High Prevalence of Unrecognized Celiac Disease in an Unselected Hospital Population in North-Eastern Brasil (Recife, Pernambuco)

To the Editor: Several European studies, making use of serological tests, have shown that the true prevalence of celiac disease (CD) may be 1:200 or even greater in different countries (1), and the most affected patients do not complain gastroenterological symptoms. A recent article has reported the presence of CD also in North America (2). In Brazil, a prevalence of CD of 1:681 was found in a population of blood donors, obviously with a lower morbidity than the general population (3). The results of another Brazilian study indicate that the spectrum of presentation is similar to that already seen in developed countries (4). In this study, we screened for CD in a large, unselected Brazilian hospital population. All subjects were enrolled at various day-clinics (excluding gastroenterology) at the Instituto Materno Infantil do Pernambuco hospital of Recife. Most subjects came from the poor slum areas (favelas) of Recife and its vicinities. We collected serum samples from 1030 Brazilian subjects: 915 children (397 girls, 518 boys; median age, 9 years; range, 1–18) seen in different pediatric day-clinics (excluding gastroenterology) and 115 female subjects (median age, 14 years; range, 1–18) presenting for gynecological problems. The sera were analysed for anti-guinea pig transglutaminase IgA antibodies (anti-gp tTg) (Eurospital, Trieste, Italy). All borderline or positive sera for anti-gp tTG were tested for anti-human tTG IgA and IgG antibodies (anti-htTG) (5), and for antiendomisium antibodies (AEA). Individuals positive for anti-gp tTg, anti-htTG and AEA underwent genotyping for HLA class II alleles and intestinal biopsy. Sixty-six individuals (64 pediatric, 2 gynecological) tested positive for anti-gp tTG, and 55 were borderline (48 pediatric, 7 gynecological). Both positive and borderline individuals (121 subjects, 112 pediatric and 9 gynecological) were tested for anti-htTG to obtain confirmation of results. Forty-seven of these 121 individuals who were positive or borderline for anti-gp tTG also tested positive for anti-htTG. We also studied anti-gp tTG positive and borderline individuals by AEA. We found 20 positive cases (19 from the pediatric and 1 from the gynecological clinic) for AEA among the 121 who tested positive or borderline for anti-gp tTG. These 20 were positive for anti-gp tTG, anti-htTG and AEA. In conclusion, 20 out of 1030 screened subjects (1.9%) tested positive for all 3 CD-related serum autoantibodies. The HLA studies showed the presence of HLADRB1*03*07, and HLADQB1*02 alleles in these 20 subjects. Diagnosis of CD in all 20 patients was confirmed by intestinal biopsy. There was total villous atrophy (Marsh 3c) in 10 out of 20 cases, and in 6 cases the grade was Marsh 3a and in 4 Marsh 2c. Although we used a highly selective CD screening method, considering suitable for intestinal biopsy only those individuals who tested positive for both serological (anti-gp tTG, anti-htTG, AEA) and HLA class II genotyping tests, we found a very high prevalence of CD (1:52). Indeed, this prevalence could be considered as high as 1:22 if we were to include 27/1030 subjects testing positive for anti-htTG but negative for AEA. Some celiacs could be eventually present in this last group of patients. Unfortunately, these individuals were lost at follow-up and we were unable to perform either intestinal biopsy or HLA genotyping. The clinical features of both the pediatric and gynecological populations who entered the study are as follows: 11 out of 20 patients positive for all three CD serological tests showed clinical features typically related to CD such as anemia, abdominal pain and insulin dependent diabetes. Nine patients out of 20 were characterized by other complaints not classically related to CD, such as pneumonia and kidney disease. There was no correlation between antibody serum titre and grade of intestinal lesion and presence of classic CD-related symptoms. The prevalence of CD in our study was significantly higher than that (1:681) found in the study performed among Brazilian blood donors (3). The difference in our findings may be explained by the higher levels of sensitivity of our serological test due to the introduction of tTG antibodies (5). Furthermore, we studied a selected hospital population affected by several diseases, some of them CD-related, while the blood-donor study analyzed a selected healthy population. The subjects diagnosed as Celiacs presented with complains not directly attributable to CD or with signs and symptoms possibly dependent on unrecognised CD but usually attributed, in that setting, to malnutrition or to infections. This finding is all the more worrying given that in developing countries undiagnosed CD could exacerbate any underlying illness present such as anaemia and diarrhea. In conclusion, CD has a very high prevalence in hospitalized patients in north-eastern Brazil and may act as an important unrecognized comorbidity factor. Other studies are needed to examine the real impact of undiagnosed CD on morbidity and mortality in developing countries and to evaluate the feasibility of introducing a gluten-free diet. Chiara Trevisiol Katia Galeao Brandt Giselia Alves Pontes Silva Sergio Crovella Alessandro Ventura *University of Trieste, Trieste, Italy †Universitade Federal de Pernambuco, Recife, Brazil