No-biopsy approach to the diagnosis of coeliac disease

In 1983, endomysial antibodies (EMAs) were found in the serum of patients with coeliac disease (CD) and dermatitis herpetiformis and with high sensitivity and specificity proved an almost ideal diagnostic test. The enzyme tissue transglutaminase (TTG) was identified as the endomysial antigen in 1997 and led to a further test that could be quantified and also exhibited high diagnostic accuracy. These tests allowed large scale screening studies that showed CD to be one of the most common chronic disorders in the western world with a prevalence of about 1%.¹

Upper gastrointestinal endoscopy is invasive and for many patients requires sedation or occasionally general anaesthetic, is uncomfortable and interpretation of small intestinal biopsy specimens obtained is not always straightforward.¹ The procedure is expensive and time consuming for patients and endoscopists. It was estimated that the cost of diagnosing CD in children could be reduced by 95% if endoscopies were omitted.² Similar savings are likely to apply in adult practice. Therefore, the next step was to ask whether serological tests could be used to establish the diagnosis of CD without recourse to small intestinal biopsy. Studies based on a defined level of TTG antibodies indicated that this was the case and about 50% of paediatric and adult patients could be diagnosed without biopsy.¹ The case was so compelling that the principle was incorporated into the European Society for Paediatric Gastroenterology Hepatology and Nutrition (ESPGHAN) guidelines for the diagnosis of CD in children in 2012 and modified in 2020. However, gastroenterologists in adult practice have been reluctant to go down this route. The COVID-19 pandemic that severely restricted the use of endoscopy to acquire small bowel biopsies has accelerated a no-biopsy approach to diagnosis.³

An investigation from Finland⁴ explored the value of the serological approach in large groups with variable pre-test probabilities. A cohort (421) considered at high-risk of CD because of features consistent with the diagnosis, a cohort (2357) at moderate-risk was composed of family members of CD cases and a low-risk cohort (4722) drawn from randomly selected people representing the older Finnish population made up the series. The prevalence of CD in these groups was considered to be up to 50%, 7.5% and 2%, respectively. Triple criteria for CD comprised TTG antibodies >X 10 upper limit of normal (ULN), positive EMAs and the presence of Human Leucocyte Antigen (HLA)-DQ2/DQ8. The diagnosis was based on intestinal biopsy. For each of the cohorts, the positive predictive value (PPV) for the triple criteria was found to be 100% indicating that the diagnosis can be determined irrespective of the pre-test probability. Altogether, 90 of 274 (33%) patients with CD could have avoided biopsy. There was very good agreement between EMA positivity and TTG antibodies >X 10 ULN, so this test appears to be superfluous as does determining HLA status. No comorbidities occurred in those who would have avoided an endoscopy.

The accuracy of a no-biopsy strategy for the diagnosis of CD was determined across three adult cohorts.⁵ Those in cohort 1, 740 patients, were assessed consecutively and prospectively in a specialist CD clinic and were considered to have a high chance of developing CD. The prevalence of CD in this group was 690/740, 93%. The PPV for TTG antibody levels of ≥10 X ULN at predicting mucosal abnormalities with villous atrophy was 98.7%. Two different TTG antibody tests were used in this cohort. When those tested by EliA Celikey (273) were analysed, the PPV was 100%. Cohort 2 comprised a group of adult patients recruited prospectively from an upper gastrointestinal endoscopy clinic and considered to have low suspicion of CD. Of 532 included in the analysis, the prevalence of CD was 17/532 (3.2%). Marsh 3 lesions were found in 9/9 of those with TTG antibody levels ≥10 X ULN, so the PPV was 100%. Cases making up cohort 3 (145) were recruited retrospectively and had elevated TTG antibody levels determined by 11 laboratories in eight countries. The prevalence of CD in this cohort was 133/145 (92%). Of 42 with TTG antibodies ≥10 X ULN, 40 had Marsh 3 lesions (PPV 95.2%).

Thus, TTG antibody levels ≥10 X ULN appear equally satisfactory at predicting Marsh 3 lesions in low and high disease pre-test probabilities.

In Frontline Gastroenterology, a service review of routine clinical practice in Scotland encompassing 14 Health Boards employing four different assays to determine the accuracy of coeliac serology against duodenal histology was undertaken and presented as a ‘real world’ experience.⁶ All patients with newly positive serology in 2016 (2019 for Grampian) were included in the audit. Information collected included age at diagnosis, tissue transglutaminase IgA antibody (TGA)-IgA antibody titres, macroscopic findings at upper gastrointestinal endoscopy, if biopsies were taken from the duodenal bulb, small bowel histological grading and appearances at colonoscopy. Questions posed by the investigators included, are ESPGHAN no-biopsy guidelines transferrable to adults? What conditions do we miss if we do not scope? Who should we scope even if they qualify for no-biopsy? How will this approach benefit patients and the endoscopy service? These are questions that are being asked by those pondering a no-biopsy strategy for the diagnosis of CD but perhaps not quite won over. This audit of practice in Scotland should go some way to allay the concerns of doubters.

Of 1429 patients recruited to the study with newly positive TGA-IgA antibodies, 1037 proceeded to small intestinal biopsy. Four had endoscopies without biopsy and for a variety of reasons, 338 were not biopsied. Only 19% had documented sampling of the duodenal bulb indicating that current guidelines are not being followed. Of 324 with TGA-IgA antibody levels ≥10 X ULN, 320 (98.7%) were diagnosed with CD. Of the four exceptions, two had equivocal biopsy findings precluding drawing any conclusion regarding their CD status. Thus, at this cut-off, the PPV was 99.38%. No important comorbidities were discovered in this group at upper or lower gastrointestinal endoscopy. Across Scotland for the four different assays employed, 31% of new positive cases could be diagnosed without recourse to biopsy.

This study confirms that ESPGHAN 2020 guidelines for the serological diagnosis of CD can safely be extended to a sizeable proportion of adults. The cut-off of ≥10 X ULN for the assays employed is robust and no important comorbidities will be overlooked. Avoiding biopsies will be welcomed by adult patients who should receive a definitive diagnosis and treatment with a gluten-free diet more expeditiously.

Evidence supporting a no-biopsy strategy is accumulating and indicates that:

• Up to about 50% of adults with CD can be diagnosed on serological criteria alone.

• TTG antibody positivity is not enough to make the diagnosis. Values in terms of multiples of the ULN that equate to characteristic histology using high performing, locally validated kits must be taken strictly into account.

• Small bowel biopsy is not obsolete but has a place when serological criteria are not met, in serologically negative patients if the diagnosis is suspected or if clinicians suspect they are not dealing with a straightforward patient.

• A serological approach reduces pressures on endoscopy services with cost savings and benefits to patients.

• Expert dieticians and clinicians interested in CD should to be involved in the management of patients.

• Guidelines regarding the diagnosis of CD in adults require amendment to reflect the research base supporting a no-biopsy approach.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.