Finding the sweet spot: Diagnosing celiac disease in children with type 1 diabetes

The advent of serologic tests has revolutionized how celiac disease (CeD) is detected and—for some—diagnosed. European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) guidelines for serologic and genetic diagnosis of CeD were revised in 20201 with simplified criteria for serologic diagnosis of CeD (i.e., tissue transglutaminase [tTG] IgA level greater than 10 times the upper limit of normal [10× ULN], and a positive endomysial [EMA] IgA titer on a different blood draw). Notably, these criteria do not apply to asymptomatic children (joint decision-making is recommended), nor to children with type 1 diabetes (T1DM) due to insufficient data. These are critical gaps because screening for CeD is recommended for first-degree relatives of CeD patients and individuals with T1DM2 irrespective of symptoms. Consequently, regardless of whether serologic criteria are met, all children with T1DM require an endoscopy to confirm a CeD diagnosis. In this issue, Jessica Rutsky et al.3 report on factors associated with a CeD diagnosis in a cohort of children and adolescents (0–22 years of age) seen at a large pediatric tertiary care center in the United States. Their study coincides with their institutional adoption of a screening protocol in which all T1DM patients were being screened for CeD both at diagnosis and every 2 years. Of the 2179 T1DM patients in their registry, 116 patients underwent an esophagogastroduodenoscopy for evaluation of CeD, with seven patients having more than one endoscopy, yielding a total of 123 surgical cases for analysis. Data, including anthropometrics, symptoms, and laboratory studies at the time of endoscopy, were collected retrospectively, and a logistic regression analysis was utilized to identify risk factors associated with a CeD diagnosis, which was based on the presence of Marsh 2 or 3 histopathology on duodenal biopsies. The authors found that the factors most predictive of CeD were the degree of tTG IgA elevation and the EMA IgA titer. In their cohort, 52/61 patients with a tTG IgA ≥10× ULN had CeD, resulting in a positive predictive value (PPV) of 85% (95% CI: 73–93). Every 10-fold increase in tTG IgA resulted in a 4.7× increased risk. Furthermore, the PPV of a positive EMA IgA was 76% (95% CI: 65–91) compared to 100% (95% CI: 83–100) if the dilution was at least 1:80. All children with an EMA IgA titer of 1:80 or greater also had a tTG ≥10× ULN. Of the 32 children who met the 2020 ESPGHAN criteria, four (12.5%) did not have histologic CeD (PPV 87.5%; 95% CI: 70–96). Half of these children had normalization of their serology while continuing a gluten-unrestricted diet. One child was presumptively treated with a gluten-free diet and another subsequently developed enteropathy. Current ESPGHAN guidelines require a positive EMA IgA in addition to a tTG IgA >10× ULN to make a CeD diagnosis without a biopsy given the high specificity of an EMA IgA and as a second positive test on a different blood sample decreases the potential for an erroneous positive initial tTG IgA.1 Jessica Rutsky et al. demonstrate how the current criteria proposed in ESPGHAN guidelines may be unsuitable for children with T1DM as 12.5% (4/32) of children who met serologic criteria did not have CeD (although one child eventually developed enteropathy on a gluten-containing diet). While some studies, such as those by Popp et al.4 and Cerqueiro Bybrant et al.,5 support applying the ESPGHAN 2020 non-biopsy approach in children with T1DM, another study by Wessels et al.6 proposed a higher threshold be used (i.e., ≥11×) in this population. Notably, in the study by Wessels et al., a positive EMA IgA did not correspond to a CeD diagnosis half of the time, although the authors do not specify whether these children also had a tTG IgA ≥10× ULN. One proposed mechanism behind the association between antibodies to tTG-2 and T1DM is the expression of tTG-2 in pancreatic islet cells, which is the site of inflammation pivotal in T1DM pathophysiology.7 This may explain why both tTG IgA and EMA IgA have lower specificity for CeD enteropathy in this population and often spontaneously normalize even while on a gluten-containing diet.8 As both tTG IgA and EMA IgA recognize tTG-2,9 it is not surprising that similar to tTG IgA, higher EMA IgA titers have higher specificity for CeD. Jessica Rutsky and colleagues provide further evidence for a higher disease-specific threshold6, 10 for serologic diagnosis of CeD in patients with T1DM. Specifically, an EMA IgA titer ≥1:80 may be more appropriate than any positive EMA IgA test. This is analogous to requiring a higher threshold for tTG IgA. Although it is difficult to fathom making (or ruling out) a diagnosis of CeD without a biopsy, it is evident that an endoscopy may be avoidable in some individuals, including those with T1DM, who meet specific, highly positive serologic cut-offs. Given the relatively frequent occurrence of antibodies to tTG in the absence of enteropathy in patients with T1DM, a higher degree of confidence will be required if serologic thresholds for diagnosis are to be established. While studies, such as the one by Jessica Rutsky et al., add to the growing literature on this important issue, future prospective and multicenter studies will be necessary to find the "sweet spot" and to validate the non-biopsy approach in this population. Jocelyn Anne Silvester receives support from the United States National Institutes of Health National Institute of Diabetes, Digestive and Kidney Disease (K23 DK119584). Jocelyn Anne Silvester has served on the advisory boards for Takeda Pharmaceuticals and Teva Pharmaceuticals and received research support from Takeda Pharmaceuticals, Celiac Disease Foundation, and the United States National Institutes of Health National Institute of Diabetes, Digestive and Kidney Disease. The remaining author declares no conflict of interest.