Microbiome and pediatric leukemia, diabetes, and allergies: Systematic review and meta-analysis

Abstract

Background: Despite the different pathologies and genetic susceptibilities of childhood ALL, T1DM and allergies, these conditions share epidemiological risk factors related to timing of infectious exposures and acquisition of the gut microbiome in infancy. We have assessed whether lower microbiome diversity (Shannon Index) and shared genus/species profiles are associated with pediatric ALL, allergies, and T1DM.

Methods and findings: Literature search was performed using PubMed, Embase, Cochrane, and Web of Science databases. Case-control, meta-analyses, and cohort studies were considered for inclusion. Inclusion criteria: (i) subjects age 1-18 years at diagnosis, (ii) reports effect of microbiome measured prior to/at time of diagnosis/first intervention (iii) outcome of ALL, allergies, asthma, or T1DM, (iv) English text. Exclusion criteria: (i) age < 1 or >18 years at diagnosis, (ii) Down Syndrome-associated ALL, (iii) non-English text, (iv) reviews, pre-print, or abstracts, (v) heavily biased studies. Abstract and full text screening were performed by two independent reviewers. Data extraction was performed by one reviewer following PRISMA guidelines. Data were pooled using a random-effects model. Eighty-eight studies were included in the analysis, with seventy-seven in the qualitative analysis and 54 in the meta-analysis. Cases were found to have lower alpha-diversity than controls in ALL (SMD:-0.78, 95%CI:-1.21, -0.34), T1DM (SMD:-1.26, 95%CI:-3.49, 0.96), eczema (SMD:-0.34, 95%CI:-0.56, -0.12), atopy (SMD:-0.06, 95%CI:-0.34, 0.22), asthma (SMD:-0.37, 95%CI:-1.16, 0.42), and food allergy (SMD:-0.11, 95%CI:-0.63, 0.41).

Conclusions: These results highlight similarities in the microbiome diversity and composition of children with ALL, T1DM, and allergies. This is compatible with a common risk factor related to immune priming in infancy and highlights the gut microbiome as a potentially modifiable risk factor and preventative strategy for these childhood diseases.

Fig 1. Flow chart of study selection. Created with Biorender.com.

Fig 2. Alpha-diversity (Shannon Index) in cases compared to controls: meta-analysis.

Abbreviations: SMD = standard mean difference, 95% CI = 95% confidence interval. Square data markers indicate the degree of difference in Shannon Index between cases and controls with the lines through the markers indicating 95% CIs. The size of the square data markers indicates the weight of the study. The diamond data marker indicates the overall pooled effect based on the included studies using a random effects model. Created with meta package in R and Biorender.com.

Fig 3. Bacterial composition of disease types compared to described bacterial composition of delivery modes and breastfeeding status.

Abbreviations: T1DM = type 1 diabetes mellitus, Cesarean = cesarean section delivery, Non-BF = non-breastfed, ALL = acute lymphoblastic leukemia, Food A = food allergy, Vaginal = vaginal delivery, Brst-Fed = breastfed. Clustering heatmap corresponding to the percentage of studies reporting increased or decreased relative abundance of cases compared to controls by disease types, delivery mode (A), and breastfeeding status (B). The horizontal direction are the disease types, delivery mode (A), and breastfeeding status (B), and the longitudinal direction are the bacterial genera. Studies reporting microbiome composition by disease type (ALL, T1DM, eczema, atopy, asthma, and FA) were identified by systematic review. Studies reporting microbiome composition by delivery mode and breastfeeding status were identified by non-systematic literature review[–,–102], and therefore are separated from the more highly curated disease-related studies on the left of each heatmap. Created with ComplexHeatmap package in R and Biorender.com.