Is Gluten Challenge Really Necessary for the Diagnosis of Coeliac Disease in Children Younger Than Age 2 Years?

Abstract

Objective: 

In the diagnosis of coeliac disease (CD), gluten challenge is recommended for children under the age of 2 years at initial biopsy. The aim of the study was to investigate the diagnostic yield of gluten challenge in this group of children.

Patients and Methods: 

We included children aged 2 years or younger who were analysed for possible CD and who had villous atrophy at initial small bowel biopsy in the period 1993–2004. We subsequently identified all patients who underwent a complete gluten challenge.

Results: 

We identified 333 children with possible CD. In 100 children (30%), a gluten challenge was performed, with the diagnosis being confirmed in 97. Retrospectively, in 2 of the 3 children without mucosal relapse, data available before gluten challenge did not justify the initial diagnosis of CD. In the third patient, transient gluten intolerance could not be excluded. At first biopsy, the 2 children without mucosal relapse had negative serological parameters, whereas the third patient had IgA anti-gliadin antibodies, but no IgA anti-endomysium antibodies (EMA). Indeed, all of the patients with EMA at diagnosis had a relapse at gluten challenge.

Conclusions: 

Routine gluten challenge in children younger than 2 years at initial diagnosis of CD has an extremely low diagnostic yield. We suggest that routine gluten challenge in this group of patients is not necessary when patients have villous atrophy in combination with EMA. Therefore, a revision of the current diagnostic criteria has to be considered.

Diagnostic criteria for coeliac disease (CD) were established for the first time in 1969 by the European Society for Paediatric Gastroenterology, Hepatology, and Nutrition (ESPGHAN) at the Interlaken symposium ^(1,2). At that time, CD was defined as a permanent disorder that required 3 criteria: an abnormal jejunal mucosa on a gluten-containing diet, clear improvement of villous structure when taking a gluten-free diet, and deterioration of the mucosa during gluten challenge. In 1990, these criteria were revised because the need for routine gluten challenge had been questioned ⁽³⁾. The revision was partially based on a large study by Guandalini et al ⁽⁴⁾, who studied 2523 patients diagnosed by strict adherence to the ESPGHAN protocol. In a small percentage of patients (∼5%, 123/2523 patients) in this study, following gluten challenge, another diagnosis was made such as cow's milk–sensitive enteropathy and transient gluten intolerance. The mean age of these nonconfirmed coeliac patients was 8 months. Obviously, other causes of enteropathy can occur in this young age group, so the revised criteria recommended gluten challenge only for children aged 2 years or younger at diagnosis. Gluten challenge was not mandatory at diagnosis either in children aged 2 years or older or adults.

Re-exposing a child to gluten for a considerable period has its disadvantages. It can cause symptoms such as diarrhoea, abdominal pain, and irritability. It can also be potentially harmful to the child's growing potential ⁽⁵⁾. Furthermore, small bowel biopsy causes considerable distress, whereas general anaesthesia or sedation during biopsy has a potential risk of side effects and complications. In addition, the actual yield of gluten challenge in patients younger than 2 years of age at initial biopsy seems to be low, as even in this age group only a few diagnoses have to be revised ⁽⁴⁾.

Finally, since the publication of the ESPGHAN recommendations in 1990, the reliability of serological tests for CD has improved considerably. In a child with unequivocal histology and abnormal serology, at present many paediatricians will not use routine gluten challenge to confirm the diagnosis in young children. Because no evidence is available to support this change in attitude, we investigated the results of gluten challenge in children aged 2 years or younger at the initial diagnosis of CD to determine whether gluten challenge could also be abandoned in this group of children.

MATERIALS AND METHODS

We retrospectively investigated data of children aged 2 years or younger at initial biopsy who were analysed for possible CD between 1993 and 2004 in 4 Dutch children's hospitals (Departments of Paediatric Gastroenterology of the University Medical Centre, Utrecht; Academic Medical Centre, Amsterdam; VU Medical Centre, Amsterdam; and Juliana Children's Hospital/Haga Teaching Hospital, The Hague, the Netherlands) and who had villous atrophy at initial small bowel biopsy. We identified all of the patients who completed a gluten challenge before October 1, 2006. The methods and results of these gluten challenges were evaluated.

A second small intestinal biopsy was taken before gluten challenge. Regarding the actual challenge, 3 centres (University Medical Centre, VU Medical Centre, and Juliana Children's Hospital) adhered to the same gluten challenge protocol, using initially an intake of at least 175 mg/kg/day of gluten powder, in addition to an otherwise gluten-free diet, with at least 375 mg/kg/day of gluten powder after 2 weeks, with a maximum of 20 g of gluten daily ^(6,7). Dietary instructions were given by a specialised nurse or dietician. In one centre (Academic Medical Centre), no gluten powder was given and patients started with a regular gluten-containing diet. After at least 3 months of gluten challenge, a third small intestinal biopsy was taken, as mucosal relapse is to be expected within 3 months in most patients ⁽⁸⁾. In children with clear clinical symptoms, this final biopsy was done earlier. CD was considered to be proven if the mucosa showed villous atrophy (Marsh IIIa or higher) ⁽³⁾.

In the 3 patients without mucosal relapse, all small bowel biopsies taken were revised by pathologists with experience in the diagnosis of CD (F.J.T.K., P.G.J.N.). In all, in 100 patients who underwent a gluten challenge and in the 233 who did not, the results of serological tests at initial diagnosis were retrieved from the medical file, if possible. Furthermore, we also investigated the number of pathologists involved in the assessment of the biopsies obtained from the gluten-challenged patients.

RESULTS

We studied the data of 333 children who were investigated for CD before the age of 2 years and had villous atrophy at initial small bowel biopsy (174 patients in the University Medical Centre, Utrecht; 89 in the Academic Medical Centre, Amsterdam; 38 in the Juliana Children's Hospital/Haga Teaching Hospital, The Hague; and 32 in the VU Medical Centre, Amsterdam, the Netherlands).

Out of 333 children, a full gluten challenge according to the ESPGHAN recommendations was performed in 100 children (38 boys, 62 girls) ⁽³⁾. Clinical presentation and serological and histological results were similar in both the challenged and nonchallenged groups (Table 1). In both groups, the vast majority of patients presented with classical symptoms of malabsorption such as failure to thrive since gluten introduction and diarrhoea (challenged vs nonchallenged: 81% and 87%, respectively; NS). Only a minority of patients had atypical symptoms such as failure to thrive since birth or severe anaemia (challenged vs nonchallenged: 5% and 4%, respectively), whereas presentation could not be retrieved in 14% of challenged versus 9% of nonchallenged patients. Serological results could also be retrieved for the majority of patients (challenged vs nonchallenged: 89% and 78%, respectively) and were compatible with CD in almost all patients both in the challenged and nonchallenged groups. Results for histological evaluation were also similar in both groups. Microscopy was typical for CD in 98% and 97%, respectively. Histology was inconclusive in the remaining 2% to 3% of patients. In these 7 patients villous atrophy was found, but combined with a substantial increased amount of neutrophiles in the lamina propria or no increase in the number of intraepithelial lymphocytes.

TABLE 1: Clinical presentation, serological and histological results in 2 groups (gluten challenge vs no gluten challenge)

A typical clinical presentation, combined with abnormal serology and typical histology was found in 69 (69%) of challenged patients and 154 (66%) of nonchallenged patients. Either an atypical presentation, normal serology or inconclusive histology was present in 12% of challenged and 11% of nonchallenged patients. For 19 (19%) patients in the challenged group either the clinical presentation or the serology could not be retrieved. Similarly for 53 patients (23%) in the nonchallenged group one of these parameters was not available.

During the study period, only 100 of the 233 patients had a gluten challenge, with most challenges being performed in the University Medical Centre in Utrecht (68/174; 40%) and the Academic Medical Centre in Amsterdam (23/89; 26%), these centres indicating that they tried to comply with the ESPGHAN recommendations. The other 2 centres (Juliana Children's Hospital in The Hague and the VU Medical Centre in Amsterdam) expressed a more liberal view, which was reflected in the lower number of challenges at these centres, respectively, 6 out of 38 (16%) and 3 out of 32 (10%). The difference in the challenge rate between these 2 groups of centres is significant (P < 0.001).

The reason to refrain from performing gluten challenge was not stated in the files of 169 of 233 patients (73%). In 34 (14%) patients, either the parents or the paediatrician did not want challenge with gluten or considered this unnecessary. Furthermore, in 26 patients (11%) a gluten challenge was started but not completed mostly because symptoms recurred, whereas in 4 patients the gluten challenge was started but patients were not included in this study because the challenge was completed after October 1, 2006 (n = 2) or the biopsies were taken in other hospitals (n = 2).

The mean duration of gluten challenge was 4.2 months (range 0.7–26.7 months). The mean age at gluten challenge was 3.7 years (range 2.0–8.7 months) and the mean age at initial biopsy was 1.4 years (range 0.8–2.0 months). At the second small intestinal biopsy, performed before gluten challenge, a completely normal intestinal mucosa (Marsh 0) was found in 85 patients. In 15 patients, mucosal histology was more or less abnormal with increased intraepithelial lymphocytes (Marsh I) in 8 patients, crypt hyperplasia (Marsh II) in 3 patients, and villous atrophy (Marsh III) in 4 patients. After gluten challenge, villous atrophy (Marsh III) was found in 97 patients (97%), with the histology deteriorating in the 4 patients with villous atrophy before the challenge had started. In only 3 children (3%) the small bowel biopsy showed no abnormalities before and after, respectively, 3, 9, and 15 months of gluten challenge (Table 2). These 3 children did not have any symptoms during gluten challenge. In 2 of the 3 children, serology was negative at the time of diagnosis and before the introduction of a gluten-free diet. In 1 patient IgA anti-gliadin antibodies (IgA-AGA) were positive (24 U/mL, normal value <4 U/mL), whereas IgG anti-gliadin antibodies (IgG-AGA) and IgA anti-endomysium antibodies (EMA) were negative.

TABLE 2: Characteristics of the 3 patients with villous atrophy at initial biopsy but who did not relapse after gluten challenge

All serology results of 97 challenged patients are shown in Table 3. The 3 patients, who did not relapse, were not included in this table. In 86 of the 97 children who relapsed after gluten challenge, serology results were available for analysis; in 2 patients no serology was performed and in 9 patients the results could not be retrieved. Abnormal serological results for all tests performed at initial diagnosis were seen in 77 of the 86 children. In 51 of 57 patients who relapsed and were tested for EMA, this test was positive (sensitivity 89%, 95% confidence interval [CI] 79–95%), whereas in 75 of 78 patients, IgA-AGA was positive (sensitivity 96%, [CI 89–99%]). In 7 children who relapsed (mean age at first diagnosis 17.4 months, range 14–21 months) IgA-AGA and/or IgG-AGA could be detected at the time of initial diagnosis, whereas EMA and/or IgA anti-reticuline antibodies (IgA-ARA) were negative. In only 1 child, 11 months old at initial diagnosis, all serological tests done (IgA-AGA and EMA) were negative, whereas total serum IgA was normal.

TABLE 3: Serology results at first biopsy in the 97 patients who relapsed after gluten challenge

During the total study period, in none of the 4 participating hospitals was a single pathologist dedicated to view the biopsies of all patients who had undergone a gluten challenge. In fact, it turned out that more than 20 pathologists were involved in the assessment of these biopsies.

DISCUSSION

We retrospectively evaluated the data of 333 children who were diagnosed with CD before the age of 2 years. This study shows that only 30% of these children had undergone a formal gluten challenge as proposed by the 1990 ESPGHAN guidelines. More important, the initial diagnosis of CD was rejected in only 3 of the 100 children in whom gluten challenge was performed. In these 3 nonrelapsers highly sensitive and specific serology (EMA, IgA-ARA) was negative at diagnosis, although 1 patient had positive IgA-AGA.

Only a few studies, in which a biopsy at initial diagnosis was performed in at least 90% of the patients, have investigated the percentage of children who did not relapse at gluten challenge. In these studies, the percentage of nonrelapsers varied between 3% and 7%, and these children were usually diagnosed with cow's milk–sensitive enteropathy or transient gluten intolerance ^(4,9–11). The largest study, by Guandalini et al ⁽⁴⁾, investigated 2523 children and found 5% of nonrelapsers with a mean age of 8 months (age range not published). This study initiated the 1990 ESPGHAN recommendation to only perform routine gluten challenge in children <2 years at diagnosis. Subsequently, Danielsson et al ⁽¹⁰⁾ performed a study similar to ours in 62 children younger than age 2 years. In this study 3 of 62 (4.5%) patients did not relapse. In contrast, O'Halloran et al ⁽¹²⁾ reported that 25% of 35 children who were diagnosed under the age of 2 years showed no mucosal relapse after at least 6 months of gluten-containing diet. However, the initial diagnosis of CD was unreliable in this study because not all patients had an initial biopsy, and in one third (13/40) of nonrelapsers of all ages the histological abnormalities at initial diagnosis were only mild. The authors even suggested that retrospectively in these 13 patients, a gluten-free diet may not have been necessary anyway. It should be noted that in none of these studies were serological investigations done.

In contrast, in our study, serology results were available for the vast majority of patients. All patients with a positive EMA and/or IgA anti-tissue transglutaminase antibodies (TTG) at initial diagnosis relapsed at gluten challenge. A further 6 patients with CD proven at gluten challenge had a normal or inconclusive EMA at the time of the initial biopsy. This results in a sensitivity of 89%, which is somewhat lower than published values that generally exceed 95%, but is in agreement with studies that indicate that EMA is less sensitive in children younger than 2 years of age ^(13,14). In this group, IgA-AGA and/or IgG-AGA are almost always abnormal, as was also found in our study. However, generally, the sensitivity of these 2 markers is considerably lower as compared with EMA and TTG ⁽¹⁵⁾. They may also be false-positive, as was indeed the case in our patient 3. In the other 2 non-relapsers all serology done was negative, consistent with the normal mucosa found after gluten challenge.

There are some limitations to the present study. First, our study had a retrospective design. Also mucosal relapse may take a long time and a second control biopsy after several years of gluten intake was not performed in the 3 patients who did not relapse ^(10,11,16). However, in 2 of these 3 nonrelapsers, the initial diagnosis of CD was undoubtedly incorrect and the third patient had no clinical symptoms during 9 years on a gluten-containing diet.

Furthermore, we tried to find out the reason why a large group of patients was not challenged, but the cause could only be identified in 27% of cases. Although we had expected that in the challenged group the diagnosis may be less solid, our data (Table 1) show that no differences existed in clinical, serological, and histological characteristics between the challenged group and the nonchallenged group. Therefore, it seems that the patients who were challenged are not a select group and that the results probably can be extrapolated to the entire group. In fact, at the onset of the study, physicians in 2 centres (University Medical Centre, Utrecht, and Academic Medical Centre, Amsterdam) indicated that they tried to comply with the ESPGHAN recommendations and persuade parents to have their children challenged. In these centres, the challenge rate was 40% and 26%, respectively. In the 2 other centres (Juliana Children's Hospital and VU Medical Centre) some doubt existed whether a challenge for children younger than 2 years of age was really necessary. Consequently, the challenge rate in these centres at 16% and 10% was significantly lower (P < 0.001).

A further limitation is the considerable number of pathologists who investigated the biopsies, which may have resulted in interobserver differences and reduced the personal experience for each pathologist. However, this phenomenon is probably common in daily practice. In fact, the mean number of cases handled by each pathologist in this study (<1 challenged patient per pathologist per year) is probably not that different from the situation in smaller centres. Therefore, our results may be more generally applicable.

Because all our patients with a typical presentation, abnormal serology, and villous atrophy at presentation relapsed when a challenge was done, we suggest that routine gluten challenge should no longer be mandatory in children younger than 2 years at diagnosis, if clinical symptoms are indeed compatible with CD and the characteristic histological changes (Marsh III ⁽¹⁷⁾) in the intestinal biopsy sample are found in combination with EMA and/or TTG. Symptoms should disappear and serology should normalise after initiation of the gluten-free diet. However, gluten challenge may still be useful in children with aspecific clinical symptoms, equivocal histological changes in the small bowel biopsy sample, conflicting serological test results, and incomplete disappearance of symptoms or no improvement of serology during gluten-free diet. Individual circumstances should determine the precise approach to a particular patient. A revision of the current diagnostic criteria established by ESPGHAN should be considered.

Acknowledgments

We thank Winny van der Hoeven, Janet Feenstra, Gideon Hajer, and Irene Wilson, WKZ/UMC Utrecht, and Joyce Goris-de Geus, IJsselmeer Hospital, Lelystad, and Wieke Verbeek, VUMC, Amsterdam, the Netherlands, for their help. We are grateful to the paediatricians who referred their patients to our hospital.