Sources of dietary gluten in the first 2 years of life and associations with celiac disease autoimmunity and celiac disease in Swedish genetically predisposed children: The Environmental Determinants of Diabetes in the Young (TEDDY) study

Abstract

Background: High gluten intake is associated with increased risk of celiac disease (CD) in children at genetic risk.

Objectives: We aimed to investigate if different dietary gluten sources up to age 2 y confer different risks of celiac disease autoimmunity (CDA) and CD in children at genetic risk.

Methods: Three-day food records were collected at ages 6, 9, 12, 18, and 24 mo from 2088 Swedish genetically at-risk children participating in a 15-y follow-up cohort study on type 1 diabetes and CD. Screening for CD was performed with tissue transglutaminase autoantibodies (tTGA). The primary outcome was CDA, defined as persistent tTGA positivity. The secondary outcome was CD, defined as having a biopsy specimen showing Marsh score ≥ 2 or an averaged tTGA level ≥ 100 Units. Cox regression adjusted for total gluten intake estimated HRs with 95% CIs for daily intake of gluten sources.

Results: During follow-up, 487 (23.3%) children developed CDA and 242 (11.6%) developed CD. Daily intake of ≤158 g porridge at age 9 mo was associated with increased risk of CDA (HR: 1.53; 95% CI: 1.05, 2.23; P = 0.026) compared with no intake. A high daily bread intake (>18.3 g) at age 12 mo was associated with increased risk of both CDA (HR: 1.47; 95% CI: 1.05, 2.05; P = 0.023) and CD (HR: 1.79; 95% CI: 1.10, 2.91; P = 0.019) compared with no intake. At age 18 mo, milk cereal drink was associated with an increased risk of CD (HR: 1.16; 95% CI: 1.00, 1.33; P = 0.047) per 200-g/d increased intake. No association was found for other gluten sources up to age 24 mo and risk of CDA or CD.

Conclusions: High daily intakes of bread at age 12 mo and of milk cereal drink during the second year of life are associated with increased risk of both CDA and CD in genetically at-risk children.

Keywords: HLA; TEDDY; celiac disease; children; gluten foods.

FIGURE 1

Flowchart of Swedish children in TEDDY study with ≥1 measure for tTGA and ≥1 three-day food record collected at age 6–24 mo. CD, celiac disease; CDA, celiac disease autoimmunity; HLA, human leucocyte antigen; TEDDY, The Environmental Determinants of Diabetes in the Young; tTGA, tissue transglutaminase autoantibodies.

FIGURE 2

Summary plot of the estimated HRs and their related 95% CIs by Cox regressions of the association between daily intake of gluten-containing food groups assessed from 3-d food records at ages 6–24 mo and CDA (n = 487) in Swedish children (n = 2088) at genetic risk. Depending on the percentage of consumers (having an intake >0 g/d) at each age, intake variables were modeled as binary (if ≤50% were consumers; 0 g/d, >0 g/d) or categoric (if >50% were consumers; 0 g/d, median intake or less in participants without CDA or CD at the visit, greater than median intake in participants without CDA or CD at the visit) to represent no, low, and high intake. Included covariates in the analyses were human leucocyte antigen risk group, sex, having a parent or sibling with CD, and energy and gluten intake assessed by the respective food record. *Statistically significant result. Supplemental Table 4A–E summarizes detailed results. CD, celiac disease; CDA, celiac disease autoimmunity.

FIGURE 3

Summary plot of the estimated HRs and their related 95% CIs by Cox regressions of the association between daily intake of gluten-containing food groups assessed from 3-d food records at ages 6–24 mo and CD (n = 242) in Swedish children (n = 2088) at genetic risk. Depending on the percentage of consumers (having an intake >0 g/d) at each age, intake variables were modeled as binary (if ≤50% were consumers; 0 g/d, >0 g/d) or categoric (if >50% were consumers; 0 g/d, median intake or less in participants without CDA or CD at the visit, greater than median intake in participants without CDA or CD at the visit) to represent no, low, and high intake. Included covariates in the analyses were human leucocyte antigen risk group, sex, having a parent or sibling with CD, and energy and gluten intake assessed by the respective food record. *Statistically significant result. Supplemental Table 4A–E summarizes detailed results. CD, celiac disease; CDA, celiac disease autoimmunity.