Microscopic colitis: when the colonoscopy looks normal and the diarrhoea stays

A working group of gastroenterologists, pathologists and basic scientists produced a European guideline on microscopic colitis in 2021 following AGREE II, with a systematic literature review, GRADE assessment and Delphi voting.

It defines the three histological subtypes and explicitly describes the endoscopic picture as normal or near normal. The condition is classified as an inflammatory bowel disease, with recommendations on budesonide, bile acid binders, immunomodulators and biologics.

For you this means: this diagnosis is not a fringe phenomenon and not a label of convenience. It has a European guideline of its own, and it is made histologically, not endoscopically.

A Danish group around Bjørnbak analysed a consecutive cohort and re-assessed the histology of the unclear cases: 168 lymphocytic, 270 collagenous and 101 incomplete colitides.

Clinically the groups could not be told apart. In 48 percent of the collagenous and 24 percent of the lymphocytic cases there were mixed pictures with features of the respective other subtype. 15 percent of the incomplete forms were reclassified on re-assessment.

For you this means: if the report says collagenous or lymphocytic, that says something about the picture under the microscope. About the course and about the treatment options it says far less than the name suggests.

A team around Tome used the Rochester Epidemiology Project to capture every new case in Olmsted County between 1 January 2011 and 31 December 2019, with chart review and adjustment to the US population.

268 people were newly diagnosed. The median age was 64 years, the range 19 to 90. 77 percent were women. The adjusted incidence was 25.8 per 100,000 person-years, the prevalence at the end of 2019 was 246.2 per 100,000. No time trend appeared.

For you this means two things. First: other inflammatory bowel diseases move in the same order of magnitude. Second, and this matters more to me: the youngest affected person was 19. Anyone with watery diarrhoea at 34 is not too young for this diagnosis.

Tong and colleagues searched Medline, Embase and ISI Web of Science up to September 2014, screened 1972 citations and included 25 epidemiological studies.

The pooled incidence was 4.14 per 100,000 person-years for the collagenous and 4.85 for the lymphocytic form. The ratio of women to men was 3.05 to 1 in the collagenous and 1.92 to 1 in the lymphocytic form. The median age at diagnosis was 64.9 and 62.2 years. The rise in new cases before the year 2000 flattened afterwards in the USA, Sweden and Spain.

For you this means: the condition has not suddenly exploded. Part of the earlier rise is likely to rest on the fact that someone started looking for it at all.

Hjortswang and colleagues surveyed 116 people with collagenous colitis by post with four quality of life questionnaires and a one-week stool diary.

Those with a mean of three or more stools per day, or a mean of at least one watery stool per day, reported markedly worse quality of life. From that came the definition still used today: remission means a mean of fewer than three stools and fewer than one watery stool per day.

For you this means: there is a recognised threshold, and it was calibrated on quality of life, not on a laboratory number. One watery stool a day is enough to count as disease activity.

Guagnozzi, Arias and Lucendo searched MEDLINE, EMBASE, SCOPUS and conference abstracts, screened 227 references and included 26 studies with 5099 adults.

39.1 percent of people with confirmed microscopic colitis also met the criteria for a functional bowel disorder. Conversely, microscopic colitis was found in 9.8 percent of diarrhoea-predominant irritable bowel syndrome, compared with 1.3 percent in the constipation-predominant and 1.9 percent in the mixed type.

To be fair, this belongs with it: the difference between these irritable bowel subtypes was not statistically significant. For you this means: roughly one in ten people carrying the label diarrhoea-predominant irritable bowel could have microscopic colitis. The heterogeneity between studies was very high, though, and the confidence interval runs from 4.4 to 17.1 percent.

Virine, Chande and Driman analysed the biopsies of 101 consecutive people with confirmed microscopic colitis, 52 collagenous, 42 lymphocytic, 7 combined, from 2017 and 2018.

They counted how often a sample from a given segment carried the diagnosis: ascending colon 96.9 percent, transverse colon 95.7 percent, sigmoid 90.9 percent, caecum 90.0 percent, descending colon 85.0 percent, rectum 82.2 percent, splenic flexure 75.0 percent. Ascending and descending colon taken together captured 100 percent of cases.

For you this means: from the rectum alone, around 18 out of 100 cases would not have been picked up in this series. Not because anyone had been sloppy, but because the change is unevenly distributed.

Malik and colleagues searched PubMed, Web of Science, Scopus and Cochrane up to October 2020 and found three retrospective cohort studies with 356 people and 1854 biopsies in total.

The included papers recommended 4, 4 and 9 biopsies, a pooled mean of 5.67 with a spread of 2.89. The highest diagnostic yield lay in the ascending and descending colon. The authors recommend six biopsies, three each from the ascending and the descending part.

For you this means: there is a concrete number in the specialist literature. It is not an instruction you bring along, it is knowledge that makes a conversation easier.

Carrasco-Labra and colleagues searched MEDLINE and EMBASE up to April 2017 for the American Gastroenterological Association, included 38 diagnostic accuracy studies and appraised them with QUADAS-2 and GRADE.

Faecal calprotectin in the range of 50 to 60 micrograms per gram reached a pooled sensitivity of 0.81 and a specificity of 0.87 for separating organic from functional causes. CRP came in at 0.73 and 0.78. IgA transglutaminase for coeliac workup lay at 0.79 to 0.99 sensitivity and 0.90 to 0.99 specificity. The certainty of evidence ranged from low to moderate.

For you this means: these values are signposts, not gatekeepers. They aim at the whole group of organic causes, not at microscopic colitis in particular.

Verhaegh and colleagues analysed the British Clinical Practice Research Datalink, with cases from 1992 to 2013, matched by age, sex and general practice, using conditional logistic regression.

1211 cases faced 6041 controls, mean age 63.4 years, 73.2 percent women. For current use the adjusted odds ratios were 3.37 for proton pump inhibitors, 2.03 for SSRIs and 1.86 for non-steroidal anti-inflammatory drugs. Statins were not associated. Concurrent use of proton pump inhibitors and non-steroidal anti-inflammatory drugs showed the strongest link, around fivefold. A duration of use of 4 to 12 months raised the risk further.

For you this means: there are substance groups where the question of a temporal connection is worth asking. And it means that statins, despite their reputation, did not stand out in this well adjusted analysis.

Jaruvongvanich, Poonsombudlert and Ungprasert searched MEDLINE and EMBASE up to May 2018 and pooled seven studies with 262,312 participants, two cohorts and five case control studies.

For people who currently smoke, an odds ratio of 2.99 emerged compared with people who have never smoked. For people who smoked in the past it was 1.63, with a heterogeneity of zero percent. The funnel plots were symmetrical, no publication bias appeared.

For you this means: the risk falls markedly after quitting, but it does not disappear at once. A second meta-analysis of eight observational studies with 1461 affected people confirmed this independently, with an odds ratio of 3.58 for current smoking.

Aziz and colleagues searched six databases up to January 2021 and found five studies with 2589 people on the frequency of microscopic colitis in refractory coeliac disease, plus 21 studies with 7186 people on the reverse question.

The pooled prevalence was 4.5 percent and 6.7 percent. For people with refractory microscopic colitis there was a markedly higher chance of a coeliac diagnosis, odds ratio 8.12.

For you this means: if one of the two diagnoses is established and the symptoms persist, the other belongs checked. They do not exclude one another, they occur together more often than chance would suggest.

Kafil and colleagues included twelve randomised controlled trials with 476 participants for Cochrane, searching up to November 2016, with the risk of bias tool and GRADE assessment.

On budesonide 9 milligrams daily over 6 to 8 weeks, 81 percent, meaning 38 of 47 people, reached a clinical response, compared with 17 percent, meaning 8 of 47, on placebo. The relative risk was 4.56, the certainty of evidence low. Histologically, 72 against 17 percent improved. For maintenance: 68 against 20 percent. In the pooled studies, side effects were not more frequent than on placebo, relative risk 1.18.

That does not mean there are none. In the same Cochrane data, nausea, vomiting, abdominal pain, neck pain, excessive sweating and headache are reported on budesonide. Budesonide is a corticosteroid preparation. With longer use, bone metabolism, blood sugar, blood pressure and eye pressure belong under medical observation, and treatment is not stopped abruptly but tapered with medical guidance.

For you this means: of five people on budesonide, four responded in these studies, on placebo not even one. That is a clear difference, and Cochrane still rated the certainty of this evidence only as low, because the studies were small.

Chande and colleagues included five randomised trials with 149 participants, searching up to August 2016.

On budesonide 9 milligrams daily over 6 to 8 weeks, 88 percent responded clinically compared with 38 percent on placebo, relative risk 2.03, certainty of evidence low. Histologically 78 against 33 percent. Beclometasone dipropionate against mesalazine reached 84 against 86 percent remission after 8 weeks, but after 12 months only 26 against 20 percent.

For you this means: in the lymphocytic form too, budesonide is the best documented option. And the last figure shows the core problem of this condition. Reaching remission is one thing, holding it is another.

Münch and colleagues first treated openly over 8 weeks, then randomised the people in remission to low-dose budesonide, a mean of 4.5 milligrams daily, or placebo over 12 months, followed by 6 months without therapy.

In the open phase 84.5 percent, meaning 93 of 110, reached clinical remission. The median time to that point was 10.5 days. After one year 61.4 percent on budesonide were still in remission compared with 16.7 percent on placebo. In the therapy-free follow-up 82.1 percent of those previously treated relapsed. Quality of life was maintained on therapy, suspected side effects occurred in 7 of 44, none of them serious.

That figure comes from 44 people over twelve months. It says something about this group in this period and nothing about rare or late events. Corticosteroid treatment over months remains a decision that is medically supervised and monitored.

For you this means: the effect sets in at a median of about eleven days, not after months. And after stopping, symptoms often come back. Both belong told together.

Tome and colleagues pooled 35 studies, eleven randomised and 24 observational, with 1657 people on flare treatment and 146 on maintenance.

On maintenance therapy the pooled remission rate was 84 percent, with very high heterogeneity. After stopping, the pooled relapse rate was 53 percent. On safety endpoints there was no difference under maintenance therapy compared with placebo or other substances, examined for bone metabolism, high blood pressure, rises in blood sugar, cataract and glaucoma.

For you this means: roughly half get symptoms again after stopping. That is not a failure, it is the known course. And on the safety endpoints examined, these data showed no difference from the comparison group. That statement has its limits too: 146 people on maintenance therapy are too few to make rare events visible, and the observation periods were limited. No licence for unmonitored long-term use follows from these data, but rather the need for medical monitoring.

Levy and colleagues compared 221 people with microscopic colitis, 112 collagenous and 109 lymphocytic, with 306 people from screening colonoscopy and additionally with the US SEER data.

For tubular adenomas there was no difference, odds ratio 1.07, and none for villous adenomas either, odds ratio 1.26. Two or more flares compared with a single one changed nothing either, odds ratio 0.83 for bowel cancer. Compared with the SEER data there was no statistically raised cancer risk.

For you this means: in these data no raised bowel cancer risk could be detected, unlike what is described for ulcerative colitis and Crohn's disease. But not detecting something and ruling it out are two different things. The confidence intervals of this work are very wide, reaching from 0.17 to 9.42 for villous adenomas and from 0.20 to 3.39 for bowel cancer. It comes from a single centre, covers 221 people and was collected retrospectively.

This finding changes nothing about the recommended screening intervals. Anyone invited to bowel cancer screening goes.

Khalili and colleagues identified all 14,333 histologically confirmed cases in Sweden between 1990 and 2017 through SNOMED codes from 28 pathology departments and compared them with 68,700 people from the population.

The adjusted hazard ratio for overall mortality was 1.17. After additional adjustment for the burden of accompanying conditions the difference disappeared: 0.98, with an interval from 0.94 to 1.02. Deaths from gastrointestinal causes remained raised, hazard ratio 1.68, as did deaths from infections, hazard ratio 1.42. Cancer-related mortality was not raised but lower, with a hazard ratio of 0.83.

The authors summarise it this way: mortality was raised overall, but the difference could largely be traced back to the burden of accompanying conditions. For you this means: after that adjustment, the condition itself does not drive overall mortality in these data. Two cause groups still remained raised, which is why their figures stand above.