To cite this article: Thomsen SF, Duffy DL, Kyvik KO, Skytthe A, Backer V. Relationship between type 1 diabetes and atopic diseases in a twin population. Allergy 2011; 66: 645–647.
Aim: To study the association between type 1 diabetes and atopic diseases in a twin population.
Methods: We performed record linkage between questionnaire-defined atopic dermatitis, asthma and hay fever, and hospital discharge diagnoses of type 1 diabetes in 54 530 Danish twins, 3–71 years of age.
Results: The age- and sex-adjusted risk of atopic dermatitis was decreased in subjects with type 1 diabetes compared with nondiabetic subjects, (2.1%vs 9.9%), odds ratio (OR) = 0.23 (0.07–0.71), P = 0.011, whereas asthma and hay fever were not significantly associated with type 1 diabetes. Within twin pairs discordant for type 1 diabetes, the diabetic twin had a lower risk of atopic dermatitis relative to the nondiabetic co-twin. Genetic factors for atopic dermatitis and type 1 diabetes were negatively correlated (r = −0.30), P = 0.0009.
Conclusions: These findings substantiate the Th1 vs Th2 cell dichotomy for type 1 diabetes and atopic dermatitis, and indicate an inverse association between genetic factors for these disorders.
Atopic diseases are characterized by a Th2-driven immune response, whereas type 1 diabetes is mediated by a Th1 response. As these responses are mutually inhibitory, type 1 diabetes and atopic diseases are in theory inversely related in the individual (1). Past decades have witnessed a considerable increase in the prevalence of atopic diseases and type 1 diabetes, which could indicate an influence of common etiological factors (2). Notably, widespread lifestyle changes in westernized affluent societies have lead to changes in housing conditions, smaller family size, and decreased exposure to certain infections, which are speculated to have played a causative role in the increasing rates of atopic and autoimmune diseases (3, 4).
Several epidemiological investigations have examined the association between type 1 diabetes and atopic diseases but with conflicting results. Most have been case–control studies with this type of study’s many inherent biases and to a much smaller extent cohort and observational studies. When restricting the analysis to studies of high-quality, a meta-analysis from 2003 of 11 epidemiological studies of type 1 diabetes and atopic diseases in children found evidence of an inverse association between asthma and type 1 diabetes [overall random effects OR = 0.82 (0.68–0.99)] that also extended into the other atopic diseases; atopic dermatitis, OR = 0.82 (0.62–1.10); rhinitis, OR = 0.97 (0.82–1.16); and any atopic disease, OR = 0.84 (0.65–1.08) although not statistically significant (5). The association between type 1 diabetes and atopic diseases has not been studied systematically in twins, and therefore, we conducted such analysis to investigate the genetic and environmental architecture underlying this association.
The population studied herein comprised twins born between 1931 and 2000 who were registered in the Danish Twin Registry (6). Details of the ascertainment procedure are given in a previous publication (7). In brief, the present investigation included questionnaire responses on atopic diseases obtained from nationwide multidisciplinary questionnaire studies. For twins born between 1983 and 2000 (3–20 years of age), the following questions were used: ‘Do you have, or have you ever had/has your child ever had, asthma/hay fever/eczema in the folds of your/your child’s elbows or knees?’, whereas for twins born 1931–1982 (20–71 years of age) the following questions were used: ‘Do you have, or have you ever had, asthma/eczema in the folds of your elbows or knees?’, ‘I know, or I think that I am allergic to dust, pollen or animal dander’ (defined ‘atopy’) and ‘I get, or I have previously gotten itchy eyes and nose or sneezing when exposed to pollen, rooms with carpets or after vacuum cleaning’ (defined ‘rhinoconjunctivitis’) (8). Twin zygosity was determined using four questions of similarity and mistaken identity, which assign zygosity correctly in more than 96% of the cases (9). In total, 54 530 twin individuals participated in the questionnaire studies and responses were cross-linked with hospital discharge diagnoses of type 1 diabetes.
Diabetes diagnoses were obtained from the Danish National Patient Registry. This registry records all hospitalizations in Denmark since 1977. Based on the 8th and the 10th revisions of the International Statistical Classification of Diseases and Related Health Problems, subjects with at least one of the following discharge diagnoses were considered type 1 diabetic cases: Diabetes mellitus, insulino dependente (dia24900 through dia24909), and Diabetes mellitus ex insulino dependens (E10.0 through E10.9) and not a concomitant type 2 diabetes diagnosis.
The risk of atopic diseases according to diabetes status was calculated with logistic and conditional logistic regression analysis (co-twin control analysis) (10). Latent factor models of genetic and environmental effects were fitted to the raw data following the methods described by Neale & Cardon (11). These are customary twin analyses where the variation in the susceptibility to a disease is partitioned into genetic and environmental sources of variance (12). Bivariate analyses were conducted between type 1 diabetes and atopic diseases (atopic dermatitis) to obtain estimates of the correlation between genetic and environmental effects for these diseases (12). Data were analyzed with the statistical packages spss 16.0 (SPSS Inc., Chicago, IL, USA) and Mx (13).
Significant sex differences were observed for atopic dermatitis, hay fever, atopy, and rhinoconjunctivitis, whereas age was significantly associated with all of the studied traits (Table 1).
|n = 19 748||n = 34 782||n = 25 742||n = 28 786||n = 54 530|
|Atopic dermatitis||3213 (16.3)||2076 (6.1)||2038 (8.0)||3251 (11.4)||5289 (9.8)|
|Asthma||2208 (11.2)||3004 (8.7)||2476 (9.7)||2736 (9.6)||5212 (9.6)|
|Hay fever||1421 (7.2)||–||787 (8.0)||634 (6.4)||1421 (7.2)|
|Atopy||–||6199 (18.2)||2701 (17.4)||3498 (18.9)||6199 (18.2)|
|Rhinoconjunctivitis||–||7295 (21.4)||3051 (19.6)||4244 (22.9)||7295 (21.4)|
|Type 1 diabetes||42 (0.21)||101 (0.29)||80 (0.31)||63 (0.22)||143 (0.26)|
In the age group 3–20 years, the prevalence of all three atopic diseases was lower in subjects with type 1 diabetes compared with nondiabetic subjects but only significantly so for atopic dermatitis (Table 2). This was also the case for any atopic disease (11.9%vs 28.0%), OR = 0.35 (0.14–0.89), P = 0.020. The result remained significant after adjustment for age, sex, and zygosity, OR = 0.32 (0.13–0.82), P = 0.017.
|Nondiabetic||Type 1 diabetic||OR (95% CI)|
|3–20 years (n = 19 748)|
|Atopic dermatitis||3211 (16.3)||1 (2.4)||0.13 (0.02–0.91)|
|Asthma||2204 (11.2)||4 (9.5)||0.84 (0.30–2.34)|
|Hay fever||1420 (7.2)||1 (2.4)||0.31 (0.04–2.29)|
|20–71 years (n = 34 782)|
|Atopic dermatitis||2074 (6.1)||2 (2.0)||0.31 (0.08–1.26)|
|Asthma||2994 (8.7)||9 (9.0)||1.03 (0.52–2.05)|
|Rhinoconjunctivitis||7270 (21.4)||24 (23.8)||1.15 (0.72–1.81)|
|Atopy||6184 (18.2)||14 (14.3)||0.74 (0.42–1.32)|
In the age group 20–71 years, results were less consistent but there was a clear indication of an inverse relationship between atopic dermatitis and type 1 diabetes.
When combining all ages, and after adjustment for age, sex, and zygosity, the risk of atopic dermatitis remained significantly decreased in subjects with type 1 diabetes compared with nondiabetic subjects, (2.1%vs 9.9%), OR = 0.23 (0.07–0.71), P = 0.011.
Co-twin control analysis showed that among a total of 91 twin pairs discordant for type 1 diabetes, the sex-adjusted risk of atopic dermatitis was lower in the diabetic twin relative to the nondiabetic co-twin, (2.2%vs 5.5%), OR = 0.42 (0.08–2.21), P = 0.309. In DZ twins, the sex-adjusted risk of atopic dermatitis was further decreased in the type 1 diabetic twin relative to the nondiabetic co-twin, (1.4%vs 7.1%), OR = 0.21 (0.02–1.83), P = 0.159, suggesting genetic confounding between the two diseases.
Variance decomposition showed that additive genetic factors accounted for 81% of the susceptibility to type 1 diabetes and 83% for atopic dermatitis, whereas shared environment did not significantly influence the disease susceptibility. The sex- and age-adjusted genetic correlation between type 1 diabetes and atopic dermatitis was −0.30, P = 0.0009, whereas the environmental correlation was 0.52, P = 0.079.
This study showed that type 1 diabetes and atopic dermatitis are inversely related in the individual. Notably, atopic dermatitis is over four times less common in type 1 diabetic subjects compared with nondiabetic subjects. Co-twin control analysis substantiated these findings in that within twin pairs discordant for type 1 diabetes, the diabetic twin had a lower risk of atopic dermatitis relative to the nondiabetic co-twin. These effects were more pronounced among DZ twin pairs, consistent with some degree of genetic confounding between the traits. Following this, we observed that genetic effects for type 1 diabetes were inversely associated with genetic effects for atopic dermatitis, obtaining a genetic correlation of −0.30.
Our finding of an inverse association between atopic dermatitis and type 1 diabetes is in keeping with previous findings among child singleton populations (5). Particularly, two large case–control studies, one by Olesen and co-workers in Danish children, 3–15 years of age, and one by Rosenbauer et al. in German preschool children, found a significantly decreased risk of atopic dermatitis in subjects with type 1 diabetes (14, 15). We also saw this tendency for asthma and hay fever in the age group 3–20 years, but in accordance with the two aforementioned studies, these outcomes were not statistically significantly related to type 1 diabetes.
We cannot rule out some degree of diagnostic mix up between type 1 and type 2 diabetes in our study. Notably, the diagnosis of type 1 diabetes among adults could reflect insulin-dependent type 2 diabetes. Also some subjects with type 1 diabetes with early onset among the oldest individuals could have escaped our attention because we only retrieved diagnoses from 1977 onwards. Type 1 diabetes was only present in 143 subjects (0.26% of the entire sample), which limited the number of informative families.
The diagnoses of atopic diseases relied on self-report, which tend to produce both false positive and false negative cases. Particularly, the diagnosis of atopic dermatitis is difficult to establish in older individuals because the disease mainly presents in early childhood. Furthermore, we used flexural eczema as proxy for atopic dermatitis, which is a less precise indicator compared with established diagnostic criteria (16).
Although atopic dermatitis among older adults is rare and also was rarer in the 1930s–1950s when these subjects were children, the low prevalence likely also reflects recall bias. This was exemplified by our observation that only 2.7% of the subjects older than 50 years of age had a history of atopic dermatitis. Subjects older than 50 years of age that report atopic dermatitis are likely individuals with persistent or severe disease, which could have contributed to the positive association between type 1 diabetes and atopic dermatitis among these ages.
We conclude that atopic dermatitis and type 1 diabetes are inversely related in the individual and that this effect is consistent throughout life. On the contrary, a statistically significant association between asthma and hay fever, and type 1 diabetes could not be confirmed. Our findings substantiate the Th1 vs Th2 cell dichotomy for type 1 diabetes and atopic dermatitis in childhood. We extend previous findings by showing that genetic factors induce a negative correlation between type 1 diabetes and atopic dermatitis.
SFT is sponsored by a Danish Medical Research Council grant.