The Association of Coeliac Disease and Microscopic Colitis

A Large Population-based Study

M. Stewart; C. N. Andrews; S. Urbanski; P. L. Beck; M. Storr


Aliment Pharmacol Ther. 2011;33(12):1340-1349. 

In This Article


An association between CD and MC has been suggested; however, whether this association occurs within a distinct patient population or whether it occurs as a general phenomenon has not been determined. Furthermore, no attempts have been made to identify the typical features of the coincident patient cohort. In this study, we found that the risk of each condition is increased by more than 50-fold in patients with the alternative condition. We also made the striking observation that CD and MC concomitance is highest in middle-aged women, a group not commonly regarded as a population at risk for MC.

Dubois and colleagues first published a cohort of 21 coeliac disease patients who also had colorectal biopsies demonstrating LC in 1989.[23] Since then, there have been a number of case reports and cohort studies suggesting a link between CD and MC.[8,11,14–16,23,24] The largest cohort reported by Green et al. included 1009 biopsy-proven CD patients. These patients were primarily self-referred to a single physician at a specialized New York coeliac disease centre. Eleven of the 1009 patients were diagnosed with CC and 33 patients were diagnosed with LC. By comparing these cases to MC incidence data from Olmsted County, Minnesota, published by Pardi et al.,[25] the authors concluded that concurrent diagnosis with both diseases occurred at a rate of 45-times that seen in the general population.[26] Comparing data from one highly selected referral centre to another population limits the extent to which these rates can be more generally applied; however, their rate of disease concurrence was similar to the SIRs that we present in this article.

The incidence of CD in our study population was adjusted based on age and gender data from the 2006 national Census. The age-standardised and gender-standardised incidence of CD ranged from 10.4 to 15.7 per 100 000 population. These rates are higher than the 2–13 per 100 000 population previously reported from Olmsted County and New Zealand populations.[27,28] Both these studies demonstrated significant increases in CD incidence in the final 2 years of each study suggesting that the incidence of CD may be increasing. We have confirmed this trend with an average of 8.6% annual incidence growth between 2004 and 2008.

The age-standardised and gender-standardised incidence of MC in our population ranged from 16.9 to 26.2 per 100 000 population, which is significantly higher than the incidence rates of both ulcerative colitis and Crohn's disease in the same population (11.0 and 16.5 per 100 000 population, respectively).[29] We have also documented an average incidence growth of 12% annually, which is in keeping with a trend first described by Pardi et al. between 1985 and 2001.[25] We speculate that the continued increase in the incidence of MC may in part be related to greater physician awareness.

Interestingly, the increase in MC was driven largely by increased diagnosis of LC with relatively stable incidence of CC. The divergence in incidence growth was in the context of increased lower endoscopies and colon biopsies, which would presumably affect both diagnoses. This increase in LC could also be related to increased sensitivity among pathologists to the diagnosis of LC vs. CC or it could represent a difference in the pathogenesis of LC and CC. Although nonsteroidal anti-inflammatory medications and selective serotonin reuptake inhibitors are well known triggers for CC,[5,30] LC has been linked to the use of sertraline, beta-blockers and bisphosphonates.[5] Given our limited understanding of the pathogenesis of MC, it is certainly possible that one or more environmental triggers may exist for LC that are unrelated to CC. Although MC is more common in women, it is interesting that the increased diagnosis of LC was most striking in men, with a 232% increase in incidence over the study period. Any conclusions based on this trend need be tempered, as most of this growth was seen in final year of the study. Further studies are needed to confirm and further characterise this trend.

This study clearly shows that MC and CD occur together at a rate over 50-times that seen in the general population. Our suspicion for disease concomitance should be particularly heightened in women between the ages of 40 and 60 years presenting with unexplained diarrhoea. Furthermore, we can predict that patients with CD can expect to be diagnosed with MC at a rate of approximately 1% per year; a rate significantly higher than the 0.02% annual incidence in the general population.

Previous estimates of disease coincidence have originated from highly selected patient populations and, as such, have limited generalisability. The strength of this study is that it presents disease incidence rates that reflect actual clinical practice in an urban North American setting. It is striking that our study identified middle-aged women as the group with the most frequent disease concomitance, an age group in which MC is less frequently considered. In keeping with this observation, middle-aged women with CD and diarrhoea as a presenting or persistent symptom despite gluten restriction should undergo lower endoscopy with biopsies for MC.

Despite the increased use of serological testing, duodenal biopsy continues to be considered the gold standard for CD diagnosis.[31] Although multiple biopsies increase diagnostic yield, the number of duodenal biopsies that should be taken is unclear.[31,32] Similarly, there are no well accepted guidelines for the diagnosis of MC. Some have suggested that two or more biopsies should be taken from the right, transverse, descending and sigmoid colon.[33] Others have argued that the majority of cases of MC can be picked up on sigmoidoscopy, whereas other studies have found that MC can be limited to the right and transverse colon and that sigmoidoscopy can miss 25–35% of cases.[34–37]

This study does have some limitations. The retrospective design did restrict the availability of certain data, such as the number of biopsies obtained for each patient and the Marsh stage for patients diagnosed with CD. Further details on the patient population, including family history of other disease states, more details on past medical history and medication history would have added to our analysis. Other data, such as the stated indication for the procedure, are also subject to the limitations of retrospective analysis as they may not necessarily be recorded until after the endoscopy is performed. Microscopic colitis is generally felt to be an under-diagnosed condition. Patients presenting with additional upper gastrointestinal symptoms or those undergoing upper endoscopy may be more likely to undergo lower endoscopy and biopsy. This may result in an overestimate of the degree of disease association relative to the reference population. Conversely, we were unable to exclude the possibility that incident cases of MC or CD had a previous histological diagnosis prior to 2004 or in another jurisdiction. This could result in overestimation of incidence rates; however, it would also underestimate the degree of disease concomitance. The biopsy rates presented in this study were surprisingly low, as we would expect a higher portion of patients undergoing upper endoscopy to have duodenal biopsies obtained. This could also result in fewer patients with concomitant disease being identified. These study limitations highlight the urgent need for prospective studies, as the disease association might be higher than predicted.

Our data did allow us to make some general observations about endoscopy and biopsy patterns, specifically regarding patients in whom dual upper and lower biopsies were obtained. We were able to confirm that over the course of the study there were approximately 60 full-time gastroenterologists and surgeons performing endoscopy in the CHR, however, this retrospective data did not allow for individual physician biopsy rates to be calculated or compared. Similarly, although it would have added to our analysis, we were not able determine rates of MC or CD diagnosis for individual pathologists.

Although an association between CD and MC is clearly evident, the nature of this association is far from clear. Coeliac disease is known to result from an immune cell-mediated reaction to dietary proteins in genetically predisposed hosts; however, the aetiology of LC and CC remains uncertain. In addition, it has been suggested that paucicellular lymphocytic colitis (PLC), a condition with a clinical picture similar to LC, but where the histological criteria are not fulfilled, should be considered along the spectrum of MC. However, Fernández-Bañares et al. found that it is a distinct entity based on different pathophysiological mechanisms.[38] To confound matters, small bowel intraepithelial lymphocytosis with no evidence of villous atrophy or overt CD has been reported in patients with MC as well as other chronic autoimmune conditions such as Crohn's disease.[13,16,39,40] Recent reports suggest that increased colonic intraepithelial lymphocytosis without lamina propria inflammation or epithelial damage, can be seen in a substantial proportion of patients with active CD.[41,42] Furthermore, Fine et al. report that colonic histopathology in patients with refractory sprue may be indistinguishable from lymphocytic colitis.[41] Whether these associated conditions represent different manifestations of the same underlying disease process or simply share a similar histopathologic endpoint is yet to be determined. The findings by Fine et al. may offer some explanation for the degree of association between MC and CD demonstrated by this study, especially in light of most patients being diagnosed with both conditions on the same day.

This study has demonstrated that the incidence of MC is still rising which suggests that the condition may still be under-diagnosed. The reasons for the remarkable rise in LC incidence in men remain unclear and deserve attention. Although less dramatic then in MC, the incidence of CD also continues to rise, especially in women. It is notable that the increased CD diagnosis is clearly not driven by the numbers of biopsies or procedures performed and thus likely reflects a true increase in disease incidence. In agreement with previous publications, there is significant overlap in the diagnoses of CD and MC. Importantly, our study identifies middle-aged women, an age group where MC is rarely considered, as high risk for diagnosis of both conditions. This novel association may change our clinical practice. Colonoscopy with biopsies to rule-out MC should be performed in middle-aged women with CD, especially when diarrhoea is resistant to a gluten-free diet.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.