-analyses, standardized effect sizes (i.e., odds ratios, standardized mean difference or Hedges G) and confidence intervals were extracted. For systematic reviews the number of studies with a significant effect or association, relative risk, odds ratio or effect size were abstracted. When a systematic review described multiple exposures (for example micronutrients and pollutants) only those results relevant for the current review were extracted. Additionally, some meta-analyses and systematic reviews reported on multiple dietary patterns, therapeutic diet or nutrient. In the current review, the results are discussed per dietary pattern, therapeutic diet or nutrient. Therefore, it can be the case that a meta-analysis or systematic review is discussed multiple times. It is also important to note that some original studies were included in a multitude of meta-analyses and/or systematic reviews and that the results of this original study are thus 'counted' more than once in the weight of evidence (also see analyses). Characteristics of the meta-analysis and systematic reviews are presented in the tables and summarized in the text.
For the update of review of reviews and the review of original studies the following data were extracted from each included original study: design, number and characteristics of participants, the ASD measuring instrument, the nutritional component or intervention, the treatment in the control group, name of the first author, year of publication and results. Characteristics of the original studies and their results were presented in tables and summarized in the text. Note that an original study is defined as a study which was not included in any of the meta-analyses or systematic reviews.

Analyses
For the review of reviews, the results of the meta-analyses and systematic reviews were integrated in a self-developed qualitative way. If there were multiple meta-analyses on the same topic, we described how many of those meta-analyses found an effect in the same direction and what the size of the effect or the association was. For systematic reviews we examined the direction and consistency of the findings of the systematic reviews, and based upon this, we made a final assessment. An effect or association was considered significant if 2/3 or more of the included studies found an effect or association in the same direction (i.e., beneficial (+)/no effect (0)/unfavorable (−)) and if there were at least five studies in which the effect/association was studied. If the results in a systematic review or meta-analyses were mixed (i.e., <2/3 studies in the same direction), and there were at least five studies in which the effect/association was studied, this was indicated with +/−. If the effect/association was studied < 5 studies no conclusion was drawn.
For the update of review of reviews with original studies, the results of original studies were summarized qualitatively. Original studies in which the same nutritional factor (i.e., dietary pattern, specific food product, therapeutic diet or nutrient) was investigated were taken together and it was determined how many of these studies showed an effect or association in the same direction in the same manner as the review of review method.
If no meta-analyses or systematic reviews were available on a specific nutritional factor, but there were original studies available the results of the original studies were summarized in a review of original studies. The approach was equal to that of the review of reviews, an effect or association was considered significant if 2/3 or more of the included studies found an effect or association in the same direction (i.e., beneficial/neutral/unfavorable) and if there were at least five studies in which the effect/association was studied.
The quality of the original studies describing an RCT or cohort study was assessed using a quality assessment tool for quantitative studies [23]. The quality of cross-sectional studies was not assessed because this type of studies scores low on level of evidence (see Table 1). The components that were tested are: selection bias, research design, correction for confounders, blinding, validity and reliability of measuring instruments, and participant drop-out. Each component was assessed based on two questions that could be scored as "yes", "no", or "not clear". The scores on the two questions resulted in a judgment per component as strong, moderate, or weak. The judgments of all components together gave a final score of the quality of the study. A study was qualified as strong when no component was assessed as weak, as moderate when one component was assessed as weak and as weak when two or more components were assessed as weak.

Overall Strength of Evidence
In this review, we used the following qualifications for the overall strength of evidence: very strong, strong, moderately strong, weak, very weak, insufficient evidence and no evidence. In Table 1, the different qualifications and the way in which the strength of the evidence is indicated in the text are reported. Since no criteria for strength of evidence for reviews-of-reviews and meta-analyses exist, this qualification has been developed for the knowledge synthesis that formed the basis for this review, drawing on already existing guidelines and criteria [24][25][26][27]. Note that this strength of evidence qualifications takes together the results of meta-analyses, systematic reviews and original studies. Additionally, conclusions were only drawn and the associated strength of evidence was only given when 2/3 or more of the included studies found an effect or association in the same direction and if there were at least five studies in which the effect/association was studied. If this is not the case no conclusion and no strength of evidence is given.
Based on the results of the meta-analyses, systematic reviews and original studies found, conclusions were drawn, and recommendations were made for research and clinical practice. Conclusions were only drawn if there were one or more meta-analyses/systematic reviews with a minimum of five original studies included or at least five original studies that studied a particular association or effect. Recommendations for practice were only given when there was moderately strong to very strong evidence for the effect of a nutrition factor on ASD symptoms.

Results
In total, 5 meta-analyses, 29 systematic reviews and 27 original studies were found that focused on nutrition during childhood and childhood ASD symptoms. We located studies with a focus on therapeutic diets and intake of specific food products, fatty acids and micronutrients. No meta-analyses, systematic reviews or original studies with a focus on dietary patterns for ASD symptoms were found. All meta-analyses, systematic reviews and original articles were published in English. In total, 71 unique studies were included in the meta-analyses and systematic reviews, amounting to 98 unique studies included in the current review.
Below, we first report studies on therapeutic diets and intake of specific food products, followed by studies on fatty acids and finishing with studies on micronutrients. For each nutrition factor (i.e., therapeutic diets and intake of specific food products, fatty acids and micronutrient), we first describe the number of meta-analyses, systematic reviews and original studies which were located focusing on a specific nutrition factor. Subsequently, we report the results of the meta-analyses and systematic reviews, then we report the results of the original studies and we finish with the conclusion on the effect of the specific nutrition factor on childhood ASD symptoms.
In Supplementary File S2 the original studies included per meta-analysis and systematic review are reported. Additionally, it shows the meta-analyses and/or systematic reviews in which each original study was included.

Gluten and/or Casein Free Diet
Fifteen systematic reviews were found that included a total of 20 unique studies, in which the effect of a gluten and/or casein free diet on ASD was investigated [13,17,[28][29][30][31][32][33][34][35][36][40][41][42][43] (see Table 2 and Supplementary File S2). Approximately half of the studies included in the systematic reviews showed a beneficial effect of the GFCF diet on one or more of the ASD outcome measures which were included in the study. However, as all studies utilized different outcome measures, no specific conclusion regarding specific outcome variables can be described. Moreover, many of these studies had a weak research design (i.e., open label or single blind) and mostly included very few participants (i.e., varying between 10 and 150, with the majority of studies including fewer than 20 participants).
In addition, eight original studies were found [44][45][46][47][48][49][50][51] (see Table 3). Six of these eight original studies were intervention studies of moderate or weak quality [44][45][46][47]50,51] and two were observational studies following children who were already on a GFCF diet [48,49]. In three of the six intervention studies beneficial effects on the Childhood Autism Rating Scale (CARS) score and behavior as noted by parents were reported [47,50,51]. It is, however, important to note that these three studies did not include a control group. In the two observational studies parents reported an improvement in the behavior of their children after the start of a GFCF diet on respectively the CARS [49] and characteristic ASD behaviors and social behaviors [48]. This improvement was significantly better than those that did not eliminate gluten and casein completely [48], but not significantly better than those that were not on a GFCF diet. The chance of information bias seems high in these studies. NA + = 2/3 or more of the included studies found a positive effect or association, − = 2/3 or more of the included studies found a negative effect or association, 0 = 2/3 or more of the included studies showed no effect or association was shown, +/− = the results in a systematic review or meta-analyses were mixed (i.e., <2/3 studies in the same direction), and there were at least five studies in which the effect/association was studied, ADHD = attention, deficit (hyperactivity) disorder, ASD = autism spectrum disorder, ATEC = Autism Treatment Evaluation Checklist, CARS = Childhood Autism Rating Scale, NA = not appropriate, RCT = randomized controlled trial, sig.= significant, SR = systematic review. 1 Chanyi diet is a traditional Chinese diet in which the intake of some foods which cause "internal heat" are limited. Table 3. Studies concerning therapeutic diets and specific food products for children with ASD, that are not included in the meta-analyses and systematic reviews in Table 2. Parents who eliminated all gluten and/or casein reported bigger improvement in ASD behavior, physiological symptoms and social behavior after the start of the diet than those that did not eliminate everything. Children with food allergies, food sensitivity and gastro-intestinal problems showed a bigger improvement with GFCF than children without those problems.   Overall, fewer than 2/3 of studies showed an effect in the same direction, and hence no conclusion regarding the effect of the GFCF diet for the treatment of childhood ASD symptoms can be drawn.

Ketogenic Diet
Five systematic reviews were found in which the ketogenic diet in children with ASD was discussed [37][38][39][40][41] (see Table 2). These five systematic reviews only included a total of four unique studies. All four studies utilized the CARS (among others) as outcome measure and in all studies some, but not all participants on the ketogenic diet showed improvements on the CARS. However, a large proportion of the children either did not want to participate or dropped out during the study, making sampling and attrition bias very likely. In addition, the studies were very small (all ≤ 45 participants) and three studies did not include any control, and one study compared the ketogenic diet to the GFCF diet.
There were no original studies looking at the effect of the ketogenic diet on ASD symptoms in children.
As there were fewer than five studies available, no conclusion can be drawn regarding the effect of the ketogenic diet on ASD symptoms in children.

Other Diets and Specific Food Products
One study that looked at the Chanyi diet was included in two systematic reviews [41,43]. This is a traditional Chinese diet that avoids products that produce "internal heat", such as meat, fish and ginger (see Table 2). In this study significant fewer social communication problems and a significant positive change in executive control of repetitive, inflexible and hyperactive behavior was shown for the Chanyi diet group, but not in the control group in which children consumed their usual diet.
In five systematic reviews a total of two different studies looking at the effect of camel milk on ASD were reported (see Table 2) [13,17,[41][42][43]. In one of those two studies a significant effect was reported, but not in the other. One additional original study was found that looked at the effect of camel milk on ASD symptoms [52] (see Table 3). Here, positive effects of camel milk on the CARS and on the Social Responsiveness Scale (SRS) were shown, especially for raw camel milk.
Two original studies were found in which the effect of tea containing GABA and L-theanine on ASD symptoms [53] and the relation between sugar based beverage consumption and ASD symptoms [54] were investigated (see Table 3). Five out of nine participants improved on ASD symptoms while consuming the tea containing GABA and L-theanine compared to the placebo tea [53]. The adjusted odds ratio for the Clancy Autism Behavior Scale increased with increasing sugar-based beverage consumption [54].
As there were fewer than five original studies related to the Chanyi diet, camel milk, tea containing GABA and L-Theanine, and sugar-based beverage consumption for ASD symptoms, no conclusions can be drawn on the effects of these diets or specific food products on ASD symptoms in childhood.

Type of Paper Nutritional Factor Outcome Design(s) of Orginal Studies Effects Final Conclusion
De Crescenzo 2020 MA [20] PUFA supplementation Outcomes determined to be highly relevant for children and adolescent with ASD, determined by expert panel.

RCT 1
PUFAs were superior compared to placebo in reducing anxiety in individuals with ASD (SMD = −1.01 *, very low certainty of evidence). PUFAs worsened quality of sleep compared to a healthy diet (SMD = 1.11 *, very low certainty of evidence). PUFAs were not better than placebo in reducing aggression, hyperactivity, adaptive functioning, irritability, restricted and repetitive interests and behaviors and communication.
Omega-3 PUFA supplementation ASD 6 RCT 2 Omega-3 supplementation was sig. more effective than placebo in treating the following symptoms and/or function groups: language (g = 0.313 *) and social-autistic (g = 0.311 *) And sig. more effective for the following clinical domains: core symptoms (g = 0.268 *) and associated symptoms (g = 0.276 *).  N-3 fatty acids supplementation Core symptoms of ASD or related symptoms 1 RCT, 2 trials 3/3 studies showed improvement (1 trend, 2 unclear) + + = 2/3 or more of the included studies found a positive effect or association, − = 2/3 or more of the included studies found a negative effect or association, 0 = 2/3 or more of the included studies showed no effect or association was shown, +/− = the results in a systematic review or meta-analyses were mixed (i.e., <2/3 studies in the same direction), and there were at least five studies in which the effect/association was studied, * = significant results, ABC = Aberrant Behavior Checklist, ASD = autism spectrum disorder, ATEC = Autism Treatment Evaluation Checklist, BASC = Behavior Assessment System for Children, MA = meta-analysis, MD = mean difference, PUFA = polyunsaturated fatty acid, RCT = randomized controlled trial, sig. = significant, SMD = standardized mean difference, SR = systematic review, VABS = Vineland Adaptive Behavior Scale; 1 One study was executed in only adolescents, this study is excluded here, but is included in the MA. 2 One study was executed in only adolescents, this study is excluded here, but is included in the MA. 3 In one study the average age was 14.6 years, this study is excluded here, but is included in the MA. 4 Both manuscripts reported the same study with the same results.
In four of the five meta-analyses some small positive significant effects of fatty acid supplementation on ASD symptoms were shown [20,55,56,58], although the domains on which a beneficial effect was shown differed: anxiety [20], language [58], socialautistic [58], core symptoms of ASD [58], associated symptoms of ASD [58], hyperactivity [56], lethargy [55,56], stereotypical behavior [56] and daily living [55] (see Table 4). However, in two of these meta-analyses also some detrimental effects of fatty acid supplementation were shown. De Crescenzo showed a detrimental effect on sleep quality [20], and Horvath showed a detrimental effect on externalizing behavior and social skill [55]. In none of the meta-analyses evidence for heterogeneity or publication bias was found [20,55,56,58]. In one meta-analysis with only two RCTs no significant effect of N-3 fatty acids supplementation on ASD was shown [57]. Note that in the vast majority of studies included in the meta-analyses a rather limited number of participants was included (i.e., varying between 9 and 77 participants).
In addition, two original studies were found (see Table 5). In both studies, significant beneficial effects of N-3 fatty acid supplementation on ASD symptoms were shown [64,65]. Table 5. Studies concerning N-3, N-6 or PUFA supplementation for ASD, that are not included in the meta-analyses and systematic reviews in Table 4.  As fewer than 2/3 of studies showed an effect in the same direction, no conclusion can be drawn about the effect of N-3, N-6 fatty acid or PUFA supplementation on ASD symptoms during childhood.

ASD 7 RCT
Vitamin supplementation was sig. more effective than placebo in treating the following symptoms and/or function groups: global severity (g = 0.464 *), language (g = 0.351 *), stereotypies, restricted and repetitive behaviors (g = 0.531 *), behavioral problems and impulsivity (g = 0.402), and hyperactivity and irritability (g = 0.426 *). And sig. more effective for the following clinical domains: core symptoms (g = 0.308 *), associated symptoms (g = 0.308 *) and clinical global impression (g = 0.403 *). +   No effect supplement on ASD symptoms (expressive and receptive language, general behavior, eye contact and sociability). 0 + = 2/3 or more of the included studies found a positive effect or association, − = 2/3 or more of the included studies found a negative effect or association, 0 = 2/3 or more of the included studies showed no effect or association was shown, +/− = the results in a systematic review or meta-analyses were mixed (i.e., <2/3 studies in the same direction), and there were at least five studies in which the effect/association was studied, *

Vitamin D
Five systematic reviews were found, studying the relation between vitamin D and ASD [17,21,61,66,67] (see Table 6). The systematic reviews included a total of five unique studies in which the effect of vitamin D supplementation on ASD symptoms was investigated were included. Note that one of these five studies has been retracted due to concerns regarding the credibility of the data [72]. Two of the four remaining studies showed beneficial effects of vitamin D supplementation on the Aberrant Behavior Checklist Scale (ABC) score, and ABC and CARS respectively. One study did not show differential change of ASD measurement scores for the supplemented and non-supplemented group. Lastly, one study only showed an improvement on the self-care subscale of the Developmental Disability-Children's Global Assessment Scale. However, these four studies were of poor quality (i.e., many were open label studies and included few participants).
In addition, five original studies (reported in seven manuscripts), in which the relation between vitamin D and ASD symptoms was investigated, were found [73][74][75][76][77][78][79] (see Table 7). In two larger scale RCTs, significant effects of vitamin D supplementation on respectively irritability and hyperactivity [76][77][78] and on stereotypical behavior were found [79]. In one cohort and one case control study, no significant associations were found between vitamin D supplementation at 2.5 years of age and ASD diagnosis at a mean age of 5.1 years [73]. In addition, no difference was found in the use of vitamin D drops during infancy between children with ASD and children without ASD [74]. Table 7. Studies concerning vitamin/mineral supplementation for ASD, that are not included in the meta-analyses and systematic reviews in Table 6. The SRS-social awareness subscale improved sig. greater for the DHA group and the combined group compared to the placebo. There was a trend in greater improvement for the SRS social communicative functioning score for the combined group and the SRS total score for the DHA group both compared to placebo. Sig. greater reduction on ABC irritability and hyperactivity for vitamin D and DHA group and a trend for a greater reduction for combined group all compared to placebo. Sig. greater reduction on ABC lethargy for DHA group compared to placebo.
Evidence for interaction between baseline inflammation marker and treatment response was shown, children with elevated inflammation benefited more from supplementation.  ABC and other ASD checklists and questionnaires 1 participants improved with folic acid, 1 participants showed no effect, 2 participants showed unclear effects.  As fewer than 2/3 of studies showed an effect in the same direction, no conclusion can be drawn about the effect of vitamin D supplementation on ASD symptoms during childhood.

Vitamin B6 + Magnesium
Six systematic reviews in which the effect of vitamin B6 supplementation with or without magnesium on ASD symptoms was investigated [17,41,60,[68][69][70] (see Table 6) were found. In these systematic reviews a total of five unique studies in which the effect of vitamin B6 in combination with magnesium on ASD symptoms was studied were included. Additionally, one study which looked at vitamin B6 and magnesium both separately and combined was included [89], and lastly one study in which only the effect of vitamin B6 was studied was included [90].
For the six studies in which the combined effect of vitamin B6 and magnesium was investigated, a beneficial effect was shown in four studies and no effect in the two others. For the two studies in which vitamin B6 was supplemented alone, a positive effect was found in one study and no significant effect in the other study. It should be mentioned that the number of participants included in the studies was very small (i.e., 37 participants in the largest study), and the exact outcome measures were unclear.
No conclusion can be drawn on the influence of vitamin B6, alone or in combination with magnesium on ASD symptoms, as fewer than 2/3 of studies showed an effect in the same direction.

Iron
One original study was found in which the effect of ferrous sulfate supplementation in children with ASD, insomnia and low iron stores was investigated [85]. No differences in changes in ASD behavioral measures (i.e., Aberrant Behavior Checklist irritability scale; Swanson, Nolan, and Pelham-IV; Repetitive Behavior Scale-Revised) was shown after supplementation between the ferrous sulfate group and the control group.
No conclusion can be drawn about the effect of iron supplementation on ASD symptoms in childhood due to a too-limited number of studies (i.e., <5 studies).

Folic Acid
In two systematic reviews one unique study was found that investigated the effect of folic acid supplementation [17,61]. In this study verbal communication, daily living skills, irritability, lethargy, stereotyped behavior hyperactivity, inappropriate speech, total score on the ABC, and internalizing problems improved more in the folic acid supplementation group than in the placebo group.
Additionally, one systematic review reported one trial in which the effect of folic acid in combination with vitamin B12 was investigated [41]. In this study, significant improvements on Vineland subscales were shown. Lastly, two orignal folate supplementation studies were found [81,82] (see Table 7). These studies, however, were of poor quality and did not show unambiguous results.
No conclusion can be drawn about the effect of folic acid supplementation for ASD symptoms due to a too-limited number of studies (i.e., <5 studies).

Zinc
We found one original study in which the effect of daily zinc supplementation on CARS score was investigated [88]. In this study, a significant decrease in CARS score was shown. However, no control group was included in this study.
No conclusion can be drawn about the influence of zinc supplementation on ASD symptoms in childhood due to a too-limited number of studies (i.e., <5 studies).

Vitamin B12
Seven systematic reviews identified a total of two unique studies in which the influence of vitamin B12 on ASD symptoms was investigated [13,17,[41][42][43]60,61] (see Table 6). In both studies an improvement in Clinical Global Impression was reported (probably only significant in one study).
As noted earlier, one systematic review identified one trial in which the effect of combined vitamin B12 and folic acid was investigated. This study did show some beneficial improvements on Vineland sub scales [41].
No conclusion can be drawn about the effect of vitamin B12 alone or in combination with folic acid on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies).

Amino Acids
In one systematic review seven studies in which the influence of amino acids on ASD symptoms was studied, were included [60] (see Table 6). Four of the seven studies included in this systematic review showed a positive effect of amino acid supplementation on ASD symptoms and especially irritability. It is, however, important to note that different amino acids were used in the studies; in four studies N-acetylcysteine was supplemented, in two d-Cycloserine and in one N,N-dimethylglycine.
No conclusion can be drawn about the effect of amino acid supplementation on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies) for each specific amino acid.

L-Carnitine and L-Carnosine
In three systematic reviews, a total of three studies that studied the influence of L-carnitine on ASD were included [13,42,71] (see Table 6). All three studies showed a significant favorable effect of L-carnitine supplementation on ASD outcome measures such as CARS and ABC. In addition, Brondino et al. reported one study on the influence of L-carnosine on ASD symptoms [41]. This study showed a significant beneficial effect of L-carnosine supplementation on the Gilliam Autism Rating Scale (GARS) score when comparing the intervention group with the placebo group.
We lastly located one original study reporting on the influence of 500 mg of L-carnosine per day on GARS score; in this study no significant change in autism severity was shown after supplementation [87].
No conclusion can be drawn about the effect of L-carnitine or L-carnosine supplementation on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies).

Multivitamins
One meta-analysis and three systematic reviews included a total of two unique studies in which the effect of multivitamin supplementation on ASD symptoms was investigated [41,43,58,68]. In one of the two studies beneficial effects of supplementation on ASD symptoms were shown. The other study did not show significant beneficial effects on ASD symptoms.
No conclusion can be drawn about the effect of multivitamin supplementation on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies).

Vitamin A
One study in which the focus was on vitamin A supplementation was found in a systematic review, another study was located as an original study [43,80] (see Tables 6 and 7). Only one of these studies [43] found a significant improvement in the CARS score and Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria.
No conclusion can be drawn about the effect of vitamin A supplementation on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies).

Other Nutrients
In the systematic reviews, a single study was included that looked at the influence of vitamin C [41,60], inositol [60], and flavonoids [41], see Table 6.
In one original study the relation between thiamine deficiency early in life and scores on ASD questionnaires later in life was investigated [83] and in another original study the effect of rerum (a supplement consisting of chondroitin sulphate, vitamin D3 and oleic acid) on ASD was investigated [84], see Table 7. Lastly, in one original study a personalized treatment regime to correct nutritional derangements and its effect of CARS was investigated [86].
No conclusion can be drawn about the effect of vitamin C, inositol, flavonoid, multivitamin, thiamine deficiency, rerum or a personalized nutritional program on ASD symptoms during childhood due to a too-limited number of studies (i.e., <5 studies).

Discussion
The goal of the current scoping review was to summarize the scientific literature, including identification of knowledge gaps, regarding the relation between nutrition and ASD symptoms in childhood. The current review shows the availability of studies in which therapeutic diets, specific food products, fatty acids and micronutrients for the treatment of childhood ASD symptoms were investigated. There were, however, no studies available on the relation between dietary patterns (i.e., the habitual total food intake) or macronutrients and childhood ASD symptoms. In addition, there was a large variability in nutritional aspects under study, size and quality of studies, and thereforewe could thus not draw conclusions for any of the therapeutic diets, specific food products, fatty acids or micronutrients. The lack of consistent findings derived from methodologically strong studies makes it impossible to formulate practical nutrition guidelines for the treatment of childhood ASD.
Although 98 studies were available, the focus of the majority of studies was on a few specific nutritional factors. For example, there were 28 studies (29%) with a focus on the GFCF diet and 13 with a focus on fatty acids (13%). Most nutritional factors were investigated in <5 studies (e.g., ketogenic diet, iron, folic acid and zinc), and therefore, no conclusions could be drawn about those specific nutritional factors. It is, thus, clear that there are major gaps in knowledge regarding the relation between childhood nutrition and ASD symptoms during childhood. Among others, the relation between dietary patterns and ASD symptoms, but also the causality between supplementation of various nutrients (including iron, zinc and magnesium) and ASD symptoms, remains unclear.
Simultaneously, the amount of studies available on the relation between nutrition and ASD symptoms during childhood suggests there is much interest in this topic. However, many of the studies were of poor methodological quality, which limits the evidence base for the relation between nutrition and ASD symptoms during childhood. This is even true for nutritional factors such as the GFCF diet and fatty acids, which have been studied in >5 studies. However, as the studies were of such limited methodological quality, no firm conclusions could be drawn.
The methodological quality of studies was limited by a number of factors, which should be considered for future studies. First, although many of the included studies only included children with a clinician-based diagnosis of ASD using standardized tools such as the DSM criteria, International Statistical Classification of Diseases and Related Health Problems (ICD) criteria, Autism Diagnostic Interview-Revised (ADI-R) or Autism Diagnostic Observation Schedule (ADOS), some studies did not report how the diagnosis was made or utilized other tools which might not be suitable for diagnosis such as CARS (for on overview of the tools used see Supplementary File S3 Tables S1 and S2). It is of course important that to assess the effect of nutrition factors on ASD in children, the actual participants be children with clinician diagnosed ASD. While ASD symptoms were mostly assessed by the use of an ASD measurement, a multitude of instruments were used, and often multiple ASD measurement instruments were used within a single study (for overview of used measurement instruments see Supplementary File S3 Tables S1 and S2). Not all ASD measurement instruments used measure the same concept, nor do they necessarily measure core-features of ASD. For example, the ADOS, CARS and SRS were developed as autism diagnostic tools and measure core-features of ASD, but many other scales measure associated symptoms, such as behavioral problems (i.e., ABC), adaptive behavior (Vineland Adaptive Behavior Scale (VABS)) or clinical global impression (CGI). If a scale does not measure a core symptom of ASD, the symptom might not be present in all participants included in the trial and the results of the trial are, thus, not valid. According to the European Medicine Agency, trials should first demonstrate efficacy on at least one core symptom of ASD. Only then, efficacy on other associated symptoms may be claimed [91]. Additionally, many of the studies utilized parent-reported outcomes, mostly as standalone outcome, but in some cases in combination with clinician-reported outcomes, for the assessment of the child's ASD symptoms (for overview of used measurement instruments see Supplementary File S3 Tables S1 and S2). And even though the agreement on the presence or absence of ASD symptoms is moderate to low between clinicians and parents, there are indications that parents and clinician do differ in their evaluation of specific behaviors [92]. It thus important that both clinician and parent-reports are included in studies. However, using multiple ASD measurements and, with that, analyzing multiple outcomes can lead to Type 1 errors if no correction for multiple testing is applied. Finally, many of the studies included very few participants and utilized an open or single blind design, which both lead to the introduction of biases into the study.
Future studies should use more rigorous methodology. Researchers should, for example, utilize validated instruments that measure core-features of ASD, predetermine one primary clinically relevant endpoint and correct for multiple testing if using multiple outcomes. Additionally, studies should include a sufficient number of participants, and utilize double blind research designs if possible (e.g., for supplements). Moreover, it is advisable to choose comparable designs (e.g., form of supplementation, measuring instruments, etc.), so comparisons between studies is easier, allowing more reliable pooling of data in meta-analyses and systematic reviews.
It has been noted that ASD is a heterogeneous disorder in its clinical representation, severity, and most probably also in its etiology [8]. Additionally, psychopathologic comorbidities such as anxiety, depression, ADHD and intellectual disabilities are often present [93]. Taking these facts into consideration, it seems very likely that there is not a one-size-fits-all treatment for every person with ASD, and personalization is most likely needed. It might be that there are specific subgroups with certain biochemical or clinical features in which certain dietary interventions are specifically effective. To determine this, more research is needed and baseline characteristics such as baseline severity of ASD, comorbidities and demographics should be reported to reduce the heterogeneity of subgroups [94] and to enable elucidation of personalized treatment approached.
Furthermore, many of the included studies focus on the effect of isolated nutrients on ASD. A combination of nutrients may be more likely to evoke an effect than one isolated nutrient [95,96], because it matches the physiological requirement of the body better by representing the array of nutrients which are present in natural food and human diets. It would thus also be advisable to investigate the effect of combinations of nutrients. Additionally, others have postulated that dietary patterns may be more predictive of disease risk than single nutrients [97]. To the best of our knowledge, there are no studies available investigating the relation between dietary patterns and ASD. This relation does warrant research.
The main strength of the current review is its broad scope, investigating a wide range of nutrition factors related to ASD in children. A comprehensive search strategy was used, and rigorous methods were applied for ranking the quality of studies.
This review also has a number of limitations. We utilized a broad search strategy, as we wanted to include a broad range of studies investigating the relation between nutrition in relation to ASD in childhood. However, even with this broad search strategy it might be the case that we missed studies. Although, as we used a conservative number of studies (i.e., a minimum of five studies) with an effect in the same direction to draw a conclusion, it does not seem likely that studies that we missed could have changed the final conclusions. Additionally, this scoping review was a review of meta-analyses/systematic reviews and we chose to summarize the results in a qualitative manner. This is a common method in the execution of reviews of meta-analyses/systematic reviews. However, its main disadvantage is that original studies are counted multiple time (i.e., if an original study is included in multiple meta-analyses/systematic reviews). Consequently, the results of some studies were given more weight in the strength of evidence allocation. In future analyses this should be corrected for. In classifying the strength of the evidence (see Table 1), we took the strength of research designs as a starting point. For meta-analyses and systematic reviews, we did not consider their quality in the strength of evidence, which makes further refinement in the strength of evidence impossible.

Conclusions
The evidence for the relation between dietary patterns, therapeutic diets, and nutrients and ASD symptoms in childhood is generally insufficient and research suffers from methodological limitations. The relatively high prevalence and high social and personal costs associated with ASD in childhood justifies more research into the potential relation between nutrition and ASD symptoms in childhood. Future research should be methodologically sound, additionally studies on the relation between dietary patterns and ASD symptoms in childhood are needed. Considering, the results of this review, no clinical or practical recommendations regarding nutrition for ASD symptoms in childhood can be made.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nu14071389/s1, File S1: search strategy; File S2: Original articles in systematic reviews and meta-analyses and Original articles in each in systematic review/metaanalyses per topic. File S3: Diagnosis and effect of treatment questionnaires used (Table S1: Diagnosis and effect of treatment questionnaires used in studies included in meta-analysis and systematic reviews; Table S2: Diagnosis and effect of treatment questionnaires used in original studies).