No-biopsy strategy for coeliac disease is applicable in adult patients: a ‘real-world’ Scottish experience

Abstract

Objective

Emergency interim guidance from the British Society for Gastroenterology (BSG) states that a no-biopsy strategy is possible to diagnose coeliac disease (CD) in adults with elevated transglutaminase IgA antibody (TGA-IgA) levels. We aimed to determine if the suggested TGA-IgA ≥10× ULN is safe and robust in making the diagnosis in adult patients in Scotland. We also aimed to establish if any important co-diagnoses would be missed if no biopsy was performed.

Method

All positive coeliac serology results for patients aged >15 years in Scotland in 2016 (Grampian 2019) were accessed. Data were collected on demographics, TGA-IgA titres, D1 sampling, histology and macroscopic findings at upper and lower gastrointestinal (GI) endoscopy.

Results

1037/1429 patients with positive serology proceeded to biopsy, of which 796/1037 (76.8%) were diagnosed as CD. A total of 320/322 (99.37%) patients with TGA-IgA ≥10× ULN were diagnosed as CD giving the cut-off a positive predictive value of 99.38%. No significant co-pathology was found at endoscopy in these patients.

Conclusion

Our results show that a no-biopsy strategy using a cut-off of TGA-IgA ≥10× ULN is safe to diagnose CD and that no important pathology would be missed. The European Society for Paediatric Gastroenterology, Hepatology and Nutrition 2020 and BSG COVID-19 interim guidelines are applicable to adult patients in Scotland.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

• Coeliac disease is currently under diagnosed in Scotland.

• Waiting times for diagnostic endoscopy are long and in post-COVID times are getting longer.

• Current (2014) British Society for Gastroenterology (BSG) guidance recommends at least four duodenal biopsy specimens, including the bulb (D1).

• Existing adult guidance (BSG and National Institute for Health and Care Excellence) is that biopsy is mandatory to make the diagnosis.

• Evidence exists that a no-biopsy strategy is possible in adults.

WHAT THIS STUDY ADDS

• A no-biopsy diagnosis is safe and robust in IgA-sufficient adults in Scotland and the ≥10× ULN cut-off is safe and practical across all assays used in Scotland.

• BSG guidance is frequently not adhered to, D1 biopsies are often omitted and it is possible that many patients are missed due to non-adherence to current guidance.

• No significant co-pathology would be missed in a no-biopsy protocol.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• We recommend that biopsy of D1 (in addition to D2) is embedded into routine adult GI practice and that samples are sent in separate pots for assessment as this will increase diagnostic rate and allow ongoing audit.

• We recommend prospectively auditing diagnostic outcomes for those also being investigated via other pathways (egsuspicion of cancer due to iron-deficiency anaemia) and to also understand whether serology in other autoimmune conditions (eg type 1 diabetes mellitus (T1DM), thyroid disease) is applicable to the same management strategy.

• Selective IgA-deficient (sIgAD) patients should continue to have an endoscopic diagnosis to help evidence the applicability of a no-biopsy strategy in anti-TGA-IgG-positive patients with sIgAD.

• Benefits of the new strategy include significant cost savings and will also reduce the delay in commencing a gluten free diet (GFD) for eligible patients.

Introduction

The discovery of tissue transglutaminase IgA antibody (TGA-IgA) in 1997 as the autoantigen of endomysial IgA antibody (EMA-IgA) heralded a new era of understanding of the genetics and pathophysiology of coeliac disease (CD).¹ Current CD diagnostic criteria in adults are based on the 1990 European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) guidance.² Current British Society for Gastroenterology (BSG) and National Institute for Health and Care Excellence guidance still has a biopsy-based diagnosis as the ‘gold standard’ with some groups also rebiopsying to assess mucosal healing.^{3 4}

Since 2012, paediatric gastroenterologists in the UK have used a ‘no-biopsy’ strategy for diagnosis from the ESPGHAN CD group, accepted but modified for UK practice by the British Society of Paediatric Gastroenterology, Hepatology and Nutrition (BSPGHAN) and Coeliac UK in 2013 (CD).^{5 6} The ESPGHAN guidelines were prospectively ‘road tested’ in the ProCeDE study.⁷ The ESPGHAN 2020 guidance then confirmed the utility of the TGA-IgA ≥10× ULN cut-off but omitted HLA-DQ typing from the diagnostic criteria (it adds nothing to diagnostic certainty), with a first test using TGA-IgA and then a second separate test, EMA-IgA to confirm the diagnosis.⁸ Several studies show that serological diagnosis is as safe in adult practice, some predating ESPGHAN 2012.^9–11 In 2018–2019, Finland adopted ESPGHAN guidance to include adults and have confirmed their applicability, with 33% of adults qualifying for a no-biopsy diagnosis across all pre-test probabilities, subsequently confirming cost-effectiveness when endoscopy and HLA-DQ typing are both omitted.^{12 13} Since the project began, the COVID-19 pandemic has catalysed a change in practice in UK adults with the BSG suggesting treating patients (<55 years) with suspected CD and a TGA-IgA 10× ULN without biopsy.^14–16 Scotland has a higher CD incidence than many other parts of the UK and significant geographical differences within Scotland in the paediatric setting.^{17 18} Using retrospective data from all Scottish adult endoscopy services, we assessed diagnostic certainty across four different assays, analysed serology/ pathology correlation and adherence to 2014 BSG CD guidance.

Methods

This was a service review auditing routine clinical practice and existing pathway management, specifically accuracy of serology against duodenal histology in the adult population in Scotland. The estimated adult population in Scotland (aged 16 years and over) is 4.49 million.¹⁹ Patient data were anonymised prior to analysis. All positive coeliac serology results for Scotland from 2016 were accessed. This included results from National Health Service (NHS) Ayrshire and Arran, Borders, Dumfries and Galloway, Fife, Forth Valley, Greater Glasgow and Clyde, Grampian, Highland, Lanarkshire, Lothian, Orkney, Shetland, Tayside and Western Isles, see table 1 for individual health board screening tests. The year 2016 was chosen to allow further analysis of follow-up serology and to detail adverse outcomes (or co-morbidities) and who may subsequently have had other investigations and diagnoses (eg, iron-deficiency anaemia (IDA) and suspicion of cancer pathways, other autoimmune conditions). NHS Grampian looked at 2019 data because they were more robust and safer to use (in 2016, the kit manufacturer had changed the assay without informing clients). We included all patients with newly positive serology in 2016 (2019 in Grampian), some of whom were biopsied into 2017 (2020) due to endoscopy waiting times in order to capture all patients from a complete year.

Table 1.

Heath board screening test(s)

Region	TGA-IgA	EMA
NHS Borders	Combined IgA &IgG TTG (Orgentec, Mainz Germany) Normal range 0.1 to 15.0 U/mL	No
NHS Grampian, Orkney and Shetland	IgA TTG Phadia 250 and EMA Normal range 0.1 to 4.0 U/mL and upper limit of report>128 U/mL	Yes on same sample as TTG
NHS GGC, A&A, D&G, FV, Lanarkshire	IgA TTG (Phadia) 1000 Normal range 0.1 to 7.0 U/mL and upper limit of report>128 U/mL	Yes on same sample as TTG
NHS Highland, Western Isles	IgA TTG (Phadia Immunocap 250) IgG TTG for IgA low. Normal range 0.1 to 7.0 U/mL	Yes on same sample as TTG
NHS Lothian and NHS Fife	IgA TTG (Orgentec, Mainz Germany) Normal range 0.1 to 5.0 U/mL and upper limit of report>200 U/mL	No	Fife switched in 2017 Diasorin Liaison XL Normal range 0.1 to 8.0 U/mL and upper limit of report>128 U/mL
NHS Tayside	IgA TTG Phadia 250 and EMA Normal range 0.1 to 4.0 U/mL and upper limit of report>128 U/mL	Yes on same sample as TTG

EMA, anti-endomysial antibody; TTG, tissue transglutaminase IgA.

Data were collected on sex, age at diagnosis, TGA-IgA titres, D1 sampling, histology, use of Marsh grading, macroscopic findings at upper GI endoscopy (and at lower GI endoscopy if performed). For pathology results, we used the descriptors: normal, raised intraepithelial lymphocytes, partial villous atrophy, subtotal villous atrophy and total villous atrophy as formal Marsh grading (or similar) was not available in the majority of cases. A standard set of questions was proposed for the study team (see next section).

Spreadsheets in Microsoft excel (2016) were used to conduct data handling. Statistical analysis and graph construction were performed using Jupyter Notebook and Python V.3.10 with the Pandas, Numpy and Seaborn modules. CIs were calculated using the Exact Clopper-Pearson method in the Statsmodels Python module. TGA-IgA levels were expressed as a multiple of ULN to allow comparison between different assays. A p value of <0.05 was considered statistically significant.

Project questions

Are the ESPGHAN 2020 No-Biopsy guidelines applicable to adult patients?

Is the cut-off of ≥10× ULN of each assay robust in making the diagnosis?

What conditions do we miss if we do not scope?

Do we need to advise on a change in endoscopy and pathology practice to ensure optimal diagnostic rates?

How beneficial will this be in other ways (eg, patient friendly, reduced delays to GFD, cost-effectiveness, endoscopies saved)?

Results

Demographics

1429 patients had a newly positive TGA-IgA across all 14 health boards in Scotland. A total of 903 were women and 526 were men (1.7:1). Age range 15–95 years (mean 49 and median 48).

Indications for testing are listed in table 2.

Table 2.

Indications for testing

Indication	Number of times mentioned on request
Altered bowel habit	319
Abdominal pain	291
Iron deficiency anaemia	267
Bloating	165
Fatigue	123
Other	113
Weight loss	101
Family history	48
Reflux/dyspepsia	47
Nausea/vomiting	32
Rash/dermatitis herpetiformis	29
T1DM	11
Indication not recorded	205

T1DM, type 1 diabetes mellitus.

Endoscopy

A total of 1037 patients proceeded to biopsy and a further four patients had endoscopy without biopsy. A total of 388 patients did not have an endoscopy for a number of reasons (see table 3). Overall, 796 patients (76.8%) who underwent biopsy were diagnosed with CD (722 on histology and 74 based on clinical evidence). Only 202/1037 (19%) patients biopsied had documented sampling of D1.

Table 3.

Reasons for no endoscopy

Reason	Number of patients
Frailty/co-morbidity	52
Declined	41
Did not attend	25
Repeat serology	25
Trial GFD	11
Intolerant of endoscopy	7
Previous endoscopy	5
Family history	2
Pregnancy	2
HLA testing	1
Negative EMA	1
Dermatitis herpetiformis	1
Haemophilia	1
No information	214
Total	388

EMA, anti-endomysial antibody.

TGA-IgA results ≥ 10× ULN

Across all centres, a total of 324 patients with a TGA-IgA level ≥10× ULN proceeded to biopsy. Of those, 320 (98.7%) were diagnosed as CD. A total of 312 were diagnosed on histology and a further 8 patients had normal biopsies but were diagnosed as CD on clinical grounds (resolution of symptoms and normalisation of TGA-IgA on gluten-free diet), with the normal biopsies attributed to the patients being on GFD.

Two patients had a false-positive TGA-IgA. One patient had newly diagnosed type 1 diabetes mellitus. If the patient had been left 6 months, they would not have been scoped due to normalisation of the TGA-IgA level.²⁰ The second patient died of alcoholic liver disease before a GFD was trialled. Two further patients had equivocal findings at biopsy and no conclusion was reached as to CD status.

TGA-IgA results < 10× ULN

A total of 713 patients with a TGA-IgA level <10× ULN proceeded to biopsy. Of those, 476 (66.8%) were diagnosed as CD (410 based on histology and 66 diagnosed on clinical grounds).

The relationship between TGA-IgA and histology is outlined in the boxplot below (figure 1). It excludes ambiguous cases where the pathology report uses the following terms: duodenitis, equivocal or ‘coeliac’ without describing the histological features. The swarmplot overlay demonstrates all cases.

Figure 1.

Relationship between TGA-IgA and histology. TGA, transglutaminase.

Applying a cut-off of ≥10× ULN, the TGA-IgA test has a positive predictive value of 99.38% (97.77%–99.92%), see figure 2. Two patients were excluded for having an equivocal diagnosis with no further information on follow-up.

Figure 2.

Performance characteristics of TGA-IgA 10 x ULN at predicting coeliac disease. NPV, negative predictive value; PPV, positive predictive value; ULN, upper limit of normal.

Upper GI endoscopy

One patient from a total of 1037 biopsied was found to have CD and a tumour at endoscopy. The patient was 55 years old and presented with abdominal pain, loose stools and bloating. TGA-IgA was 52 (7× ULN). There were no red flags at presentation. The tumour was revealed to be a poorly differentiated adenocarcinoma of the duodenum, diagnosed following a Whipple’s procedure. No significant concomitant pathology was found in the group of patients with TGA-IgA ≥10× ULN who underwent upper GI endoscopy.

Lower GI endoscopy

A total of 167 patients underwent lower GI endoscopy, see table 4 for findings. Two patients were diagnosed with colorectal adenocarcinoma. One male patient, aged 68, presented with TGA-IgA 36 (5× ULN), IDA, bowel symptoms and weight loss. The other patient was woman, aged 48 with TGA-IgA 41 (2.7× ULN), IDA and loose stool. Both patients were diagnosed as coeliac. No significant pathology was found in the group of patients with TGA-IgA ≥10× ULN who underwent lower GI endoscopy.

Table 4.

Findings at lower GI endoscopy

Finding	Number of patients
Normal	115
Diverticular disease	17
Adenomas	25
Inflammation	3
Microscopic colitis	5
Adenocarcinoma	2
Total	167

Discussion

Despite the obvious limitations of our study (including its retrospective nature and lack of centralised serology and pathology analysis), we have demonstrated ‘real-world’ serology/histology correlation using four different assays and conclude that a no-biopsy strategy using a cut-off of TGA-IgA over 10× ULN is safe. Excluding the two patients with equivocal diagnoses, a total of 320/322 (99.37%) of biopsied patients with TGA-IgA ≥10× ULN were diagnosed with CD (on biopsy or clinical grounds). The ≥10× ULN cut-off has a positive predictive value of 99.38%. The ESPGHAN 2020 and BSG COVID-19 interim guidelines are applicable to adult patients in Scotland. Importantly, we would not miss any important co-diagnoses. No cancers were detected in those ≥10× ULN. One cancer was detected in a patient with TGA-IgA <10× ULN, but due to the patient’s age and TGA-IgA level, they would have a diagnostic endoscopy, even under the proposed new guidelines. IDA in the presence of positive Coeliac serology should still initiate appropriate investigation but should be actively audited to evidence the frequency of co-diagnoses.

At lower TGA-IgA levels, there are significant issues. Duodenal biopsies in 178/309 (58%) were negative for CD at up to 2 x ULN. This could be a true effect but could also be because D1 biopsies were not performed in the majority of those patients (only 19% (202/1037) had documented D1 sampling). These results indicate that biopsy practice in Scotland is suboptimal, as even existing adult guidelines are clearly not being followed. This is a likely reason for underdiagnosis, as research suggests many true coeliac diagnoses are missed due to omitting bulb (D1) biopsies.²¹

Centralised ‘second opinions’ regarding serology–histology correlations and a forum for case management for difficult cases, including those who are considered ‘non-responsive’ and/or ‘refractory’ CD should be established. The optimal regional management strategy should have expert dietitians at the centre of the service, with a named and CD-interested gastroenterologist as part of a multidisciplinary approach. Significant numbers of patients refuse biopsy or are unable to tolerate endoscopy and need practical, expert management.

Across all the assays in Scotland, we found 31% (324/1037) of new positive cases could safely be diagnosed without a biopsy. Taking the interim BSG guidance age cut-off of 55 years into account, this equates to 214/1037 (21%) patients who could potentially have avoided endoscopy with the associated benefits as the patient avoids unnecessary and invasive tests, no significant delay in starting a GFD and with obvious cost savings to the healthcare system.

We consider the ESPGHAN guidance to be robust and we are aware that all regions of Scotland have adopted the BSG 2020 COVID-19 guidance. In the long term, this new strategy will benefit the healthcare system (including significant cost savings to the NHS from avoiding expensive invasive tests) but much more importantly is patient focused and will hopefully reduce the delay in commencing a GFD for eligible patients. Within the Scottish system, this ‘joined-up’ approach will help promote standardised investigation and ongoing management. We hope this work, all stemming from the Scottish Government’s Test of change CD project,²² will spur ongoing prospective auditing of practice and importantly embed a better, more patient-focused experience.

Conclusions

The ESPGHAN 2020 and BSG COVID-19, no-biopsy interim guidelines are applicable and safe for all adult patients in Scotland with anti-TGA IgA antibodies ≥10× ULN. Adult gastroenterology services need to improve adherence to current guidelines with respect to time to endoscopy for those requiring more than serological diagnosis, to our endoscopic practice, for example, sampling D1 (which may improve diagnosis) and prospective national service audit should be used to systematically evidence applicability of the current strategy. In addition, it should be assessed whether a limit lower than ≥10× ULN could be safely applied, dependent on local serology–histology correlation. Even more importantly, patient satisfaction and feedback around the experience of the new diagnostic pathway must be gauged as part of a holistic approach to this common condition.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statements

Patient consent for publication

Not applicable.