Background
Celiac disease is a life-long immune-mediated disorder characterized by gluten-triggered inflammation and morphological damage of the small-bowel mucosa in genetically susceptible individuals [
1,
2]. The disease is one of the commonest food-related disorders with a currently estimated prevalence of 1–2 % in the Western populations [
3]. Moreover, it is overrepresented in type 1 diabetes (DM1), which shares a mutual genetic predisposition with CD [
4]. Although this association is well established particularly in children, the effect of concomitant DM1 on the clinical and histological presentation in adult celiac disease patients remains unclear. In contrast to DM1, the non-autoimmune-mediated type 2 diabetes (DM2) is not considered to be overrepresented in celiac disease [
5]. In fact, Kabbani and colleagues [
6] have recently reported DM2 to be even less frequent among celiac disease patients than in the US population in general, indicating possible protective effect. Evidently, their finding should be confirmed in other countries. In addition, the effect of concomitant DM2 on the presentation and natural history of celiac disease in adult patients is scarcely studied.
The aim of the present study was to investigate the prevalence of DM1 and DM2 in Finnish adults with celiac disease and to compare these figures with those in the general population. In addition, we compared a variety of clinical, serological and histological parameters and dietary adherence between celiac disease patients with and without concomitant DM1 or DM2.
Discussion
A novel finding of the present study is that DM1 is markedly overrepresented especially in men with celiac disease, whereas the prevalence of DM2 was practically similar to that of the Finnish population in general. Furthermore, concomitant DM1 or DM2 predispose celiac disease patients to severe comorbidities, and DM1 also to lower adherence to the gluten-free diet.
Here the overall prevalence of DM1 among celiac patients over 30 years of age was 3.8 %. This is significant overrepresentation compared to the population even in Finland where the incidence of DM1 is one of the highest in the world [
12]. In contrast to our study, the great majority of the previous studies investigating the association between celiac disease and DM1 have been conducted in children [
13,
14] of whom the characteristics of the concomitant diseases could be different. In addition, opposed to our design, the earlier studies have mainly investigated the prevalence of celiac disease in diabetic patients [
15,
16]. Another interesting finding of the present study was the substantially higher prevalence of DM1 among men with celiac disease (8.0 %) compared with the corresponding women (1.8 %). Even though, unlike most other autoimmune diseases, DM1 is known to be in general more common in men [
17,
18], this aspect has not been studied in celiac patients before, and the gender difference was surprisingly large. Interestingly, it has been observed that the male-to-female ratio of DM1 ratio differs between age groups [
17], and also that it is higher in countries where DM1 is common [
18]. The latter might partly explain the high number of DM1 men in our study population, but further studies are needed to fully explain these gender discrepancies. The prevalence of DM1 was also almost five times higher in celiac disease men aged 30–64 years compared with men being 65 years or more. The lower DM1 prevalence among older celiac patients could be due to increased mortality in DM1 [
16]. Accordingly, it has been shown that co-existing celiac disease of more than 15 years increases the risk of death in DM1 patients compared to patients with only DM1 [
19,
20]. The higher prevalence of DM1 in younger men here could also be partly due to increasing incidence of DM1 over time at population level [
21].
The overall prevalence of DM2 in our celiac disease cohort was at the same level as in Finnish adults in general. Further, the only noticeable difference in the subgroup comparisons was the lower prevalence of DM2 in celiac men aged ≥65 years compared with the population. Our results differ from those observed by Kabbani and colleagues, who recently found the prevalence of DM2 to be more than three times lower in celiac disease patients than in the US population [
6]. One explanation for the discrepant results could be differences in the disease presentation, as in Finland celiac patients have in general rather short diagnostic delay and mild clinical picture, whereas in US patients the classical malabsorptive disease with weight loss is still common [
22,
23]. Finnish patients are usually normal weight or even overweight at diagnosis and it might be argued that they thus have “normal” risk for DM2 [
24]. However, in the US study the protective effect of celiac disease was observed even after controlling for malabsorption, symptoms and BMI, indicating that there are other reasons for the different results. Another possibly affecting factor is the selection of control group and the overall prevalence of DM2 in the population. For example, the US study comprised two separate control groups, including age-, sex-, and ethnicity-matched controls and subjects of National Health and Nutrition Examination Survey [
25]. The prevalences of DM2 in these groups were 9.6 % and 9.8 %, respectively, while in our control group it was only 4.4 % in men and 3.0 % in women. Obviously there can also be differences in genetics, diagnostic criteria of DM2 and lifestyle, but the true role of these factors remains to be solved in future studies.
The clinical presentation of celiac disease was quite similar in three groups, the only significant difference being the markedly higher rate of screen-detected patients in the DM1 group. This is not surprising, as in Finland the increased risk of celiac disease in DM1 patients is well-known among physicians, and thus they are screened at low threshold [
26]. Subjects in DM1 group also had their celiac disease diagnosed at younger age, which might similarly be due to active screening. As expected, the presence of both concomitant DM1 and DM2 significantly increased the prevalence of coronary artery disease and hypertension in celiac disease patients compared with those being non-diabetic. This is in agreement with previous finding by Pitocco and co-workers that concomitant DM1 increases the risk of sub-clinical atherosclerosis in celiac disease [
27]. However, since we did not have specific non-celiac DM1 or DM2 control groups, it remains unclear whether celiac disease truly has an additional effect to the risk of cardiovascular complications, or whether their increased prevalence are caused solely by the concomitant diabetes. The prevalence of thyroidal diseases was also higher in celiac patients with DM1 compared with other groups even after controlling for age and gender. This was also quite expected, as the association of autoimmune thyroidal diseases with both DM1 and celiac disease has been well-established [
28,
29]. These disease associations are important to remember when treating patients with simultaneous celiac disease and diabetes.
Adherence to the gluten-free diet was significantly lower among celiac disease patients with a concomitant DM1 compared with the two other study groups. This result from interviews was further confirmed by the higher proportion of EmA-positivity in the DM1 group while on the diet. The lower adherence is in line with previous studies where the dietary adherence has also been relatively low among patients with the double-diagnosis of CD and DM1 [
30,
31]. Celiac disease patients with a concomitant DM1 may find it more difficult to cope with the demanding diet while already needing to concentrate to continuous daily glucose level monitoring, carbohydrate calculation and insulin dosing. Moreover, as seen in here and previously [
32], celiac patients with concomitant DM1 are often screen-detected and therefore have possibly only mild or no apparent symptoms, which might further decrease the motivation to adhere to expensive and socially restrictive gluten-free diet. There is no reason to believe that the maintenance of strict diet is less important in celiac patients with diabetes that in celiac disease altogether. These patients may need individualized support for the dietary treatment, and also careful follow-up to ensure coping with the increased disease burden. Accordingly, we have earlier shown that rigorous dietary counselling in patients with celiac disease and concomitant DM1 improves the dietary adherence to gluten-free diet [
30]. Interestingly, despite the poorer adherence, DM1 patients did not show significantly inferior mucosal recovery after a median of one year on a gluten-free diet. Nevertheless, they had the highest percentage of subtotal or total villous atrophy while on diet. Furthermore, notwithstanding the apparently equal dietary adherence, there was also a trend for slower histological recovery among subjects in the DM2 group compared with those in the CD group. Further studies are needed to establish whether concomitant diabetes predisposes celiac disease patients to slower mucosal recovery or even persistent villous atrophy on a gluten-free diet.
Major strengths of the present study are the high number of biopsy-proven celiac disease patients with heterogeneous clinical presentation and the representative control population acquired from a nationwide public health study. In addition, we were able to gather abundant clinical and histological information from each patient. Major limitations were the retrospective study design and the lack of specific information on the disease profile of DM1 and DM2. Further, the non-celiac diabetic control groups would have been needed to estimate the role of concomitant celiac disease in the increased risk of cardiovascular complications. The fact that repeat biopsy on a gluten-free diet was taken less often from DM1 and DM2 patients may cause selection bias. It would have also been very interesting to see whether there are differences between the groups in overall mortality, but unfortunately we did not have this data systemically collected. In any case, the cross-sectional study design and a relatively short follow-up time in most of the patients might have made these results somewhat unreliable. Finally, there was no exact data when the diabetes diagnosis was made with respect of celiac disease diagnosis.