Early Probiotic Supplementation and the Risk of Celiac Disease in Children at Genetic Risk

Probiotics are linked to positive regulatory effects on the immune system. The aim of the study was to examine the association between the exposure of probiotics via dietary supplements or via infant formula by the age of 1 year and the development of celiac disease autoimmunity (CDA) and celiac disease among a cohort of 6520 genetically susceptible children. Use of probiotics during the first year of life was reported by 1460 children. Time-to-event analysis was used to examine the associations. Overall exposure of probiotics during the first year of life was not associated with either CDA (n = 1212) (HR 1.15; 95%CI 0.99, 1.35; p = 0.07) or celiac disease (n = 455) (HR 1.11; 95%CI 0.86, 1.43; p = 0.43) when adjusting for known risk factors. Intake of probiotic dietary supplements, however, was associated with a slightly increased risk of CDA (HR 1.18; 95%CI 1.00, 1.40; p = 0.043) compared to children who did not get probiotics. It was concluded that the overall exposure of probiotics during the first year of life was not associated with CDA or celiac disease in children at genetic risk.


Study Population
The TEDDY study involves 6 clinical research centers located in Colorado, Georgia, and Washington in the U.S. and in Finland, Germany, and Sweden in Europe. All sites follow the same study protocol including scheduled visits every 3rd month until the age of 4 years and every 6th month thereafter until 15 years of age [25]. Between September 2004 and February 2010, the TEDDY study screened 424,788 newborns infants of whom 21,589 infants fulfilled the inclusion criteria based on the Human Leukocyte Antigen (HLA) genotyping (Supplemental Table S1). Among those eligible children, 8676 children were enrolled in the prospective cohort study. For this study we only included children with selected HLA genotypes: DR3/3, DR3/4, DR4/4, and DR4/8 who had been screened for tissue transglutaminase autoantibodies (tTGA) (n = 6520) ( Figure S1). As of July 2017, 6520 children had been followed to a median (interquartile range (IQR)) age of 8.7 (7.4-10.2) years and included for this study.
For all study participants, separate written informed consent for genetic screening and participation in the follow-up study were obtained from a parent or primary caregiver before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by local Ethics Committees and Institutional Review Boards and monitored by an External Evaluation Committee formed by the National Institutes of Health.

Screening for Celiac Disease Related Autoimmunity (CDA) and Celiac Disease
Annual screening for celiac disease starts with tTGA at 2 years of age, as previously described [26]. Children who are tTGA positive are re-tested after 3-6 months and defined as having CDA if persistently tTGA positive in two consecutive samples. In addition, children who tested tTGA positive had their serum samples retrospectively analyzed. Samples from as early as 3 months of age were available in order to determine the closest time-point of seroconversion. Caregivers to children with CDA were referred to their health care provider for further evaluation of celiac disease. For the purpose of this study, celiac disease was defined as biopsy proven (i.e., an intestinal biopsy showing a Marsh score ≥ 2) or having a persistently tTGA level of ≥100 units if a biopsy was not performed [27].

Information on Characteristics, Diet and Health of the Study Population
Information about basic demographic characteristics of the mother and her newborn baby was received from the infant screening form. A questionnaire was mailed to the home of the mother prior to the first clinic visit (3 to 4.5 months postpartum). This questionnaire contained questions regarding illnesses during pregnancy, mother's use of medications and dietary supplements, smoking status, and maternal body mass index (BMI) before pregnancy. After enrollment, families received a questionnaire on the mode of delivery and the child's early diet, including use of probiotics between 0 to 3 months of age. Information about the mother's education and the birth order of the child was received from the primary caretaker questionnaire at the 9-month clinic visit. Parents were advised to consistently maintain a diary after the first clinic visit in order to collect information on child illnesses and diet. The child's age at the start and end of probiotic supplementation and/or infant formula, as well as of each type of formula, were recorded. Probiotic exposure was defined as the timing of first probiotic introduction of either via dietary supplement or infant formula. Species of probiotics in supplements and infant formulas were examined based on the composition of the self-reported brand name products. The majority of the probiotic bacteria in dietary supplements and infant formulas taken by the TEDDY children consisted of Lactobacillus reuteri and Lactobacillus rhamnosus, although 17% of the families were not able to identify the brand name of the probiotics they used. The latter was more likely to occur during the first 3 months of age, when the information on probiotic intake was recalled retrospectively. Study nurses reviewed the questionnaires and diaries during the family's clinic visit or over the phone every 3 months to minimize missing or inaccurate information.

Statistical Methods
Time-to-event analysis with Cox proportional hazards (PH) modeling was performed to examine the overall probiotic exposure and timing in the first year of life in relation to the risk of CDA and celiac disease. The magnitudes of the associations were described by hazard ratios (HR) with 95% confidence intervals (CI). The Cox PH models were adjusted for known risk factors for celiac disease including sex, HLA genotype, family history of celiac disease (first-degree relative (FDR) with celiac disease vs. not), and country (as strata). In addition, Cox PH models were adjusted for potential confounders associated with both probiotics use and CDA or celiac disease, including birth year, mode of delivery, mother's education, duration of exclusive breastfeeding, and child's diarrhea status in the first 3 months.
Probiotic exposure in the first year of life was incorporated into the Cox PH model in two ways: (a) probiotic exposure was categorized as binary (yes vs. no): probiotics users vs. non-users, and (b) source of probiotic exposure was categorized into three groups: dietary supplements, infant formula, or none. Among probiotics users, the age at first probiotic exposure was examined as a continuous variable in the Cox PH model. Analyses were carried out using the Statistical Analysis System software (version 9.4; SAS Institute, Cary, NC, USA).

Study Population
During follow-up, 1212 (18.6%) children were identified as having CDA at a median (IQR) age of 3.3 (2.2-5.0) years (range 0.9-11.5 years), while 455 (7.0%) children were diagnosed with celiac disease at a median (IQR) age of 4.3 (3.2-6.2) years (range 1.2-12.5 years). The characteristics of the children by the status of CDA or celiac disease are presented in Table 1.

Probiotic Use
A total of 1460 children were exposed to probiotics via dietary supplements or infant formula during the first year of life. The characteristics of the children by the source of probiotic exposure in the first year of life are presented in Table 2. The participants' characteristics that were positively associated with probiotics use during the first year of life were country (p < 0.001), later birth year (p < 0.001), mode of delivery (other than Cesarean section) (p = 0.012), being the first born child (p < 0.001), older maternal age (p = 0.001), higher maternal education (p < 0.001), not smoking during pregnancy (p = 0.005), maternal probiotic use during pregnancy (p < 0.001), shorter duration of exclusive breastfeeding (p < 0.001), antibiotic use (p < 0.001), diarrhea during the first 3 months (p < 0.001), gastrointestinal infections (p < 0.001), and lower incidence of common cold during the first 3 months (p = 0.006) ( Table 2). There was a considerable increase in probiotics use by birth year in Sweden ( Figure S2) when compared to other countries where there was not as much difference in the probiotics use across the study years. a : Three children were exposed to both probiotic dietary supplements and infant formula at the same time and were not included here. b : p value from the Cochran-Mantel-Haenszel test for the association of characteristics between probiotics users and non-users during first year of life; analyses adjusted for country. c : p value from the Cochran-Mantel-Haenszel test (on categorical variables) or the analysis of covariance (on continuous variables) for the association of characteristics between the sources of first probiotic exposure; analyses adjusted for country. Data are presented as number (percentage) unless otherwise indicated.

Risk of CDA and Celiac Disease
There was no difference in the risk of CDA (HR 1.15; 95%CI 0.99, 1.35; p = 0.07) or celiac disease (HR 1.11; 95%CI 0.86, 1.43; p = 0.43) between probiotics users and non-users during the first year of life when the models were adjusted for potential confounders: country, sex, HLA-genotype, FDR with celiac disease, birth year, mode of delivery, mother's education, duration of exclusive breastfeeding, and child's diarrhea during the first 3 months (Table 3). However, probiotic exposure from dietary supplements alone when compared to no exposure was associated with a slightly increased risk of CDA when adjusted for the potential confounders (HR 1.18; 95%CI 1.01, 1.40; p = 0.043) ( Table 3). Table 3. Overall probiotic exposure, timing of first probiotic exposure by source, and the risk of celiac disease auto-immunity (CDA) and celiac disease. (Birth year, mode of delivery, mother's education, duration of exclusive breastfeeding, and child's diarrhea during first 3 months were statistically significantly (p-value < 0.05) associated with probiotic exposure during the first year of life, and with celiac disease autoimmunity (CDA) and/or celiac disease.) c Hazard ratios describe the change in the risk for every one week delay in the probiotic exposure. d Hazard ratios adjusted additionally for the source of probiotics.

CDA
No association was found between the age of the child at the time of initial probiotic exposure and the risk of the outcomes when adjusting for all the potential confounders (Table 3). However, a time-to-event analysis with smoothing splines [28,29] showed a slightly increased subsequent risk of celiac disease when a probiotic dietary supplement was introduced during the first weeks of life (Figure 1).

Discussion
The present study showed no protective association between overall probiotics use during first year of life and the risks of CDA or celiac disease in children at increased risk of T1D and celiac disease. In fact, probiotic exposure from dietary supplements during the first weeks of life was associated with a small increase in the risk of celiac disease. This finding is in contrast to a recent randomized double-blinded placebo-controlled study in Sweden in which synergistic effects of probiotics on the peripheral autoimmune response were observed in genetically predisposed children with CDA who were receiving two strains of Lactobacillus, as compared to placebo [30]. However, to our knowledge no other study has investigated the association between probiotic use and celiac disease in children in a prospective multicenter study like TEDDY.
Probiotic use in the first year of life was positively linked to various infections and antibiotic use in children. This finding could be interpreted as infectious episodes potentially acting as a confounder when studying the association between probiotic exposure and the outcomes. However, we did not find any association between infections or antibiotic use and the risk of CDA or celiac disease in this study. Probiotics use was associated with shorter exclusive breastfeeding, which could have been related to earlier introduction of gluten-containing cereals. However, there is no current evidence that early introduction of gluten could be linked to the risk of celiac disease risk [31,32]. Moreover, Figure 1. The estimated effects of age at probiotic exposure (by source of probiotics) on the log hazards of celiac disease autoimmunity (CDA, n = 281) (a,c; nonlinearity: p = 0.054 and p = 0.16, respectively) and celiac disease (n = 99) (b,d; nonlinearity: p = 0.16 and p = 0.20, respectively) from time-to-event analysis with smoothing splines on 1460 subjects who were exposed to probiotics during the first year of life.

Discussion
The present study showed no protective association between overall probiotics use during first year of life and the risks of CDA or celiac disease in children at increased risk of T1D and celiac disease. In fact, probiotic exposure from dietary supplements during the first weeks of life was associated with a small increase in the risk of celiac disease. This finding is in contrast to a recent randomized double-blinded placebo-controlled study in Sweden in which synergistic effects of probiotics on the peripheral autoimmune response were observed in genetically predisposed children with CDA who were receiving two strains of Lactobacillus, as compared to placebo [30]. However, to our knowledge no other study has investigated the association between probiotic use and celiac disease in children in a prospective multicenter study like TEDDY.
Probiotic use in the first year of life was positively linked to various infections and antibiotic use in children. This finding could be interpreted as infectious episodes potentially acting as a confounder when studying the association between probiotic exposure and the outcomes. However, we did not find any association between infections or antibiotic use and the risk of CDA or celiac disease in this study. Probiotics use was associated with shorter exclusive breastfeeding, which could have been related to earlier introduction of gluten-containing cereals. However, there is no current evidence that early introduction of gluten could be linked to the risk of celiac disease risk [31,32]. Moreover, the association between breastfeeding duration and the risk of celiac disease still remains inconclusive [33].
Administration of antibiotics was not associated with CDA or celiac disease in this study, which is in line with an earlier report from the TEDDY Study [20] in which antibiotic use between 3 months and 4 years of age was examined in relation to CDA and celiac disease, adjusting e.g. for probiotic use during the first 3 months. However, two recent studies suggest that taking antibiotics at an early age is associated with celiac disease [34,35], although neither of these two studies took potential early probiotic exposure into account.
Colic usually appears during the first 2-3 weeks of life and a common practice, especially in Sweden, involves recommending probiotic supplement drops to ease the abdominal discomfort caused by colic [36]. Unfortunately, colicky events were not recorded in this study and their role as a potential confounder could not be further evaluated. Nevertheless, the etiology of both colic [37,38] as well as celiac disease [39] has been associated with dysbiosis in gut microbiota, and both are also associated with similar health conditions [40,41].
We also speculated whether use of probiotics after the first year of life could be linked to the risk of CDA or celiac disease. Because yoghurt and other fermented milk products are frequently given to young children especially after the first year of age, probiotics in the form of dietary supplements or infant formula could seldom be counted as the first exposure at that age. Food sources of probiotics (e.g., Lactobacillus species via fermented milk and vegetable products or fortified foods), would then become more frequent in infant diet, minimizing the importance of dietary supplements as the first probiotic exposure. Furthermore, given suggestions that gut microbiota is an important part of the causal pathway of celiac disease, we have to take into consideration that gut microbiota has already been shaped by solid foods after 1 year of age [42] and the long-term modification of gut microbiota by introduction of probiotic supplements at that time would probably not be feasible.
The strength of this study is the prospective design including subjects from multiple international clinical centers, as well as using standardized and validated methods in data collection across the study centers. The mechanistic actions of probiotic bacteria could be dependent on species and doses of probiotics, for example. Since this information was not available for this study, it suffers from limitations despite its large size and prospective design. However, there is currently no definite consensus by which specific bacteria could be associated with celiac disease [22,43]. The dose of probiotics could not be studied because of the lack of information on the manufacturing processes and storage conditions [44] of the large variety of probiotic supplements and infant formulas that were used by the study participants. There was also limited information about early life events and their timing due to the retrospective collection of self-reported data at the age of 3 months.

Conclusions
This study demonstrated that the overall supplementation of probiotics in infancy was not associated with celiac disease. However, the finding that first probiotic exposure from dietary supplements during the first weeks of life was associated with the increased risk of celiac disease warrants further investigation.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2072-6643/11/8/1790/s1, Figure S1: Flowchart describing the study population, Figure S2: Probiotic supplementation (%) during the first year of life by year of birth and by country, Table S1: High risk HLA genotypes followed in TEDDY.
Author Contributions: U.U., C.A.A., and D.A. conceptualized and designed the study, interpreted the data, drafted the initial manuscript, and revised the manuscript. X.L. carried out all the data analyses, reviewed and revised the manuscript.