Celiac disease (CD), also known as gluten-sensitive enteropathy

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:567 573 Correlation of Duodenal Histology With Tissue Transglutaminase and Endomysial Antibody Levels in Pediatric Celiac Disease MATTHEW R. DONALDSON,* SEAN D. FIRTH, HOLLY WIMPEE,* KRISTIN M. LEIFERMAN,* JOHN J. ZONE,* WYATT HORSLEY,* MOLLY A. O GORMAN, W. DANIEL JACKSON, SUSAN L. NEUHAUSEN, CHRISTOPHER M. HULL,* and LINDA S. BOOK *Department of Dermatology and Division of Pediatric Gastroenterology, Department of Pediatrics, University of Utah, Salt Lake City, Utah; and Division of Epidemiology, Department of Medicine, University of California, Irvine, California Background & Aims: IgA antibodies against tissue transglutaminase (TTGA) and endomysium (EMA) are sensitive and specific markers for celiac disease (CD). Data correlating TTGA and EMA levels with degree of villous atrophy are limited. We compared duodenal histopathology in pediatric CD patients with TTGA and EMA serologies, symptoms, height, and weight. Methods: We identified 117 pediatric patients retrospectively who had serologic testing for IgA TTGA and IgA EMA and duodenal biopsies graded by modified Marsh criteria as 0 3c. Data were analyzed with Spearman rank correlation and multinomial logistic regression. Results: IgA TTGA (r.704, P <.001) and IgA EMA (r 0.740, P <.001) correlated with intestinal villous atrophy in pediatric CD patients by Spearman rank correlation. Similar correlations were found in a subset of 23 patients younger than 3 years of age. Multinomial logistic regression revealed increased probability of Marsh 3a or greater changes with increasing TTGA or EMA levels. Strongly positive antibody levels (TTGA >100 units or EMA titer >1:1280) were highly specific (>98%) for Marsh 3a or greater lesions. Among symptoms, abdominal distention and diarrhea were associated with abnormal histology. Conclusions: IgA TTGA and EMA levels correlate with duodenal villous atrophy in pediatric CD patients. IgA TTGA >100 or EMA >1:1280 were nearly always associated with CD histopathology. With further validation of this observation, strongly positive titers might be considered sufficient for diagnosis of pediatric patients at risk for CD. Symptoms, height, and weight are not reliable predictors of CD. Celiac disease (CD), also known as gluten-sensitive enteropathy (GSE), is an immune-mediated disease primarily affecting the small bowel. In genetically susceptible individuals, consumption of gluten triggers an inflammatory reaction with intraepithelial lymphocyte infiltration, crypt hyperplasia, and villous atrophy of the small bowel mucosa. 1 In children, these changes might cause malabsorption, diarrhea, abdominal distention, and abdominal pain, a broad range of non-classical findings, or be clinically silent. 2 4 Down syndrome, family history of GSE, and autoimmune disease such as type 1 diabetes mellitus (DM) and thyroiditis are associated with CD. 5,6 If left untreated, patients are at increased risk for gastrointestinal lymphoma and other malignancies. 7 Treatment with a glutenfree diet (GFD) achieves complete reversal of disease and eliminates increased risk of malignancy in most patients. 7 10 Estimates of disease prevalence vary, with Northern European or descendent populations having the highest rate at around 1%. 11,12 Data from Fasano et al 13,14 indicate that disease prevalence in average risk, predominantly white children in the United States is 1/320 and much higher in those at risk. Because of its variable, nonspecific, and often silent presentation, diagnosis of CD can be challenging. Characteristic histopathologic changes of the small bowel mucosa by biopsy are considered the gold standard of diagnosis. 5 Specimens are graded as 0 3c according to the modified Marsh criteria, with 3a or greater clearly consistent with CD, Marsh 2 possibly consistent, and Marsh 1 indicating nonspecific changes. 3 The gluten-triggered inflammatory response leads to activation of the humoral pathway and production of autoantibodies. 1,6 Assays to detect serum antibodies against tissue transglutaminase (TTGA), endomysium (EMA), and gliadin are used to screen patients for biopsy and follow response to treatment. 4,5 IgA anti-human recombinant TTGA (TTGA-HR) detected by enzyme-linked immunosorbent assay (ELISA) and IgA EMA seen on monkey esophagus (ME) or human umbilical cord (HUC) substrates by indirect immunofluorescence (IIF) are considered the most sensitive and specific serologic tests for CD. 15,16 In children, IgA TTGA-HR and IgA EMA-ME have reported diagnostic sensitivities and specificities exceeding 95%. 15,16 Patients with mild or intermediate mucosal changes, as might be observed in early CD, are frequently seronegative. This lowers the sensitivity of serology in practice. 17 20 It is well-established that increased EMA and TTGA in both children and adults correlate with abnormal small bowel histopathology. 10,17,19,21 25 Less data relating serologic levels to degree of villous atrophy are available, especially in the pediatric pop- Abbreviations used in this paper: CD, celiac disease; CI, confidence interval; DM, diabetes mellitus; EGD, esophagoduodenoscopy; ELISA, enzyme-linked immunosorbent assay; EMA-HUC, endomysial antibodies on human umbilical cord substrate; EMA-ME, endomysial antibodies on monkey esophagus substrate; GFD, gluten-free diet; GSE, gluten-sensitive enteropathy; IIF, indirect immunofluorescence; MLR, multinomial logistic regression; NASPGHAN, North American Society for Pediatric Gastroenterology Hepatology and Nutrition; ROC, receiver operating curve; TTGA-GP, tissue transglutaminase antibodies guinea pig; TTGA-HR, tissue transglutaminase antibodies human recombinant. 2007 by the AGA Institute 1542-3565/07/$32.00 doi:10.1016/j.cgh.2007.01.003

568 DONALDSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 5 ulation. Tursi et al 19 studied 119 adults and observed that the prevalence of TTGA and the mean serum value were higher in CD patients with higher grade. In a study of 30 children, Hoffenberg et al 24 reported a correlation of r 0.569 of TTGA titer and Marsh score. Information about the predictive value of clinical presentation is also limited. However, up to 1 in 33 patients referred to pediatric gastroenterology for symptoms or diseases associated with CD will have CD. 14 To investigate these relationships, we conducted a retrospective chart review of patients referred to a pediatric gastroenterology practice to be evaluated for CD. With multinomial logistic regression (MLR), we correlated histopathology from distal duodenal biopsies with IgA TTGA-HR and IgA EMA levels, height, weight, and symptoms at presentation. Materials and Methods Patients Charts of patients evaluated at a pediatric gastroenterology clinic at Primary Children s Medical Center (Salt Lake City, UT) for possible CD from 1999 2005 were reviewed. Patients were referred for evaluation because of gastrointestinal or systemic complaints suggestive of CD, family history of GSE, recent diagnosis of type 1 DM or other associated autoimmune disease, or having Down syndrome. All patients were 18 years or younger, had no previous diagnosis of CD, and were not on a GFD. Subjects were included if duodenal biopsy with pathologic interpretation, serum, and clinical records were available. After undergoing history and physical examination by a boardcertified pediatric gastroenterologist, consent from responsible guardians was obtained for esophagogastroduodenoscopy (EGD). Before or at the time of biopsy, patients had blood drawn for serologic testing. Height and weight were obtained from clinic visits or procedure records at the time of biopsy. These were converted to approximate percentiles by using a standard growth curve for age and sex and adjusted for Down syndrome as appropriate. Symptoms of abdominal pain, abdominal distention, diarrhea, constipation, and vomiting were recorded as present or absent on the basis of pre-endoscopy clinic records. Esophagogastroduodenoscopy and Pathology Endoscopy was performed with a minimum of 4 forceps biopsies of the distal duodenum. 26 Specimens were oriented before embedding by an experienced technician. Fourmicrometer sections were stained with hematoxylin-eosin and examined by 1 of 4 board-certified pathologists experienced in the diagnosis of CD and without knowledge of serologic findings. A modified Marsh grade was assigned to each specimen: 0, normal; 1, increased intraepithelial lymphocytes; 2, increased intraepithelial lymphocytes plus hyperplastic crypts; 3a, partial villous atrophy; 3b, subtotal villous atrophy; 3c, complete villous atrophy. 27,28 If specimens demonstrated features of multiple grades, the higher grade was recorded. We used the recent North American Society for Pediatric Gastroenterology, Hepatology and Nutrition (NASPGHAN) guidelines for assignment of a diagnosis of CD. 3 We considered all Marsh 3 lesions with positive serology consistent with a diagnosis of CD. Patients with Marsh 3 lesions and negative serology were considered to have CD if there were a clinical response to a GFD and no alternative diagnosis to account for the histologic abnormalities. Marsh 2 histology was considered CD in the context of positive serology. Marsh 1 lesions were considered nonspecific but possibly consistent with CD if serology was positive and symptoms responded to a GFD. Serology Serum IgA TTGA and EMA were measured by using commercial ELISA and IIF, respectively. Serum was obtained from whole blood by centrifugation at 3000 rpm for 10 minutes and frozen at 80 C with sodium azide. Semiquantitative detection of TTGA was performed by using the QUANTA Lite human recombinant TTGA IgA and IgG kits (INOVA Diagnostics, San Diego, CA). Per manufacturer recommendations, positivity was assigned at 20 units. IgA EMA were detected on cryostat sections of monkey esophagus by using fluorescein isothiocyanate conjugated anti-iga (Cappel, Cochranville, PA) as described previously. 29 Initial serum testing was performed at 1:5, 1:10, and 1:20 dilutions. When positive, IIF with serial 2-fold dilutions to 1:40,960 was carried out, and the limiting dilution, or end-point titer with positive staining, was recorded. Increased EMA were defined at 1:10 dilution or greater showing smooth muscle endomysial staining on monkey esophagus. Patients sera with positive staining at 1:5 but negative at 1:10 were considered negative. All slides were interpreted by a single, blinded, and experienced immunodermatologist (K.M.L.). Total serum IgA was determined by nephelometry by ARUP Laboratories (Salt Lake City, UT) or MININEPH human IgA kit (The Binding Site Ltd, Birmingham, UK). Per manufacturer s recommendations, patient serum was diluted to 1:11 and assayed on a mininephelometer. Serum IgA levels were recorded in g/dl. All values were referenced to manufacturer-published norms by age to evaluate for IgA deficiency. Patients with IgA deficiency were screened for IgG TTGA by using the INOVA kit as described above. 3,30,31 Statistics Data were analyzed by using Stata v9.2 (College Station, TX). For statistical analysis, EMA end-point titers were converted to an ordinal scale, with each incremental increase representing the next 2-fold dilution. Negative IIF at a 1:5 dilution was considered 0, positive (or indeterminate) staining at 1:5 was labeled a 1, positive staining at 1:10 was 2, positive at 1:20 was 3, and so on until 1:40,960 was 14. Spearman rank correlation was used to test the associations between Marsh scores and IgA EMA end-point titers and units of TTGA. Ninety-five percent confidence intervals (CIs) were calculated for Spearman rho. MLR was used to determine associations between histopathologic and serologic or clinical data. Logistic regression and post-estimation analyses of receiver operating curve (ROC), sensitivity, specificity, positive predictive values, and negative predictive values were used to evaluate cut points for TTGA as a predictor of Marsh scores. Results Patient Features A total of 117 pediatric patients who had clinical evaluation, serologic testing, and EGD with duodenal biopsy comprised our study group. We identified 80 patients with Marsh 1 or greater lesions and included 37 Marsh 0 patients evaluated during the same time period. Table 1 describes patient demo-

May 2007 DUODENAL HISTOLOGY, TTGA, AND EMA IN PEDIATRIC CD 569 Table 1. Patient Demographics, Comorbid Disease, and Symptoms Variable Marsh 0 Marsh 1 2 Marsh 3a c Number 37 24 56 Male 20 9 28 Female 17 15 28 Age (y) (mean, SD) 8.2 5.1 9.8 4.9 7.9 4.7 FHx of GSE 5 (14%) 3 (13%) 17 (30%) Comorbid disease Type 1 DM 1 (3%) 2 (8%) 14 (25%) Down syndrome 1 (3%) 0 4 (7%) Thyroiditis 0 0 2 (4%) JRA 1 (3%) 0 0 Williams syndrome 0 0 1 (2%) Autism/Asperger s 1 (3%) 2 (8%) 0 Symptom Diarrhea 18 (49%) 9 (38%) 37 (66%) Pain 20 (54%) 17 (71%) 36 (64%) Constipation 10 (27%) 7 (29%) 18 (32%) Vomiting 10 (27%) 6 (25%) 10 (18%) Distention 7 (19%) 7 (29%) 26 (46%) Height percentile 35% 28% 37% 31% 45% 28% (mean, SD) Weight percentile (mean, SD) 35% 33% 53% 35% 44% 32% Marsh 3b and most 3c patients had strongly positive IgA TTGA. Marsh 3a patients had the greatest variability with negative, positive, and strongly positive values represented (Table 2). Characteristics of Marsh 3 patients with negative serology are shown in Table 3. Of these 7 patients, 4 had confirmed CD by symptoms and response to a GFD, whereas 3 others had non- GSE disease. Analysis by MLR revealed a statistically higher probability of CD with increasing IgA TTGA-HR. Each unit of increase of TTGA-HR increased the risk of having a Marsh 3a, 3b, or 3c lesion by approximately 8%, 14%, or 10%, respectively. As shown in Table 4, the sensitivity of IgA TTGA-HR to predict Marsh 2 or higher or Marsh 3 lesions decreases as IgA TTGA-HR cut points increase. Conversely, as IgA TTGA-HR cut point increases, so do specificity and PPV. As shown in Figure 2, ROC analysis yields the highest area under the curve at the IgA TTGA-HR cut point of 40. As IgA TTGA-HR cut points increase above 40, the area under the ROC curves decreases. An IgA TTGA-HR cut point of 100 units yields 100% specificity and positive predictive value but 0% sensitivity. NOTE. Age, family history, associated disease, and symptom data were collected on all 117 patients, whereas 114 had for-age height and 115 had for-age weight percentiles recorded. SD, standard deviation; FHx of GSE, family history of GSE; JRA, juvenile rheumatoid arthritis. graphics, comorbid conditions, and symptoms by Marsh score. MLR analyses of associated conditions, symptoms, height, and weight related to Marsh grade were performed. Patients with type 1 DM had a significantly increased probability of Marsh 3b and 3c lesions (P.02). A family history of GSE was significantly associated with Marsh 3c lesions (P.05). There were no differences in height or weight percentiles for Marsh 3 patients compared with Marsh 0. Abdominal distention and diarrhea were statistically more likely in patients with Marsh 3 lesions, but there was no significant association with constipation, vomiting, or pain (data not shown). Immunoglobulin A Tissue Transglutaminase Antibodies, Human Recombinant and Marsh Grade One hundred seventeen patients had IgA TTGA-HR ELISA, with units ranging from 3 175 as plotted in Figure 1A. Mean units for each Marsh group are shown in Table 2. IgA TTGA significantly correlated with Marsh grade (Spearman rho 0.704, P.001; 95% CI, 0.599 0.785). Stratification of Marsh grades into quantitative TTGA groups (negative, 0 19 units; positive, 20 99 units; strongly positive, 100 units) is also shown in Table 2. Seven patients with Marsh 0 1 lesions were TTGA-positive, and none had 100 units. Details regarding these patients are presented in Table 3. One Marsh 1 patient very likely had CD on the basis of response to a GFD, and 2 patients likely had latent CD on the basis of a strong family history. One of 5 Marsh 2 patients had positive serology (38 units). This patient likely had CD because his abdominal pain and bloating resolved and weight increased on a GFD. All Figure 1. Scatter plots of IgA TTGA and EMA by Marsh grade. (A) IgA TTGA-HR units are plotted on the y-axis against Marsh grade along the x-axis. (B) IgA EMA titers are plotted on the y-axis against Marsh grade on the x-axis.

570 DONALDSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 5 Table 2. IgA TTGA and EMA by Marsh Group IgA TTGA-HR IgA EMA-ME Marsh N Mean units SD N, 20 U N, 20 to 100 U N, 100 U Median titer N, 1:10 N, 1:10 to 1:1280 N, 1:1280 0 37 7 6 35 (95%) 2 (5%) 0 0 32 (86%) 5 (14%) 0 1 19 14 18 14 (74%) 5 (26%) 0 0 13 (68%) 5 (26%) 1 (5%) 2 5 13 14 4 (80%) 1 (20%) 0 0 4 (80%) 1 (20%) 0 3a 22 82 58 5 (23%) 8 (36%) 9 (41%) 1:320 6 (27%) 6 (27%) 10 (45%) 3b 16 166 17 0 0 16 (100%) 1:10240 0 1 (6%) 15 (94%) 3c 18 122 65 3 (17%) 2 (11%) 13 (72%) 1:5120 2 (11%) 1 (6%) 15 (83%) SD, standard deviation. Immunoglobulin A Endomysial Antibodies and Marsh Grade One hundred seventeen patients had IgA EMA-ME with titers ranging from negative to positive at 1:40,960 as plotted in Figure 1B. Median titers are shown in Table 2. IgA EMA significantly correlated with Marsh grade (Spearman rho 0.740, P.001; 95% CI, 0.645 0.813). Patients from each Marsh group were quantitatively stratified by EMA titer (negative, 1:10; positive, 1:10 1:640; strongly positive, 1:1280) in Table 2. Table 3 describes Marsh 0 1 patients who were IgA EMA positive. One had a reproducible, atypical immunofluorescence pattern. Five of six Marsh 1 patients with increased EMA were also IgA TTGA positive. The Marsh 2 patient who responded to a GFD described above was both IgA EMA and TTGA positive. Nearly all Marsh 3b and 3c patients were strongly positive for EMA. Again, Marsh 3a patients showed the broadest distribution among negative, positive, and strongly positive groups. Analysis by MLR revealed a significantly higher probability of Marsh 3 lesions with increasing IgA EMA titers. EMA positivity at each successive 2-fold dilution increased the risk of having a Marsh 3a, 3b, or 3c lesion by approximately 54%, 130%, or 93%, respectively. Sensitivity of IgA EMA with a positive cut off of 1:10 was 73% for Marsh 3a, 100% for 3b, and 89% for 3c lesions. At a higher cut off of 1:1280, sensitivity was comparable for Marsh 3b (94%) or 3c (83%) lesions and decreased for 3a (45%). The specificity of the higher cut off for Marsh 3 lesions was 98%. As shown in Table 3, Marsh 3 patients negative for IgA EMA showed a strong concordance with IgA TTGA. Effect of Age on Histopathologic Association With Immunoglobulin A Tissue Transglutaminase Antibodies and Endomysial Antibodies Twenty-three of 117 patients were younger than 3 years of age at the time of evaluation. Spearman rank correlation comparing Marsh grade with serology yielded coefficients of 0.787 for TTGA (P.001; 95% CI, 0.555 0.905) and 0.737 for Table 3. Patients With Discordant Serology and Histology Patient Age (y) Marsh IgA EMA-ME (titer) IgA TTGA (U) GI Sx Background GFD response Dx Biopsy, serology 1 7 0 1:5 34 No FHx Latent CD 2 7 0 1:5 22 Yes No CD 3 1 0 1:80 3 Yes Down None No CD 4 13 0 1:320 3 Yes No CD 5 12 0 1:10 19 Yes EE 6 3 0 1:320 14 No FHx Latent CD 7 1 0 1:20 3 Yes EG 8 6 1 1:80 32 Yes EE 9 13 1 1:160 16 Yes FHx Latent CD 10 14 1 1:2560 68 Yes Improved CD 11 8 1 1:80 35 Yes Unknown 12 11 1 1:80 40 Yes DM Refused Unknown 13 11 1 1:160 20 No FHx Latent CD Biopsy, serology 1 2 3a 1:5 4 Yes Improved CD 2 11 3a 1:5 3 Yes IBD 3 2 3a 1:5 11 Yes Improved CD 4 13 3a 1:5 3 Yes Improved CD 5 4 3a 1:5 3 No None EG 6 2 3c 1:5 4 Yes IBD 7 4 3c 1:5 4 Yes Improved CD NOTE. Titers, presence of symptoms, response to GFD where known, and final diagnosis are indicated. For diagnosis, non-cd disease is specified, or a diagnosis of CD, no CD, latent CD, or unknown was established according to guidelines described in the Methods. GI Sx, gastrointestinal symptoms; Dx, diagnosis; FHx, family history; Down, Down syndrome; EE, eosinophilic esophagitis; EG, eosinophilic gastroenteritis.

May 2007 DUODENAL HISTOLOGY, TTGA, AND EMA IN PEDIATRIC CD 571 Table 4. Sensitivity, Specificity, Positive and Negative Predictive Values, ROC Area for IgATTGA-HR Cut Points for Marsh Scores of 2 or Greater and 3 Marsh score IgATTGA-HR units Sensitivity Specificity PPV NPV ROC area 3 10 89.3% 77.1% 78.1% 88.7% 83.2% 3 20 85.7% 88.5% 87.3% 87.1% 87.1% 3 30 83.9% 90.2% 88.7% 85.9% 87.1% 3 40 82.1% 98.4% 97.9% 85.7% 90.3% 2 10 85.3% 78.6% 81.3% 83.0% 81.9% 2 20 80.3% 89.3% 89.1% 80.7% 84.8% 2 30 78.7% 91.1% 90.6% 79.7% 84.9% 2 40 75.4% 98.2% 97.9% 78.6% 86.8% PPV, positive predictive value; NPV, negative predictive value. EMA (P.001; 95% CI, 0.466 0.881) in this subset. Patients younger than 3 years of age represented all Marsh grades, with serologies as follows: 10 of 11 Marsh 0 2 patients were seronegative for both IgA EMA and TTGA, whereas one Marsh 0 patient was IgA EMA positive and TTGA negative; all eight Marsh 3b or 3c patients were seropositive by both assays. One Marsh 3a patient was TTGA-positive (63 units) and negative for EMA. Two Marsh 3 patients were both EMA- and TTG-negative but met clinical criteria for CD (Table 3). Immunoglobulin A Deficiency Five patients were found to be IgA-deficient for age. These included two Marsh 0, one Marsh 2, one Marsh 3a, and one Marsh 3c patients. No IgA-deficient patients negative for IgA TTGA were positive for IgG TTGA. Although both Marsh 0 patients were negative for IgG and IgA TTGA, one was IgA EMA positive at a titer of 1:20. The Marsh 2 patient was negative for IgG and IgA TTGA and indeterminate for IgA EMA at 1:5. The Marsh 3a IgA-deficient patient was negative for IgA and IgG TTGA and IgA EMA. The only IgA-deficient Marsh 3c patient was positive by all 3 assays. Discussion According to recently published guidelines, serologic testing is of primary importance in screening for CD. 3 We found that increasing IgA TTGA-HR units or EMA-ME titers correlated with a higher probability of partial to complete villous atrophy in 117 pediatric patients. A similar correlation was present in children younger than 3 years, a population in whom the reliability of serology has been questioned. These findings mirror previously reported correlations between IgA TTGA-GP (guinea pig) titers and Marsh grade found in 119 adult and 30 pediatric patients. 19,24 In another study of 47 patients ranging from 3 73 years of age, increasing TTGA-GP levels and EMA-HUC titers were associated with more severe villous atrophy.25 The difficulty in making generalizations from these studies lies in lack of assay standardization. With IIF to detect EMA, results might vary with substrate type (ME vs HUC), source, and observer. 3,32 The IgA TTGA-GP ELISA used by Tursi et al has a slightly lower sensitivity and specificity compared with IgA TTGA-HR, with TTGA-HR now recognized as the standard. 16,33 However, more than one commercial IgA TTGA-HR ELISA is available. Our findings indicate both a concordance between IgA TTGA-HR and IgA EMA and a significant correlation between each assay and Marsh grade in pediatric patients. Establishing a correlation between quantitative serology and small bowel histopathology has considerable clinical utility. When considering arbitrarily chosen cutoffs of IgA TTGA 100 units or EMA 1:1280, we observed 100% and 98% specificity for Marsh 3a or higher lesions, respectively. A modified diagnostic algorithm suggesting strongly positive IgA TTGA-GP ( 100 units) be considered diagnostic without biopsy was recently proposed. 34 The authors of this proposal found that 48 of 49 patients (98%) with IgA TTGA 100 units had Marsh 2 or greater lesions. Given the sample size, risk of misdiagnosis, and morbidity of the GFD, others disagree. 35 TTGA-GP 100 units has been reported in 2 patients with normal duodenal biopsies and a third with CD histolopathology but no response to a GFD. 36 In addition, patients with Crohn s disease, ulcerative colitis, primary biliary cirrhosis, and autoimmune hepatitis can have TTGA, although markedly increased levels have not been reported. 37,38 These latter studies suggest that non-cd patients positive for TTGA are EMA-negative. In our sample of 117 patients, IgA TTGA 100 was present in 38 patients and uniformly indicated Marsh 3a or greater lesions on biopsy. In contrast to Barker et al, 34 we used a TTGA-HR assay and considered only Marsh 3a and greater lesions as positive. A similar phenomenon was seen with strongly positive IgA EMA-ME because 40 of 41 patients with a titer 1:1280 had at least Marsh 3a changes on biopsy. One Marsh 1 patient was strongly IgA EMA positive at a titer of 1:2560, IgA TTGA Figure 2. ROC. Depicting maximum area under the curve for Marsh 3 histology at cut point of TTGA-HR 40 units.

572 DONALDSON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 5 positive with 68 units, and had gastrointestinal complaints suggestive of early CD. Small bowel changes in CD can be patchy. 39,40 In rare cases of strongly positive IgA EMA or TTGA with Marsh 2 or lesser lesions, higher grade pathology might have been missed by sampling error. As demonstrated by our patients with discordant biopsies and serologies in Table 3, the diagnosis of CD requires clinical judgment and expertise. Despite serologic, histopathologic, and clinical data, the diagnosis might remain unclear in some patients. Negative serologies occurred in approximately 15% of patients with Marsh 3 lesions. Four of 7 of these were assigned a diagnosis of CD. These patients were generally younger, with 6 of 7 younger than 4 years, and IgA deficiency was present in one. Intermediate Marsh grades and partial villous atrophy have been associated with negative serology previously. 17,18,41 In one study, Dickey et al42 found that one fifth of IgA EMA negative patients had Marsh 3a or greater lesions. Others have reported sensitivities and specificities of IgA TTGA-HR or IgA EMA-ME that were lower than those typically quoted. 17,43 Villous atrophy in the context of negative serology could result from non-gse disease. False-positive biopsies might also result from overinterpretation of specimens or poor specimen orientation. We found that Marsh 2 changes in our pediatric population were uncommon, occurring in only 5 of 117 individuals. Only one of these was assigned a diagnosis of CD. Our findings are consistent with NASPGHAN guidelines indicating that Marsh 2 changes are less specific for CD, and such patients require positive serology and clinical judgment to make the diagnosis. 3 Positive serologic tests with Marsh 0 and 1 changes are also unlikely to indicate CD. Nearly 10% of our patients were in this category. Four asymptomatic patients could be reasonably classified as having latent CD because of strong family histories of GSE. Only one Marsh 1 patient with positive serology convincingly responded to a GFD and was considered to have CD. Guidelines indicate serum IgA should be performed in all patients screened for IgA-class TTGA or EMA. 3 In our population, 2 patients with histologic changes of CD were IgA-deficient. The first, graded as a Marsh 3c on biopsy, had increased IgA EMA, IgA TTGA, and IgG TTGA. The second was Marsh 3a by biopsy and was seronegative for IgG and IgA-class TTGA as well as IgA EMA. Currently, CD is thought to present with severe growth delay infrequently. 44 MLR suggested height or weight for age was not useful to predict Marsh lesions in our patients. Our findings are consistent with those of Hoffenberg et al, 45 who did not observe a difference in weight or height for age in TTGA-positive children. Although bloating and diarrhea were associated with high Marsh grades, we generally found that symptoms or growth parameters cannot reliably predict CD. 17,24 Our data show that IgA TTGA and EMA levels correlate with severity of histologic changes in a population at high risk for pediatric CD. The positive predictive value of serology becomes more favorable and clinically useful in populations with a high prevalence of CD, such as patients with Down syndrome, DM, or a family history of GSE. We found that greatly increased antibody levels predict villous atrophy on biopsy. However, we approach modifying the current diagnostic algorithm with caution. 3 More studies are needed to confirm that markedly elevated IgA TTGA ( 100 units) or IgA EMA ( 1:1280) rarely occurs in the absence of CD. Until then, duodenal biopsy remains the gold standard for confirmation of the diagnosis. When circumstances preclude small intestinal biopsy in patients with strongly positive serology, it might be appropriate and safe to treat directly with a GFD. References 1. Kagnoff MF. Overview and pathogenesis of celiac disease. Gastroenterology 2005;128:S10 S18. 2. Fasano A. Clinical presentation of celiac disease in the pediatric population. Gastroenterology 2005;128:S68 S73. 3. Hill ID, Dirks MH, Liptak GS, et al. Guideline for the diagnosis and treatment of celiac disease in children: recommendations of the North American Society for Pediatric Gastroenterology, Hepatology and Nutrition. J Pediatr Gastroenterol Nutr 2005;40:1 19. 4. Dewar DH, Ciclitira PJ. Clinical features and diagnosis of celiac disease. Gastroenterology 2005;128:S19 S24. 5. Green PH, Jabri B. Celiac disease. Annu Rev Med 2006;57:207 221. 6. Alaedini A, Green PH. Narrative review: celiac disease understanding a complex autoimmune disorder. Ann Intern Med 2005; 142:289 298. 7. Catassi C, Bearzi I, Holmes GK. Association of celiac disease and intestinal lymphomas and other cancers. Gastroenterology 2005; 128:S79 S86. 8. Pink IJ, Creamer B. Response to a gluten-free diet of patients with the coeliac syndrome. Lancet 1967;1:300 304. 9. Collin P, Reunala T, Pukkala E, et al. Coeliac disease associated disorders and survival. Gut 1994;35:1215 1218. 10. Sategna-Guidetti C, Pulitano R, Grosso S, et al. Serum IgA antiendomysium antibody titers as a marker of intestinal involvement and diet compliance in adult celiac sprue. J Clin Gastroenterol 1993;17:123 127. 11. Rewers M. Epidemiology of celiac disease: what are the prevalence, incidence, and progression of celiac disease? Gastroenterology 2005;128:S47 S51. 12. Dube C, Rostom A, Sy R, et al. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology 2005;128:S57 S67. 13. Fasano A, Berti I, Gerarduzzi T, et al. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med 2003;163:286 292. 14. Hill I, Fasano A, Schwartz R, et al. The prevalence of celiac disease in at-risk groups of children in the United States. J Pediatr 2000;136:86 90. 15. Hill ID. What are the sensitivity and specificity of serologic tests for celiac disease? do sensitivity and specificity vary in different populations? Gastroenterology 2005;128:S25 S32. 16. Rostom A, Dube C, Cranney A, et al. The diagnostic accuracy of serologic tests for celiac disease: a systematic review. Gastroenterology 2005;128:S38 S46. 17. Abrams JA, Diamond B, Rotterdam H, et al. Seronegative celiac disease: increased prevalence with lesser degrees of villous atrophy. Dig Dis Sci 2004;49:546 550. 18. Tursi A, Brandimarte G, Giorgetti G, et al. Low prevalence of antigliadin and anti-endomysium antibodies in subclinical/silent celiac disease. Am J Gastroenterol 2001;96:1507 1510. 19. Tursi A, Brandimarte G, Giorgetti GM. Prevalence of antitissue transglutaminase antibodies in different degrees of intestinal damage in celiac disease. J Clin Gastroenterol 2003;36:219 221. 20. Tursi A, Brandimarte G. The symptomatic and histologic response to a gluten-free diet in patients with borderline enteropathy. J Clin Gastroenterol 2003;36:13 17. 21. Szaflarska-Szczepanik A. Assessment of correlation between the presence of antiendomysial antibodies and small intestine mucosal villous atrophy in the diagnostics of celiac disease. Med Sci Monit 2002;8:CR185 CR188.

May 2007 DUODENAL HISTOLOGY, TTGA, AND EMA IN PEDIATRIC CD 573 22. Ozgenc F, Aksu G, Aydogdu S, et al. Association between antiendomysial antibody and total intestinal villous atrophy in children with coeliac disease. J Postgrad Med 2003;49:21 24. 23. Demir H, Yuce A, Kocak N, et al. Celiac disease in Turkish children: presentation of 104 cases. Pediatr Int 2000;42:483 487. 24. Hoffenberg EJ, Bao F, Eisenbarth GS, et al. Transglutaminase antibodies in children with a genetic risk for celiac disease. J Pediatr 2000;137:356 360. 25. Kotze LM, Utiyama SR, Nisihara RM, et al. IgA class anti-endomysial and anti-tissue transglutaminase antibodies in relation to duodenal mucosa changes in coeliac disease. Pathology 2003; 35:56 60. 26. Dandalides SM, Carey WD, Petras R, et al. Endoscopic small bowel mucosal biopsy: a controlled trial evaluating forceps size and biopsy location in the diagnosis of normal and abnormal mucosal architecture. Gastrointest Endosc 1989;35:197 200. 27. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol 1999;11:1185 1194. 28. Marsh MN. Gluten, major histocompatibility complex, and the small intestine: a molecular and immunobiologic approach to the spectrum of gluten sensitivity ( celiac sprue ). Gastroenterology 1992;102:330 354. 29. Unsworth DJ. ACP broadsheet no 149: September 1996 serological diagnosis of gluten sensitive enteropathy. J Clin Pathol 1996;49:704 711. 30. Dahlbom I, Olsson M, Forooz NK, et al. Immunoglobulin G (IgG) anti-tissue transglutaminase antibodies used as markers for IgAdeficient celiac disease patients. Clin Diagn Lab Immunol 2005; 12:254 258. 31. Korponay-Szabo IR, Dahlbom I, Laurila K, et al. Elevation of IgG antibodies against tissue transglutaminase as a diagnostic tool for coeliac disease in selective IgA deficiency. Gut 2003;52: 1567 1571. 32. Russo PA, Chartrand LJ, Seidman E. Comparative analysis of serologic screening tests for the initial diagnosis of celiac disease. Pediatrics 1999;104:75 78. 33. Wong RC, Wilson RJ, Steele RH, et al. A comparison of 13 guinea pig and human anti-tissue transglutaminase antibody ELISA kits. J Clin Pathol 2002;55:488 494. 34. Barker CC, Mitton C, Jevon G, et al. Can tissue transglutaminase antibody titers replace small-bowel biopsy to diagnose celiac disease in select pediatric populations? Pediatrics 2005;115: 1341 1346. 35. Rashid M. Diagnosing celiac disease with a positive serological test and without an intestinal biopsy. Pediatrics 2005;116: 1054 1055. 36. Freeman HJ. Strongly positive tissue transglutaminase antibody assays without celiac disease. Can J Gastroenterol 2004;18: 25 28. 37. Clemente MG, Musu MP, Frau F, et al. Antitissue transglutaminase antibodies outside celiac disease. J Pediatr Gastroenterol Nutr 2002;34:31 34. 38. Di Tola M, Sabbatella L, Anania MC, et al. Anti-tissue transglutaminase antibodies in inflammatory bowel disease: new evidence. Clin Chem Lab Med 2004;42:1092 1097. 39. Bonamico M, Mariani P, Thanasi E, et al. Patchy villous atrophy of the duodenum in childhood celiac disease. J Pediatr Gastroenterol Nutr 2004;38:204 207. 40. Ravelli A, Bolognini S, Gambarotti M, et al. Variability of histologic lesions in relation to biopsy site in gluten-sensitive enteropathy. Am J Gastroenterol 2005;100:177 185. 41. Rostami K, Kerckhaert J, Tiemessen R, et al. Sensitivity of antiendomysium and antigliadin antibodies in untreated celiac disease: disappointing in clinical practice. Am J Gastroenterol 1999; 94:888 894. 42. Dickey W, Hughes DF, McMillan SA. Reliance on serum endomysial antibody testing underestimates the true prevalence of coeliac disease by one fifth. Scand J Gastroenterol 2000;35:181 183. 43. Rostami K, Kerckhaert JP, Tiemessen R, et al. The relationship between anti-endomysium antibodies and villous atrophy in coeliac disease using both monkey and human substrate. Eur J Gastroenterol Hepatol 1999;11:439 442. 44. Catassi C, Fasano A. Celiac disease as a cause of growth retardation in childhood. Curr Opin Pediatr 2004;16:445 449. 45. Hoffenberg EJ, Emery LM, Barriga KJ, et al. Clinical features of children with screening-identified evidence of celiac disease. Pediatrics 2004;113:1254 1259. Address requests for reprints to: Linda Book, MD, Professor of Pediatrics, Chief, Pediatric Gastroenterology, University of Utah, 100 N Medical Dr, Salt Lake City, Utah 84113. e-mail: linda.book@ intermountainmail.org; fax: (801)-588-2375. Supported by NIH Grant DK50678 (S.N.) and Dermatology Foundation Career Development Grant (C.H.). The authors would like to thank all patients for participating in this study. They would also like to thank Marjorie Campbell in the Immunodermatology Laboratory at the University of Utah for technical assistance and personnel in medical records at Primary Children s Medical Center.