The prevalence of celiac disease (CD) in children and adults with type I diabetes mellitus (DM) according to studies in Europe, North America and Australia greatly exceeds the prevalence in the general population (1-14)
Most cases of CD associated with DM, however, do not present typical gastrointestinal symptoms for CD despite the presence of characteristic intestinal lesions, and these patients are therefore at risk of remaining undetected and untreated. Screening for CD can be performed by measurement of immunoglobulin (Ig)A and IgG antigliadin antibodies, IgA endomysial antibodies (EMA) and IgA tissue transglutaminase antibodies. EMA and tissue transglutaminase antibodies show high sensitivity and specificity in the various phases of CD and have proven to be effective in screening tests for atypical forms as well as for silent or latent CD (3,8,12,15). Untreated CD may be associated with a number of complications such as iron deficiency, anemia, growth retardation, osteoporosis, neuropsychiatric disorders, fertility problems and gastrointestinal malignancy (16-19). A lifelong gluten-free diet (GFD) is recommended to avoid those long-term risks.
In the past the potential benefits and risks of screening for CD and treatment with GFD have not been assessed in a critical and systematic way. A number of investigators (20-25) proposed that all children with diabetes mellitus should be screened for CD and, in cases having antibody positivity and typical lesions in the small bowel biopsy, be subsequently treated by GFD.
Few publications on the follow up of patients with diabetes and CD detected by screening exist with inconsistent outcomes (26-33).
The aim of our longitudinal study was to analyze clinical and metabolic characteristics of diabetic children with screening-detected CD compared with age-matched and age at onset matched controls with DM repeatedly negative for EMA antibodies in the same center from the time of diabetes manifestation.
Screening for CD by yearly measurement of EMA was performed in 10 pediatric diabetic centers in their clinic cohort of diabetic children since 1995 (Kosice), 1996 (Vienna, Prague, Ljubiljana, Budapest), 1997 (Groningen), 1998 (Lisbon) and 2000 (Szedged, Pecs). For cases having positive screening results, the measurement was repeated after 3 months.
For each EMA positive case two controls in the same center who were negative for EMA during the same period were chosen, stratified by age and age at DM onset.
IgA anti-EMA measurements in serum were taken by means of indirect immunofluorescence assay, using cryostat sections of monkey esophagus. Any positive readings (+ to +++++) in the immunofluorescence of a serum diluted 1:10 were considered as being positive. Total IgA levels were measured in all sera. In patients with IgA deficiency, IgG EMA was also used for screening. All patients with two positive EMA results were offered small bowel biopsy. Small bowel biopsy was mostly performed by means of upper endoscopy under general anesthesia or with a two-port, Watson-type capsule under sedation.
Biopsy specimens were analyzed according to a modified Marsh classification by the local pathologist (34), a pre-infiltrative stage or only increased intraepithelial lymphocyte counts and hyperplastic crypts (Marsh type 0, Marsh type 1 and Marsh type 2) and subtotal or total mucosal atrophy were defined as Marsh type 3 and Marsh type 4.
Height, weight, HbA1c, insulin dosage, and acute complications (ketoacidosis with need for intravenous fluid and insulin and severe hypoglycemia with unconsciousness or convulsions) were documented every 6 months after DM manifestation until at least 1 year of follow up after CD diagnosis in a standardized form and sent for evaluation to Vienna (Th Waldhoer).
Patients were advised to maintain GFD when the biopsy result was Marsh type 3 or type 4 (silent CD), but not in case of a normal biopsy result or only increased intraepithelial lymphocyte counts (Marsh type 1 or 2). At least yearly EMA measurements were continued in all cases and controls.
Patients with DM and silent CD who became EMA-negative on GFD and remained EMA-negative during the study period were thought to be compliant to GFD (compliance category 1) and patients who refused to maintain GFD or did not become EMA-negative on the suggested GFD were considered as non-compliant (compliance category 2).
For statistical analysis we used normal mixed models in SAS (SAS, Cary, NC) for describing the effect of age, group (DM and CD diagnosed/only DM), compliance (compliance 1/2) and grade (silent CD versus only DM). We tested for the effect of clinical center as a random variable. It turned out that center had no significant variance contribution; therefore it was dropped from the following models.
Values are given as mean and SEM. For comparison between groups we used two-sided t-tests.
98 diabetic patients (54 males and 44 females) were diagnosed having potential or silent CD by routine screening with EMA and subsequent small bowel biopsy (Fig. 1).
Two controls in the same center stratified by age and age-at-DM onset, which were negative for EMA during the same period (n = 195; 97 males, 98 females), were followed in a standardized way.
Clinical data of EMA-positive cases and EMA-negative controls at DM manifestation are given in Table 1.
Mean age of diabetes manifestation was 6.5 ± 4.1 years, and diagnosis of CD was made at 10.0 ± 5.4 years. EMA positivity at DM manifestation (within 90 days) was found in 20 patients (8 male; 12 female).
Biopsy results in 74 patients were equivalent to Marsh type 3 or 4 (silent CD), in 13 patients increased intraepithelial lymphocytes were found and in 7 cases a normal mucosa was observed. In 2 cases no clear result was possible and in 2 further cases no biopsy was performed; these patients were not included in the further analysis.
The mean observation period in this study after the diagnosis of CD was 3.3 ±1.9 years and after DM manifestation 7.66 ± 4.9 years for cases and 7.59 ± 4.9 years for controls (P = 0.86).
Mean HbA1c levels over the whole study period were similar in cases and controls (8.63% ± 1.45% versus 8.50% ± 1.39%; P = 0.35). Comparisons of HbA1c between cases and controls at diabetes manifestation (diag-DM), at CD diagnosis (diag CD) and at the last individual measurement during the study (follow up) showed no significant difference (Fig. 2).
There was also no difference in the frequency of severe hypoglycemia, defined as loss of consciousness and/or convulsions. In controls 25.8% reported one or more hypoglycemic attacks during the observation period compared with 20,4% of the cases (P = 0.55). One or more ketoacidotic episodes besides a possible ketoacidosis at the onset of the diabetes were observed in 4.3% in controls and in 4.06% of the cases (P = 0.84). There was no difference in the frequency of ketoacidosis at diabetes onset (9.6% versus 7.3%).
The applied insulin dosage was comparable in both groups (0.71 ± 0.19 IU/kg versus 0.72 ± 0.19 IU/kg; P = 0.45).
In cases and controls BMI-SDS showed no difference at DM manifestation and CD diagnosis. BMI-SDS increased significantly during the whole observation period (Fig. 3); this increase was observed in boys (P < 0.02) and in girls (P < 0.01).
At diagnosis of CD there was no difference for BMI in boys between cases and controls, but after diagnosis of CD further weight gain was diminished in cases compared with their controls (P < 0.05) (Fig. 4).
Female cases with silent CD had as well a lower BMI than female patients with diabetes alone (controls), but the difference did not reach statistical significance (P = 0.067).
Height at DM manifestation was not differing between cases and controls. In the follow-up height SDS decreased significantly (P < 0.05) in cases and controls and in both sexes (Fig. 5).
Influence of GFD
The 74 patients (35 boys and 39 girls) with silent CD were advised to take GFD. According to the repeated control measurements of EMA in the follow-up period (3.3 ± 1.9 years) patients were assigned to compliance category 1 (n = 33; 14 boys and 19 girls) or compliance category 2 = noncompliant (n = 25; 11 boys and 14 girls). In six patients there was no EMA follow-up measurement available and ten cases had fluctuating EMA titers, so those patients could not be categorized and were excluded from the analysis. Comparing patients with good compliance to GFD (compliance 1) and noncompliant cases (compliance 2), there was a trend to lower BMI-SDS in noncompliant cases (0.70 ± 0.25 versus 0.08 ± 0.25 P = 0.089). While height-SDS in patients, who remained EMA positive (compliance 2) during the follow up were not different.
Although it is well established that patients with untreated CD may develop complications, the natural history of undiagnosed silent or potential CD is unclear. An important argument for screening and early GFD treatment are observations in CD patients that the risk of developing further autoimmune diseases (35-37) and intestinal lymphomas (18,20) is proportional to the time of exposure to gluten, but these conclusions are based on clinically diagnosed patients with CD rather than cases detected by screening.
Short-term observational follow-up data on patients with oligosymptomatic or silent CD and DM are scarce and inconclusive concerning possible positive or negative influences of GFD on metabolic control or compliance with the diabetic treatment regimen.
In contrast to our results, glycosylated hemoglobin has been found significantly lower in children with CD and DM by several authors (26,32).This finding has been interpreted to be attributable either to impaired gastrointestinal absorption of nutrients in CD patients with imperfectly adjusted GFD or better controlled intake of carbohydrates in patients with good compliance to GFD. Improvement in metabolic control and a decrease in hypoglycemic episodes have been observed in children with CD and DM after institution of a GFD (28) and hypoglycemia and a reduction of insulin requirements were indicators for the presence of CD in children (38). In contrast to those findings, several studies (6,27,29,31,33,39) report no effect of the diagnosis of silent CD and the treatment with GFD on diabetic control although other investigators observed a deterioration in metabolic control in some patients (31). But in many patients, as in our group, adherence to a GFD could not be judged exactly, making a proper assessment difficult.
It seems to be of importance that several of these follow-up studies describe few patients or lack a matched control group. Our case-control study includes almost 100 diabetic children with diabetes and celiac disease. Comparing patients with proper adherence to GFD with patients who were clearly noncompliant to GFD, we found no difference in HbA1c during the observation period.
Another concern of untreated CD is the negative impact on growth and weight gain, resulting in short stature, failure to thrive or delayed puberty. CD-positive patients with diabetes in a large diabetes database showed significantly impaired growth and lower weight compared with diabetic patients without CD (32). The differences even increased in the long-term course, but this was mostly seen in patients younger than the age of 11, and the cohort included a mixture of patients including symptomatic CD and not only silent CD cases, as in our study. In our cohort height -SDS decreased in both groups with duration of diabetes with no obvious difference between cases and control. BMI-SDS increased after diagnosis of diabetes in cases and controls, but there was a trend to lower BMI-SDS in cases with CD and diabetes.
Some authors assume that weight increases after starting GFD because of improved resorption (6,29,31). In our cohort we observed that weight and BMI-scores did increase immediately after initiation of insulin treatment during the course of DM and at the time of CD diagnosis cases were not lower in their body weight than their controls. Comparing patients with proper adherence to GFD with patients who were clearly noncompliant to GFD, we found a trend to lower BMI-SDS, but not height-SDS, in the noncompliant patients. There is no clear explanation for the negative influence of GFD on weight gain in our cohort; it may be multifactorial.
The introduction of a second additional dietary prescription might have led the patients to a more strict dietary control, counterbalancing the observed weight gain in the controls, a known phenomenon in diabetic children and adolescents during insulin treatment compared with healthy controls. As the GFD products are less palatable and the variety of offers is limited on the market, a lower caloric intake by this mechanism may contribute to the reduction in weight gain. Malabsorption of nutrients in CD may play a role in some of the patients.
The mean onset of DM in childhood is approximately 8 years (40). CD-affected patients appear to be younger at diabetes manifestation (in our cohort, 6.5 ± 4.1 years), indicating that this group of patients seems to be predisposed to develop autoimmune disorders and may be more vulnerable to mucosal injuries. Some authors have reported similar findings (13,32,41) whereas others were unable to confirm this observation (10).
In conclusion, silent CD in diabetic children seems to have no obvious effect on metabolic control but a negative influence on weight gain even during the short observation period of this study. As GFD represents an additional burden on young patients with diabetes, extended follow-up studies are needed to document the clinical benefit of CD screening and treatment in diabetic patients.
1. Shanahan F, McKenna R, McCarthy CF, Drury MI. Coeliac disease and diabetes mellitus: a study of 24 patients with HLA typing. QJM
2. Aktay AN, Lee PC, Kumar V, et al. The prevalence and clinical characteristics of celiac disease in juvenile diabetes in Wisconsin. J Ped Gastoenterol Nutr
3. Kordonouri O, Dieterich W, Schuppan D, et al. Autoantibodies to tissue transglutaminase are sensitive serological parameters for detecting silent coeliac disease in patients with Type 1 diabetes mellitus. Diabet Med
4. Schober E, Bittmann B, Granditsch G, et al. Screening by anti-endomysium antibody for celiac disease in diabetic children and adolescents in Austria. J Pediatr Gastroenterol Nutr
5. Barera G, Bonfanti R, Viscardi M, et al. Occurrence of celiac disease after onset of type 1 diabetes: A 6-year prospective longitudinal study. Pediatrics
6. Acerini CL, Ahmed ML, Ross KM, et al. Coeliac disease in children and adolescents with IDDM: Clinical characteristics and respond to gluten-free diet. Diabet Med
7. De Vitis I, Ghirlanda G, Gasbarini G. Prevalence of celiac disease in type I diabetes: a multicenter study. Acta Paediatr
8. Fraser-Reynolds KA, Trussell RA, Butzner JD, Scott RB, Stephure DK. Use of immunoglobulin A antiendomysial antibody to screen for celiac disease in North American children with type 1 diabetes. Diabetes Care
9. Carlsson A, Axelsson I, Borulf S, et al. Prevalence of IgA-antiendomysium and IgA-antigliadin autoantibodies at diagnosis of insulin dependent diabetes mellitus in Swedish children and adolescents. Pediatrics
10. Hansen D, Bennedbaek FN, Hansen LK, et al. High prevalence of coeliac disease in Danish children with type I diabetes mellitus. Acta Paediatr
11. Arato A, Korner A, Veres G, et al. Frequency of coeliac disease in Hungarian children with type 1 diabetes mellitus. Eur J Pediatr
12. Lampasona V, Bazzigaluppi E, Barera G, Bonifacio E. Tissue transglutaminase and combined screening for coeliac disease and type 1 diabetes-associated autoantibodies. Lancet
13. Koletzko S, Burgin-Wolff A, Koletzko B, et al. Prevalence of coeliac disease in diabetic children and adolescents. A multicentre study. Eur J Pediatr
14. Smith CM, Clarke CF, Porteous LE, Elsori HE, Cameron DJS. Prevalence of coeliac disease and longitudinal follow-up of antigliadin antibody status in children and adolescents with type 1 diabetes mellitus. Pediatric Diabetes
15. Calero P, Ribes-Koninckx C, Albiach V, et al. IgA antigliadin antibodies as a screening method for nonovert celiac disease in children with insulin-dependent diabetes mellitus. J Pediatr Gastroenterol Nutr
16. Mora S, Barera G, Beccio S, et al. A prospective, longitudinal study of long term effect of treatment on bone density in children with celiac disease. J Pediatr
17. Gasbarrini A, Sanz Torre E, Trivellini C, et al. Recurrent spontaneous abortion and intrauterine fetal growth retardation as symptoms of coeliac disease. Lancet
18. Logan R, Rifkind E, Turner I, Ferguson A. Mortality in coeliac disease. Gastroenterology
19. Hadjivassiliou M, Gibson A, Davies-Jones GA, et al. Does cryptic gluten sensitivity play a part in neurological illness? Lancet
20. Catassi C, Fabiani E, Corrao G, et al. Risk of non-Hodgkin lymphoma in celiac disease. JAMA
21. Holmes GKT. Screening for coeliac disease in type 1 diabetes. Arch Dis Child
22. Freemark M, Levitsky LL. Screening for celiac disease in children with type 1 diabetes. Diabetes Care
23. Schober E, Rami B, Granditsch G, Crone J. Coeliac disease in children and adolescents with Type 1 diabetes mellitus: to screen or not, to treat or not? Horm Res
24. Spiekerkoetter U, Seissler J, Wendel U. General screening for celiac disease is advisable in children with type 1 diabetes. Horm Metab Res
25. Not T, Tommasini A, Tonini G, et al. Undiagnosed coeliac disease and risk of autoimmune disorders in subjects with type 1 diabetes. Diabetologia
26. Amin R, Murphy N, Edge J, et al. A longitudinal study of the effects of a gluten-free diet on glycemic control and weight gain in subjects with Type 1 diabetes and celiac disease. Diabetes Care
27. Crone J, Rami B, Huber WD, Granditsch G, Schober E. Prevalence of celiac disease and follow-up of EMA in children and adolescents with type 1 diabetes mellitus. J Pediatr Gastroenterol Nutr
28. Mohn A, Cerruto M, Iafusco D, et al. Celiac disease in children and adolescents with Type 1 diabetes: importance of hypoglycemia. J Pediatr Gastroenterol Nutr
29. Westman E, Ambler GR, Royle M, Peat J, Chan A. Children with coeliac disease and insulin dependent diabetes mellitus: growth, diabetes control and dietary intake. J Pediatr Endocrinol Metab
30. Lorini R, Scaramuzza A, Vitali L, et al. Clinical aspects of coeliac disease in children with insulin-dependent diabetes mellitus. J Pediatr Endocrinol Metab
31. Saukkonen T, Vaisanen S, Akerblom HK, Savilahti E. Coeliac disease in children and adolescents with type 1 diabetes: a study of growth, glycaemic control, and experiences of families. Acta Paediatr
32. Kaspers S, Kordonouri O, Schober E, et al. Anthropometric parameters, metabolic control and thyroid autoimmunity in 127 biopsy-positive children and adolescents with Type 1 diabetes and celiac disease compared to 18,470 diabetic subjects without celiac disease. A multicenter survey. Diabetologia
33. Kaukinen K, Salmi J, Lahtela J, et al. No effect of gluten-free diet on the metabolic control of type 1 diabetes in patients with diabetes and celiac disease. Diabetes Care
34. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastorenterol Hepatol
35. Toscano V, Conti F, Anastasi E, et al. Importance of gluten in the induction of endocrine autoantibodies and organ dysfunction in adolescent celiac patients. Am J Gastroenterol
36. Ventura A, Magazzu G, Greco L, Sige P. Duration of exposure to gluten and risk for autoimmune disorders in patients with celiac disease. Gastroenterology
37. Valerio G, Maiuri L, Troncone R, et al. Severe clinical onset of diabetes and increased prevalence of other autoimmune diseases in children with coeliac disease diagnosed before diabetes mellitus. Diabetologia
38. Iafusco D, Rea F, Prisco F. Hypoglycemia and reduction of the insulin requirement as a sign of celiac disease in children with IDDM. Diabetes Care
39. Andreelli F, Plotton I, Riou JP, Thivolet C. Diabetic instability and celiac disease. Diabetes Care
40. EURODIAB ACE Study Group. Variation and trends in incidence of childhood diabetes in Europe. Lancet
41. Cerutti F, Bruno G, Chiarelli F, et al. Younger age at onset and sex predict celiac disease in children and adolescents with Type 1 diabetes. Diabetes Care