Stratification of Bone Fracture Risk in Patients With Celiac Disease

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2004;2: Stratification of Bone Fracture Risk in Patients With Celiac Disease MARÍA L. MORENO,* HORACIO VAZQUEZ,* ROBERTO MAZURE,* EDGARDO SMECUOL,* SONIA NIVELONI,* SILVIA PEDREIRA,* EMILIA SUGAI,* EDUARDO MAURIÑO,* JUAN C. GOMEZ, and JULIO C. BAI* *Small Intestine Section, Department of Medicine, Hospital de Gastroenterología Dr. Carlos Bonorino Udaondo, Buenos Aires, and School of Medicine, Universidad del Salvador, Buenos Aires, Argentina; and Unidad de Soporte Nutricional y Malabsorción, Hospital José de San Martín, La Plata, Buenos Aires, Argentina Background & Aims: Our objective in this cross-sectional, case-control study was to gain insight into celiac osteopathy by examining a well-defined cohort of patients with a wide clinical spectrum of the disease. Methods: We studied 148 unselected celiac patients and 296 (1:2) age- and sex-matched controls diagnosed with functional gastrointestinal disorders. Based on the clinical history, 53% were classically symptomatic, 36% had subclinical celiac disease, and 11% were silent, detected by screening. The fracture information was obtained through an in-person interviewusing a predesigned questionnaire. Results: Classically symptomatic patients had an increased number of fractures in the peripheral skeleton (47%) compared with age- and sex-matched controls (15%; odds ratio, 5.2; 95% confidence interval, ). However, fractures in subclinical/silent celiac cases (20%) were no different from those in controls (14%; odds ratio, 1.7, ). Compared with the subclinical/silent group, a significantly greater prevalence of fractures was detected in classically symptomatic patients (odds ratio, 3.6, ). Compared with controls, celiac disease patients had significantly more fractures produced by mild trauma (P < 0.01), but there were no differences in the severity of trauma events that induced fractures. Mean bone density femoral neck z score was higher for subclinical/ silent cases compared with classically symptomatic patients (P < 0.05). Conclusions: Celiac patients showa very wide variation in fracture risk, with increased risk in classically symptomatic patients. Diagnostic and therapeutic strategies to prevent bone loss and fracture should be preferentially used in the subgroup of patients with classic clinical disease. Celiac disease (CD) patients are troubled by the presence of a metabolic bone disease that produces low bone mineral mass and bone weakness and are, therefore, at an increased risk of fractures. 1 7 These abnormalities have been frequently reported, and a general alert has been given to patients and to the medical community suggesting investigation of bone status in every new patient. 8 Earlier studies have shown that the most important and common clinical manifestations of the osteopathy associated with CD are bone pain, altered gait, retardation of growth, bone deformities, and fractures. 9 In some situations, these clinical findings are presenting features that lead to the diagnosis of CD. Although fractures are frequently reported in patients with CD, very scant information related to the true magnitude of the problem exists in the scientific literature; this is evidence that the knowledge in this area is still in its infancy. Exploring this topic in a case-control crosssectional study, we have recently shown that CD patients present a significantly higher prevalence of fractures in the peripheral skeleton compared with controls. 7 It is interesting to note that the increased rate of fracture for peripheral bones occurred most often while patients were following a gluten-containing diet. In contrast, despite an exhaustive morphometric analysis of vertebral structures, we were not able to find any significantly increased risk of developing fractures for the axial skeleton. 7 We also observed that most fractures were produced by minimal or moderate trauma and that, in contrast with the age-related osteoporotic fractures, three quarters of fractures were observed in patients younger than 50 years. As far as we know, the former study was the first that specifically addressed the skeletal condition of CD patients. However, we were not able to determine why some patients show a marked reduction in bone mass and are prone to (in some cases, multiple) fractures, whereas others with similar risk factors do not experience such events. In a retrospective analysis of that study, we did not take into consideration the well-known clinical di- Abbreviations used in this paper: CD, celiac disease; CI, confidence interval by the American Gastroenterological Association /04/$30.00 PII: /S (03)

2 128 MORENO ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No. 2 versity of the disorder. Thus, the estimation of fractures in our former study was mainly established among classically symptomatic patients and was, therefore, not representative of the entire CD population. In this context, overwhelming information in the last decade has shown that the number of subclinical and clinically silent patients is possibly 8 10-fold greater than that of classically symptomatic cases. 10,11 It has been well documented that clinically silent patients may also present with osteopenia and osteoporosis. 3,12 However, a study from our group showed evidence that asymptomatic CD patients diagnosed from family studies have lower bone involvement (in terms of decreased bone mineral density) compared with classically symptomatic cases. 3 Therefore, we hypothesized that subclinical and clinically silent cases might not display a similar propensity to bone fractures. On the basis of all these premises and the former hypothesis, we designed this study to assess the relationship between bone fractures and CD by exploring a clinically more comprehensive patient population that included cases representatives of the overall clinical spectrum of the disorder. Materials and Methods Patients We enrolled 148 unselected consecutive adult patients (117 women; age range, years) who fulfilled the standard criteria for diagnosis of CD and who were attending 2 different tertiary referral centers devoted to the diagnosis and follow-up of patients with CD. Diagnosis of classically symptomatic and subclinical CD patients was based on the clinical picture, typical histological findings (in all patients), and a concordant CD-specific serology. For those with negative serology at diagnosis, a response to a gluten-free diet was required. Asymptomatic CD cases were collected from a screening program for first-degree relatives of well-established patients and from the general population. All new patients had characteristic findings at intestinal biopsy. According to the clinical assessment at the time of diagnosis and data obtained from the history of patients, they were categorized as 1 of the following: 1. Classically symptomatic: patients with chronic diarrhea, overt malabsorption, weight loss, and malnutrition. 2. Subclinical: patients who presented only extraintestinal signs and symptoms (anemia, osteopenia, osteoporosis, aphthae, late menarche, repeated spontaneous abortions, and so on). 3. Asymptomatic (silent CD): patients diagnosed from a population-based screening program, first-degree relatives, and those in whom no symptoms, signs, or biochemical abnormalities were detected despite an intensive search. For this categorization, we considered the overall clinical history of patients, and care was taken to avoid crossover in the clinical picture. For the purposes of this investigation, the study parameters were considered according to the grouping of patients into 2 subgroups: classically symptomatic celiac cases and subclinical/silent cases. Controls Results obtained in the overall CD population and both subgroups of patients were compared with those obtained in 296 age- and sex-matched controls collected from the outpatient clinic of the Gastroenterology Hospital (2 controls for each patient). Individuals were enrolled as controls only if their final diagnosis was a functional gastrointestinal disorder. For comparisons in the study, subjects were considered as an overall control population or were subdivided into those who matched with both classically symptomatic CD and those with a subclinical/silent course. Methods Design of the study. The study was case controlled and was based on a retrospective historical review through a personal interview with a predesigned form by the physician in charge. Epidemiological data evaluated were age at enrollment, age at diagnosis of CD, number and site of fractures, age at diagnosis of fractures and its relationship to the diagnosis of CD, compliance with a gluten-free diet, and type and intensity of the trauma that produced fractures. Trauma was considered (1) severe if it involved a traffic accident, was sports related, or was caused by falling from a height; (2) moderate if the fracture resulted from slipping or stumbling or from a fall on level ground; and (3) mild when minimal trauma was involved. 13 A spontaneous fracture was considered when no evidence of trauma was reported. The same questionnaire was used in both centers, and all data reported at the time of the interview were checked with those reported in patient records. If any discrepancy was detected, patients were contacted by telephone to confirm observations. If the discrepancy still persisted and no documentation of the event was available, the patient was excluded from the study. We also excluded patients and controls with associated disorders known to produce osteopenia or osteoporosis and patients with CD diagnosed in infancy unless they were following a glutencontaining diet. Bone mineral density measurements. The bone mineral density of the lumbar spine, total skeleton, and femoral neck was measured by dual-energy x-ray absorptiometry. The procedure was performed in only 75 patients (51%) and not in controls, and areas explored were total skeleton (n 27), lumbar spine (n 68), and femoral neck (n 38). The reproducibility of measurements and the coefficient of variation of the method have been reported previously. 14,15 Results are reported as z scores or T scores that represent the number of SDs that an individual value is separated from the corresponding mean normal value corrected for sex and age or that from a young adult, respectively. Osteopenia and osteoporosis

3 February 2004 RISK OF FRACTURES IN CELIAC DISEASE 129 Table 1. Demographic Data and Clinical and Densitometric Findings of CD Patients Grouped According to Their Categorization in Classically Symptomatic and Subclinical/Asymptomatic Cases Characteristic Classic CD patients Subclinical/silent patients P value No. of patients (%) 78 (53) 70 (47) Sex (female/male) 62/16 55/15 Age at diagnosis, yr, mean (range) 42 (12 75) 36 (13 79) 0.02 Age at study, yr, mean (range) 44 (18 77) 38 (17 81) Antibody-positive results (%) AGA IgA AGA IgG EmA IgA Body mass index (mean SEM) Age at menarche, yr (mean SEM) Age at menopause, yr (mean SEM) Bone mineral density z score (mean SEM) LS TS FN T score LS TS FN LS, lumbar spine; TS, total skeleton; FN, femoral neck; AGA, antigliadin antibodies; EmA, endomysial antibodies; Ig, immunoglobulin. were diagnosed when z scores were lower than 1 and 2, respectively. On the basis of T scores, osteopenia and osteoporosis were diagnosed when values were lower than 1or 2.5, respectively. Statistics Data are reported as mean values SEM. The proportion of patients and controls with fractures was compared by calculation of odds ratios and 95% confidence intervals (CI). Comparison of results among different groups was performed with the 2 test or the Fisher exact test for qualitative variables; with the Student t test for quantitative variables that followed a normal distribution; or with the Mann Whitney ranked sum test for data that failed the normality test. A P value 0.05 was considered statistically significant. The actuarial percentage of patients with fractures was estimated by Kaplan Meier curves, and differences were assessed by the Mantel Haenszel log-rank test. Multivariate logistic regression analysis was performed to determine which factors would predict fractures: sex, age, age at diagnosis, time on a glutenfree diet, and so on. Statistical analysis was performed with Statistix 7.0 (Analytical Software, Tallahassee, FL) and Epi Info 2.3. Results Clinical Characteristics of Patients A total of 148 CD patients were enrolled in the study: 78 (53%) patients were considered classically symptomatic, 53 (36%) were considered subclinical patients, and 17 (11%) were considered silent cases. Table 1 shows some epidemiological and clinical features of patients and controls at study entry and at the time of diagnosis of CD grouped as classically symptomatic and subclinical/silent cases and their respective controls. Patients with classic symptoms were older than subclinical/ silent cases (P 0.002) at entry to the study and at the age of diagnosis of CD (P 0.02). The mean time from the onset of symptoms to diagnosis of CD could be estimated only for classically symptomatic and subclinical patients. Patients with classic clinical symptoms showed a significantly more prolonged course as undiagnosed. Surprisingly, despite a clear cutoff between symptom subgroups, no differences were detected in body mass index between classic cases and subclinical/silent cases. Classic CD patients had an older mean age at menarche and an earlier menopause. These differences did not reach statistical significance. However, although this study did not collect data on the gynecologic history of the control population, the results are in agreement with those formerly reported by our group 16 and confirm that classic symptomatic CD is associated with a reduction of the fertile period of women. Frequency of Fractures and Other Considerations Overall, 51 patients (34%) from the CD population presented 57 fractures. On the basis of the categorization of patients, those with a classic clinical picture had significantly more fractures (n 37, 47%) compared with their respective controls (n 23, 15%; odds ratio, 5.2; 95% CI, ; P ). However, no difference was detected between subclinical/silent patients

4 130 MORENO ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No. 2 Table 2. Data on Fractures in Patients and Controls Categorized According to Clinical Characteristics Characteristic Classic patients Controls Subclinical/silent patients Controls No. of patients with fractures (%) 37 (47) 23 (15) a 14 (20) 20 (14) b Total number of fractures Fractures in women at ages 50 yr, n (%) 15/27 (55) 11/18 (61) 7/11 (64) 10/14 (71) Type of trauma producing fractures, no. cases (%) Mild 4 (10) 0 4 (29) 0 Moderate 18 (49) 12 (52) 6 (42) 9 (45) Severe 15 (41) 11 (48) 4 (29) 11 (55) a P 0.05; b P 0.004, patients vs. controls. (n 14, 20%) and controls (n 20, 14%; odds ratio, 1.7; 95% CI, ; P not significant; Table 2). The comparison of both subgroups of patients showed significantly more cases with fractures among classic cases (odds ratio, 3.6; 95% CI, ; P 0.001). Four patients in the classically symptomatic group and 2 of their subclinical/silent counterparts had more than 1 fracture. This was similar to the observations for both control groups. The proportion of fractures produced in women before age 50 was similar for both subgroups of patients and their respective controls. No differences were detected in the proportion of fractures in the premenopausal period (before the age of 50 years) in comparing classical patients with subclinical/silent cases, on the one hand, and both subgroups with their respective controls, on the other (Table 2). It is interesting to note that further data on the type of trauma that induced the fractures suggests that bone weakness may be involved in fractures. Compared with control subjects (0%), significantly more (14%) fractures occurred in the overall celiac population as a consequence of mild trauma (P 0.01). When patients were stratified according to the severity of the clinical picture, no difference was observed in the type of fracture (mild, moderate, or severe) when both subgroups of patients were compared with their respective controls (Table 2). The number of fractures at each skeletal site (either as a whole population or grouped according to their clinical presentation at diagnosis) was compared with those of control populations. Although the most frequently affected sites of fractures in CD patients were the upper extremities, this observation was not different from that in the control population (Table 3). Thus, 50% (29/57) of fractures in CD patients and 49% in control subjects (23/47) were produced in the arms, forearms, and hands (P not significant; Table 3). Furthermore, no differences were observed when patients were categorized according to their clinical compromise at diagnosis in classically symptomatic cases vs. controls (54% vs. 52%, respectively; P not significant) and subclinical/silent cases vs. controls (44% vs. 45%, respectively). We also compared the prevalence of fractures in CD patients and controls by considering the decade of life when the event was produced (Figure 1). Patients with classic CD had significantly more fractures in the fifth and sixth decades of life (P 0.05; Figure 1B). In contrast, subclinical/silent patients and controls did not show differences by decade (Figure 1D). When we compared both subgroups of patients, patients with classic CD had a significantly increased prevalence of fractures before 20 years of age (P 0.05) and at the fifth decade of life (P 0.01; Figure 1C). Patients with fractures were significantly older at study entry than those without fractures (P 0.05 and P for classically symptomatic and subclinical/ silent cases, respectively). Furthermore, among subclinical/silent patients, those who experienced fractures had CD diagnosed at significantly older ages than those without fractures (P 0.006; Table 4). No other differences were detected between subgroups of patients when those with and without fractures were compared (in terms of clinical features, densitometric measurements, time of fractures, or presence of a positive CD-related serology). The bone mineral density of patients was compared with population-based data and reported as a z score. For Table 3. Distribution of Fractures by Skeletal Site Reported Among Subgroups of Patients and Controls Skeletal site Classic patients Controls Subclinical/ silent patients Controls Hands/fingers Distal forearm/forearm Elbow Arm Clavicle Thoracic/lumbar spine Ribs Pelvis Femur/knees Other leg Feet/toes Face All sites

5 February 2004 RISK OF FRACTURES IN CELIAC DISEASE 131 Figure 1. Prevalence of fractures in patients and controls according to the time of the event. (A) All CD patients vs. controls; (B) classic CD patients vs. their controls; (C) classic CD vs. subclinical/silent patients; and (D) subclinical/silent cases vs. their respective controls. the overall CD population, although the lumbar spine z score was in the range of osteopenia ( ), those values for the total skeleton ( ) and femoral neck ( ) were at the lower end of the normal range. Whereas densitometric measurements at the total skeleton and lumbar spine (L2 to L4) levels were not different between subgroups, classically symptomatic patients had significantly lower bone density at the femoral neck level (P 0.05; Table 2). According to T score values, 63%, 62%, and 59% of patients had either osteopenia or osteoporosis at the lumbar spine, total skeleton, and femoral neck levels, respectively. We did not find significant differences in T scores when we compared patients with and without fractures ( vs for the lumbar spine, respectively; vs for the total skeleton, respectively; and vs for the femoral neck, respectively). Finally, we estimated the proportion of patients who were free of fractures according to Kaplan Meier analysis (Figure 2). There we observed that classically symptomatic patients, but not subclinical/silent cases, differed significantly from controls (P ; Figure 2B). Discussion Recently, our group suggested an increased prevalence of fractures in the peripheral skeleton of CD patients. 7 It is interesting to note that these findings in peripheral bones did not extend to the axial skeleton, where a meticulous morphometric analysis of radiographs was used to determine the presence of vertebral deformities (wedge and compression). Although these observations are in concordance with the known increased fracture risk based on densitometric bone mass estimations, some recent data suggest that a large-scale study and a more profound analysis of variables are still required. Two publications contrast with the former assumption. First, a preliminary study from England, published only in abstract form, did not find increased fracture rates for celiac patients. 13 More recently, a study from Denmark did not detect an increased prevalence of

6 132 MORENO ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No. 2 Table 4. Comparison of Clinical Parameters Among Different Subgroups of Patients and Their Respective Control Groups Classical patients Subclinical/silent patients Characteristic With fracture Without fracture With fracture Without fracture No. of patients (%) 37 (47) 41 (53) 14 (20) 56 (80) Sex (female/male) 27/10 35/6 11/3 44/12 Age at study, yr, mean (range) 49 (22 77) 42 (18 75) a 48 (22 81) 35 (17 73) b Age at diagnosis, yr, mean (range) 44 (19 75) 40 (12 65) 46 (21 79) 33 (13 70) c Age at onset of symptoms, yr (mean SEM) Body mass index, kg/cm 2 (mean SEM) Age at menarche, yr (mean SEM) Age at menopause, yr (mean SEM) Bone mineral density, z score (mean SEM) LS TS FN Antibody-positive results (%) AGA IgA AGA IgG EmA IgA LS, lumbar spine; TS, total skeleton; FN, femoral neck; AGA, antigliadin antibodies; EmA, endomysial antibodies; Ig, immunoglobulin. a P 0.05; b P 0.004; c P fractures among 1021 CD patients. 14 Even though comparison among studies is not possible, we suggest that this disparity in results may be due to differences among the study populations (in terms of clinical behavior) or to different methodologies used in the studies. In this context, our former study was based on a population of patients diagnosed and evaluated more than a decade ago, when suspicion and diagnosis of CD were largely focused in patients with classic malabsorptive symptoms and when less severe cases were rarely included. 7 Therefore, an overrepresentation of severe (classically symptomatic) cases limits the generalization of the former risk estimation. Thus, we hypothesized that the risk of fracture in CD should be categorized according to the severity of the clinical compromise of patients. The present study showed that fractures of the peripheral skeleton are a real problem for CD patients. However, this observation seems to be restricted to a subgroup of cases. Thus, we observed that, compared with an age- and sex-matched population, CD patients with a classic clinical course (basically, chronic diarrhea and malabsorption) have a significantly higher prevalence of fractures in the peripheral skeleton. It is very interesting to note that patients who present with subclinical or silent forms of the disease have a significantly lower prevalence of fractures compared with classically symptomatic patients and have a rate similar to their own ageand sex-matched controls. Once again, we detected that compared with control subjects, significantly more fractures in patients were produced as a consequence of mild trauma. An interesting finding of our study was the significantly lower age of subclinical/silent patients compared with classically symptomatic patients. This observation might suggest a bias because this subgroup did not have enough follow-up time when they were at risk for fractures. However, this does not seem to be the case when we make a detailed analysis of the prevalence of fractures according to decades of life (Fig. 1C and D). It is interesting to note that this analysis shows that subclinical/silent cases have a reduced prevalence of fractures in all decades and that these patients have very few fractures produced in the fourth and fifth decades of life. Both findings suggest that the low prevalence of fractures in this subgroup seems to be a constant phenomenon that occurs throughout the life span. Some findings of this study support the hypothesis that the increased prevalence of bone fractures is a consequence of bone weakness induced by the degree of activity of CD. In this context, the most relevant findings concordant with this pathogenesis are, first, a significantly higher number of fractures as a consequence of mild trauma and, second, that the bone mineral density measurement, for bones of plane architectural design, was significantly lower for classically symptomatic patients in comparison with subclinical/silent cases. The increased risk for bone fracture in celiac patients with classic symptoms should be viewed in concordance with the prevalence of the metabolic disturbances and the bone damage severity (measured by densitometry) and correlated with the clinical presenting pattern of CD. Thus, compared with classically symptomatic pa-

7 February 2004 RISK OF FRACTURES IN CELIAC DISEASE 133 diagnosed patients have very few clinical features or even a silent course. 10,11 Even though the mechanisms involved in bone weakness in CD patients are not completely established, pathophysiological findings seem to support our assumption of a direct relationship among the degree of clinical compromise, the magnitude of immunopathologic features, and the propensity for bone fractures. In concordance, the presence of impaired calcium and vitamin D absorption (the metabolic pathway) was shown to be lower in subclinical/silent cases. 5,6,19 Furthermore, the degree of immunologic compromise (in terms of T- and B-dependent mechanisms) is less severe in the less affected patients compared with classically symptomatic patients. 20,21 In conclusion, our case-control, cross-sectional study based on a population of CD patients that included the widest clinical spectrum of the disease confirmed that most patients with severe CD have a significantly increased fracture risk. Furthermore, the clinical manifestations of the disease seem to be a better predictive factor than that provided by bone mass measurements. We did not find a special skeletal localization for the excess of fractures in patients compared with controls. We hypothesize that this increased fracture risk might be related to the magnitude of the inflammatory state that occurs in individual patients. Our data question the necessity and effectiveness of screening for fracture risk in the entire CD population by using bone mass measurement. We suggest that this procedure be restricted to patients with a classically symptomatic clinical course. Figure 2. Kaplan Meier estimates of the proportion of patients free of fractures in the CD population and controls (A), classically symptomatic patients and their controls (B), and subclinical/silent cases and their controls (C). NS, not significant. tients, bone mineral density in CD was shown to be significantly less affected in asymptomatic patients (e.g., relatives diagnosed by screening), 3 patients with subclinical forms, 15 treated CD patients, 15 suboptimally treated individuals, 17 and untreated patients with dermatitis herpetiformis. 18 Therefore, we believe that the relative preponderance of the problem must be balanced with the estimated prevalence of classically symptomatic patients in the overall CD population. Recent epidemiological studies have shown that for each classically symptomatic patient detected in the general population, 8 to 10 other References 1. Caraceni M, Molteni N, Bardella M, Ortolani S, Nogara A, Bianchi PA. Bone and mineral metabolism in adult celiac disease. Am J Gastroenterol 1988;83: Valdimarsson T, Toss G, Ross I, Lofman O, Strom M. Bone mineral density in coeliac disease. Scand J Gastroenterol 1994; 29: Mazure R, Vazquez H, Gonzalez D, Mautalen C, Pedreira S, Boerr L, Bai JC. Bone mineral affection in asymptomatic adult patients with celiac disease. Am J Gastroenterol 1994;89: McFarlane XA, Bhalla AK, Reeves DE, Morgan LM, Robertson DAF. Osteoporosis in treated adult coeliac disease. Gut 1995; 36: Corazza GR, Di Sario A, Cecchetti L, Tarozzi C, Corrao G, Bernardi M, Gasbarrini G. Bone mass and metabolism in patients with celiac disease. Gastroenterology 1995;109: Gonzalez D, Mazure R, Mautalen C, Vazquez H, Bai J. Body composition and bone mineral density in untreated and treated patients with celiac disease. Bone 1995;16: Vazquez H, Mazure R, Gonzalez D, Flores D, Pedreira S, Niveloni S, Smecuol E, Mauriño E, Bai JC. Risk of fractures in celiac disease patients: a cross-sectional, case-control study. Am J Gastroenterol 2000;95: Marsh MN. Bone disease and gluten sensitivity. Time to act, to treat and to prevent. Am J Gastroenterol 1994;89:

8 134 MORENO ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No Cooke WT, Holmes GKT. Clinical presentation. In: Cooke WT, Holmes GKT, eds. Coeliac disease. London: Churchill Livingstone, 1984: Catassi C. Screening of coeliac disease. In: Maki M, Collin P, Visakorpi JK, eds. Coeliac disease. Proceedings of the 7th International Symposium on Coeliac Disease. Tampere, Finland: Celiac Disease Study Group, 1997: Gomez JC, Selvaggio G, Viola M, Pizarro B, La Motta G, De Barrio S, Castelletto R, Echeverría R, Sugai E, Vazquez H, Mauriño E, Bai JC. Prevalence of celiac disease in Argentina: screening of an adult population in the La Plata area. Am J Gastroenterol 2001; 96: Mustalahti K, Collin P, Sievanen H, Salmi J, Maki M. Osteopenia in patients with clinically silent celiac disease warrants screening. Lancet 1999;354: Thomason R, Coupland CA, Holmes GKT, Logan RFA. Do coeliacs suffer more fractures than the general population: a populationbased survey of the fracture experience of older celiacs (abstr). Gut 1997;41:A Vestergaard P, Mosekilde L. Fracture risk in patients with celiac disease, Crohn s disease, and ulcerative colitis: a nationwide follow-up study of 16,416 patients in Denmark. Am J Epidemiol 2002;156: Bai JC, Gonzalez D, Mautalen C, Mazure R, Pedreira S, Vazquez H, Smecuol E, Siccardi A, Cataldi M, Niveloni S, Boerr L, Mauriño E. Long-term effect of gluten restriction on bone mineral density of patients with celiac disease. Aliment Pharmacol Ther 1997; 11: Smecuol E, Mauriño E, Vazquez H, Pedreira S, Niveloni S, Mazure R, Boerr L, Bai JC. Gynecological and obstetric disorders in coeliac disease. Frequent clinical onset during pregnancy and puerperium. Eur J Gastroenterol Hepatol 1996;8: Walters JRF, Banks LM, Butcher GP, Fowler CR. Detection of low bone mineral density by dual energy X ray absorptiometry in unsuspected suboptimally treated celiac disease. Gut 1995;37: Mautalen C, Gonzalez D, Mazure R, Vázquez H, Lorenzetti MP, Mauriño E, Niveloni S, Pedreira S, Smecuol E, Boerr LA, Bai JC. Effect of treatment on bone mass, mineral metabolism, and body composition in untreated celiac disease patients. Am J Gastroenterol 1997;92: Mora S, Barera G, Beccio S, Proverbio MC, Weber G, Bianchi C, Chiumello G. Bone density and bone metabolism are normal after long term gluten-free diet in young celiac patients. Am J Gastroenterol 1999;94: Fornari MC, Pedreira S, Niveloni S, Gonzalez D, Diez R, Vazquez H, Mazure R, Sugai E, Smecuol E, Boerr L, Mauriño E, Bai JC. Preand post-treatment serum levels of cytokines IL-1, IL-6, and IL-1 receptor antagonist in celiac disease. Are they related to the associated osteopenia? Am J Gastroenterol 1998;93: Sugai E, Cherñavsky A, Pedreira S, Smecuol E, Vazquez H, Nivelini S, Mazure R, Mauriño E, Ravinovich G, Bai JC. Bone-specific antibodies in sera from patients with celiac disease: characterization and implications in osteoporosis. J Clin Immunol 2002; 22: Address requests for reprints to: Julio C. Bai, M.D., Department of Medicine, Dr. C. Bonorino Udaondo Gastroenterology Hospital, Av. Caseros 2061, 1264 Buenos Aires, Argentina. jbai@intramed. net.ar; fax: (54) The authors thank Christina Surawicz, from the University of Washington, for her generous help in preparing the manuscript. This study was partially funded by the 2002 Research Award Ramón Carrillo-Arturo Oñativia from the Ministerio de Salud, Comisión Nacional de Programas de Investigación Sanitaria, Argentina.