It has become increasingly evident that celiac disease (CD) is more common than previously thought (1). Owing to the diffusion of sensitive serological markers, such as the antigliadin antibody (AGA) and the anti-endomysium antibody (EmA) tests, a number of atypical presentations have been described, and the existence of asymptomatic (silent) CD cases has come to our attention (2-3). This high CD polymorphism poses a number of problems (4).
In order to investigate the “celiac iceberg” (5), we conducted a pilot study of the screening for CD by means of the AGA test in a population sample of healthy students living in central Italy. Herein we present the final results of this screening of 6,315 subjects. An estimate of the overall CD prevalence in this age group is reported, together with comment on issues related to CD screening.
THE PROJECT “COELIAC DISEASE IN THE YEAR 2000: EXPLORING THE ICEBERG”
Patients and Methods
Between March 1992 and June 1994, 8,690 students of both sexes, aged 11 to 16 years and attending compulsory secondary school in the provinces of Pesaro-Urbino and Ascoli Piceno (central Italy), were invited to take part in the screening project. There were five known (biopsy-proven) cases of CD in this population. These five patients were on a gluten-free diet and did not take part in the screening. After giving informed consent, 6,315 students were screened for CD (73% of the eligible population). The study group had a mean age of 12.8 ± 1.0 years.
The screening procedure has previously been described (6). During school time, samples of whole blood were obtained for the first-level screening test. This was the determination of both serum IgG-AGA and IgA-AGA in either a capillary or venous blood sample. AGA was measured by enzyme-linked immunoassay (Alfagliatest; Eurospital, Trieste, Italy) on duplicate 5-μl samples. Subjects with a positive IgG-AGA and/or IgA-AGA test were recalled for the EmA test and for total serum IgA determination. IgA class EmA was measured by indirect immunofluorescence, using monkey esophagus as the antigenic substrate (Endomisio; Eurospital). The total serum IgA assay was performed in order to rule out a selective IgA deficiency. An intestinal biopsy was finally recommended for subjects fulfilling at least one of the following criteria: (a) a positive EmA test, (b) a positive IgA-AGA test, or (c) a positive IgG-AGA test associated with selective IgA deficiency. The jejunal biopsy was performed at the hospital gastroenterology clinic using a Watson pediatric capsule. The diagnosis of CD was made according to the revised ESPGAN diagnostic criteria (7).
The first-level AGA screening was normal in 6,177 of 6,315 subjects (97.8%). After the second-level evaluation, screening results were considered negative in 95 of 138 subjects who showed either normal results on the second level tests or isolated IgG-AGA positivity. An intestinal biopsy was performed in 43 of the 44 subjects who met the criteria for the third-level investigation. Only a 12-year-old boy was not available for intestinal biopsy, but he was considered to have CD since he showed positive results in the IgG-AGA, IgA-AGA, and EmA tests. Twenty-eight of these 43 subjects (17 girls, 11 boys) had flat mucosa on intestinal biopsy. Most of these subjects had no major symptoms, and none of them was under medical surveillance (Table 1). Sixteen of the screening-detected cases of CD were also HLA typed, showing the following genetic patterns: DQw2 in 13 cases, DR3 in eight cases, and DR7 in six cases.
The prevalence of undiagnosed CD in this study group was one in 218 subjects (4.59 per 1,000; 95% confidence interval [CI], 2.91 to 6.26). The prevalence of CD for the eligible population was calculated on the assumption that disease prevalence was the same in eligible subjects who did not take part in the study. The estimated prevalence of CD in the eligible population (8,690 students), including the five known cases of CD, was one in 193 subjects (5.17 per 1,000; 95% CI = 3.65 to 6.68). The estimated ratio of known to undiagnosed cases of CD in the eligible population was 1:9.
SEROLOGICAL CD SCREENING: WHICH TEST?
Different serological tests, such as the AGA, EmA, and antireticulin antibody (ARA) assays, are available for CD screening. In our study, the combined IgG-AGA and IgA-AGA determination was chosen because of the high sensitivity and proper standardization of the cutoff values. A large pediatric study of untreated subjects with CD and controls showed a sensitivity of ≥96% and a specificity of 97% for combined IgG-AGA and IgA-AGA measurements (8). Although the EmA test is considered the most reliable screening method for CD at present (9), its applicability to large screening programs is questionable. Patients with selective IgA deficiency cannot be detected by the EmA test, since this is an IgA class antibody. This problem requires consideration, since it has been established that subjects with selective IgA deficiency carry an increased risk of CD. In our study, one patient who was IgG-AGA positive, but IgA-AGA and EmA negative, was found to be affected with both selective IgA deficiency and CD.
HOW COMMON IS CD IN THE GENERAL POPULATION?
In this study group, which constitutes the largest population sample screened for CD by a single center, the CD prevalence among teenagers living in central Italy was close to one in 200 subjects. Some patients can remain undiagnosed after AGA screening. The prevalence of 1 in 200 subjects in this age category represents the minimum value. In this area, therefore, CD appears to be the most common chronic disorder in this age group; it is seen much more frequently than other pediatric diseases, such as hypothyroidism, type I diabetes mellitus, cystic fibrosis, or familial hypercholesterolemia.
In a recent multicenter study, the prevalence of registered CD cases in Italy was about one of 1,000 subjects. Our results clearly show that clinically diagnosed cases represent just the tip of the CD iceberg. It has been reported that CD prevalence changes with time or shows wide variations between neighboring countries, such as Finland and Sweden or Italy and Greece (10). Since the number of undiagnosed people greatly exceeds the number of patients diagnosed with CD, the highly variable CD prevalence found in previous epidemiological surveys could primarily depend on the different strategies adopted for case ascertainment.
A NEW IDENTIKIT FOR CD
The clinical picture of CD has undergone change. Typical CD manifestations, such as chronic diarrhea or malnutrition, were present in a minority of cases, while vague intestinal complaints, such as poor appetite or recurrent aphtous stomatitis, were not uncommon. Iron deficiency, a well-known CD-associated disorder secondary to jejunal iron malabsorption, was found, with or without anemia, in 16 of 28 patients with CD detected by screening, suggesting that an accurate CD diagnostic workup is mandatory for patients with unexplained iron deficiency, especially when they are resistant to oral iron therapy. We noticed that irritability, tendency to depression, and unsatisfactory school performance were common findings among these patients. However, we did not perform a standardized psychosocial investigation that could clarify this clinical impression.
WHY DO MANY PATIENTS WITH CD REMAIN ASYMPTOMATIC?
The reason that many CD patients remain symptom free is still unknown and can only be a matter of speculation. It might be that the extent of mucosal damage is smaller in silent than in symptomatic CD. In this case the absorptive defect of the proximal small bowel could have no clinical consequences, owing to the compensatory function of the distal jejunal mucosa. The results of the intestinal permeability test with lactulose-mannitol in our study support this view. While the lactulose-mannitol urinary ratio is usually increased in typical CD (11), 11 of 21 patients with screening-detected CD showed a normal value (12).
SHOULD WE TREAT ASYMPTOMATIC CASES?
It is our current policy to recommend the standard gluten-free diet for the treatment of silent CD cases, for at least two reasons. First, in our study most patients with apparently silent CD were not completely asymptomatic. This is evident from the follow-up visits, when many patients reported subtle benefits after starting the gluten-free diet. The most common changes were gains in weight and height, appetite and mood amelioration, better school performance, and physical fitness. Second, clinically silent CD causes the same jejunal damage and has the same immunological pattern and genetic background as typical CD. At present we have no reason to suggest a different therapeutic strategy for these patients.
A number of long-term CD complications have been described, such as infertility, osteoporosis, and lymphoma (13). The treatment of asymptomatic CD may become primarily an issue of prevention. Further studies are required in this area, because the impact of CD-associated risk is still unclear.
This study confirms that CD is much more common in Italy than previously thought and that most cases remain undiagnosed unless actively identified through serological screening tests. Efforts should be made to improve the understanding of this condition among doctors and other health workers, since this would allow detection of some undiagnosed cases. It also appears advisable to recommend the inclusion of a screening test for CD, such as the AGA or EmA assay, in routine hematological investigations during childhood. Further studies are needed to map the prevalence of CD in different countries.
1. Auricchio S, Greco L, Troncone R. What is the true prevalence of coeliac disease? Gastroenterol Int
2. Ferguson A, Arranz E, O' Mahony S. Clinical and pathological spectrum of coeliac disease: active, silent, latent, potential. Gut
3. Corazza GR, Frisoni M, Treggiari EA, et al. Subclinical celiac sprue. J Clin Gastroenterol
4. Challacombe D. When is a coeliac? Lancet
5. Logan RFA. Problems and pitfalls in epidemiological studies of coeliac disease. In: Auricchio S, Visakorpi JK, eds. Epidemiology of coeliac disease
. Basel: Karger, 1992:14-24. (Common food intolerances,
6. Catassi C, Rätsch IM, Fabiani E, et al. Coeliac disease in the year 2000: exploring the iceberg. Lancet
7. Working Group of the European Society of Paediatric Gastroenterology and Nutrition. Revised criteria for diagnosis of coeliac disease. Arch Dis Child
8. Bürgin-Wolff A, Berger R, Gaze H, Huber H, Lentze MJ, Nusslé D. IgG, IgA and IgE gliadin antibody determination as a screening test for untreated coeliac disease in children, a multicentre study. Eur J Pediatr
9. Carroccio A, Iacono G, Montalto F, et al. Immunologic and absorptive tests in celiac disease: Can they replace intestinal biopsies? Scand J Gastroenterol
10. Greco L, Mäki M, Di Donato F, Visakorpi JK. Epidemiology of coeliac disease in Europe and the Mediterranean area. In: Auricchio S, Visakorpi JK, eds. Epidemiology of coeliac disease
. Basel: Karger, 1992:25-44. (Common food intolerances,
11. Catassi C, Pierani P, Natalini G, Gabrielli O, Coppa GV, Giorgi PL. Clinical application of a simple HPLC method for the sugar intestinal permeability test. J Pediatr Gastroenterol Nutr
12. Catassi C, Fabiani E, Ratsch IM, et al. Is the sugar intestinal permeability test a reliable investigation for coeliac disease? Gut
13. Holmes GKT. Long-term high risks for unrecognized coeliac patients. In: Auricchio S, Visakorpi JK, eds. Epidemiology of coeliac disease. Basel: Karger, 1992:105-18.(Common food intolerances,
Heymans: I think we have some experience with the sugar absorption test, using lactulose-mannitol. We use it in a hyperosmolar solution, which improves the results dramatically. We haven't published them yet, but we have seen differences in the results of the test according to the osmolarity of the solutions. Maybe van Elburg should tell you about this.
van Elburg: We are still using the same components: 5 g of lactulose and 2 g of mannitol, but we make the substance hyperosmolar by adding 40 g of sucrose. When we used that solution, we had 83% sensitivity for CD. I do agree that the EmA test is the most sensitive, but in our experience the lactulose-mannitol test in a hyperosmolar solution also is a good alternative.
Catassi: As you know, some groups use hyperosmolar solutions, and others did not. We prefer isotonic solutions, because we feel that they are more physiological. When we perform the intestinal permeability studies in typical CD cases, we always find an increase in the lactulose-mannitol ratio. Our data were published in the Journal of Pediatric Gastroenterology and Nutrition (Catassi et al. 1991;12:209-12).
Heymans: We found that about one third of the first-degree family members of CD patients had normal histology results on small-bowel biopsy, but abnormal lactulose-mannitol permeability test results. Thus, even if we don't find severe destruction of the mucosa, we do find abnormal intestinal permeability for sugars in those persons who are at risk.
van Elburg: In addition, we found a correlation between lymphocyte infiltration in the small bowel and abnormal results on the lactulose-mannitol test. So what we find in family studies is that we possibly recognize latent CD in a large subgroup of family members. We are disappointed about the results of the AGA tests. One of the problems is that we lack international standards for these tests.
Feighery: We have heard a lot of discussion about the prevalence of CD and about whether or not CD is underdiagnosed, and I have no doubt that in Ireland underdiagnosis is responsible for the variations in prevalence. The data we have just heard are similar to the data from Northern Ireland. Unfortunately, the data are not yet supported by biopsy studies, but on the EmA test almost 1% of some 5,000 individuals had positive results. If the EmA test is as specific for CD as it appears to be, there are large numbers of undiagnosed patients with CD. Very recently we found that 7% of 100 persons with insulin-dependent diabetes were EmA positive, so I am convinced that there are hordes of undiagnosed patients.
Peña: Professor Feighery, you work in Dublin. Do you think that the big differences in prevalence of CD between Galway and Dublin are based on problems with diagnosis?
Feighery: It is difficult to comment on that specifically, but I have no doubt that the screening tests we are using now are vastly better than the ones used before. I think that when you screen and diagnose properly, you will find many more unrecognized patients with CD in our community.
Holmes: I agree with all that. I think that CD is grossly underdiagnosed, and what we need to do is have screening programs, either within hospitals or within private practices, so that cases of CD can be picked up. It is important to diagnose CD, because we can treat it. In my hospital we have a biochemist who is interested in CD, so every time he finds, for example, a low serum calcium, he performs IgA-AGA and EmA tests. That has found a number of truly silent cases. I agree with what is said from Ireland; we recently screened 2,000 insulin-dependent diabetics using IgA-AGA and came up with a prevalence of CD of one in 50. So that is a group that can benefit from screening, because half of the patients did complain of symptoms. Every time they went to the diabetes clinic, however, they were told that diabetes was causing the diarrhea-the old diabetic diarrhea story, which I suspect is wrong, because once these people were established on a gluten-free diet, they improved enormously. I think that an organized screening program is needed to investifate groups like this.
Ascher: I just wanted to inform you that there is a working group within the framework of ESPGAN trying to standardize serological markers for CD, so there will be standardized methods in the future. I also wanted to comment a bit on this beautiful presentation. I think that when we try to answer the question of whether we should treat asymptomatic patients with CD, it is important to realize that the intestine is not only an absorptive organ but also an important immunological organ and a hormone-producing organ. CD might have many effects, and I wonder what the long-term effect is of chronic intestinal inflammation over many years. Just one question for Dr. Catassi: Did you look at the calcium density of the skeleton in your cases?
Catassi: Unfortunately, we did not. The group in Milan looked at this feature in CD, and they found lower calcium contents (Mora et al., Am J Clin Nutr 1993;57:224-8).
Mearin: Dr. Catassi, you say that the answer to the question of whether we should treat asymptomatic CD is yes, but I am not convinced. Maybe I will be convinced at the end of this meeting, but not yet. Two of the reasons you give for treating asymptomatic disease are that patients will feel better once they are treated and that you are protecting them from bad things in the future. But I wonder whether the reason for their feeling better is indeed the treatment or that the gluten-free diet has a placebo effect. It is common clinical practice that every diet helps in some way. The people following the diet think that it is going to help, and so it helps. Furthermore, how sure are we that those asymptomatic patients have the same risks of, say, malignancy as those with symptomatic disease?
Catassi: I am not sure about this, and no one is. For this reason these subjects have to be studied in the future.
van der Meer: Dr. Catassi, I am very interested in your study. Did you find comparable results using AGA and EmA tests, and what is your opinion about the sensitivity and specificity of both?
Catassi: I did not show these data, because it would have taken too long. In this study we could not evaluate the sensitivity of these tests, because we would have had to take biopsies from the whole group. We have the feeling that the EmA test is almost 100% sensitive. In all the CD cases we found, except for one showing IgA deficiency, the EmA test gave positive results. The AGA tests works well, in my opinion, if you determine both IgA and IgG. You will usually find at least one of them to be abnormal. If I had to choose between IgA and IgG AGA tests, I would choose IgG AGA, which is, in my opinion, much more sensitive.
van der Meer: There is some advantage in their lower specificity, in that antibodies to gliadin can also be present in Crohn's disease and in ulcerative colitis.
Catassi: That's right. In the general population this problem is smaller.
Greco: Among 52 adults with unspecified sideropenic anemia, we found 16 cases of CD. In answer to Luisa Mearin, we identified 26 asymptomatic women who were treated with a gluten-free diet. After 1 year I asked them how they felt, and they were experiencing new life, including renewed sexual activity.
Heymans: We are largely dealing with pediatric patients, among whom you don't see dermatitis herpetiformis. But if you see dermatitis herpetiformis in adults, you also see malignancies in those who have minimal proximal small-intestinal lesions. Therefore, the risk of malignancy for people who are gluten-sensitized and have mild lesions exists. This is one ethical concern that always makes me put a person with CD on a gluten-free diet.
Peña: Now that you have mentioned the problem of dermatitis herpetiformis, in Tampere, Finland, there are 700 to 800 patients with the disease. But in the Netherlands, even after extensive searching, you will not find many such patients. Do you agree, professor van der Meer?
van der Meer: Yes.