Page 77 - MI-2-4

Microbes & Immunity Dietary emulsifiers affect the presence of AIEC

(Supplementary File). The food additives questionnaire was of variance (PERMANOVA) test. Differentially abundant
validated in a CD survey across Australia, Hong Kong, and taxa and functional modules were identified using a linear
mainland China to identify the exposure to food additives mixed model with the geographic region as a random
in CD patients and healthy controls. 11,20 Our study also effect. The batch effect of microbiome data was adjusted
22
incorporated a secondary dataset of CD patients and healthy by the MMUPHin method. A sensitivity analysis was
controls from Hong Kong as an urban cohort. The subjects performed to verify the robustness of differentially
8
in the urban cohort were recruited from the Prince of Wales abundant taxa using the adjusted microbiome data. After
Hospital in Hong Kong (population density = 6582.6/km ). selecting the significantly different genera, we validated the
2 21
All participants had not been exposed to antibiotics, discrimination ability of selected microbial genera using
probiotics, or prebiotics in the past three months before the random forest model. Five-fold cross-validation was
enrollment. All participants gave informed consent, and the applied during the model training. The model was trained
study was conducted in accordance with the Declaration of in 80% of CD patients from two cohorts and validated
Helsinki. The identification of AIEC presence, sample DNA on the remaining 20% of CD patients and those patients
extraction, and 16S amplicon sequencing were detailed from different regions. We also compared the functional
in the Supplementary Methods section. The study was differences in CD patients with and without AIEC presence
approved by the Research Ethics Committee of the First using a linear mixed model.
Affiliated Hospital of Kunming Medical School (reference
no. 2017.L.15-1). 3. Results
3.1. The presence of AIEC was significantly
2.2. Statistical analysis associated with CD risk and carrageenan intake
Characteristics of CD patients with and without AIEC The study design is illustrated in the Graphical Abstract.
presence were reported. Data were presented as counts for A total of 112 subjects, including 72 CD patients and 40
categorical variables with percentages, and the mean or healthy controls from a rural area (Yunnan, China), were
median for continuous variables with standard deviation recruited (Figure S1). AIEC was detected in 20.83% of
or interquartile range. In univariate analysis, the Wilcoxon CD patients and 12.50% of healthy controls (p=0.270,
rank sum test was applied to determine the statistical Table 1). Among CD patients, AIEC presence was
significance for continuous variables, and Pearson’s Chi- significantly associated with lower educational attainment
squared test was used to identify the statistical difference (p=0.023), with 33.0% of AIEC-positive patients having
for categorical variables. The food additives difference no formal education compared to 7.0% of AIEC-negative
between CD patients with and without AIEC presence was patients (Table 2). Multivariate logistic regression showed
calculated using the smd package to obtain the Standardized a significant association between AIEC presence and
Mean Difference. Multivariate logistic regression was increased CD risk (adjusted Odds Ratio [aOR] = 7.50, 95%
used to assess the relationship between the risk factors confidence interval [CI]: 1.04–54.23, p=0.046, Figure 1A
and outcome, with confounders adjusted. We evaluated and Table S1). Low education level (middle school) was
the association between CD and AIEC presence and the positively associated with CD risk (aOR = 8.20, 95% CI:
association between AIEC presence and consumption of 1.35–49.71, p=0.022), whereas body mass index (BMI) was
food additives in the rural cohort. Finally, the impact of negatively associated with CD risk (aOR = 0.77, 95% CI:
urbanization on AIEC prevalence was analyzed in CD 0.65–0.9, p=0.001). CD patients consumed more aluminum
patients from rural and urban regions. silicate (8382 mg/year vs. 2092 mg/year, p=0.038) and
titanium dioxide (127151 mg/year vs. 29664 mg/year,
2.3. The 16S amplicon sequencing analysis
The taxonomy annotation and functional prediction for Table 1. AIEC prevalence in the urban and rural cohorts
the mucosal microbiome sequencing data are detailed in
the Supplementary Methods section. For alpha diversity Area AIEC‑positive AIEC‑positive AIEC‑negative p‑value †
analysis, Shannon diversity and Observed Features were rate (%)
calculated using the operational taxonomic units table that Rural CD 20.83 15 57 0.270
was rarefied to 10,000 sequences per sample. In addition, Rural HC 12.50 5 35
Bray–Curtis distance was calculated for all samples, and the Urban CD 30.00 18 42 0.003
analysis of similarities (ANOSIM) test was used to identify Urban HC 7.14 4 52
the statistical difference in beta diversity. The explanation Notes: p-value was calculated according to Pearson’s Chi-squared test.
†
of host factors on the microbiome composition variation Abbreviations: AIEC: Adherent-invasive Escherichia coli; CD: Crohn’s
was identified by the permutational multivariate analysis disease; HC: Healthy controls.

Volume 2 Issue 4 (2025) 69 doi: 10.36922/MI025230051