Celiac disease (CD) is an immune-mediated disorder that occurs worldwide, with a prevalence of approximately 1% of the general population (1). First- and second-degree relatives of individuals with CD experience increased risk of developing CD, with reported prevalence at 4.5% to 10% among screened first-degree relatives (2,3).
The etiology of CD is not fully understood. Gluten, the dietary offender in CD, is required to initiate the immune process that, upon ingestion, leads to small-intestinal inflammation and villous atrophy. The genotype HLA-DQ2 or -DQ8 is required for an individual to be susceptible to CD (4); however, because approximately one-third of the Western population carries these alleles (5) but only 1 in 30 individuals with a susceptible genotype develops CD, attention has been paid to environmental risk factors, particularly infant feeding practice (6). It has been suggested that breast-feeding (duration and breast-feeding at the time of gluten introduction) protects against CD (6); however, other studies found no such association (7–9). These latter negative studies (7–9) investigated CD only in childhood; hence, it cannot be excluded that breast-feeding merely delays disease onset because CD may be diagnosed in adulthood (10). Furthermore, some studies have found an increased risk of CD when gluten is introduced early (<3 months (11) or <2 months (12)) or late in life (>7 months (11)), although a recent study by our group found no such association (9).
The infant feeding practice is of concern when studying CD pathogenesis. We hypothesized that maternal CD per se may affect infant feeding patterns. Adherence to infant feeding guidelines may be higher in women with CD because these mothers may have a greater awareness of CD risk factors and an aim to avoid CD in their offspring. Maternal CD may thus confer longer breast-feeding duration and alter CD risk in their high-risk offspring. We therefore examined infant feeding patterns primarily in mothers with CD. To do so, we used prospectively collected data from the population-based cohort study ABIS. Mothers with a CD diagnosis before childbirth were considered exposed, whereas mothers with no prior CD diagnosis made up the unexposed population. To evaluate whether an association between maternal CD and infant feeding patterns was restricted to maternal CD, we also examined the relation between paternal CD and offspring infant feeding pattern in a subanalysis.
The ABIS Cohort
Parents of all of the children born in southeast Sweden between October 1997 and October 1999 were invited to participate in the ABIS study. A questionnaire was completed at the time of infant birth. Additionally, parents were asked to prospectively complete a diary during their child's first year of life. In total, 21,700 children were born during the study period. The parents of 17,055 (79.6%) children gave informed consent to participate, and 16,286 mothers completed the birth questionnaire. Diary data were collected for 9849 children. In the questionnaire, data on covariates were obtained. We defined high maternal age as mothers giving birth at age >30 years. A high education level was defined as >12 years of study (equivalent to postsecondary or tertiary education in Sweden).
Maternal CD and Feeding Data
In the birth questionnaire, parents were asked to report whether they, or their infant's siblings or grandparents, experience immune-mediated disorders including CD, diabetes mellitus, or rheumatoid arthritis. In the diary, the date of breast-feeding cessation was reported, as well as data on the introduction of different foods into the infant's diet. Complete data regarding breast-feeding duration were available for 9644 children (97.9%), on gluten introduction into the infant's diet in 9529 children, and complete data on breast-feeding duration and gluten introduction in 9414 children. Except wherever explicitly stated, all analyses are based on these 9414 children, constituting 96% of children for whom diaries were completed (9414/9849).
We included all of the mothers for whom breast-feeding duration and gluten introduction data were completed in the diary. Forty-eight mothers had a record of a CD diagnosis confirmed through small-intestinal biopsy showing villous atrophy reported in the ABIS study. The CD diagnosis was established before giving birth in 22 mothers who thus constituted the exposed population. Mothers who did not report a history of CD (n = 9366), and mothers who received the CD diagnosis after giving birth (n = 26), constituted the control population (total n = 9392). We did not exclude mothers with later CD from the study population because their CD was not identified at the start of follow-up when exposure was defined. Follow-up started at the time of giving birth and ended at breast-feeding cessation or gluten introduction (outcome) or at 1 year of age (since diary record-keeping ended at that age). Cox proportional hazards model was used to investigate the risk of weaning among mothers with CD compared with reference mothers. In a second Cox model we investigated the association between maternal CD and age at gluten introduction. Finally, using logistic regression, we assessed the association between breast-feeding at gluten introduction and maternal CD diagnosed before infant birth. In a post-hoc analysis, we examined the effect of paternal CD on reported breast-feeding duration.
The ABIS study was approved by the research ethics committees of the Medical Faculty, Lund University (Lu 83–97), and the Faculty of Health Sciences, Linkoping University (Li 287–96). All of the participants gave written informed consent to participate in the study.
Mothers with CD and reference mothers were on average 30 years old when they gave birth to an ABIS child (Table 1). Mean age (at maternal celiac diagnosis) was 23.3 years. Additional background characteristics are given in Table 1.
Both higher maternal education level (unadjusted hazard ratio [HR] 0.8, 95% confidence interval [CI] 0.7–0.8) and older maternal age (unadjusted HR 0.7, 95% CI 0.7–0.8) were protective against early weaning. There was no difference in breast-feeding duration between boys and girls (data not shown).
Some 63% of children were breast-fed for at least 9 months (Table 2). We found no association between maternal CD and early weaning (unadjusted HR 0.9, 95% CI 0.5–1.6). This risk estimate did not change when we adjusted for maternal age, infant sex, and maternal education level (adjusted HR 1.0, 95% CI 0.6–1.7). Excluding mothers with undiagnosed CD (n = 26) from the control group had little effect on the estimate. In a separate analysis, we found no association between undiagnosed maternal CD (diagnosed after giving birth to their ABIS child) and breast-feeding duration (HR 0.6, 95% CI 0.3–1.1). In a post-hoc analysis, we examined the effect of paternal or parental CD before delivery on breast-feeding duration. No significant association was found between paternal CD (HR 0.5, 95% CI 0.1–1.9, based on 6 fathers with CD before childbirth) or between any parental CD (HR 0.8, 95% CI 0.5–1.3) and breast-feeding duration (unadjusted analyses).
Gluten was most often introduced in months 5 and 6 in both offspring to mothers with CD and in offspring to reference mothers, and maternal CD was not associated with infant's age at gluten introduction (Table 3; HR 0.9, 95% CI 0.6–1.3). Adjusting for maternal age, infant sex, and maternal education level had little effect on the estimate (HR 0.8, 95% CI 0.6–1.3).
A majority of mothers were still breast-feeding at the time of gluten introduction (19/22 [86%] mothers with CD vs 7715/9392 [82%] reference mothers), and no difference in the proportions of mothers that were breast-feeding at gluten introduction was seen between the 2 groups (P = 0.78, Fisher exact test). Using logistic regression, we found no association between maternal CD and breast-feeding at the time of gluten introduction (odds ratio [OR] 1.4, 95% CI 0.4–4.7). No association between any parental CD and breast-feeding at the time of gluten introduction was seen (OR 1.3, 95% CI 0.5–3.8).
In a post-hoc sensitivity analysis we also used logistic regression to examine the relation between maternal CD and the prevalence of early weaning (defined as ≤90 days) or late weaning (defined as >179 days). No statistically significant associations were found (OR 0.4, 95% CI 0.1–3.3; OR 1.6, 95% CI 0.6–4.0).
Maternal or any parental CD does not seem to influence the age at gluten introduction or breast-feeding duration in the infant. The majority of parents in this study adhered to infant-feeding guidelines issued by the Swedish pediatric society, recommending that breast-feeding should be ongoing when gluten is first introduced into the infant's diet (13).
An important strength of the present study is its prospective design and that the study population represents an unselected, general population. Parents filled out the diary at home during the child's first year of life, thereby minimizing the risk of recall bias. Furthermore, the population-based approach enables generalization to the general population.
A weakness of the present study is that the subjects represented merely 43% (9414/21,700) of the total birth cohort. It cannot be excluded that participation in the study may have resulted in bias by factors such as socioeconomic position and diagnosed CD, and these factors may be associated with infant-feeding habits.
We found no association between infant-feeding practice and maternal CD; thus, infant-feeding practice is unlikely to contribute to the increased risk of CD seen in offspring to parents with CD (3). CD is considered a multigenetic disorder, with HLA-DQ2 and -DQ8 being prerequisite for the disease. Recent genome-wide association studies have also pointed to >100 genes involved in CD (14). In addition to genetic risk, we cannot exclude that factors not studied in the present work, such as the timing of introduction of nutrients other than gluten, may be involved in CD pathogenesis.
The association between parental CD and early weaning was of somewhat higher magnitude for maternal CD compared with paternal CD (unadjusted HR 0.9, 95% CI 0.5–1.6 vs unadjusted HR 0.5, 95% CI 0.1–1.9); however, these CIs overlap. We therefore cannot conclude, based on the present study, that there is a difference between the impact of maternal vs paternal CD on breast-feeding duration.
The prevalence of CD among mothers in this study (0.5%) was comparable with that of previous studies on diagnosed CD (15) but lower than in a recent study in which adults were screened for CD (1.8%) (16). There is no general screening for CD in Sweden, and we cannot extrapolate our infant feeding findings to mothers with undiagnosed CD who remained undiagnosed until the end of follow-up. That some women with undiagnosed CD were likely included as false-negative controls in this study will not influence the risk estimate more than marginally because they constitute only a small fraction of controls.
The relatively low number of women with CD in this study is because we used data from the ABIS study. This should be viewed in the context of the benefits of using this cohort. It provided prospectively recorded data on infant feeding, which could not be as reliably and accurately collected using retrospective techniques. Additionally, the ABIS study provided accurate information on infant feeding in a representative sample of women without CD, with little risk of differential bias as inclusion and collection techniques were identical for those with and without the disease.
The breast-feeding duration was similar in mothers with CD and mothers without such a diagnosis. If any, there was a tendency that mothers with CD were more likely to breast-feed at the time of gluten introduction (OR 1.4). We cannot rule out that in a larger sample, this association would have been statistically significantly positive, although such an association cannot explain the increased risk of CD in the offspring because that is more often seen in offspring not breast-fed at gluten introduction (17).
In conclusion, feeding patterns do not seem to vary between offspring to mothers with CD and to those without. Infant feeding patterns are therefore unlikely to explain the increased risk of CD among offspring (2,3) to parents with CD.
We are grateful to all of the families participating in the ABIS project and to all of the staff members at Mother and Baby Health Centres where the questionnaires and blood samples were collected.
1. Green PH, Cellier C. Celiac disease. N Engl J Med
2. Fasano A, Berti I, Gerarduzzi T, et al. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med
3. Rubio-Tapia A, Van Dyke CT, Lahr BD, et al. Predictors of family risk for celiac disease: a population-based study. Clin Gastroenterol Hepatol
4. Trynka G, Wijmenga C, van Heel DA. A genetic perspective on coeliac disease. Trends Mol Med
5. Sollid LM, Thorsby E. HLA susceptibility genes in celiac disease: genetic mapping and role in pathogenesis. Gastroenterology
6. Akobeng AK, Ramanan AV, Buchan I, et al. Effect of breast feeding on risk of coeliac disease: a systematic review and meta-analysis of observational studies. Arch Dis Child
7. Ascher H, Krantz I, Rydberg L, et al. Influence of infant feeding and gluten intake on coeliac disease. Arch Dis Child
8. Roberts SE, Williams JG, Meddings D, et al. Perinatal risk factors and coeliac disease in children and young adults: a record linkage study. Aliment Pharmacol Ther
9. Welander A, Tjernberg AR, Montgomery SM, et al. Infectious disease and risk of later celiac disease in childhood. Pediatrics
10. Green PHR, Stavropoulos SN, Panagi SG, et al. Characteristics of adult celiac disease in the USA: results of a national survey. Am J Gastroenterol
11. Norris JM, Barriga K, Hoffenberg EJ, et al. Risk of celiac disease autoimmunity and timing of gluten introduction in the diet of infants at increased risk of disease. JAMA
12. Greco L, Auricchio S, Mayer M, et al. Case control study on nutritional risk factors in celiac disease. J Pediatr Gastroenterol Nutr
13. Axelsson I, Finkel Y, Michaelsen KF, et al. Complementary food to breastfed infants. Introduction can wait until the age of six months but not longer. Lakartidningen
14. Dubois PC, Trynka G, Franke L, et al. Multiple common variants for celiac disease influencing immune gene expression. Nat Genet
15. Dube C, Rostom A, Sy R, et al. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology
2005; 128 (4 suppl 1):S57–S67.
16. Walker MM, Murray JA, Ronkainen J, et al. Detection of celiac disease and lymphocytic enteropathy by parallel serology and histopathology in a population-based study. Gastroenterology
17. Ivarsson A, Hernell O, Stenlund H, et al. Breast-feeding protects against celiac disease. Am J Clin Nutr