Etiologies and Predictors of Diagnosis in Nonresponsive Celiac Disease DANIEL A. LEFFLER, MELINDA DENNIS, BRIAN HYETT, EOIN KELLY, DETLEF SCHUPPAN, and CIARAN P. KELLY Department of Gastroenterology, Beth Israel Deaconess Medical Center, Boston, Massachusetts See Barone MV et al on page 1245 for companion article in the April 2007 issue of Gastroenterology. Background & Aims: Nonresponsive celiac disease (NRCD) is a common problem affecting from 7% to 30% of celiac patients. Because NRCD comprises varied and poten- tially morbid entities, efficient and cost-effective patient care requires knowledge of the specific causes of this dis- order. The aim of this study was to determine the common etiologies of NRCD in a tertiary referral center. Methods: All cases of biopsy examination–proven celiac disease (CD) seen at our institution over the preceding 5 years were included in this study. NRCD was defined as a failure to respond to at least 6 months of treatment with a gluten-free diet or the re-emergence of symptoms or laboratory abnor- malities typical of CD while still on treatment with a gluten- free diet. Results: A total of 113 patients with NRCD meeting the earlier-described criteria were seen from a total of 603 patients with CD (19%), however, among patients for whom we provided primary specialist care the incidence of NRCD was 10% (P < .001). Gluten expo- sure was the most common cause of NRCD (36%), fol- lowed by irritable bowel syndrome (22%), refractory CD (10%), lactose intolerance (8%), and microscopic colitis (6%). The mean immunoglobulin A tissue transglutami- nase level in the gluten-exposed group was 67 vs 17 U/mL (normal, <20) for other diagnoses (P < .05). Weight loss and male sex were highly predictive of refractory CD (P < .05 and < .001, respectively). Conclusions: NRCD is a common phenomenon affecting 10%–19% of celiac patients. A limited number of etiologies account for the majority of cases. Clinical factors may be used to guide evaluation. C eliac disease (CD) is a small-intestinal inflammatory dis- ease defined by characteristic histologic changes including villous atrophy and increases in intraepithelial lymphocytes. CD is triggered by gluten proteins from wheat, rye, and barley, in genetically predisposed individuals who carry the human lymphocyte antigen (HLA)-DQ2 or -DQ8. 1 Mostly because of the availability of accurate serology-based tests, 2–5 increasing numbers of individuals are being diagnosed with CD, therefore clinicians must be aware of the frequency and common etiolo- gies of incomplete response to gluten withdrawal. Although the majority of individuals with CD have substan- tial improvement within the first few weeks of gluten with- drawal, between 7% and 30% continue to have symptoms or clinical manifestations suggestive of CD despite being on a gluten-free diet. 6,7 This clinical problem, which encompasses many distinct diagnoses, is known as nonresponsive celiac dis- ease (NRCD). NRCD may be defined further as primary if there is initial failure to respond to a gluten-free diet or secondary if signs, symptoms, or laboratory abnormalities consistent with CD re-emerge after initial normalization while maintaining a gluten-free diet (Table 1). In smaller studies the most common cause of NRCD was unintentional gluten intake, accounting for approximately 50% of cases of NRCD, 6,8 but etiologies can vary greatly to include lym- phoma, 9 small-intestinal bacterial overgrowth (SIBO), 10,11 micro- scopic colitis, 12 pancreatic insufficiency, 8,12 disaccharidase defi- ciency, 12 and irritable bowel syndrome (IBS), 8 all of which may present similarly but require very different therapies. Because NRCD comprises a number of potentially severe conditions with disparate treatment and prognosis, efficient and cost-effective care of patients with this syndrome may be challenging. In contrast to prior investigations, our data allow differen- tiation of patients referred from outside practices and those receiving primary gastroenterology care though our referral center. In addition, the large number of consecutive patients evaluated allows for a more definitive determination of clinical characteristics, especially as correlated to refractory CD. We sought, therefore, to better define the prevalence of NRCD in current clinical practice in the United States, to identify the range of specific etiologies for this disorder, and to determine which, if any, clinical factors are predictive of the final etiology. Methods A database of all patients seen at our institution from January 1, 2000, to April 1, 2006, coded for CD under the International Classification of Diseases 9th edition code 579.0 was compiled and predetermined clinical data were recorded. From this list, 603 patients were found to have biopsy exami- nation–proven CD. Individuals without definitive evidence of CD in the form of duodenal biopsy examination, or skin biopsy examination in cases of dermatitis herpetiformis, were not included in this study. HLA typing was performed for patients in whom there was doubt regarding the validity of the diagno- sis. Analysis of tissue transglutaminase (tTG) titers was per- formed by enzyme-linked immunosorbent assay with recombi- nant human antigen (INOVA Quanta Lite human-tTG immunoglobulin [Ig]A; San Diego, CA: sensitivity, 94%; speci- ficity, 99%). For patients with CD, electronic medical records that included all clinician notes, laboratory data, and results of Abbreviations used in this paper: CD, celiac disease; EATL, enterop- athy associated T-cell lymphoma; ELISA, enzyme-linked immunosor- bent assay; HLA, human lymphocyte antigen; IBS, irritable bowel syndrome; Ig, immunoglobulin; NRCD, nonresponsive celiac disease; SIBO, small intestinal bacterial overgrowth; tTG, tissue transglutami- nase. © 2007 by the AGA Institute 1542-3565/07/$32.00 doi:10.1016/j.cgh.2006.12.006 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:445– 450