Background: In a developing country, many conditions other then celiac disease (CD) can give rise to villous atrophy. We therefore assessed the role of immunoglobulin A (IgA)–antigliadin antibody (AGA) in addition to the ESPGHAN criteria in the diagnosis of CD in 104 Indian children.
Methods: Consecutive children with suspected CD were evaluated over 3 years with an intention to diagnose CD. Complete hemogram, d-xylose absorption test, endoscopic duodenal biopsy, and IgA–AGA titers were performed in all. CD was diagnosed on the basis of modified ESPGHAN criteria irrespective of IgA–AGA positivity (>5 U/mL), and those diagnosed were put on gluten-free diet and were monitored regularly. Children with suspected CD who had a normal duodenal biopsy result were taken as controls.
Results: The mean age of 50 children with CD was 6.3 ± 2.6 years, with a male to female ratio of 3:2. The mean duration of symptoms was 3.4 ± 2.2 years. Predominant symptoms were pallor in 96%, failure to thrive in 92%, and diarrhea in 80%. On follow-up (19.6 ± 8 months), symptoms subsided within 16 ± 9.8 days, and patients showed significant weight gain (mean weight at diagnoses and at last follow-up visit were 66% and 86% of expected, respectively;P < 0.001) and height gain (mean height at diagnoses and at last follow-up visit were 88% and 94% of expected, respectively;P = nonsignificant). The control group comprised 47 children with a mean age of 6.9 ± 3 years. On comparing CD with controls, diarrhea, anemia, low weight, and stunting were significantly (P < 0.001) more frequent in patients with CD. Sensitivity and specificity of AGA at a cutoff value of 5 U/mL were 94% and 91.5% and at 10 U/mL 88% and 100%, respectively. Follow-up AGA test was performed in 42 of 47 positive cases. All showed significant decrease in AGA titer, and 29 (70%) had a negative test result.
Conclusions: Indian children with CD are true cases of CD. They present late, diarrhea is absent in 20% of cases, and AGA test results show 88% of children without false-positive results at a cutoff value of 10 U/mL. However, AGA test with 94% sensitivity at a cutoff value of 5 U/mL can be used as screening test to select suspected cases for further workup.
Division of Pediatric Gastroenterology, Department of Gastroenterology, Postgraduate Institute of Medical Education and Research, Chandigarh, India
Received May 21, 2001; accepted January 24, 2002.
Address correspondence and reprint requests to Dr. B. R. Thapa, Department of Gastroenterology, Postgraduate Institute of Medical Education and Research, Chandigarh-160012, India (e-mail: firstname.lastname@example.org).
This article is accompanied by an editorial. Please see Celiac Disease: An Emerging Global Problem. Emanuel Lebenthal and David Branski. J Pediatr Gastroenterol Nutr 2002;35:472–474.
Celiac disease (CD) was believed to be a disease in Europe and in Europeans living outside Europe. However, there are sporadic reports of CD from Arabia (1–3) and India (4–7). Celiac disease was reported from India in 1966 (4), and subsequently Nelson et al. (8) reported a series of 17 immigrant Asian children with CD from Birmingham. However, there has been doubt recently regarding the existing literature on CD from Asia (5) because diagnoses were based on villous atrophy only (1–4,6), and in developing countries many conditions other than CD can give rise to severe villous atrophy (7). In the modified ESPGHAN criteria (9), only biopsy changes and clinical response to gluten-free diet (GFD) are sufficient to make a diagnosis of CD, but in developing countries we should have some additional criteria to distinguish CD from other causes of villous atrophy. Criteria in addition to the modified ESPGHAN criteria could be either gluten challenge or antibody test. The former is cumbersome and requires the patient's and parents' cooperation, whereas antibody tests are simple. When antibody test results are positive at the time of diagnoses in a child with a typical small intestinal mucosa and when they disappear in parallel to a clinical response to GFD, weight is added to the diagnoses of CD that may now be said to have been finally established.
Ideally one should use endomysium antibody (EMA) as it is shown to be the most sensitive and specific autoantibody in CD (10). However, detection of EMA is cumbersome (indirect immunofluorescence technique), observer dependent, and costly. It has recently been shown that tissue transglutaminase is the main endomysial autoantigen in CD, and CD serum immunoglobulin A (IgA) reacts with tissue transglutaminase (11). However, this test is not yet widely available in developing countries. On the other hand, the antigliadin antibody (AGA) test is performed by enzyme-linked immunosorbent assay and is therefore inexpensive and easy to perform, and a quantitative result of this test helps to compare serial measurements. Therefore, in a developing country, where cost is a major factor, we should look for a more inexpensive but effective test. AGA assay is ideally suited for a developing country.
In 1989, Khoshoo et al. (7) from India showed that the AGA test was useful in distinguishing CD from other causes of villous atrophy. However, in that study the number of CD cases was small (n = 11), and a cutoff value of AGA, which could be used to screen children for further workup, was not given. There is no other study about serology in CD from Asia. We prospectively studied the role of IgA–AGA in the diagnosis of CD in North Indian children.
MATERIALS AND METHODS
From September 1997 to December 1998, 104 children were evaluated with a suspected diagnosis of CD in our Pediatric Gastroenterology unit. Children who presented with chronic diarrhea, failure to thrive, or pallor were investigated with an intention to diagnose CD. Complete hemogram, serum IgA-AGA assay, endoscopic duodenal biopsy, d-xylose test, and fecal fat (72 hours) estimation were performed. Clinical details were recorded in a proforma. Failure to thrive was defined as weight and/or height less than the 5th percentile for age. Children with villous atrophy were put on a GFD irrespective of AGA status and underwent follow-up regularly. Children underwent follow-up every 2 to 3 months. During follow-up visits, anthropometric parameters, clinical response, and compliance to GFD were recorded. CD was diagnosed on the basis of modified ESPGHAN criteria (8), i.e., the presence of villous atrophy with unequivocal clinical response to GFD. Children with CD underwent follow-up until October 2000. Children who did not show good response to GFD within 6 weeks were excluded from the CD group during analysis. Response to GFD was defined as subsidence of symptoms and weight gain within weeks of starting GFD. A repeat AGA test was conducted in CD cases at least after 3 months of starting GFD when possible. Children who were evaluated with a suspicion of CD but who were found to have normal mucosal biopsy results (i.e., normal villous architecture) were taken as controls. Sensitivity, specificity, and positive and negative predictive value of AGA (IgA) at various cutoff levels were calculated between cases and controls.
IgA–AGA assay was conducted by micro enzyme-linked immunosorbent assay technique as described previously (12,13) with a commercial kit provided by The Binding Site Ltd. (United Kingdom) with both positive and negative control. Values were expressed as units per milliliter, and a cutoff value of 5 U/mL was taken for defining positive and negative results.
Values are expressed as mean ± SD. The χ2 and paired and unpaired t tests were used for comparison. Significance was taken as P < 0.05.
Of the 104 children, 57 were initially diagnosed to have CD, and the remaining 47 were controls. Of the CD cases, seven did not show a response to GFD and were excluded. The mean (± SD) age of these 50 children with CD was 6.3 ± 2.6 years (range, 2.5 ± 12 years), and the male-to-female ratio was 3:2. The clinical presentations of these children are given in Table 1. Ten children (20%) did not have diarrhea at presentation; all had failure to thrive, six also had anemia, and three had constipation. The mean age of onset of symptoms was 3 ± 2 years, and the mean duration of symptoms was 3.4 ± 2.2 years (range, 1 month to 11 years), indicating a significant delay in referral.
Results of investigations are given in Table 1. The d-xylose test was conducted in 47 cases and 44 controls, and the 72-hour fecal fat estimation was conducted in 35 cases and 20 controls. Anemia was defined according to World Health Organization criteria (14) as follows: hemoglobin less than 11 g/dL in children younger than 6 years and less than 12 g/dL in those older than 6 years. The mean hemoglobin concentration in the CD group was 7.7 ± 2 g/dL. Hyposplenism, indicated by a high platelet count (≥6 × 105/μL) was seen in three children. The control group comprised 47 children. The mean age was 6.9 ± 3 years (range, 1.5–14 years), and the male-to-female ratio was 3:2. The diagnosis of control children were as follows: short stature in 24, nonspecific diarrhea in 14, Giardia in 5, anemia in 2, recurrent abdominal pain and cow's milk protein allergy in 1 each. A comparison between CD cases and controls are given in Table 1. Although there was no difference in the age at presentation, sex distribution, and duration of symptoms, there were significant differences in clinical presentations and malabsorption parameters. Children with CD presented more frequently with diarrhea, and they were more frequently stunted (height < 5th percentile) and underweight (weight < 5th percentile) than controls. Sensitivity, specificity and positive and negative predictive values of the AGA test are given in Table 2. AGA titers between cases and controls are shown in Figure 1. Duodenal biopsy results showed villous atrophy in all cases (subtotal in 41 and partial in 9) with elongated crypts. In addition, all cases had marked increase in intraepithelial lymphocytes and lymphomononuclear cell infiltrates in the lamina propria. All children in the control group had a normal crypt–villous ratio, but 19 (38%) of them had chronic inflammatory cell infiltrates in lamina propria (nonspecific inflammation).
The mean follow-up duration after starting GFD was 19.6 ± 8 months (range, 4–36 months). Seven of 57 children who were started on GFD did not show response within 6 weeks, five of them were IgA–AGA-negative, and two had AGA titer between 5 and 10. They were finally diagnosed as cow's milk protein allergy in two, protein energy malnutrition in two, and Crohn's disease, bacterial overgrowth, and Giardiases in one each. The remaining 50 children showed a dramatic response to GFD. Their symptoms subsided within a mean period of 16 ± 9.8 days (range, 4–30 days), and all showed significant weight gain (weight at diagnoses and at last follow-up visit were 66% ± 14% and 86% ± 11% of expected, respectively;P < 0.001) and height gain (height at diagnoses and at last follow-up visit were 88% ± 5% and 94% ± 5% of expected, respectively;P = nonsignificant). On follow-up, 17 (34%) had poor compliance to GFD at some point in time. Comparison of symptoms, weight, height, and AGA positivity at the time of diagnosis and on follow-up are shown in Figure 2.
The repeat AGA test could be conducted in 42 of 47 (89.4%) positive cases after a mean follow-up duration of 11.6 ± 6 months (range, 3–24 months). Results of repeat AGA test depending on the duration of GFD period is given in Table 3. There was a significant decrease in antibody titer (88 ± 104 to 9 ± 14.6 U/mL;P < 0.001) in all but two patients. Repeat AGA test in 29 patients (70%) had negative results at a cutoff value of 5 U/mL, and 32 (80%) had negative results at a cutoff value of 10 U/mL. There was no significant difference in the duration of follow-up (on GFD) before repeat AGA test between those who had negative versus those who had positive AGA test results (10.8 ± 5.4 vs. 12.7 ± 4.6 months).
In 1888, Gee (15) provided the first clinical description of CD. However, until the 1950s it was not known that dietary cereals were harmful to children with CD. In 1953, Dicke et al. (16) documented the relation between the ingested wheat and the degree of steatorrhea in children with CD. In India, there was doubt about the existence of this disease until 1966, when Walia et al. (4) first reported 10 children from Delhi with CD. The subsequent reports of CD in India are mainly from North India (6,7,17), where wheat is a staple food. In all of these series, diagnosis of CD was established on the basis of villous atrophy on intestinal mucosal biopsy. In a series of 121 children with protracted diarrhea from Delhi, Khoshoo et al. (7) showed that 32 had severe villous atrophy caused by enteropathogenic Escherichia coli (n = 8), Giardia (n = 8), cow's milk protein intolerance (n = 5), Salmonella (n = 4), bacterial overgrowth (n = 4), and transient gluten intolerance (n = 3). The median age of these children at the time of diagnoses was 21 months, while the mean age of children in the current study with villous atrophy at the time of diagnoses is 6.3 years. It is therefore unlikely that villous atrophy in our children would be because of these conditions, which are more common in those younger than 2 years (9). However, 7 (12%) of our cases who were initially diagnosed as CD because of villous atrophy did not respond to GFD, indicating that other conditions can give rise to villous atrophy. In these cases, the AGA test result was either negative or in low titer (<10 U/mL). Therefore, AGA at a cutoff value of 10 U/mL can easily distinguish CD from other causes of villous atrophy.
The only study of serology (AGA) from Asia was from Delhi by Khoshoo et al. (7), who had shown that the mean IgG and IgA–AGA titers were significantly higher in patients with CD than any other group (includes protracted diarrhea of non-celiac causes, acute gastroenteritis, protein energy malnutrition, and normal children). They also showed that villous atrophy as a result of non-celiac causes can be distinguished from CD on the basis of AGA assay. In our study, we have shown that the AGA test can be used for screening because of its high sensitivity (94%) at a cutoff value of 5 U/mL. If the IgA–AGA test result is positive, then there is a 92% chance of getting a villous atrophy on mucosal biopsy. In the presence of positive IgA–AGA result, villous atrophy indicates CD as shown by Khoshoo et al (7). In our series, 50 of 57 (88%) children with villous atrophy not only fulfilled modified ESPGHAN criteria (9), but also proved that in developing countries, addition of IgA–AGA positivity will make the diagnosis more definite. Moreover, in 70% of our cases who tested positive, AGA test results became negative while on a GFD, further substantiating the diagnosis of CD. Sensitivity and specificity of the IgA–AGA test in our series is better than what have been reported from our country (7) but similar to those reported from the West (18–21). In a strongly suspected case of CD who has villous atrophy, a negative IgA-AGA test result should raise the suspicion of selective IgA deficiency. There is a 10-fold increased risk of CD in IgA-deficiency cases (22). In such case, the IgG–AGA test is useful. Three of our cases who had negative AGA test results could have had underlying selective IgA deficiency. In the control group, there were four children who had AGA positivity with a normal villous architecture. They may be the cases of potential CD who need close follow-up (23). However, it has been shown that the likelihood of subsequent development of CD is much higher in those potential CD who are EMA-positive than those who are only AGA-positive (24). Therefore, we should be careful in labeling false-positive or false-negative serologic test results in a patient with suspected CD.
Apart from the initial AGA positivity, the subsequent decrease in titers and negativity in most of the cases on GFD proves beyond doubt that these are cases of CD. It has been shown that the IgA–AGA test is the most reliable serologic marker for monitoring the initial response to gluten withdrawal as it disappears faster (usually by the sixth month) than EMA. In a study by Cataldo et al. (25) in 688 children with CD, it was shown that at 6 months of GFD, 40% of patients were still EMA-positive, whereas only 12% were positive for IgA–AGA, and these figures at 12 months were 8% and 0.3%, respectively. In our study, 70% cases had negative IgA–AGA results on follow-up. In the remaining 30%, although there was a significant decrease in the titers, AGA test results were still positive, indicating poor dietary compliance. Moreover, persistent AGA positivity was not related to the duration of the GFD period, indicating that these were a result of poor compliance only. Similar results were recently reported by Chartrand et al (20), who also showed that in all 11 cases of CD, there was a significant decrease of AGA titers, but 3 of them were still positive after 10 months on a GFD.
The clinical picture of our children with CD is almost similar to other series reported from India (6). Twenty percent children with CD did not have diarrhea in our series. Therefore, we should suspect and investigate children with refractory anemia and failure to thrive even if there is no diarrhea. Our series highlights the late onset of the disease and delayed referral as compared with the West. Late onset of the disease may be a result of delayed weaning and the late introduction of gluten. Delayed presentation is mainly a result of the lack of awareness about the disease among parents and pediatricians.
In developing countries, facilities for performing intestinal biopsy are limited to a few centers only. Therefore, we need to have a simple, inexpensive, and effective test that will help to screen suspected cases of CD for further workup. In our study, although the d-xylose test was as sensitive as the AGA test, its specificity was low (46%). On the other hand, IgA–AGA at a cut off value of 5 U/mL was highly sensitive and can be used to select patients for further workup.
In conclusion, Indian children with CD are real cases of CD. They present late, diarrhea is absent in 20%, and the AGA test detects 88% of children without false positivity at a cutoff value of 10 U/mL. However, the AGA test with 94% sensitivity at a cutoff value of 5 U/mL can be used as screening test to select suspected cases for further workup. IgA–AGA assay at the time of diagnosis and on follow-up should be added to diagnose CD in developing countries.
1. Al-Hassany M. Celiac disease in Iraqui children. J Trop Pediatr 1975; 21:178–9.
2. Bitar JG, Salem AA, Nasr AT. Celiac disease from the middle east. J Med Liban 1970; 23:423–44.
3. Rawashdeh MO, Khalil B, Raweily E. Celiac disease in Arabs. J Pediatr Gastroenterol Nutr 1996; 23:415–8.
4. Walia BNS, Sidhu JK, Tandon BN, Ghai OP, Bhargava S. Celiac disease in North Indian children. Br Med J 1966; 2:1233–4.
5. Bhatnagar S, Bhan MK. Persistent diarrhea and malabsorption in children. Ind J Gastroenterol 1993; 12:111–5.
6. Yachha SK, Misra S, Malik A, Nagi B, Mehta S. Spectrum of malabsorption syndrome in North Indian children. Ind J Gastroenterol 1993; 12:120–5.
7. Khoshoo V, Bhan MK, Puri S, et al. Serum anti-gliadin antibody profile in childhood protracted diarrhea due to celiac disease and other causes in a developing country. Scand J Gastroenterol 1989; 24:1212–6.
8. Nelson R, McNeish AS, Anderson CM. Celiac disease in children of Asian immigrants. Lancet 1973; i:348–50.
9. Walker-Smith JA, Guandalini S, Schmitz J, Schmerling DH, Visakorpi JK. Revised criteria for diagnosis of celiac disease. Arch Dis Child 1990; 65:909–11.
10. Maki M, Collin P. Celiac disease. Lancet 1997; 349:1755–9.
11. Dieterich W, Ehnis T, Bauer M, et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997; 3:797–801.
12. Hed J, Lieden G, Ottoson E, et al. IgA antigliadin antibodies and jejunal mucosal lesions in healthy blood donors [letter]. Lancet 1986; II:215.
13. Lindh E, Ljunghall S, Larsson K, Lavo B. Screening for antibodies against gliadin in patients with osteoporosis. J Int Med 1992; 231:403–6.
14. WHO. Technical Rep Ser 1968, No. 405.
15. Gee SJ. On the celiac affection. St. Bartholomew's Hospital Reports 1888; 24:17.
16. Dicke WK, Weijers HA, Van DeKamer JH. Celiac disease II. Acta Paediatr (Upssala) 1953; 42:34.
17. Khoshoo V, Bhan MK, Jain R, et al. Celiac disease as cause of protracted diarrhea in Indian children. Lancet 1988; 1:126–7.
18. Ascher H, Hahn-Zoric M, Hanson LA, Kilander AF, Nilsson LA, Tlaskalova H. Value of serologic markers for clinical diagnosis and population studies of celiac disease. Scand J Gastroenterol 1996; 31:61–7.
19. Grodzinsky E, Jansson G, Skogh T, Stenhammar L, Falth-Magnusson K. Anti-endomysium and anti-gliadin antibodies as serological markers for celiac disease in childhood: A clinical study to develop a practical routine. Acta Pediatr 1995; 84:294–8.
20. Chartrand LJ, Agulnik J, Vanounou T, Russo PA, Baehler PA, Seidman EG. Effectiveness of antigliadin antibodies as a screening test for celiac disease in children. Can Med Assoc J 1997; 157:527–33.
21. Stahlberg MR, Savilahti E, Viander M. Antibodies to gliadin by ELISA as a screening test for childhood celiac disease. J Pediatr Gastroenterol Nutr 1986; 5:726–9.
22. Collin P, Maki M, Keyrilainen O, Hallstrom O, Reunalo T, Pasternack A. Selective IgA deficiency and celiac disease. Scand J Gastroenterol 1992; 27:367–71.
23. Troncone R, Greco L, Mayer M, et al. Latent and potential celiac disease. Acta Paediatr Suppl 1996; 412:10–4.
24. Murray JA. The widening spectrum of celiac disease. Am J Clin Nutr 1999; 69:354–65.
25. Cataldo F, Ventura A, Lazzari R, Balli F, Nassimbeni G, Marino V. Antiendomysium antibodies and celiac disease: Solved and unsolved questions. An Italian multicenter study. Acta Paediatr 1995; 84:1125–31.
Keywords:© 2002 Lippincott Williams & Wilkins, Inc.
Celiac disease; Antibody; Gliadin; Child