Microscopic colitis : epidemiology, death and associated disorders

Abstract: Microscopic colitis (MC) is the most recently recognized inflammatory condition of the large intestine. MC is an umbrella term for two disease entities, namely lymphocytic colitis (LC) and collagenous colitis (CC). These subtypes are distinguished by their histopathological presentation and cannot be separated based on clinical observation or symptoms. The most prominent clinical presentation is watery, non-bloody diarrhea. However, a proportion of patients also suffer from abdominal pain, weight loss, fecal incontinence and reduced quality of life. MC primarily affects the elderly and shows a clear female preponderance with some 2/3 of patients being women. Large scale, epidemiological research of the disease has historically been hampered by insufficient awareness. Therefore, using data from Swedish health care registers, this thesis aims to elucidate temporal patterns of MC as well as the association with mortality, cancer and celiac disease (CD). In study I, we examined the validity of having a MC diagnosis recorded in Swedish regional pathology registers. Through manual review of medical charts (n=211), carried out by two independent reviewers, we computed a positive predictive value for MC of 95%. Thus, we concluded that Swedish pathology registers are a reliable source for identifying patients with MC. In study II, we identified every patient with a first-time diagnosis of MC recorded in Swedish pathology registers from 1995-2015. Using this cohort, we examined temporal trends, age distribution and sex differences in MC. As expected, a majority of patients (72%) were female and mean age at diagnosis was 60.2 years. Incidence rates increased appreciably from 1995 to 2012, after which rates have stabilized. The mean age-standardized incidence rate from 2006 to 2015 was 10.5 cases/100,000 person-years, with a female to male incidence rate ratio of 2.4, adjusted for age and calendar period. When analyzing age-specific incidence, incidence rates increased up to 75-79 years after which they declined. Furthermore, we estimated that during a life-time 1 in 115 women and 1 in 286 men are expected to be diagnosed with MC. In study III we examined mortality in patients with MC. This was done using a matched cohort study design where each exposed individual (MC) (n=14,333) was matched according to age, sex, county of residence and year of biopsy to five reference individuals from the general population. During the study period (1990 to 2017) patients with MC had a higher probability of death. However, after adjustment for comorbidities the association vanished. Thus, we concluded that the increased risk of death is attributable to the burden of concomitant disease. Study IV aimed to investigate the association between MC and cancer. Again, this was done using the matched cohort design described above. In total, we identified 11,758 patients diagnosed with MC between 1990 and 2016 with MC that were matched to 50,828 reference individuals. After adjustments for the matching variables and comorbidities (CD and diabetes), we estimated an adjusted hazard ratio (aHR) of 1.08 (95%CI=1.02-1.16) for overall cancer. In secondary analyses, we found a decreased probability of colorectal cancer (aHR, 0.52 (95%CI_0.40-0.66)). The same pattern was observed for gastrointestinal cancers overall (aHR, 0.72(95%CI=0.60-0.85)). In study V we examined the association between CD and MC. Using the same matched cohort study design described above, we identified 45,267 patients with CD and 224,568 reference individuals between 1990 and 2016. 456 patients with CD were diagnosed with MC during the study period compared to 198 reference individuals that developed MC during the same period. These figures correspond to an aHR of 11.5 (95%CI=9.3-13.7). However, as the proportional hazards assumption was violated, the main result should not be interpreted as the probability for a CD patient to develop MC at any instant during the study period compared to the reference population, but rather a mean aHR based on all lengths of follow up. However, as the increased risk remained even after >10 years of follow up, our results indicate that the concomitance of these diagnoses should be considered if symptoms persist or reoccur despite a gluten free diet.