Discussion
Inflammatory bowel disease (CD and UC) and celiac disease are the two most common immune-mediated gastrointestinal diseases and share common genetic risk variants. However, there have been few previous studies that have examined the association between the two diseases; and in particular, clinical outcomes of IBD in the presence of coexisting celiac disease have not been previously well-characterized.
Our data show that patients who have both celiac disease and UC are more likely to have pancolitis as opposed to more limited colonic inflammation, despite having a shorter duration of UC disease than our non-celiac UC controls. This is similar to a few previous studies that have examined other coexisting autoimmune diseases in UC. For example, UC patients with primary sclerosing cholangitis (PSC) are similarly more likely to have pancolitis as opposed to patients with isolated UC. A previous case series of IBD patients with celiac disease similarly found four out of five patients with UC and celiac disease to have pancolitis. Thus, patients with UC and coexisting celiac disease or PSC may represent a unique autoimmune-related phenotype of UC. However, further studies with other concomitant autoimmune diseases would be essential to define this further. We also identified a trend toward increased use of IMMs among the celiac-UC cohort. However, there was no significant difference in other IBD medications, hospitalizations, or surgeries. These findings differ from some of the small early case reports that postulated that patients with a combination of UC and celiac disease were more likely to have severe disease requiring colectomy. However, as the study by Yang et al. was also from a referral center but had no non-celiac UC controls for comparison, it is difficult to estimate if the rate of colectomy in that exceeded that in non-celiac UC patients.
We also did not find any effect of coexisting celiac disease in influencing disease phenotype or natural history in CD. A previously published review of eight cases found that CD manifested in atypical locations when in conjunction with a celiac disease diagnosis; however, our study shows no significant difference in CD location in the presence of celiac disease. The lack of significant differences in the clinical outcomes of patients with celiac disease and CD may argue against an effect of the shared risk loci between CD and celiac disease in influencing the natural history of disease. It may also be indicative that the association between immunological response seen against luminal microbial antigens and severe disease may not extend to dietary antigens such as gliadin.
An interesting observation was that patients in the celiac-IBD group were less likely to have been ever-smokers than controls, which is consistent with previously published studies, have also found lower smoking rates in patients diagnosed with celiac disease and have suggested a protective role of cigarette smoking. Smoking has similarly been found to have a protective effect on UC and PSC, again highlighting the possibility of a common link in the pathogenesis of these immune-mediated diseases. Our celiac-IBD patients were more likely to be afflicted with other autoimmune diseases than non-celiac IBD controls, highlighting the shared genetic susceptibility many of these patients may have. Indeed, shared genetic risk loci have been described between celiac disease and CD, Type I diabetes, and rheumatoid arthritis.
An important finding from our study is that while the prevalence of celiac disease in IBD patients within the Partners database was similar to what was expected in the general North American Caucasian population the prevalence of IBD among those with celiac disease was considerably higher than would be expected in the general population as has been reported previously. This suggests that in patients with celiac disease, particularly in those with non-response to gluten-free diet, one should consider work up for possible underlying IBD as this may exist concurrently in such patients. There is limited data to compare prevalence of celiac disease among other autoimmune diseases. We found a lower prevalence of celiac disease in Type I diabetics than has been reported elsewhere (0.7% vs. 4–9%). The true prevalence of the comorbidity of celiac disease with other autoimmune disorders such as RA and SLE has not been previously well-established and merits further analysis in larger validated disease cohorts, and where possible with prospective structured screening for celiac disease. Ludvigsson et al. recently identified a threefold increase in risk for SLE among those with biopsy-proven celiac disease. There are a few potential explanations for the potentially higher frequency of celiac disease among IBD compared with other autoimmune diseases. First, this could represent an ascertainment bias. As celiac disease and IBD are both managed commonly by gastroenterologists, there may be a lower threshold to obtain celiac serology testing or endoscopic biopsies in such patients compared with those managed by non-gastroenterologist physicians. Second, the higher frequency of celiac disease among IBD compared with other autoimmune diseases could also be consequent to the greater degree of genetic overlap between celiac disease and IBD than that observed with other diseases. The higher prevalence of IBD in those with celiac disease, but not celiac disease in those with IBD, may be due to the multifactorial etiopathogenesis of both diseases in which the development of disease requires a combination of environmental and genetic influences, and shared genes contribute a differential fraction of the disease risk for both diseases.
We believe that there are few clinical implications to our findings. First, reassuringly we did not identify a more aggressive phenotype of disease in those with coexisting celiac disease and IBD. Consequently, we do not suggest any need to modify current practice patterns in the management of IBD in this cohort of patients with the exception of recognizing the higher frequency of pancolitis among celiac-UC patients. Second, recognizing a possible higher prevalence of IBD among those with celiac disease, celiac disease patients with persistent gastrointestinal symptoms, despite a gluten-free diet, should also potentially trigger evaluation for underlying IBD (in addition to standard practice of evaluating for inadvertent gluten ingestion as well as underlying refractory celiac disease or lymphoma).
There are a few limitations to our study. First, referral bias is possible in studies at large tertiary centers. However such bias is unlikely to affect our results because both the case and controls groups were taken from the same pool of patients in the Partners database. As with any retrospective study, the level of past documentation affects completeness and quality of data collected. In our study, due to gaps in data, we were not able to consistently determine whether IBD or celiac disease was diagnosed first, or the symptoms/signs that led to diagnosis of celiac disease. It is possible that the order in which a patient is afflicted with the diseases may have some effect on outcome and further investigation is warranted in this area. Owing to the retrospective nature of the study, it was not possible to obtain serologic testing for all of the included patients; therefore, there is a possibility that some patients in our control group have "silent" or asymptomatic celiac disease. However, given the expected prevalence of this being less than 1%, it is unlikely to have influenced our results. Finally, we also acknowledge the possibility of false-positive findings due to simultaneous comparison of multiple patient characteristics and outcomes. As this is an exploratory study with limited prior literature in this area, we did not correct for multiple hypothesis testing within the context of this study. Consideration for this should be given in larger cohorts further characterizing and confirming the phenotype of celiac-IBD patients.
In summary, to our knowledge, this is the first case–control study examining clinical outcomes of patients with concurrent diagnoses of celiac disease and IBD to non-celiac IBD controls. While patients with celiac disease and UC may be more likely to have pancolitis, it is not clear that having the two diseases simultaneously leads to worse outcomes. However, while celiac disease is not more common in those with CD or UC, coexisting IBD seems to occur more commonly in patients with celiac disease.