What Is Known
- The new 2012 guidelines for diagnosis of coeliac disease are based on evidence, but they have not been clinically validated yet.
- Diagnostic biopsy may be omitted in selected patients.
What Is New
- This is one of the first multicentre studies to validate the new European Society for Pediatric Gastroenterology, Hepatology, and Nutrition (ESPGHAN) diagnostic guidelines for coeliac disease.
- We included a large number of patients diagnosed as having coeliac disease based on the 1990 criteria.
- According to the 2012 ESPGHAN guidelines, small bowel biopsy may have been omitted in 52% of the symptomatic individuals with serologic and HLA available data.
- The new ESPGHAN guidelines can be safely used in clinical practice.
The strict criteria for diagnosis of coeliac disease (CD) in children were established by the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition (ESPGHAN) in 1969 (1).
In 1990, these criteria were modified so that the gluten challenge (GC) was mandatory only for infants <2 years at first biopsy to exclude other causes of enteropathy and whenever the initial diagnosis was uncertain (2). Also for the first time, elevated anti-gliadin antibodies at disease onset, followed by antibody vanishing after gluten withdrawal, was considered to support the diagnosis (3,4). Further development of anti-endomysial antibodies (EMA), however, in the late 1980s (5,6), followed by tissue transglutaminase antibodies (TG2A) in the 1990s (7), and more recently deamidated gliadin peptide antibodies (DGPA), represented a true revolution because they all displayed sensitivities and specificities for the diagnosis of CD >95% (8–10). Moreover, a correlation between duodenal histopathology and TG2A levels in paediatric patients with CD has been reported, higher levels being associated with more severe villous atrophy (11–13).
Thus, in the last few years, critical voices have repetitively proposed that, in view of their high diagnostic accuracy, CD-specific antibody testing might replace the small bowel biopsy (SBB) in the diagnostic procedure for CD, at least for patients with high TG2A (14,15). A recent publication, even proposes an SBB would be required only for patients with discordant antibody results (15).
In this scenario, an ESPGHAN working group on the diagnosis of CD presented a detailed evidence-based report on antibody testing in CD (10), together with new guidelines for the diagnosis of CD in children and adolescents (16). In addition, the working group developed a new and broader definition of CD as a systemic disorder accordingly, diagnosis should not rely on 1 single parameter, that is, histology, but on a combination of clinical symptoms, CD-specific autoantibodies, histology, and genetics (16). In summary, the new guidelines state that SBB may not be mandatory for the diagnosis of CD in symptomatic patients carrying the HLA DQ2 and/or DQ8 haplotypes, and with TG2A >10 times the upper limit of normal (ULN; >10 times), and positive EMA. In addition, they establish that GC is no longer obligatory in children who underwent an SBB before the age of 2 years but only in unclear cases.
In view of the extreme repercussions of these modifications, the working group considered that a period of implementation and testing is mandatory before these guidelines definitively replace the previous ESPGHAN 1990 guidelines. Notwithstanding, few studies have been published to date assessing the efficiency of these guidelines (17,18). The aim of our work was to evaluate the new diagnostic criteria for the diagnosis of CD independently of histology in a large retrospective multicentre study in Spain.
Members of the Spanish Society of Paediatric Gastroenterology, Hepatology and Nutrition were asked via the Internet (internal e-mail distribution list) to participate in a retrospective multicentre study. The paediatric gastroenterology units (PGUs) that joined the study were asked to include all consecutive paediatric patients—age range 0.6 to 15.9 years—who, because of a suspected diagnosis of CD, based on clinical and/or serological data, had undergone an SBB between January 1, 2000, and June 30, 2005, for infants <2 at enrollment, and between January 1, 2000, and June 30, 2009, for older children. The established time frame aimed at allowing a complete diagnostic procedure to be concluded, including response to GC when mandatory.
For patients to be included, a minimum 2-year follow-up after diagnosis was required plus having a definite diagnosis as established by the reporting centre; diagnosis of CD was based on the 1990 ESPGHAN criteria (2), and noncompliant patients were discarded.
Clinical, serological, histological, and genetic data were recorded in an anonymous case report form (CRF, https://links.lww.com/MPG/A492). The same data were collected from all of the children who had undergone a GC.
Children diagnosed as having CD were divided into 3 groups based on their presenting signs and symptoms: malabsorption syndrome (MS) with classical symptoms defined as chronic diarrhoea and/or weight loss and/or abdominal distension; other gastrointestinal (GI) symptoms; nonclassical symptoms with isolated signs or symptoms such as iron deficiency (low ferritin level), short stature, dental enamel defects, recurrent oral aphthous ulcers, increased level of liver enzymes, dermatitis herpetiformis, and so on; asymptomatic children with an increased risk of CD: type 1 diabetes mellitus (T1DM), first-degree relatives with CD, or others.
Histological lesions were graded according to the Marsh or Marsh-Oberhuber classification (19). Whenever another classification was reported, repeat grading was performed by the coordinators according to the histological findings described by the local pathologists. Those SBBs that were not properly orientated, as reported by the investigator, were discarded and corresponding patients excluded from further evaluation.
Several methods for serology were employed depending on the centre and on the time period. Information on the corresponding method together with the ULN values applied by the local laboratory were included in the CRFs.
For HLA, depending on the technical means available in each centre, children were tested either for DQ2/DQ8 by serology or for DRB1/DQA1/DQB1 haplotype by polymerase chain reaction.
The study coordinators excluded those patients in whom no definite diagnosis was stated or who did not fulfill the 1990 ESPGHAN criteria (2) for the diagnosis of CD. Accordingly, those patients were not challenged, although the first SBB was performed before the age of 2, or older children with doubtful initial diagnosis in whom no confirmatory challenge had been performed were not included.
Finally, because our aim was to evaluate the new 2012 ESPGHAN criteria, patients with TGA2 plus EMA and HLA data available were identified and analysed separately. For practical purposes in a previous study of our group (17), a triple test (TT) was defined as positive, if TGA2 was ≥10 times the ULN, EMA was positive, and HLA was DQ2 and/or DQ8. TT was considered to be negative whenever any of these 3 conditions was not fulfilled.
When assessing a response to GC, relapse was defined as clinical plus serological response and/or histological relapse.
In addition, we also applied in our population the score system published together with the 2012 ESPGHAN guidelines (16) so as to assess its utility for the diagnosis of CD.
The scoring evaluates symptoms, antibodies, HLA, and biopsy findings, each contributing once (positive or negative result or not performed). To make a diagnosis of CD, a sum of 4 points is required (with punctuation from −1 to 2 points).
All of the CRFs were reviewed by the study coordinators to clear out inconsistencies. Data were included in a database for the statistical analyses. The study protocol was approved by the ethics committee of the Hospital Universitari i Politècnic La Fe (Valencia, Spain).
Demographic and clinical characteristics were summarized with frequencies and percentages, means and standard deviations, or medians and ranges. Sensitivity and specificity were estimated using cross-tabulation tables for the serological markers of our sample and also for the TT plus symptoms in those patients with available data. Ninety-five percent confidence intervals were computed for all estimations.
Thirty-three PGUs from 12 Spanish autonomous regions, covering all the different geographical areas of Spain, participated. Of the 33 PGUs, 16 included >60 patients per centre (Fig. 1).
Initially, 2647 patients were reported, but 470 were excluded: 110 because no definite diagnosis was stated and 360 because the 1990 ESPGHAN criteria (2) for the diagnosis of CD were not fulfilled. Thus, 2177 children were considered for evaluation: 2126 were patients with CD and 51 were patients without CD. The clinical presentation pattern is shown in Figure 2.
Focusing on histology (Table 1), CD was diagnosed in 69 patients with Marsh 2 (M2) lesion and in 2054 patients with Marsh 3 (M3) lesion. In 3 patients <2 years of age at the time of the first SBB, 1 Marsh 0 (M0), 1 Marsh 1 (M1), and 1 with a nonvalid biopsy sample, CD was confirmed by histological relapse after GC.
Of 1467 patients with CD reporting HLA data, 28 were DQ2/DQ8 negative, that is, 1.9%. Twenty-six patients of the 28 were diagnosed with age >2 years; 13 had MS at presentation, and 26 had a M3 lesion; the remaining 2 had M2. All but 1 were TG2A positive, in 18 TG2A being ≥10 times the ULN. In 10 patients, EMA was performed, and in all of them the result was positive including the only patient negative for TG2A.
Evaluation of the New ESPGHAN 2012 Diagnostic Criteria for CD for Omitting the SBB
Data for TG2A, EMA, and HLA were reported in 751 patients; thus, these were eligible to evaluate the ESPGHAN 2012 diagnostic guidelines for CD, 640 (86%) were symptomatic, 50% of them presenting an MS; 111 (14%) were asymptomatic, 62 patients having been screened because of a first-degree relative with CD and 14 because they had T1DM. Of these 751 patients, 731 had CD, and 20 were patients without CD.
TG2A levels >10 times the ULN, together with positive EMA plus the HLA DQ2 and/or DQ8 haplotypes, that is, a positive TT, were found in 336 symptomatic and 69 asymptomatic patients as shown in Figure 3.
As it turned out that all of the symptomatic children with positive TT had a diagnosis of CD according to the new diagnostic guidelines, an SBB could have been avoided in all of them (336 patients), and no false diagnosis of CD would have ensued. These represent 52% of the 640 symptomatic children with serology and HLA data. In the asymptomatic subgroup, 65 of 69 patients had confirmed CD, and in 4 patients CD was excluded.
A total of 346 of 751 patients, 330 patients with CD and 16 patients without CD, were negative for the TT (Fig. 3). Nine of the 330 patients with CD not fulfilling the criteria for a positive TT were DQ2/DQ8 negative, all with EMA positive and 8/9 with TG2A positive (in 2 of these, TG2A was ≥10 times the ULN). Sixteen of 330 patients were negative only for TG2A, 12 were negative only for EMA, and 8 were negative for both serological markers. Four additional patients, all with a diagnosis of CD, had immunoglobulin A (IgA) deficiency.
Considering only the IgA-sufficient patients with CD, 24 were negative for TG2A and 20 for EMA, thus resulting in sensitivity for the diagnosis of CD of 96, 6% and 97, 2%, respectively. The 24 IgA TG2A negative children were >2 at the time of the first SBB.
In the group of patients without CD, 4 of 20 (20%) had TG2A ≥10 times the ULN and positive EMA, at a 1/40 or 1/80 dilution. They carried the DQ2 haplotype, with histological lesion as M1, but were asymptomatic individuals pertaining to a screening study. Several biopsy samples were obtained by endoscopy in each patient, and TG2A became negative on follow-up.
When estimating sensitivity and specificity of the TT plus symptoms (data obligatory to permit omitting the SBB), we obtain a sensitivity of 50% (95% confidence interval 47–54) and a specificity of 100% (95% confidence interval 76–100).
Evaluation of the New ESPGHAN 2012 CD Guidelines for GC
In our population, GC was carried out in 152 children, of whom 74 were <2 years at the time of the first biopsy. Investigators reported initially a total of 440 patients diagnosed before the age of 2 years, but only 74 had undergone a GC as recommended in the 1990 ESPGHAN criteria; thus, accordingly, only these were included in the final study group. Of these 74 children, all but 8 relapsed (Table 2): in 7 children, no serological data were available at diagnosis; histological findings were not conclusive; HLA was not DQ2/DQ8. In 1 patient, GC is still ongoing and, after 2 years of gluten intake, neither clinical nor serological relapse has been ascertained. Because parents are reluctant to accept a postchallenge control biopsy, no histological data of the mucosa are available. At the time of the first SBB, this HLA DQ2-positive patient had GI symptoms, TG2A ≥10 times the ULN and M3b lesion. So although no EMA was performed, according to the new ESPGHAN criteria, GC would not have been mandatory. Consequently, avoidance of GC would have implied a risk of overdiagnosis of 1.3% (1/74).
A GC was also performed in 78 children >2 years of age at the time of the first SBB, 36 were ages 2 to 3 years, and 42 were >3 years. Four of them did not relapse; 3 were ages 2 to 2.5 years at the time of the first biopsy, and the fourth was 6.65 years old. In all of them, TG2A antibodies at initial evaluation were negative, and, accordingly, GC was mandatory to confirm the diagnosis (Table 2).
Application of a CD Diagnosis Score System
When applying the diagnostic score proposed in the 2012 ESPGHAN guidelines, 22 patients without CD (22/51 = 43%) obtained a score of ≥4, and thus in these 22 patients a diagnosis of CD would have been considered. In the group of patients with CD, 77 of 2126 (3.6%) patients obtained a score <4; thus, with this scoring system, a diagnosis of CD would not have been achieved.
By selecting only the 751 patients with full data, that is, reporting HLA plus serology results (Table 3), the score results differ: 7 of 20 patients with CD obtained a score of ≥4; thus, the risk of overdiagnosis would be 30% as compared with 43% in the full series of patients. In the group of patients with CD, only 10 of 731 would score <4, which means that only 1.3% of patients would not have been detected (Table 3).
This is one of the first large studies to support the new 2012 ESPGHAN criteria for diagnosis of CD in children and adolescents, which determines that, in certain circumstances, the diagnosis CD can be safely established without performing SBB and thus with no histology evaluation. Accordingly, SBB can be omitted in HLA DQ2/DQ8 symptomatic patients with TG2A antibody levels >10 times the ULN and positive EMA (16). The joint British Society for Pediatric Gastroenterology, Hepatology, and Nutrition and Coeliac UK guidelines for the diagnosis and management of CD in children, which were recently published, also support this attitude (20).
To comply with the inclusion criteria, 470 patients were excluded. A total of 25% of them had no clear diagnosis, and most are still on follow-up in the corresponding PGUs; thus, to assign them either to the CD or to the non-CD group would have supposed a risk of bias. In the remaining patients, either M0 or M1 was reported at the first SBB, or GC was not performed.
On the contrary, only patients with a minimum follow-up of 2 years were included, and although patients with an indefinite diagnosis may be the most interesting ones, for instance in view of potential CD, they require an extended and specific evaluation, which is beyond the scope of the present study. A few patients in whom histologic evaluation was not reported because of bad orientation of the sample were also discarded. Although correct orientation and handling of biopsy samples is crucial for correct histological evaluation (21), we cannot be completely sure about the quality of biopsies of included cases because no revision by an independent observed was asked for. Our aim was indeed to assess whether the new criteria works well in real life and not only in ideal conditions or highly specialized centres.
Although the diagnosis of CD in HLA DQ2/DQ8 individuals may be a matter of concern, these represent only a 1.9% of the total reported patients with CD, which is in keeping with most publications (22). Also, we must consider the possibility of a false-negative result moreover because some results dated >14 years backwards when reliability of HLA testing in some settings could be questionable. The clinical picture, serology data, and histology, however, strongly support the diagnosis of CD in those patients.
TG2A are measured only in relative units, which are kit specific; so numerical values for high results show considerable variations. Also, the diverse TG2A tests use different calculation methods and different positive controls; thus, the results between them are not strictly comparable. Thus, because of this lack of standardization of TG2A methods, 10 times the ULN is a different value depending on the commercial kit. The majority of routinely used commercial kits, however, by using calibration curves, can provide numerical values that are proportional to antibody concentration and detect high antibody levels. Notwithstanding, this is a true shortcoming not only of our study but also of the new guidelines themselves and of all publications, which claim to adopt a definite cut off value of TG2A to omit the SBB.
By strictly applying the new ESPGHAN criteria to our population, SBB could have been avoided in a high proportion of symptomatic patients and moreover without any risk of overdiagnosis (100% specificity). We must keep in mind that the complete information on the TT, however, was available only in about one third of the patients, and this may again be a potential bias of our results. Still, only patients who completely fulfil each one of the conditions as established by the 2012 ESPGHAN diagnostic guidelines are candidates to be diagnosed without histological evaluation. Complying with the necessary items suggested by the 2012 guidelines to omit the SBB offers a high degree of security as in our study the obtained specificity is 100% and this in a large heterogeneous population. We must keep in mind that there are also a large number of patients with suspected CD, however, in whom SBB is still indispensable for the diagnosis of CD.
Indeed, as in asymptomatic patients, and regardless of serology, SBB is always mandatory, none of the children in whom CD had been excluded complied with the new criteria to skip the first SBB. In 4 asymptomatic patients, and despite TG2A being >10 times the ULN and positive EMA, a M1 lesion was found; so it was not possible to confirm CD. Because they pertained to the high-risk groups with CD, follow-up is mandatory considering that CD can develop later in life as shown in the article by Kurppa et al (18); in their study, 7 of 8 EMA-positive individuals with initially normal mucosa developed villous atrophy upon follow-up; 6 of them with symptoms worsened while maintaining a gluten containing diet.
A retrospective study from our group involving 150 children with suspected CD was the first to give support to the new CD diagnostic guidelines and concluded that in the selected children, diagnosis of CD could be safely established independently of histological findings (17).
In 2012, the utility of the new criteria was evaluated by applying the human red blood cell TG2A antibody test (RBC-TG2ab) to 3031 relatives of patients with CD (23). By using 5 times the ULN as cut off point, 99% of these were found to be also positive for EMA, and 94% displayed histological lesions compatible with CD. The authors consider that these results reinforce the view that a positive predictive value of a biopsy is far <100% and that positivity for EMA is a strong predictor of subsequent CD; they conclude that these outcomes support the new ESPGHAN criteria and that these criteria could be extended to the adult population.
According to the new guidelines, GC is no longer obligatory in all patients when a SBB has been performed before the age of 2 years but is now restricted to unclear cases. In our study, GC was not routinely performed in subjects <2 years at the time of the first SBB, even before the 2012 new guidelines came into force. The fact that only 16% of initially reported cases of patients <2 years had a GC is noteworthy and is in keeping with the results of a survey carried out among ESPGHAN members from different European countries about their opinion on the 1990 ESPGHAN criteria (24). The majority claimed for a change in the challenge policy and acknowledged that they were no longer compliant with the 1990 criteria on the challenge rules (24). In keeping with our inclusion criteria, only those patients who strictly complied with the 1990 ESPGHAN guidelines, however, were included in the study and accepted for evaluation. Seven of the 8 patients who did not relapse had no conclusive data at the time of the first SBB. Therefore, in all of them, diagnosis of CD was doubtful, and in accordance with the new criteria, GC was mandatory. Yet in 1 remaining patient, applying the new ESPGHAN guidelines, a firm diagnosis of CD at the time of the first SBB would have been established; thus, GC was not mandatory. Truly, histological relapse cannot be discarded because the parents did not authorise a control biopsy, and also serological and/or clinical relapse can develop later; however, we must admit a potential minimum risk of overdiagnosis in our population (1.3%). Although our data support that GC can be restricted to selected patients independently of the age at the time of the first SBB, an evaluation in a larger study population and with a longer follow-up is needed to reach definitive conclusions.
Application of the proposed CD diagnostic score (16) would have meant missing diagnosis in a small proportion of patients with CD. Yet more relevant, because of a high proportion of overdiagnosis, the score would show a low specificity. These results, however, have to be taken with caution because this is a retrospective study, and applying this score prospectively could display a higher efficacy. Moreover, for a better specificity performance, our results seem to recommend its application only when data from all of the included parameters are available.
According to the results of this retrospective study, the new ESPGHAN 2012 criteria for the diagnosis of CD can be safely used with no risk of overdiagnosis, provided they are strictly applied. Confirmation by a prospective study, however, would reinforce this conclusion.
1. Meeuwisse GW. Diagnostic criteria in coeliac disease. Acta Paediatr Scand
2. Walker-Smith JA, Guandalini S, Schmitz J, et al. Revised criteria for diagnosis of coeliac disease. Arch Dis Child
3. Ribes-Koninckx C, Giliams JP, Polanco I, et al. IgA antigliadin antibodies in celiac and inflammatory bowel disease. J Pediatr Gastroenterol Nutr
4. Troncone R, Ferguson A. Anti-gliadin antibodies. J Pediatr Gastroenterol Nutr
5. Grodzinsky E, Jansson G, Skogh T, et al. Anti-endomysium and anti-gliadin antibodies as serological markers for coeliac disease in childhood: a clinical study to develop a practical routine. Acta Paediatr
6. Korponay-Szabo I, Kovacs J, Czinner A, et al. High prevalence of silent celiac disease in preschool children screened with IgA/IgG antiendomysium antibodies. J Pediatr Gastroenterol Nutr
7. Troncone R, Maurano F, Rossi M, et al. IgA antibodies to tissue transglutaminase: an effective diagnostic test for celiac disease. J Pediatr
8. Prause C, Ritter M, Probst C, et al. Antibodies against deamidated gliadin as new and accurate biomarkers of childhood coeliac disease. J Pediatr Gastroenterol Nutr
9. Hill ID, Dirks MH, Liptak GS, et al. Guideline for the diagnosis and treatment of celiac disease in children: recommendations of the North American Society for Paediatric Gastroenterology, Hepatology and Nutrition. J Pediatr Gastroenterol Nutr
10. Giersiepen K, Lelgemann M, Stuhldreher N, et al. ESPGHAN Working Group on Coeliac Disease Diagnosis. Accuracy of diagnostic antibody tests for coeliac disease in children: summary of an evidence report. J Pediatr Gastroenterol Nutr
11. Barker CC, Mitton C, Jevon G, et al. Can tissue transglutaminase antibody titers replace small-bowel biopsy to diagnose celiac disease in select paediatric populations? Paediatrics
12. Donaldson MR, Firth SD, Wimpee H, et al. Correlation of duodenal histology with tissue transglutaminase and endomysial antibody levels in paediatric celiac disease. Clin Gastroenterol Hepatol
13. Dahlbom I, Korponay-Szabó IR, Kovács JB, et al. Prediction of clinical and mucosal severity of coeliac disease and dermatitis herpetiformis by quantification of IgA/IgG serum antibodies to tissue transglutaminase. J Pediatr Gastroenterol Nutr
14. Mubarak A, Wolters VM, Gmelig-Meyling FH, et al. Tissue transglutaminase levels above 100 U/mL and celiac disease: a prospective study. World J Gastroenterol
15. Bürgin-Wolff A, Mauro B, Faruk H. Intestinal biopsy is not always required to diagnose celiac disease: a retrospective analysis of combined antibody tests. BMC Gastroenterol
16. Husby S, Koletzko S, Korponay-Szabo I, et al. ESPGHAN guidelines for the diagnosis for coeliac disease in children and adolescents. An evidence-based approach. J Pediatr Gastroenterol Nutr
17. Klapp G, Masip M, Bolonio M, et al. Coeliac disease: the new proposed ESPGHAN diagnostic criteria do work well in a selected population. J Pediatr Gastroenterol Nutr
18. Kurppa K, Ashorn M, Iltanen S, et al. Celiac disease without villous atrophy in children: a prospective study. J Pediatr
19. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol
20. Murch S, Jenkins H, Auth M, et al. Joint BSPGHAN and Coeliac UK guidelines for the diagnosis and management of coeliac disease in children. Arch Dis Child
21. Taavela J, Koskinen O, Huhtala H, et al. Validation of morphometric analyses of small-intestinal biopsy readouts in celiac disease. PLoS One
22. Donat E, Planelles D, Capilla-Villanueva A, et al. Allelic distribution and the effect of haplotype combination for HLA type II loci in the celiac disease population of the Valencian community (Spain). Tissue Antigens
23. Kurppa K, Salminiemi J, Ukkola A, et al. Utility of the new ESPGHAN criteria for the diagnosis of celiac disease in at-risk groups. J Pediatr Gastroenterol Nutr
24. Ribes-Koninckx C, Mearin ML, Korponay-Szabó IR, et al. ESPGHAN Working Group on Coeliac Disease Diagnosis. Coeliac disease diagnosis: ESPGHAN 1990 criteria or need for a change? Results of a questionnaire. J Pediatr Gastroenterol Nutr