Interaction Between Dietary Iron Intake and Genetically Determined Iron Overload: Risk of Islet Autoimmunity and Progression to Type 1 Diabetes in the TEDDY Study

Childhood type 1 diabetes (T1D) is a chronic multifactorial immune-mediated disease with an increasing incidence (1). There is increasing evidence that a β-cell stress response to environmental factors contributes to the loss of immunological tolerance and triggers β-cell autoimmunity (2,3).

Iron is essential for β-cell function and insulin secretion (4). However, iron overload is toxic for β-cells, as iron catalyzes the formation of reactive oxygen species (5). Balanced iron homeostasis is important for maintaining both β-cell and immunological health (6,7). Under normal circumstances, elevated iron levels stimulate hepcidin secretion from the liver. Hepcidin then inhibits iron export through the cellular iron transporter ferroportin from iron storage sites such as liver, gut, and macrophages. Genetic iron overload can be caused by mutations in HFE (HighFE2⁺) and TMPRSS6 (rs855791-G allele) (8,9), which regulate hepcidin, and in SLC40A1 (rs11568351-C allele), which codes for ferroportin 1.

Therefore, we investigated the association between energy-adjusted iron intake and development of persistent islet autoimmunity (IA), its initiation with either insulin autoantibodies (IAA-first) or GAD antibodies (GADA-first), and progression to T1D in The Environmental Determinants of Diabetes in the Young (TEDDY) study (10). Furthermore, we investigated whether relationships were modified by 1) a genetic risk score (GRS) based on three single nucleotide polymorphisms (SNPs) with the strongest association with high iron status (increased ferritin and transferrin saturation levels) in adult genome-wide association studies and 2) preselected SNPs located in iron metabolism genes.

TEDDY is a prospective cohort study in Sweden, Finland, Germany, and the U.S. From 2004 to 2010, TEDDY enrolled 8,676 newborn infants who carried one of the eligible HLA-DR-DQ genotypes with high risk for T1D. The details of screening and follow-up have been previously published (11). In total, 7,770 children were included in our analyses (Supplementary Fig. 1). The study was approved by local institutional review boards or ethics committees and is monitored by an external evaluation committee formed by the National Institutes of Health.

Iron intake was assessed using a 24-h recall and repeated 3-day food records (from 6 to 36 months) filled out before each visit by the caregiver (12) (Supplementary Material).

Serum samples were drawn every 3 months for the first 4 years of participation and then every 6 months thereafter, unless autoantibodies developed, in which case quarterly visits and blood draws were continued (13). The persistent development of IA was defined as the presence of confirmed positive findings for IAA; GADA, specifically isoform GAD-65; or IA2 autoantibody on two or more consecutive visits (14) (details are provided in the Supplementary Material).

SNPs located in iron metabolism genes were genotyped using the Illumina ImmunoChip (15) (Supplementary Table 1). We created a GRS based on three SNPs associated with circulating iron status (see details in the Supplementary Material).

Cox proportional hazards regression was used for our main analyses of IA during the first 10 years of life (all TEDDY participants had turned 10 years by 29 February 2020, when we initiated this study) and progression to T1D as outcomes (see details in the Supplementary Material). Participants with positive tissue transglutaminase autoantibody (TGA+) findings were excluded in a post hoc analysis.

Intake of iron was adjusted for total energy intake by using the residual method (16) stratified by country (i.e., the mean residual was 0 for each country) and visit of food record collection in the first 3 years (Supplementary Fig. 2). Energy-adjusted iron intake was considered as 1) a continuous variable (mg/day) and 2) three categories based on first and third quartiles (Q1 and Q3) of energy-adjusted iron intake as follows: low, less than or equal to Q1; moderate, between Q1 and Q3; and high, greater than or equal to Q3. Specifically, the quartiles of residuals were calculated per visit to consider the potential variations of the residuals across the visits. The magnitude of the associations was described by hazard ratios (HRs) with 95% CIs. Descriptions of iron intake and iron SNP interactions and model adjustments are presented in-depth in the Supplementary Material.

Data were analyzed using SAS 9.4 statistical software (SAS Institute, Cary, NC). Two-tailed P < 0.05 was considered statistically significant.

In the first 10 years of life, 769 (9.9%) children developed IA, representing 291 (3.7%) with IAA-first and 334 (4.3%) with GADA-first. Of those, 1,288 were TGA+. The median (Q1–Q3) age at onset of IA was 3 (1.5–6) years (1.8 [1–3.8] years for IAA-first, 4.4 [2.3–7.3] years for GADA-first). Among the 769 children with IA, 152 were TGA+ and 315 developed T1D. Median (Q1–Q3) age at onset of T1D was 6.3 (3.3–9.3) years (Supplementary Table 2). The median (Q1–Q3) follow-up time was 10.7 (5.3–12.5) years.

Increased iron intake (mg/day) was associated with increased risk of GADA-first (adjusted HR 1.03 [95% CI 1.00; 1.06], P = 0.04). High (greater than or equal to Q3) versus moderate intake (between Q1 and Q3) was associated with increased risk of GADA-first (1.43 [1.07; 1.92], P = 0.02), whereas low (less than or equal to Q1) versus moderate intake was not (Table 1 and Supplementary Table 7). Iron intake (per mg/day or categorical) was not associated with the development of overall IA, IAA-first, or progression to T1D (Table 1). In a sensitivity analysis, iron intake was nonlinearly associated with development of GADA-first (U-shaped association; P value for nonlinearity = 0.03) (Fig. 1) and progression to T1D (P value for nonlinearity = 0.03) with and without TGA+ (Supplementary Fig. 3).

The main effects of the associations between SNPs and GRS on the risk of IA, IAA-first, and GADA-first and progression to T1D are shown in Supplementary Table 3. For GRS ≥2 iron risk alleles, the HR for IAA-first was 0.77 (95% CI 0.45; 1.31 [P = 0.33]) for low versus moderate intake, but 1.71 (1.14; 2.58 [P = 0.01]) for high versus moderate iron intake (Table 2). This association was driven by TMPRSS6 (rs855791-G allele) (1.71 [1.07; 2.75], P = 0.03) for high versus moderate iron intake (Table 3A and Supplementary Table 4B).

There was a stepwise increasing risk for SLC40A1 (rs11568351-C allele) (adjusted P_interaction = 0.02) (Supplementary Table 5) for progression to T1D for high versus moderate iron intake with an HR of 2.25 (95% CI 1.36; 3.71 [P = 0.01]) for one risk allele and 6.42 (1.59; 25.99 [P = 0.01]) for two risk alleles with and without TGA+ (Table 3B and Supplementary Tables 8 and 9).

Dietary vitamin C interacted with iron intake on the risk of GADA-first (P_interaction = 0.02). In children with high iron intake, higher dietary vitamin C intake (1 mg/day increase) was associated with an increased risk of GADA-first (HR 1.002 [95% CI 1.000; 1.004], P = 0.014); vitamin C may, to some degree, enhance nonheme iron absorption (Supplementary Material).

This prospective study revealed novel associations between iron intake during the first 3 years of life and the risk of β-cell–specific autoimmunity. Iron intake was associated in a U-shaped fashion with risk of GADA-first. Furthermore, higher iron intake was only associated with IAA-first and progression to T1D when stratified by genetic polymorphisms, which increase intestinal iron absorption and impair cellular iron release.

The effects of early-life iron exposure on childhood T1D risk have been investigated with inconsistent results (17–19). Compared with previous studies, we investigated a high-risk T1D cohort with more detailed early-life iron exposure with energy-adjusted dietary and supplemental iron intake. These data were collected at multiple visits throughout the first 3 years of life and allowed us to investigate time to autoantibody development and progression to T1D, including gene-diet interactions. Murine models of T1D also have supported gene-diet interactions, where downregulation of early-life hepcidin in pancreatic islets may result in accelerated IA, which is further aggravated if mice are fed a high-iron diet (7).

A strength of this study is that TEDDY is a large prospective multicenter study of children genetically predisposed to T1D with the longitudinal dietary data and development of IA and T1D, as well as immunogenetic data. Another strength is that we were able to investigate IAA-first and GADA-first separately, which is important because they may reflect different disease processes. IAA usually appears during the 1st to 2nd year of life, whereas GADA usually appears at 3–5 years of age or even later (10).

Our results may be confounded by yet-unknown dietary patterns associated with iron intake, (e.g., vegetarian diet, fortified food, high-meat diet, processed food, celiac disease); however, to our knowledge, it remains unexamined whether such dietary factors are associated with IA and T1D risk. Currently, the TEDDY study lacks measures of plasma iron parameters, which could have provided insight into endogenous circulating iron concentrations.

In conclusion, our novel findings indicate that iron may contribute to the initiation of IA. However, because the observations were exploratory and made in children with high-risk HLA haplogenotypes only, the data should be interpreted cautiously. Further research will be needed to confirm our findings in population-based cohorts.

Acknowledgments. The authors thank the TEDDY study participants for their generous contributions to this study.

Funding. The TEDDY study is funded by grants U01 DK63829, U01 DK63861, U01 DK63821, U01 DK63865, U01 DK63863, U01 DK63836, U01 DK63790, UC4 DK63829, UC4 DK63861, UC4 DK63821, UC4 DK63865, UC4 DK63863, UC4 DK63836, UC4 DK95300, UC4 DK100238, UC4 DK106955, UC4 DK112243, UC4 DK117483, U01 DK124166, and U01 DK128847 and contract HHSN267200700014C from the National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Allergy and Infectious Diseases, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institute of Environmental Health Sciences, Centers for Disease Control and Prevention, and JDRF. This work is supported in part by the National Institutes of Health/National Center for Advancing Translational sciences clinical and translational science awards to the University of Florida (UL1 TR000064) and the University of Colorado (UL1 TR002535) and by Fabrikant Vilhelm Pedersen og Hustrus Mindelegat grant 17798.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.U.T., X.L., Y.K., T.M.-P., S.K., J.M.N., C.E., and J.S. researched data and wrote the manuscript. X.L. conducted the statistical analyses. U.U., S.M.V., J.M.N., M.R., W.H., S.R., Å.L., A.-G.Z., J.T., and J.K. designed the study. U.U., S.M.V., M.R., W.H., J.Y., J-X.S., B.A., S.R., C.A.A., Å.L., A.-G.Z., J.T., J.K., and H.M.P. supervised the study design and data analysis and assisted with writing the manuscript. All authors critically reviewed and approved the final version of the manuscript. X.L. and J.S. are the guarantors of the work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.