Diagnostic significance of HLA-DQ typing in patients with previous coeliac disease diagnosis based on histology alone

Authors


Dr I. R. Korponay-Szabó, Department of Pediatrics, University of Debrecen, Nagyerdei krt 98, Debrecen, H-4012, Hungary.
E-mail: ilma.korponay-szabo@uta.fi

Abstract

Summary

Background

Coeliac disease is strongly associated with human leukocyte antigen (HLA)-DQ2 or DQ8 genotypes. The diagnosis is based on demonstrating crypt-hyperplastic villous atrophy, endomysial or transglutaminase antibodies and correlation of disease activity with gluten intake.

Aim

To evaluate the clinical utility of HLA-DQ typing, when coeliac disease diagnosis had previously been established solely by histology.

Methods

HLA-DQ alleles, endomysial and transglutaminase antibodies were investigated and histology slides reviewed in 70 patients diagnosed 2–25 years earlier by small-intestinal biopsy but without measuring endomysial or transglutaminase antibodies. Patients without DQ2 or DQ8 or without unequivocal villous atrophy were followed-up on free diet by using serology and biopsies.

Results

All 40 endomysial/transglutaminase antibodies positive patients carried DQ2 or DQ8, and 39 of them had severe villous atrophy. Only 56% of patients without endomysial or transglutaminase antibodies positivity had DQ2 or DQ8 (P < 0.001). Seropositivity and relapse developed in 4 of 11 DQ2 positive but in none of 15 DQ2 and DQ8 negative patients on long-term gluten exposure.

Conclusions

Coeliac disease diagnosis based solely on histology is not always reliable. HLA-DQ typing is important in identifying DQ2 and DQ8 negative subjects who need revision of their diagnosis, but it does not have additive diagnostic value if endomysial positivity is already known.

Introduction

Coeliac disease is a T-cell-mediated chronic inflammatory disease of the small intestine, which occurs in 1% of the European population. In genetically susceptible persons, small bowel mucosal inflammation, crypt hyperplasia and villous atrophy develop after ingestion of gluten, present in wheat, rye and barley.1 Active coeliac disease is characterized by disease-specific antibodies against endomysium (EMA) or tissue transglutaminase (anti-TG),2, 3 and their presence in serum3, 4 or locally in the gut5 is of high diagnostic value. Most common symptoms are weight loss, pale offensive diarrhoea or constipation and abdominal bloating, but clinical symptoms may even be absent or only involving extraintestinal organs (e.g. osteoporosis). The intolerance towards gluten lasts life-long, but elimination of gluten from the diet results in complete remission, thus in disappearance of symptoms, antibodies and histologic abnormalities.2, 4

Coeliac disease is a strongly inheritable disease with 10% prevalence among first-degree relatives and at least 75% concordance in monozygotic twins compared with 11% in dizygotic twins.6 Approximately, 90–95% of patients carry the human leukocyte antigen (HLA)-DQ2 heterodimer composed of DQA1*05 and DQB1*02 molecules, either encoded in cis position together with DRB1*03 or in trans with DRB1*05/07.7 The remaining patients usually have DR4;DQ8 haplotypes (DQA1*0301, DQB1*0302 alleles) with extremely few exceptions.8 Patients negative for both HLA-DQ2 and −DQ8 are very unlikely to suffer from coeliac disease, because these molecules are necessary to present the antigens to T cells.7, 9 Gluten peptides are substrates for tissue transglutaminase enzyme that transforms glutamine residues to negatively charged glutamic acid residues by deamination. With these negative charges, DQ2 and DQ8 molecules (which have positively charged binding pockets) will be able to bind and present them to CD4+ T cells. In Hungary, the first HLA study in patients with gluten enteropathy was performed by Kárpáti et al. in 1985,10 and subsequent reports further confirmed the strong association of coeliac disease with DQ2 and DQ8 molecules both in Hungarian children and adults.11, 12

The presence of HLA-DQ2 or DQ8 is necessary but not sufficient for the development of coeliac disease, because they also occur in 20–30% of the general European population. Moreover, also non-HLA genes have an important but yet unclarified contribution in disease development.13 However, HLA-DQ typing may have clinical relevance in estimating the risk of family members, or to evaluate the probability of coeliac disease in uncertain cases. A low prevalence of DQ2 or DQ8 was found among patients who started a gluten-free diet without a confirmatory small bowel biopsy.14 The gold standard of coeliac disease diagnosis is today small bowel histology, but its evaluation may have several pitfalls.15, 16 Some features, such as elevated intraepithelial lymphocyte count or shortening of the villi, are non-specific changes, and also may occur in other diseases, such as nutritive allergy or postinfectious damage.17 In this study, we investigated the value of HLA-DQ typing in cases where the diagnosis of coeliac disease was based on an earlier pathology statement but information on the presence of EMA or anti-TG was lacking.

Methods

Patients and study design

Patients with a coeliac disease diagnosis prior to 2002 were traced from patient files of the Department of Pediatrics, University of Debrecen and subjects with a histology-based diagnosis but without EMA or anti-TG antibody results before diet were enrolled in the study. Altogether 78 patients met the inclusion criteria and 70 of them (current median age 13 years, range: 3–31) presented for clinical examination and could be enrolled. They received their original diagnosis when they were in median 3 years old (range: 1–28) with a gluten-free diet prescribed 2–25 years ago. Initial pathology statements and whenever available, original biopsy slides were reviewed by one pathologist (L.T.). Serum EMA and anti-TG antibodies and HLA-DQ haplotypes were determined from blood. Dietary history on the length and strictness of gluten exclusion after diagnosis as well as current gluten intake were recorded. Current diet was classified as strict gluten-free diet, non-strict diet and free gluten intake based on interview with the patients or their parents, and on current EMA/anti-TG results.

Patients with symptoms or seropositivity, or in whom the original biopsy material was insufficient or unsuitable for diagnosis, underwent a new small intestinal biopsy. Patients with histology findings compatible with active coeliac disease were prescribed a strict gluten-free diet, received a detailed dietary education and were followed-up clinically and by EMA and anti-TG tests.

In cases, where the initial diagnosis was uncertain and current gut histology did not support the presence of coeliac disease, a free gluten intake was suggested. Serum EMA and anti-TG antibodies were determined at the intervals of 3–6 months. A new biopsy was performed if the antibodies became positive or symptoms developed. If neither antibodies nor symptoms appeared, a small intestinal biopsy was performed after 2 years of free gluten intake to exclude coeliac disease in line with the original diagnostic criteria the European Society of Paediatric Gastroenterology (ESPGAN) for establishing the diagnosis of coeliac disease by histology, formulated in Interlaken in 1969.18 The flowchart of the study is presented in Figure 1.

Figure 1.

 Flowchart of the study. Patients with endomysial antibody positivity (EMA) were also positive for transglutaminase antibodies HLA-DQ.

Blood for HLA-DQ typing was also obtained from 40 healthy control subjects with the same ethnic origin as patients. They were either hospital employees or visitors who were unrelated to the patients.

HLA-DQ typing

Genomic DNA was isolated from buffy coats of EDTA-anticoagulated blood using QIA amp blood mini kit (QIAGEN GmbH, Hilden, Germany) according to the instructions of the manufacturer. Polymerase chain reaction (PCR)-based HLA-DQ typing was performed (Olerup-SSP), using low-resolution kit (HLA-DQ Low–bulk; GenoVision, Oslo, Norway). In cases when this examination did not give unambigous results, subtyping was performed with DQB1*02, DQB1*03 subtyping kits or DQA1 typing kit from GenoVision. All the examinations were performed according to the instructions of the manufacturer. HLA genotypes were determined on the basis of the PCR pattern obtained by electrophoresis in 2% agarose gel.

Determination of EMA and anti-TG antibodies

IgA and IgG class EMA were investigated from serum by indirect immunofluorescence by using human umbilical cord substrate as described elsewhere.19 The starting serum dilution was 1:10 in phosphate-buffered saline. EMA was also checked on monkey oesophagus by using 1:2.5 serum dilution.20 Anti-TG antibodies were measured by ELISA by using human recombinant antigen expressed in Escherichia coli21 following the incubation protocol described in Reference 3.

Histology evaluation

Villous height/crypt depth ratio was determined from well-oriented sections and intraepithelial lymphocytes were counted. If needed, new sections were cut from the original blocks. Histology lesions partial villous atrophy grade II-III, subtotal and total villous atrophy (villous height/crypt depth ratio < 1) were considered compatible with coeliac disease diagnosis.15, 17, 18, 20 Only histology results while the patient was on a gluten-containing diet were considered relevant for final diagnosis and included in the comparison with HLA-DQ typing results.

Results

At the time of enrollment, only 31 of the 70 patients (44.3%) followed a strict gluten-free diet. These patients were negative for EMA and anti-TG in serum at normal serum total IgA levels. From the other 39 patients, 29 were on a non-strict and 10 on a free diet. Both EMA and anti-TG were detectable in 27 (69.2%) of these 39 patients, but none had clinical symptoms. Clinical symptoms registered at the time of initial diagnosis did not differ in EMA/anti-TG positive and negative patients (Table 1).

Table 1.   Clinical symptoms at diagnosis
  EMA or anti-TG positive patients (n = 40*), n (%)Patients without EMA or anti-TG positivity (n = 30), n (%)
  1. * Twenty-seven patients were endomysial (EMA) and transglutaminase antibody (anti-TG) positive at enrollment, nine had historical EMA positivity and four developed EMA and anti-TG during follow-up.

Indigestion27 (67.5)24 (80.0)
Abdominal bloating20 (50.0)11 (36.7)
Diarrhoea17 (42.5)15 (50.0)
Failure to thrive15 (37.5)14 (46.7)
Anaemia11 (27.5)7 (23.3)
Diabetes mellitus1 (2.5)0 (0)

All 27 EMA and anti-TG positive cases and nine further patients, who were currently negative for serum EMA but had a positive EMA result at some time after diagnosis, carried the DQ2 or DQ8 heterodimers. Anti-TG determinations were introduced only after 2000, thus results were not available for all patients with anamnestic EMA positivity. However, all anti-TG positive patients were also positive for EMA. DQ2 or DQ8 was also found in 56% of patients without any documented EMA or anti-TG positive results. In 15 patients, neither DQ2 nor DQ8 alleles were present and these patients did not have single alpha or beta chains of these heterodimers either (Table 2). The DQ2 or DQ8 heterodimers occurred in 23% of healthy Hungarian control subjects indicating that the presence of these alleles alone does not prove the existence of the disease.

Table 2.   HLA-DQ2 and DQ8 positivity in patients having or not having positive serum endomysial (EMA) or transglutaminase (anti-TG) antibody results at the time of enrollment
 EMA or anti-TG+ n = 36 (%)EMA and anti-TG −n = 34 (%)Total n = 70 (%)
  1. From all patients 47 carried DQ2 in cis, four DQ2 in trans and four carried DQ8.

DQ2+ or DQ8+36 (100)19 (55.8)55 (78.6)
DQ2− and DQ8−015 (44.2)15 (21.4)

Altogether 79 previous histology samples from 65 patients were available for re-evaluation (Table 3). Further, 16 detailed pathology descriptions for missing samples were reviewed. There was a high frequency of poorly oriented specimens (21%, Figure 2) and misleading statements based on dissecting microscopy findings (12%). In three cases, the specimen was not intestinal tissue but stomach.

Table 3.   Comparison of initial and revised histology data of 79 historical small bowel biopsy specimens from 65 patients diagnosed with coeliac disease 2–25 years earlier
Initial pathology descriptionInitial pathology statementInitial clinical interpretationSecond opinion after revisionFinal diagnosis after follow-up*Number of samplesHLA-DQ2 or DQ8+
  1. * Based on clinical, serology, and new histology data during gluten intake.

  2. † Normal villous structure in further sections in five of these samples.

  3. ‡ Patient developed endomysial antibody positivity during gluten intake but refused a new biopsy.

  4. § More than one samples were reviewed per patient.

Severe villous atrophyCoeliac diseaseCoeliac diseaseSevere villous atrophy with crypt hyperplasia in well oriented specimen (Marsh IIIb-IIIc)Coeliac disease4141
Severe or partial villous atrophyCoeliac diseaseCoeliac diseasePartial villous atrophy with crypt hyperplasia (Marsh IIIa)Coeliac disease66
Severe or partial villous atrophyCoeliac diseaseCoeliac diseasePoor orientationCoeliac disease44
Severe or partial villous atrophyCoeliac diseaseCoeliac diseasePoor orientation†Not coeliac disease105
Minor changes in the villi or epitheliumCoeliac diseaseCoeliac diseaseNormalNot coeliac disease71
Normal villous architectureNormalCoeliac disease based on stereomicroscopyNormalNot coeliac disease43
StomachStomachCoeliac disease based on stereomicroscopyStomachNot coeliac disease30
Insufficient materialUnsuitable for diagnosisCoeliac disease based on symptoms or stereomicroscopyUnsuitable for diagnosisNot coeliac disease30
Insufficient materialUnsuitable for diagnosisCoeliac disease based on clinical findingsUnsuitable for diagnosisProbably coeliac disease‡11
Total    79§61§
Figure 2.

 (a) Poorly oriented initial small bowel biopsy sample from a 5-year-old girl raisinig the suspicion of villous atrophy and elongated crypts (arrow). (b) Same specimen after correct orientation shows normal villous structure.

At the time of DQ testing, 32 new biopsies were also performed. Based on data from old and new biopsies, severe villous atrophy with elevated intraepithelial lymphocytes and crypt hyperplasia was unequivocally present in 35 out of the 36 patients with EMA positivity and DQ2 or DQ8 genotypes (97%), but only in eight out of 19 cases (42%, P < 0.001) with DQ2 or DQ8 genotypes and unknown EMA status before diet. However, negative results during restricted gluten intake could not be used to exclude the disease. In the 15 patients without DQ2 or DQ8 alleles, only slight changes without severe villous atrophy were found. None of the patients without EMA positivity had the diagnosis verified by gluten reintroduction according to the 1969 ESPGAN criteria.

In order to evaluate the possibility of coeliac disease in uncertain cases, a follow-up on unrestricted free gluten intake was suggested to 11 DQ2/DQ8 carrier patients who never had EMA positive results before and to all 15 DQ2 and DQ8 negative patients. Of these, 21 agreed to follow-up biopsies according to the ESPGAN protocol. Four of the DQ2 positive patients developed clinical symptoms and EMA positivity within 6 months and three of them evinced subtotal villous atrophy in the biopsy specimen obtained after gluten reintroduction. One patient refused the control biopsy after relapse. All other patients remained symptom-free and none developed EMA, anti-TG or histologic abnormalities at the time of follow-up biopsy after 2 years on gluten (Figure 1). All 15 DQ2 and DQ8 negative patients tolerated gluten well for a follow-up of 2.3–7 years on free gluten consumption, thus the negative predictive value of the DQ typing result was 100%.

The collected histology evidence for and against the diagnosis of coeliac disease is presented in Table 4 in relation to HLA-DQ results. According to data summarized from past and prospectively performed biopsies, presence of coeliac disease has been verified in 39 of the 70 patients and is very probable in further 9 (altogether 68.6%); excluded in 14 and highly unprobable in 8. The presence of HLA-DQ2 or DQ8 had 100% sensitivity for coeliac disease, but the specificity was only 85% (Table 3). However, the specificity of the initial pathology statement (75%) was even lower (Table 3).

Table 4.   Cumulative histology evidence pro and against coeliac disease in patients originally diagnosed by only histology. Results are based on extended follow-up and new biopsies on gluten-containing diet in uncertain cases. All endomysial (EMA) and anti-TG antibody (anti-TG) positive cases carried DQ2 or DQ8
HistologyEMA or anti-TG+ DQ2 or DQ8+EMA and anti-TG− DQ2 or DQ8+EMA and anti-TG− DQ2 and DQ8-
  1. * Patient refused new gluten exposure or new biopsy.

  2. SVA, subtotal villous atrophy; PVA, partial villous atrophy; CD, coeliac disease.

SVA3570
PVA410
Slight changes000
Normal0514
Unsuitable for diagnosis1*2*1*
Total401515
Supporting CD diagnosis (%)39 (97.5%)8 (53.3%)0 (0%)

Discussion

Coeliac disease can be diagnosed on the basis of severe villous atrophy and presence of coeliac disease-specific autoantibodies (anti-TG or EMA) in serum.2, 22 Our study confirmed the reliability of this policy, as all patients meeting both these criteria also carried the correct genetic HLA-DQ background required for the disease. Moreover, the combination of seropositivity and villous atrophy had similarly high value if registered at different times or at later stage after initial diagnosis. None of our patients had initial EMA results from the time of the original diagnosis, but dietary lapses allowed us to register EMA positivity in approximately 50% of these patients. Anti-TG positivity could be utilized in the same way, as the results were identical with those of EMA determinations.

In cases without documented EMA (or anti-TG) positivity, the diagnosis of coeliac disease is less certain. The histologic abnormalities are not specific for the disease17, 18 and there can be frequent pitfalls in the pathology evaluation, such as poor sample quality or orientation.16 In our series, a high frequency of misleading dissecting microscopy results was found, and these were later misinterpreted by the clinicians as proof of villous atrophy. These difficulties were recognized as early as in 1969, when ESPGAN has formulated the classical criteria for coeliac disease diagnosis, also known as Interlaken criteria.18 The Interlaken criteria rely on the gluten-dependency of the disease and require at least three biopsies, showing that the histologic abnormalities heal after a gluten-free diet, but return if the patient starts to eat gluten again. Although these criteria seem today outdated, a new and controlled gluten exposure can have utmost clinical importance in doubtful cases.23, 24 In fact, our series represented such a patient group, as subjects with currently negative EMA results and unknown EMA status before diet had only low prevalence of HLA DQ2 or DQ8 alleles.

The group of young adults on a gluten-free diet and with a coeliac disease diagnosis from their early childhood before EMA and anti-TG had been introduced represents a special clinical challenge. Their diagnosis often does not meet current diagnostic standards, the initial clinical symptoms could have been non-specific and some of the histology findings in early childhood could also be due to other diseases, such as cow's milk protein enteropathy or transitory damage after infections.14 In this context, determination of HLA-DQ can be the most valuable test to exclude coeliac disease.

In our series, none of the DQ2 and DQ8 negative patients relapsed on long-term gluten exposure or developed EMA or anti-TG antibodies. Although a few DQ2 and DQ8 negative coeliac patients are known in the literature, only four such cases were described in a large European study among approximately 1000 coeliac cases and none of those had the diagnosis proven by gluten challenge.8 Such exceptional cases may carry rare mutations in such DQ alleles which normally do not confer susceptibility to coeliac disease but which make them able to present gliadin to T cells. A similar phenomenon has been described in type-1 diabetes mellitus.25

In general, only DQ2 and DQ8 negative HLA-DQ typing results can be utilized clinically as arguments against coeliac disease. Presence of DQ2 or DQ8 is common in the general population and does not prove the presence of the disease. In the absence of EMA/anti-TG positivity and convincing histology, some DQ2 or DQ8 carriers were also able to tolerate gluten on the long term in our study. It has been shown that an initially negative EMA finding has a high negative predictive value for coeliac disease even in patients with villous atrophy.20, 24 Although our patients did not have initial EMA results, they remained negative for EMA and did not develop signs of villous atrophy at the biopsy after free gluten exposure for more than 2 years. In some cases in the literature, histologic relapse was observed after as long as 14 years.26 Thus, the original two-year rule of the Interlaken criteria to exclude coeliac disease18 might not be valid nowadays. However, for a group, more than 2 years observation time seems to be sufficiently safe to label these patients as probably non-coeliacs.

The gluten-free diet is a safe and effective treatment for coeliac disease, but not easy to follow, especially in case of adolescents and young adults. Most patients tend to relax the diet after some time without continuous support.27 We observed a high rate of dietary non-compliance in our patient cohort, which indicates that the family did not take the old diagnosis seriously. All these patients were clinically silent, but those with EMA/anti-TG positivity had ongoing villous damage. In fact, 21 of the 27 patients who were EMA positive because of diet transgressions achieved seronegativity after a confirmatory new biopsy and appropriate dietary education (data not shown).

The PCR typing of alfa and beta chains of HLA-DQ molecules is simple and cost-effective. It is still a question whether it is necessary to examine also HLA-DR,-DP or HLA-A,-B,-C or not. DQ2 in cis is transmitted in the conserved HLA-A1,Cw7,B8,DR3 haplotype in European subjects,7 whereas different HLA-A and -B alleles were found in coeliac disease patients in India.28 Also other studies suggest that only DQ molecules have additive role in the inheritance of the disease.29 Thus, typing of HLA-DQ seems to be sufficient for clinical purposes. In accordance with previous studies,14 all EMA positive patients were found to carry the HLA-DQ2 or DQ8 molecules, thus HLA-DQ typing in EMA positive patients is not necessary. In our study, all anti-TG positive patients were also positive for EMA and HLA-DQ2 or DQ8. In fact, anti-TG and EMA tests measure the same antibodies,30 and discrepancies are very rare when using appropriate methods. Thus, positivity for anti-TG probably also makes HLA-DQ typing unnecessary.

In conclusion, our findings support the clinical utility of HLA-DQ typing in cases where the diagnosis of coeliac disease is uncertain.

Acknowledgements

This study was supported by the ETT518/2003 grant from the Hungarian Ministry of Health and by the Research Fund of Tampere University Hospital, Finland.

Ancillary