597 - Oral and Nasal microbiome analyses of pediatric patients for the development of novel diagnostic and prognostic biomarkers of Autism Spectrum Disorders (ASD)

Poster Number: 597
Publication Number: 597.309

Naina Mehta, Arnold Palmer Hospital for Children, Orlando, FL, United States; Eric M. Jaryszak, Pediatrix Medical Group, Maitland, FL, United States; Devendra I. Mehta, Florida State University College of Medicine, Orlando, FL, United States; Holly M. Winbigler, Orlando Health, Orlando, FL, United States; William Morgan, Arnold Palmer Hospital for Children, Orlando, FL, United States; Joseph Manipadam, Orlando Health, Maitland, FL, United States; Florence George, Florida International University, Miami, FL, United States; Chirajyoti Deb, Translational Research, Orlando Health, Orlando, FL, United States

Background: Human gut, oral, and nasal microbiomes have been shown to have some association with ASD. However, we still do not have well characterized microbiota or their functions useful for accurate ASD diagnosis and therapeutic interventions.

Objective: We are investigating the compositional and functional roles of key bacterial species of both oral and nasal microbiome of clinically characterized patient cohorts to develop more appropriate diagnostics and prognostic tests for pediatric ASD.

Design/Methods: In this IRB approved prospective cross-sectional study we are exploring microbiome profile differences in oral and nasal samples from pediatric patients via next-generation sequencing (NGS) and bioinformatic analyses of bacterial 16S rRNA gene amplicons. Three groups of pediatric patients were recruited at Orlando Health: 1. Autism group (ASD)- (n=26): clinically diagnosed with ASD; 2. Control Group A (n=18) –Non-autistic patients with neurotypical cognitive skills; and 3. Control Group B (n=41) –Non-autistic patients with similar but non-ASD morbidities (Table 1A).

Results: Oral samples showed higher species level variation compared to nasal samples from all the three study cohorts. Granulicatella elegans, Haemophilus parahaemolyticus, Porphyromonas endodontalis, Alloprevotella LT608321_s, Abiotrophia defectiva, Porphyromonas JF239777_s, and Leptotrichia PAC001350_s species were more prevalent in the ASD group oral samples (Table 1B; Figure 1A). The phylum Fusobacteria was abundant only in the Autism group among nasal samples (Figure 1B). The Gemella haemolysans group was prevalent in the ASD group nasal samples (Table 1B; Figure 1C).
Logistic regression analysis with the bacterial species prevalence data identified Granulicatella elegans and Haemophilus parahaemolyticus in oral samples, as well as Gemella haemolysans and Moraxella nonliquefaciens in nasal samples, to predict ASD (Figure 1 & Figure 2). A Kruskal-Wallis H test analysis identified functional biomarkers differentiating the ASD and control groups. The KEGG Rap1 signaling pathway orthologs (ko4432; ko4456) were significantly (p=0.011) abundant in the ASD group compared to the controls (Table 1C). Shannon diversity index data were similar for both oral and nasal samples between the ASD and control groups (p= >0.05). Conclusion(s): We have identified novel microbiome taxonomic profile and functional differences in oral and nasal bacterial samples from ASD and control groups, suggesting significant predictive links to ASD. The taxonomic and functional biomarkers are expected to help developing better diagnostic and prognostic tools.
Table 1. Demographic and clinical characteristics of the three study cohorts – Autism group, Control Group A & Control Group B; Relative abundances and frequency of prevalent bacterial phyla and species in oral and nasal samples from the three study cohorts mentioned above.Table 1A: Demographic and clinical characteristics of the study the three study cohorts – Autism group, Control Group A, & Control Group B. Age in (y) is average ± standard deviation. Inclusion criteria: Autism group: Patients with established diagnosis of Autism. Control Group A: Patients who are clinically non-Autistic or neurotypical patients with normal cognitive skills. Control Group B: Non-Autistic patients with additional behavioral and developmental morbidities. Defined to have at least one morbidity from chronic illnesses such as asthma, GERD, Diabetes, Down Syndrome, other syndromes, and any previous non nasal surgery. Any cognitive dysfunction and/or neurodevelopmental disorders such as: ADHD, Learning Disabilities, Cognitive Impairment, Developmental Delays, Developmental Language Disorder, Developmental Motor Coordination Disorder, Expressive receptive language disorder, Sensory Processing Disorders, Anxiety, Depression, Oppositional Defiant Disorder, Disruptive mood dysregulation Disorder, Social Communication Disorder, Mood disorder. All available clinical data will be presented. Exclusion Criteria: Patients with the following symptoms will be excluded for both Autism and Control groups: Acute sinusitis, currently being treated with oral antibiotics, Current use of antibiotics, Emergent cases, Nasal surgery after the age of 14, Bleeding diathesis as reported in history by parents and/or medical history, Acute oral or upper respiratory tract infections past 3 days as reported in history by parents and/or medical history.
Table 1B: Relative abundance, and frequency of prevalent bacterial taxa in oral and nasal samples from the three study cohorts – Autism group, Control Group A, and Control Group B. Relative Abundance is defined as a normalized percentage of counts per sample. Relative Abundance (in %) is average ± standard deviation. Frequency is defined as % of samples per group in which organism was detected. Most prevalent organisms for each group determined by highest raw relative abundances. * indicates organism was not considered prevalent in its respective group while prevalent in others. Zero (0) indicates organism was not at all detected in its respective group.
Table 1C: Functional biomarkers: Mean relative abundances of 2 KEGG pathways and 4 KEGG orthologs involved in pathways identified using Kruskal-Wallis H test analysis. KEGG: Kyoto Encyclopedia of Genes and Genomes. Rap1: Ras-related protein 1. Note: genetic defects that both up and/or down regulate Ras/Rap signaling pathways (ko04015) have been linked with several mental disorders associated with learning disability. RAC serine/threonine-protein kinase (k04456) signaling has been linked to ASD, and dendritic abnormality has been noted in rat models. MAPK3 (k04432) is located on the Autism-linked Human 16p11.2 Locus. EzBioCloud (https://www.ezbiocloud.net/) functional biomarker discover uses the PICRUST tool to make inferences from the 16S data using the KEGG database, so please refer to the PICRUST paper (https://www.nature.com/articles/nbt.2676) which explains how these functional predictions can be made from 16S data. Detailed plausible implications of the identified functional pathways and modules will be discussed.
Figure 1. Comparative analyses of relative abundances of the major bacterial species in oral and bacterial phyla and species in nasal samples from the three study cohorts (Autism, Control Group A, & Control Group B).Figure 1. Comparative analyses of relative abundances of the major bacterial species in oral and bacterial phyla and species in nasal samples from the three study cohorts (Autism, Control Group A, & Control Group B). A. Species level relative abundance comparison in oral samples from Autism, Control Group A, and Control Group B patients. B. Phylum level relative abundance comparison in nasal samples from Autism, Group A, and Group B patients. Phylum Fusobacteria was only detected at a higher than 1% relative abundance in nasal samples from Autism group. C. Species level relative abundance comparison in nasal samples from Autism, Group A, and Group B patients.
Alpha diversities were comparable among the three study cohorts. No major differences were observed in the control groups A and B, in terms of bacterial species diversities in both oral and nasal samples. Species richness (Chao 1) difference between Control Group A and B oral samples was found to be significant (p= < 0.05). Descriptive statistics of the differential profiles of identified bacterial species and phyla are presented in Table 1B and plausible functional biomarkers are presented in Table 1C. Statistical analyses with the differentiating and disease (autism) predictive bacterial species are presents in Figure 2. With the data obtained using both oral and nasal samples from the current sample size (three study cohorts) we were able to discover some significant taxonomic (bacterial species specific) and functional biomarkers (biochemical and metabolic pathways) useful for differentiating and predicting Autism compared to the control subjects. We are hoping that with increased number of subjects we will be able to identify more of potential taxonomic and functional biomarkers for diagnosis and therapeutic interventions of autism.