Meals, Microbiota & Mental Health of Children & Adolescents

Overview

Recent studies indicate that the interplay between diet, intestinal microbiota composition, and intestinal permeability might impact mental health. The aim of thisl study is to compare diet, intestinal microbiota, intestinal permeability, and related metabolic factors among children and adolescents diagnosed with mental health disorders and control groups and identify potential relationship patterns. All children and adolescents referred to the outpatient clinic at the Child and Adolescent Psychiatry Department at The National University Hospital in Reykjavik Iceland will be offered to participate (n=15) (age 5-15 years). Two control groups will be used; the same parent siblings close in age (n=x) as well as age and sex-matched children from the same postal area (n=15). A three-day food diary, rating scales for mental health and multiple questionnaires will be completed as well as faecal sample, buccal swab, urine, saliva and blood samples will be collected. This is a novel approach as more multidimensional transdisciplinary studies including longitudinal observational data have been called for as a basis for lifestyle treatment options for improving mental health and wellness.

The current pilot study will be used to test the feasibility of the methods used. If the methodology is found to be feasible, the n=40+ cases and controls/siblings in the pilot study applied for here will be included in the larger study. For those cases who are invited to participate and decline participation, information on final medical diagnosis of the child will be collected from journals, but no other information. In the current pilot study (information from first 20 cases, controls and siblings) the participants will be asked if they are willing give reasons for not participating (while at the same time assured that they absolutely do not need to do so). Questionnaires on diet, mental health and background data will be web based with help of the research data management company Smart Trial (https://www.smart-trial.co) transforming them into a readable electronic format. This procedure has been thoroughly quality-checked and will secure proper storage of data. Testing the dietary programs for nutrient calculations from food and supplement intake data, based on the Icelandic food composition database (ISGEM)1,2 as well as software for mental health questionnaires analysis. While on the waiting list to come to the outpatient clinic, the guardians and child will receive invitation letters. If interested in participating after a follow-up telephone call, a home visit follows from the project deputy of UNR, where the informed consent is signed. Contact information to the child´s teacher is also received at this stage. A kit including tools for collecting biological samples with good instructions as well as a booklet for the three-day dietary recording will be provided. The questionnaires collecting additional information such as the short validated frequency on food and supplement intake, the ICEFFQII, designed for the Icelandic food environment3,4, food aversions, physical ailments or diagnosis ROMEIV5-7, atopic diseases iFAAM8 and medication as well as other important background factors are completed online with each participant having codes used for unlocking the questionnaires. The mental health questionnaires used in this study have all been validated internationally as well as in Iceland. These are; 1) Teacher and parent-reports: The ADHD Rating Scales9-11, the Child Behavior Checklist (TRF and CBCL) a family of symptom checklists developed to provide information on a broad range of emotional and behavior problems.12-15; the High-Functioning Autism Spectrum Screening Questionnaire (ASSQ)16 and the Social Responsiveness Scale (SRS)17-21. 2) Self-reports for children \>10y: The Children's Depression Inventory (CDI)22-26 and the Multi-Dimensional Anxiety Scale for Children (MASC)27,28-33. Furthermore, if allowed by the guardians, information on prescription medication dispensed for the child from birth to current date will be retrieved from Icelandic Medicine Registry (IMR). Further information on the questionnaires used can be found in Appendix 1. After the collection of data from the case, a randomly selected child, matched by gender and age, will be found within the same postal area by Social Sciences Research Institute (SSRI) at the University of Iceland and is invited to participate by sending an introduction letter, followed by a phone call. The SSRI will find 5 controls for each case that has finalized the study as participation rate might be low. These will be called in the sequence provided by the SSRI. The biological samples are preferably to be collected towards the end of the 3-day food registration, with help of guardians, according to standard procedure using dedicated devices, including stabilization factors, and stored cold (4-8°C). The kit is then collected from both cases and controls by the researchers along with the food diary when close Reykjavik, Akureyri and surroundings. Participants can also deliver the samples either to BUGL or the Nutritional Unit, both at Landspitali University Hospital. For other locations the samples will be sent out including guidelines on how to resend them to the researchers after their collection. The urine samples will be aliquoted into smaller tubes for storing at a laboratory unit at the University Hospital or Akureyri Hospital in a -80°C freezer along with the other biological samples. Over the year of recruitment, diagnosis of all patients coming to BUGL will be recorded to be able to detect potential selection biases in data collection. Work on the input of data will be kept to the minimum through the web-based questionnaires where information will be entered directly into data sheets. Information on current diet will be calculated for meal frequency, food frequency and food intake as well as energy, macro- and micronutrient intake from both food and supplements using the three-day food diary and a food frequency questionnaire (ICEFFQII)3 and information on food aversions. Physical ailments are evaluated using the diagnostic questionnaire for pediatric functional gastrointestinal disorders for children and adolescents (ROMEIV)34 and a questionnaire on atopic reactions for children (iFAAM)8. Additionally, information from the parent's questionnaires on history of pregnancy and birth, breastfeeding, history of food sensitivity, defecation frequency, traumas during lifetime, socio-economic factors, physical activity, sleeping habits and anthropometric data will be collected. Quality checks will be performed throughout the period as well as simple descriptive statistical analysis. Some of the questionnaires on mental health need special software to run which will be bought for this study. Furthermore, work on the input of data will be kept to the minimum through the web-based questionnaires where information will be entered directly into data sheets. Mental health status of cases is based on best expert clinical diagnoses usually collected during a long period of time, often with semi-structured clinical interviews. These longitudinal data contain both dimensional and categorical data (with diagnoses), with comprehensive and multi-informant clinical information (patient, parents, school etc.). Information on medication of the child from birth on will be retrieved from Icelandic Medicine Registry (IMR). All samples will be measured at the same time point at the end of the study period and are sent to their respective locations for analysis. Samples will be randomized with regard to case-control status before any measurements are performed. The following will be measured: 1. Fecal samples, saliva and buccal swab: Microbiome Shotgun analyses will be performed according to previous published protocols targeting 16S rDNA71-73and genotyping will be performed on human DNA isolated from buccal swab samples using PCR based methods. High quality DNA from swab and fecal samples will be extracted using the commercially available QIAamp PowerFecal DNA extraction kit. DNA isolation will be followed by PCR amplification using universal prokaryotic primers. Sequencing will be performed using an Illumina high-throughput platform and pair-end modality using standard protocols. The resulting sequences will be analyzed using the end-to-end microbiome analysis pipeline QIIME 2 for sequence quality control, chimera removal, assigning sequences to samples and defining sequence variants (SVs). SVs will be assigned to taxonomy using 16S rRNA reference databases, and phylogenetic diversity metrics. Interactive visualization and statistics will be computed using QIIME 2 to interpret and compare microbiome compositions. The data sets of microbial diversity data gathered from the gut samples, will be used for the construction of a microbiota database. Genotyping will be performed on human DNA isolated from buccal swap samples using PCR based methods. 2. Urine sample: A morning urine sample is collected at home in a tube with peptide inhibitors. The urine is divided into two samples, one for analysis of targeted analysis of food related peptides on a HPLC reverse phase (C-18) chromatography that gives total peptide amount related to creatinine. The other urine sample will used for untargeted metabolomics profiling (LC-ESI-qTOF-MS). 3. Blood samples: Inflammatory biomarkers will be analyzed by multiplex immunoassays based on the mesoscale discovery platform (e.g. CRP, interleukins) and nutritional status (haemoglobin, ferritin, folic acid, Vitamin B12, 25-hydroxy-vitamin-D, ⍵-3 fatty acids by kits in the Landspitali laboratory. Measurements of zonulin from serum will be performed by assay based on the method of competitive ELISA.35 For each type of omics and biomarker data (microbiota, metabolomics, food related peptides, inflammatory factors, nutrients) appropriate preprocessing and quality control will be coordinated and performed by the data generating groups as indicated above. All data will be appropriately normalized and transformed, checked for potential batch effects and outliers removed. The effects of potential confounders such as age, sex, body mass index, Bristol stool scale (for intestinal microbiome), medication or diet will be investigated. For each datatype we will compare the case and two control groups, but also look at associations between the different variables separately as well as in relation to sub-groups, using multivariate models including adjustments for confounder variables. Benjamini-Hochberg false discovery rate will be used to correct for multiple hypothesis testing. For oral and intestinal microbiome data we will perform principal coordinates analysis to investigate the compositional differences between samples. The case and control groups will be compared with regard to microbial diversity (alpha- and beta-diversity) and taxonomic abundances. Here we will apply methods developed for microbiome data to account for sparsity and under sampling of the microbial community, such as implemented in the R package.36 Correlations between microbial species will be mapped using sparCC37, which is designed for compositional data such as microbial relative abundances, and compared between case and control groups to reveal potential differences in the microbiome community structure. For the dietary data the FFQs will be analyzed through PCA identifying dietary patterns in the groups, while the three day registration will be used to calculate a more exact intake of food items, calculated forward into nutrients (energy, macronutrients (protein, fat, carbohydrates/fibers)) and micronutrients (vitamins, minerals and other important factors). Further, the dietary data as well as information on the nutritional status, inflammatory factors and omics data, will be used to study associations with oral and fecal microbiota. The same analysis will also be made in the larger subgroups for example children with ADHD or autism. Genotype data on study participants will be used to investigate how host gene anchors might predict or interact with particular microbiota. Here we will focus on genetic variants previously associated with microbiota composition or psychiatric disorders. Integrating different types of high-dimensional omics data is a challenging task and multiple approaches can be undertaken. A computational framework for such analysis38 has been revealed39, based on dimensionality reduction of the omics data using co-abundance clustering, followed by cross-omics correlation analysis that is focused on features that differ between the cases and controls. In addition, we will apply other types of omics integration approaches as recently reviewed40,41, such as robust sparse canonical correlation, co-inertia or Procrustes analysis, to identify shared patterns across omics and dietary data in the context of mental health.