Autism spectrum disorder (ASD) is a complex of neurodevelopmental conditions with increasing prevalence. A large proportion of children with ASD are reported to have food selectivity. They refuse to eat a diet based on a variety of foods and consume a narrower food repertoire consisting of about two-thirds the number of foods as typically developing children [1
]. Sensory selectivity, based on tactile/texture, gustatory, and olfactory oversensitivity, is proposed to contribute to the development of food selectivity [1
]. Additionally, children with ASD are often subjected to some type of dietary intervention and elimination of some foods or food groups [3
]. These feeding habits raise the question of nutritional adequacy [1
]. As an example, it was shown that children with ASD consume fewer proteins and more carbohydrates than the general population [4
], and their diets may be low in micronutrients [4
]. The most frequently omitted food group is vegetables, followed by fruits [6
]. Evidence shows that feeding problems are associated with gastrointestinal (GI) and intestinal microbiota change in children with ASD. It has been established for the general population and confirmed using animal models that the consumption of particular diets shifts gut microbiota to specific bacterial genera [7
]. In particular, a high-fat diet deceases α-diversity, Firmicutes/Bacteroidetes
abundance and increases Alistripes
]. Supplementation of omega-3 fatty acids has been shown to temporarily shift intestinal microbiota towards bacterial species producing short-chain fatty acids [9
]. Dietary intake of non-fermentable fibre is reported to increase such genera as Helicobacter
, while decreasing genera such as Lactobacillus
and TM7 genera Incertae Sedis
. This leads to changes in the metabolic profile that are beneficial for the prevention of autoimmune diseases [10
]. Recent meta-analysis including studies with different kinds of dietary fibre concluded that fibre intake leads to a higher faecal abundance of Bifidobacterium
spp. but does not affect α-diversity [11
The amount of a specific macronutrient in the diet, as well as its source, is important for the microbial configuration of host intestines. Type, composition and quantity of dietary proteins are associated with specific intestinal microbiota, thus they influence microbial metabolites in the intestine, and modulate the function of intestinal barrier and host immune defence [12
]. High animal protein intake increases the Bacteroides/Prevotella
ratio and even changes the microbial enterotype of the host. Furthermore, the nature of food intake affects intestinal microbiota. but in response, the intestinal microbiota itself can regulate eating habits, stimulating the intake of foods beneficial to them and suppressing competitors [13
]. Thus, the feeding habits of children with ASD raise several questions including the role of nutritional inadequacy of the diet, the connection of food composition with GI disorders, and the consequences in terms of a change in their intestinal microbiota.
In addition to the above mentioned, it is known that gut microbiota in children with ASD is different compared with control children [14
]. The ratio of the main phyla Bacterodetes
switches, and genera such as Clostridium, Sutterella, Lactobacillus
, Nitriliruptor, Youngiibacter, Methanomicrobiales, Bilophila
change in abundance. This change displays an association with behavioural manifestations. This association suggests bidirectional influence of microbiota and brain, explaining the recent interest in the microbiota–gut–brain axis [17
]. The different pathways involve the leakage of bacterial metabolites into the blood, and these, through cascades of reactions, alter the tight junctions in the blood–brain barrier, cortex, hippocampus, amygdala and cerebellum [18
]. A recent review of nutritional interventions for gut microbiota modulation in ASD [19
] has revealed that in this area, the gaps in knowledge are greater than what is known.
Taken together, the specific microbiota reacts to an individual’s diet in a personalised way [20
]. The interplay between diet, microbiota and host, in children with ASD along with their distinct microbiota, would probably be different from that of neurotypical children.
Thus, this aim of our study to make a step towards elucidating the connection of eating disorders and specificity of food intake, with intestinal microbiota composition in children with ASD, keeping in mind that some microbiota changes might be associated specifically with their neurodevelopmental disorder.
3. Materials and Methods
The study included 62 boys, of which 46 had ASD and 16 were control non-autistic children. The characteristics of the groups are included in Table 1
. Written informed consent was obtained from parents of all participating children.
Children with ASD were recruited from the Academic Research Centre for ASD (ARCA) based at the Institute of Physiology, Faculty of Medicine Comenius University in Bratislava, Slovakia. Psychological evaluation of children with ASD was performed using the ADOS-2 (Autism Diagnostic Observation Schedule, Second Edition) [38
] and ADI-R (Autism Diagnostic Interview, Revised) [39
] behavioural assessment scales, which are internationally accepted gold standards for the diagnosis of ASD. ADOS-2 evaluated the behaviour in domains of social affect (SA), distinguishing communication (COM) and reciprocal social interaction (RSI), restricted and repetitive behaviour (RRB) scores, and total raw score (Total). ADI-R was evaluated in the areas of qualitative abnormalities in reciprocal social interaction (A) and communication (B), as well as in restricted, repetitive and stereotyped patterns (C) of behaviour. All subjects involved in the study met the criteria for ASD using both diagnostic tools. The diagnosis of ASD met the criteria for DSM-V (The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition). Control subjects were recruited from local kindergartens and schools and had no psychiatric conditions according to their parent interview. Written informed consent was obtained from parents of all participating children. The protocol was approved by the Ethics Committee of the Comenius University Faculty of Medicine and the University Hospital, APVV 15-0085 approved on 13.06.2016. The study conformed to the code of ethics stated in the Declaration of Helsinki.
Data on gastrointestinal (GI) status and mealtime behaviours were evaluated based on the parental questionnaires. The prevalence and frequency of gastrointestinal symptoms (abdominal pain, bloating, constipation, diarrhoea, hard stools, pain during defecation, voluminous stools) were obtained. Based on their prevalence, the GI score was calculated; the higher the value, the more prevalent the GI problems were. BMI was calculated as body weight in kilograms/square root of height in meters. BMI standard deviation scores (BMI-SDS) were calculated using the reference data of the Slovak population in order to determine the deviation in BMI from the mean BMI of the general population of children of the same age and gender [40
]. The prevalence and frequency of adverse mealtime behaviours was recorded, including selectivity in food intake, as well as anger, crying or self-injuries associated with food intake. Child insistence on having the food prepared and served in the same manner was defined as stereotyped behaviour during mealtimes. The score of mealtime problems was calculated based on their prevalence. Higher values indicated more severe mealtime problems. Data on modifications of the child’s diet were recorded, as well as information about the use of food supplements in the last 12 months.
Of the total 62 individuals, data from 30 children with ASD and 16 controls were collected about the typical diet of the subjects by a self-administered food frequency questionnaire (FFQ) validated for the Slovak population that included 85 food items. Parents of children were requested to indicate the frequency of consumption of each food item (with options including times per day, times per week, times per month, or almost never) together with the portion size. Average daily consumption of selected food types (g/day) was calculated. Food intake data served for the calculation of the nutrient intake of the individuals, vegetable and fruit intake was adjusted for seasonality. For the conversion of food intake into nutrient intake, the Slovak food composition database was used (Slovak Food Composition Bank. Revision 2004 (2004). Food Research Institute, Bratislava 2004) (http://www.vup.sk/en/index.php?navID=25?start
). Calculations were performed using Microsoft Excel standard formulas.
Stool samples were collected by parents at home in sterile flasks. Parents were given a detailed explanation of the procedure and stored the samples at + 4 °C after collection. Samples were delivered to the laboratory within 4 hours and divided into 2 aliquots, one of which was immediately frozen at −80 °C for future DNA purification of intestinal microbiota assessment and the other frozen at −20 °C for calprotectin investigation.
DNA was extracted from frozen stool samples using a commercial extraction system (QIAamp DNA Fast Stool Mini Kit, Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The DNA concentration was determined using the NanoDrop 2000 Spectrophotometer (ThermoFisher Scientific, MA, USA).
High-throughput sequencing was performed for the DNA libraries. The PCR amplification of V1-V9 region of 16S rRNA using primer set 27f / 1492r was carried out using the 27f (5’-AGA GTT TGA TCM TGG CTC AG-3’) and the 1492r (5’-CGG TTA CCT TGT TAC GAC TT-3’) primers (Lane, 1991). In this process, 3-50 ng of total input DNA in 20 µL volume PCR reaction was amplified with 4µl of 5x HOT FIREPol Blend Master Mix (Solis BioDyne, Tartu, Estonia), 0.4 µL (10 µM) of each primer (final concentration 0.2 µM) and milli-pore water. PCR conditions were as follows: initial denaturation 95 °C/15 min, cycling 25× (95 °C /20 sec, 60 °C/30 sec, 72 °C/2 min), final polymerization 72 °C/10 min. Amplicons were column-purified (Zymo DNA Clean and Concentrator-5, Zymo Research, Irvine, CA, USA) according to standard protocols and quantified fluorometrically with Qubit™ dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Amplicon sequences were fragmented by a transposon-based approach (Nextera XT, Illumina San Diego, CA, USA) and low-cycle PCR and mutual indexing of the fragments was performed. Fragment size selection and purification with 1.8x AMPure XP beads yielded final DNA libraries that were verified using the Agilent 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany) and quantified using the Qubit 2.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). The 4 nM pool of libraries was further diluted to 10pM and sequenced on Illumina MiSeq platform (Illumina, San Diego, CA, USA) with 2 × 300 bp paired-end sequencing at the Comenius University Science Park (Bratislava, Slovakia). Library sequence data were quality checked using FastQC, Andrews, 2010, a quality control tool for high throughput sequence data available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc
. Further data processing, including trimming, 16S analysis and visualization, was performed with Geneious (Biomatters Ltd, Auckland, New Zealand). For visualizing and clustering of multivariate data using principal component analysis (PCA) ClustVis (https://biit.cs.ut.ee/clustvis/
] and MicrobiomeAnalystR [42
) were applied. For diversity calculations, the Bray-Curtis Index and Permanova test were used.
Data were presented as mean ± SEM values and p values lower than 0.05 were accepted as significant. Since the data distribution matched Gaussian distribution for the correlation analysis, the Pearson correlation coefficient was applied. For all statistical analyses, GraphPad Prism 5 and Microsoft Excel 2016 were used. Data on food intake did not pass the normality test, therefore they are presented as median ±95% CI (confidence interval), and their differences were tested by the Mann–Whitney test. For the testing of categorical variables, the chi-square test and Fisher’s exact test were used.
High carbohydrate, protein and fat intake corresponded to the consumption of more than 180, 40 and 50 g/day, respectively. High fibre, fresh vegetables and fresh fruit intake corresponded to the consumption of more than more than 10, 40 and 150 g/day, respectively. High fermented milk or white bakery intake corresponded to the consumption of more than 100 or 40 g/day, respectively. High omega-3 (linolenic, eicosapentaenoic, docosahexaenoic acids) intake corresponded to the consumption of more than 0.75 g/day.
Interest in the modulation of intestinal microbiota by specific food intake has increased in recent years. The degree to which eating habits influence gut microbiota in children with ASD is not clear as of yet. Children with ASD frequently have eating and GI disorders, which influence the intestinal microbiota and clinical manifestations and vice versa, with the microbiota influencing eating habits, GI status and behavioural disorders. Our study shows that nutritional inadequacies in children with ASD could be hidden behind normal BMI. Food selectivity, found more frequently in children with ASD, resulted in different micronutrient intake as compared with controls. Although the core microbiota composition was independent of diagnosis or specific nutrient intake, ASD group microbiota was characterised by Dichelobacter, Nitriliruptor and Constrictibacter, while Diaphorobacter and Nitratireductor were typical for the control group. For “picky eaters”, the gut microbiota was more diverse at the phylum level, distinct at the genera level and characterized by specific genera (Escherichia/Shigella and Salmonella, Clostridium XlVa, Anaerofilum). Changes in gut microbiota were similar in children with and without ASD. Higher carbohydrate intake changed beta diversity only in children with ASD, for example by decreasing Bacteroides abundance. Fibre intake was similar in groups, but increasing it in children with ASD decreased microbial diversity. Fresh fruit and vegetable intake were significantly higher in neurotypical children compared to children with ASD, and their consumption increased microbiota diversity exclusively in the control group. High intake of fermented milk products had strong effects, that were similar in both groups: they increased Lactobacillus and Fusicatenibacter abundance and decreased microbial diversity. Distinct microbiota were found when omega-3 consumption with food was increased in both groups. Aside from bacteria characterising the microbiota in autism, the reflection of microbiota to nutrients intake was similar in both groups, with the exception of fibre intake.
Food selectivity, as well as the consumption of fermented milk products, total fat, omega-3, animal/plant protein resulted in similar changes in the intestinal microbiota of children with and without autism. However, the effects of carbohydrates, fibre, fruits, vegetables intake were different.
Thus, although food intervention in children with ASD is difficult, such changes could help to alter the intestinal microbiota in such a way as to improve GI and immune status.