Refractory coeliac disease and how to avoid them

Once a diagnosis of coeliac disease has been established, clinical and mucosal healing is usually reached within 12 months as a result of gluten withdrawal (i.e. adoption of a GFD). However, some coeliac patients respond slowly to a strict GFD and continue to experience symptoms and incomplete recovery after 12 months. In these slow responders, full recovery may occur after 18–24 months.¹⁵ The appropriate approach in this situation is a cautionary surveillance and ‘wait-and-see’ follow-up strategy, which avoids unnecessary invasive tests and potentially harmful treatments (e.g. immunosuppressive drugs and steroids).

Patients with coeliac disease feel the burden of long-term dietary restriction of gluten. Indeed, data indicate that only 60% of coeliac disease patients adhere strictly to a GFD.¹⁶ The remaining 40% of patients inadvertently, or (more commonly) willingly, reintroduce a significant amount of gluten into their diet, which causes the persistence of villous atrophy and gastrointestinal/extraintestinal symptoms. The lack of adherence to a GFD is reflected by high titres of anti-transglutaminase (TG2) antibodies, which continue to be detected in the serum of these patients. By contrast, RCD patients have either a negative or minimal increase (in approximately 10% of cases) in anti-TG2 antibodies.¹⁵ Physicians should advise all patients with coeliac disease about the risks of voluntarily introducing gluten, in terms of complications (such as RCD, enteropathy-associated-T-cell lymphoma [EATL], small bowel adenocarcinoma and ulcerative jejunoileitis), whereas resuming a strict GFD normalizes clinical and morphological (duodenal histology) features, preventing the possible occurrence of the above-mentioned complications.

Patients who have a nongluten-dependent villous atrophy can sometimes be mistakenly diagnosed as having seronegative coeliac disease. A thorough diagnostic work-up should be undertaken as it may identify conditions other than coeliac disease that are responsible for villous atrophy. Alternative causes include infections (e.g. with Giardia lamblia (giardiasis) or HIV), autoimmune enteropathy (characterized by anti-enterocyte autoantibodies), drug-induced enteropathy (e.g. caused by angiotensin II-receptor blockers [e.g. olmesartan], mycophenolate mofetil and NSAIDs), common variable immunodeficiency, small intestinal bacterial overgrowth (SIBO), Crohn’s disease, Whipple disease and eosinophilic gastroenteritis, among others.¹⁷ A simplified algorithm for the differential diagnosis of seronegative villous atrophy is shown in figure 2. Prior to labelling a patient as having RCD, it is mandatory to verify whether the initial diagnosis of coeliac disease was appropriate. 

Figure 2 | Evaluation of persistent villous atrophy in seronegative coeliac disease. An algorithm we developed for the evaluation of persistent villous atrophy in initially seronegative coeliac disease patients on a gluten-free diet (GFD). Exclusion of coeliac disease should prompt the investigation of other nongluten-dependent causes of villous atrophy. CVID, common variable immunodeficiency; EATL, enteropathy-associated-T-cell lymphoma; RCD, refractory coeliac disease; SIBO, small intestinal bacterial overgrowth. 

Biopsy samples should be adequately oriented in the endoscopy room in order to avoid obtaining false-positive results (i.e. a wrong interpretation of villous atrophy supporting the lack of response to a GFD). This is a critical issue that can be avoided by correct longitudinal orientation (along the length of the villi) of the biopsy samples using appropriate devices (i.e. a cellulose acetate filter). Endoscopists should take at least n=4 biopsy samples from the second part of the duodenum and n=2 from the duodenal bulb (the latter at the 9 o’clock and 12 o’clock position to maximize the histopathological yield). Histopathological reports, indicating persistence of villous atrophy, should be the result of adequately embedded and sectioned biopsy material according to the orientation given in the endoscopy room.¹⁵ 

Physicians should be alert to avoiding the incorrect diagnosis of nonresponsive coeliac disease as RCD. Patients who have nonresponsive coeliac disease may complain of a wide array of symptoms, including bowel habit abnormalities, abdominal pain, bloating as well as nausea, vomiting and gastro-oesophageal reflux disease (GORD) symptoms. All such symptoms and/or signs are ascribable to other diseases (table 2) that frequently overlap with coeliac disease. The differential diagnosis of nonresponsive coeliac disease versus RCD is based on thorough histopathological evaluation of duodenal biopsy samples, with a normal villi cytoarchitecture visible in patients with nonresponsive coeliac disease compared with marked changes in patients with RCD.⁶

A GFD can lead to clinical improvement in a proportion of coeliac disease patients who still have a mild villous atrophy at histopathology (lesion 3a according to the Marsh–Oberhüber classification).¹⁸  In these cases, the improvement of intestinal damage from grade 3b/3c (at diagnosis) to grade 3a (at follow-up) should be considered an indication of a positive outcome. Although there are no long-term studies, it is likely that this subset of patients will have complete intestinal villous regrowth over time.¹⁹ 

Although not believed to be strictly necessary in all cases, taking intestinal biopsy samples remains the gold standard to verify the status of mucosal improvement in coeliac disease patients who are on a strict GFD. We suggest physicians avoid recommending histopathological assessment prior to 12 months from initiation of the GFD. This interval is necessary to allow for regrowth of the intestinal mucosa. Should biopsy samples be taken in the 3–6-month after starting a GFD, the risk of still obtaining histopathological findings of severe villous atrophy is quite high, thus leading to a misdiagnosis of RCD.¹⁵ 

RCD can be classified clinically as primary and secondary subtypes. Primary RCD encompasses those patients who have no clinical/histopathological improvement from the time a GFD is begun; secondary RCD includes patients who experience sudden clinical worsening after many years of a very good response to a GFD.^1–3 The mechanisms underlying these two phenotypes of RCD are largely unknown. Primary RCD patients are easily recognizable because of the absence of any clinical response to a GFD. By contrast, physicians should be aware that having a good clinical and histological response to a GFD for many years does not rule out the possible occurrence of secondary RCD. In this context, low levels of haemoglobin (<11 g/dl) and albumin (<3.2 g/dl) that are associated with symptom recurrence may be indicative of secondary RCD.⁴ These cases should be reassessed by taking duodenal biopsy samples as early as possible to confirm the late refractoriness to a GFD. 

A critical aspect of the possible evolution of patients with RCD type 2 to EATL is that it occurs more commonly in patients of advanced age and if there has been late diagnosis of coeliac disease. In this context, increased levels of lactate dehydrogenase and -2-microglobulin suggest that RCD has evolved to EATL.²⁰ Imaging tests are of paramount importance in order to allow for early recognition of EATL, which is a life-threating complication of RCD type 2.¹⁵ Once the diagnosis of EATL is suspected, physicians should recommend positron emission tomography (PET), which is the best exam to identify lymphoproliferative foci throughout the small intestine and their possible extension to the bone marrow. Entero-MR or entero-CT is needed to identify small intestinal lesions and, therefore, guide endoscopic assessment via DBE, which is useful for obtaining biopsy samples for accurate histopathological analysis. Compared with DBE, VCE appears to have a low diagnostic yield (it does not enable the taking of biopsy samples) and could be harmful for patients with EATL or ulcerative jejunoileitis because of the possible risk of device retention.