Childhood asthma has previously been linked to individual foods, nutrients, and broad diet-quality metrics, but the role of specific early-childhood dietary patterns and their microbial or metabolic correlates is not well defined. This study used data-driven approaches to identify co-consumption patterns at age 3 and evaluated associations with concurrent and later childhood asthma/wheeze, while also characterizing related gut microbiome and stool/plasma metabolomic profiles.
The primary cohort analyzed food frequency questionnaire data collected at age 3 from participants in the Vitamin D Antenatal Asthma Reduction Trial (VDAART). Food-frequency variables were log-transformed, energy-residualized, and standardized before dimensionality reduction.
Principal component analysis (PCA) was used to derive dietary patterns (PCs) representing common food co-consumption axes. Associations between dietary PCs and age-3 asthma/wheeze were tested with covariate-adjusted logistic regression. Prospective associations were assessed using age-6 asthma/wheeze as the outcome.
To seek external support, the authors evaluated leading PC food-cluster proxies in the National Health and Nutrition Examination Survey (NHANES) 2021–2023 among children aged 2–3 years, and performed sensitivity analyses in broader age windows (2–5 and 2–8 years).
Selected PCA-derived components were tested for associations with the gut microbiome and with stool and plasma metabolomic features to identify biological correlates of diet-associated asthma signals.
The PCA revealed multiple components describing contrasts in early-childhood eating patterns. Notably:
These PCs served as the primary exposure variables tested against asthma/wheeze outcomes and downstream microbiome/metabolome features.
Among PCA-derived patterns, PC3 demonstrated the strongest positive cross-sectional association with age-3 asthma/wheeze. The reported odds ratio per 1-standard-deviation increase in PC3 was 1.42 (P = 0.00109). Other components, including PC1, were characterized but PC3 emerged as the leading signal associated with concurrent asthma/wheeze.
Age-3 dietary PCs were also prospectively associated with age-6 asthma/wheeze. When the full set of age-3 PCs was included in prospective models, the overall dietary-pattern set improved model fit in permutation testing: likelihood-ratio statistic = 20.3 with an empirical P = 0.033. This prospective analysis included 394 cases and 397 controls, indicating that early-childhood dietary patterns had associations that persisted to age 6 in this sample.
The study sought external support for the identified food-cluster signals using NHANES 2021–2023 data. A proxy representing the PC3 food cluster was positively associated with current asthma among children aged 2–3 years in NHANES, with an odds ratio of 1.59 (95% confidence interval, 0.95–2.67). Sensitivity analyses extended the evaluation to children aged 2–5 and 2–8 years. The reported NHANES association is directionally consistent with the VDAART findings, though the confidence interval includes the null.
PC3 was linked to variation in the gut microbiome and to differences in stool and plasma metabolomic profiles. Metabolomic features that associated with PC3 included annotations related to steroid sulfate, vitamin E–related, nucleoside-related, and lipid-related metabolites. These molecular signatures provide biological context for the diet–asthma associations, suggesting potential metabolic pathways that accompany the processed meat/fried-food dietary axis.
The authors interpret their results to mean that early-childhood dietary signals associated with asthma/wheeze are better captured as patterns of co-consumed foods rather than as isolated single-food effects. The identified PC3 (processed meat/fried-food axis) was consistently associated with concurrent and prospective asthma/wheeze and showed directionally consistent external support in NHANES.
Limitations reported or implied by the manuscript include that this work is a preprint and has not undergone peer review, so findings should not be used to guide clinical practice. Specific details on covariates included in adjusted models, effect sizes for other PCs, and full metabolite and microbiome feature lists or statistical adjustment methods beyond what was summarized in the abstract are reported in the full manuscript; readers should consult the complete preprint for full methodological detail.
The study reports that data described in the manuscript, a code book, and analytic code will be made available upon request pending application and approval. The research was conducted within the VDAART clinical trial (NCT00920621). The authors stated that IRB approval was obtained from the participating centers and that written informed consent was provided by study subjects.
In this preprint cohort analysis, early-childhood dietary patterns—particularly a processed meat/fried-food axis (PC3)—were associated with increased odds of childhood asthma/wheeze cross-sectionally at age 3 and prospectively at age 6. These diet patterns also correlated with gut microbiome composition and stool/plasma metabolomic signatures, including steroid sulfate, vitamin E–related, nucleoside-related, and lipid-related metabolites. The authors conclude that modeling food co-consumption patterns can reveal diet–asthma associations that are not apparent when examining single foods alone. As a preprint, these findings await peer review and should not be used to change clinical practice at this time.