Screening and Treatment Outcomes in Adults and Children With Type 1 Diabetes and Asymptomatic Celiac Disease: The CD-DIET Study

Celiac disease (CD) is an immune reaction–mediated condition triggered by gluten, a protein found in wheat, barley, and rye. CD risk is increased in patients with type 1 diabetes and is characterized by a broad spectrum of presentations, including gastrointestinal symptoms, growth alterations, and anemia, but many patients are asymptomatic (1–4).

Serologic screening for CD has high sensitivity and specificity, but few reports have evaluated CD rates and the impact of treatment with a gluten-free diet (GFD) in asymptomatic CD (aCD) patients with type 1 diabetes. The Celiac Disease and Diabetes-Dietary Intervention and Evaluation Trial (CD-DIET) prospectively screened a large cohort of patients aged 8–45 years with subsequent engagement in a randomized, dietary treatment trial to evaluate changes in HbA_1c and glycemic variability.

Patients aged 8–45 years with type 1 diabetes (≥1 year duration) were approached to undergo CD testing at 22 diabetes centers in Ontario, Canada, as part of the CD-DIET study (clinicaltrials.gov, NCT01566110) (5). Exclusion criteria included CD symptoms (gastrointestinal symptomology or evidence of growth impairment, anemia, or osteoporosis), a previous CD diagnosis, and pregnancy.

Eligible participants (Supplementary Fig. 1) were screened using tissue transglutaminase IgA (TTG-IgA; Inova Diagnostics). CD-seropositive participants (TTG-IgA ≥30 chemiluminescent units [CU]) progressed to a gastroenterology consultation with centralized biopsy confirmation (Marsh score ≥2).

Interested participants with biopsy-confirmed CD were randomized in a 1:1 ratio to a GFD or a gluten-containing diet (GCD) and followed for 1 year. Study visits with a dietitian occurred every 3 months using a standardized educational curriculum. Dietary adherence was also evaluated through quantification of dietary gluten intake and repeat TTG-IgA testing. Safety laboratory tests were evaluated (see Supplementary Material) with monitoring of growth, symptomology, and pregnancy. GCD participants who became symptomatic were transitioned to a GFD with the completion of outcome measures. A Data Safety Monitoring Committee evaluated adverse events. The primary outcome of CD-DIET was the change in HbA_1c from baseline to 12 months. HbA_1c was assessed centrally at baseline and at 6 and 12 months, with additional testing by local laboratories. Continuous glucose monitoring (CGM) was conducted using the iPro2 CGM system (Medtronic, Inc.) for blinded assessment.

Analysis was by intention to treat. Longitudinal differences in HbA_1c between groups were analyzed using the difference in differences approach and confirmed using linear mixed-effects regression. CGM results were summarized using standard parametric methods, while paired t tests were used to assess within-group differences in HbA_1c and postprandial glucose levels. All analyses were completed using R Statistics 3.5.1.

A sample size of n = 60 (30 patients per arm) was determined to be sufficient for our between-group analysis to reach 80% power at a 5% significance level under two-sided conditions. Paired analysis of HbA_1c in the GFD arm, a sample size of n = 25 before versus after pairs, reached a power above 80% using a two-sided paired t test with significance set at P = 0.05.

Overall, 2,585 patients consented and 2,387 subjects completed CD screening, including 1,298 adults (54.4%, 19–45 years) and 1,089 children (45.6%, 8–18 years) (see Supplementary Material). Higher rates of any prior CD serologic screening were seen in pediatric participants (n = 475, 43.6%, 95% CI 40.7–46.5%) as compared with adult participants (n = 89, 6.9%, 95% CI 5.5–8.3%) (odds ratio [OR] 10.5, P < 0.0001) (Fig. 1A). Adults had higher CD seropositivity rates (n = 88, 6.8%; 95% CI 5.4–8.2%) compared with children (n = 51; 4.7%; OR 1.48, 95% CI 3.4–5.9%, P = 0.035), but rates did not differ within each age-group on the basis of previous CD screening status. Biopsies were completed in 104 participants, and 82 were confirmed by biopsy with CD positivity rates of 4.2% (n = 54; 95% CI 3.1–5.3%) and 2.6% (n = 28; 95% CI 1.6–3.5%) for adult and pediatric participants, respectively (OR 1.64, P = 0.042) (Fig. 1B).

The 51 participants were randomized into the intervention phase, with 27 to the GFD group and 24 to the GCD group. Adherence to the assigned diet, as assessed by serial measures of gluten intake and TTG antibody titers, was deemed excellent (Fig. 2A). Ultimately, we observed no longitudinal difference in HbA_1c between the GFD group and GCD group (+0.14% [1.5 mmol/mol], 95% CI −0.79 to 1.08, P = 0.76). Within the GFD group, an increase in HbA_1c of +0.30% (3.3 mmol/mol) was observed over 12 months (95% CI 0.04–0.57%, P = 0.028) (Fig. 2B).

Using blinded CGM, no differences were seen in the percentage of time spent in hypoglycemic, euglycemic, or hyperglycemic ranges at 12 months. Despite this, postprandial glycemia was higher with the GFD relative to premeal levels at both the 2-h (+1.6 mmol/L, 95% CI 0.7–2.6, P = 0.0015) and 4-h (+1.5 mmol/L, 95% CI 0.4–2.7, P = 0.014) time points, while glucose levels of the GCD arm returned to premeal levels (Δ = 0.1 mmol/L after 4 h; 95% CI −0.8 to 1.0) (Fig. 2C). No differences were observed between groups in growth parameters (including height, weight, or BMI in children and adults; Supplementary Figs. 2 and 3), total daily insulin dose, or total daily insulin dose per kilogram.

Hypoglycemic adverse events that were nonsevere were reported with the GCD group (N = 3, 12.5%), with none reported with the GFD group (P = 0.097).

This study describes screening and treatment outcomes from a large prospective assessment of adult and pediatric patients with type 1 diabetes with aCD. We observed high rates of aCD with 1.5-fold higher rates in adults as compared with children. This result is consistent with increasing rates of CD in adults (6) and changes in the symptom profile toward a more subtle and less classical malabsorptive presentation with increasing age (7). In addition, age-related increases in autoimmune comorbidities are also well recognized, with recent T1D Exchange clinic registry data confirming rising rates of autoimmune comorbidities from adolescence into adulthood (8,9).

Adult participants in our study were also less likely to be screened for CD than children. This may reflect a lack of clarity about CD screening recommendations in adults and the absence of outcome studies in this population. Current North American diabetes guidelines advise consideration of screening patients with type 1 diabetes for CD, but actual details regarding screening tests, timing, and intervals are only explicitly described for children (10,11).

Dietary intervention with a GFD did not result in a difference in HbA_1c between randomized groups but led to greater postprandial glucose excursions relative to a GCD. These changes are consistent with mucosal recovery, which can be prolonged (1–3 years), with subsequent improvements in absorption (2,12). Nonetheless, transition to a GFD was safe with no significant differences observed in hypoglycemic time or adverse events.

Strengths of this report include the large number of patients screened and the rigor in delineating an asymptomatic cohort. In addition, participants in the intervention phase, who were provided with frequent dietitian-based support, exhibited clear differences in gluten intake and follow-up TTG serology. Limitations of this study include a small sample size and the heterogeneity of participant characteristics that limits generalization of these findings. In addition, our data do not represent the prevalence rates, as we were challenged by enumeration of the complete population at risk in the clinics where screening was completed. It was also our experience that despite facilitated and no-cost access to care, many asymptomatic participants were overwhelmed by the diagnostic process and by the prospect of managing these two conditions, each with significant lifestyle implications.

In summary, diabetes clinicians should be aware of CD as an important autoimmune comorbidity of type 1 diabetes that is frequently asymptomatic. In addition, clinical vigilance is warranted during dietary transition, as GFD treatment may impact short-term glycemic variability. Given the additional risks for CD-related microvascular and bone complications, longitudinal studies are also needed to assess the clinical impact of this double diagnosis across the patient’s life span (13–15).

Acknowledgments. This trial was funded by the JDRF Canadian Clinical Trial Network. We are grateful to the research staff at clinical sites and to patients and their families who participated in this trial.

Duality of Interest. F.H.M. served on advisory boards for Eli Lilly and Insulet. E.A. was employed as a dietitian with CD-DIET. Z.P. received grants, speaker fees, and consulting fees from Abbott, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, Janssen, Lexicon, Merck, Medtronic, Novo Nordisk, Pfizer, and Sanofi. A.P. received consulting fees for education-related activities and advisory boards from Medtronic, Sanofi, Novo Nordisk, Janssen, AstraZeneca, Boehringer Ingelheim, and Eli Lilly. A.A. received research support from Boehringer Ingelheim and AstraZeneca, is listed as an inventor on a patent application (WO 2015/128453) from Boehringer Ingelheim, has received an unrestricted educational grant from Eli Lilly, and has received honoraria from Novo Nordisk, Eli Lilly, Boehringer Ingelheim, Abbott, and Dexcom. B.A.P. received personal fees from Medtronic, Eli Lilly, Abbott, Novo Nordisk, Insulet, Sanofi, Janssen, and Boehringer Ingelheim, as well as grants from the Bank of Montreal and Boehringer Ingelheim, all outside the scope of the submitted work. C.S.Z. is currently a coinvestigator for a Children’s Hospital of Eastern Ontario site of the Medtronic multicentre randomized controlled trial of home use of a hybrid closed-loop system (no monetary gain) and participated on an Eli Lilly Canada Advisory board once in February 2019 (honorarium directed to institution). J.G. received honoraria for speaking from Abbott, AstraZeneca, Boerhinger Ingelheim, Dexcom, Eli Lilly, Insulet, Janssen, Merck, Novo Nordisk, and Sanofi and is on advisory boards for AstraZeneca, Boerhinger, Dexcom, Eli Lilly, Janssen, Merck, Novo Nordisk, and Sanofi. M.D.B. received honoraria from Shire and Allergan; served on advisory boards for AbbVie, Takeda, Janssen, Lupin, and Gilead; and received grants from AbbVie, Novo Nordisk, and Gilead. R.L.H. participated in clinical trials (in the last 2 years) for AstraZeneca, Boehringer Ingelheim, Eli Lilly, and Novo Nordisk. R.A. has received research grants from Sanofi, Novo Nordisk, Janssen, AstraZeneca, Becton Dickinson Technologies, Boehringer Ingelheim, Eli Lilly, Zealand, Xeris, Medpace, Kowa, Insulet, Dexcom, Bausch Health, Tandem Diabetes, and Bayer and consultancies and personal fees from Sanofi, Novo Nordisk, Janssen, AstraZeneca, Becton Dickinson Technologies, Boehringer Ingelheim, Eli Lilly, HTL Strefa, Gilead, and Merck. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. F.H.M., A.B.M.C., E.A., and M.A.M. wrote the first draft of the report and contributed to the study design, statistical analysis, and data interpretation. A.B.M.C., K.C.J., C.M., F.S., H.A.L., Z.P., A.P., A.A., B.R.S., B.A.P., C.S.Z., D.R.M., D.K., E.H., G.M., J.G., K.B., M.L.L., M.C., M.D.B., N.A.S., O.L., P.H.G., P.B., R.L.H., R.A., S.E.K., and W.G.P. were involved in the data collection and analyses. F.H.M., K.C.J., E.A., C.M., H.A.L., Z.P., A.P., A.A., B.R.S., B.A.P., C.S.Z., D.K., G.M., J.G., M.L.L., P.H.G., R.L.H., R.A., S.E.K., and M.A.M. recruited patients into the study. K.C.J., C.M., F.S., H.A.L., Z.P., A.P., A.A., B.R.S., B.A.P., C.S.Z., D.R.M., D.K., E.N.D.M., E.H., G.M., J.G., K.B., M.L.L., M.C., M.D.B., N.A.S., O.L., P.H.G., P.B., R.L.H., R.A., S.E.K., and W.G.P. contributed to the drafting and review of the report. F.H.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 2019 European Association for the Study of Diabetes Conference (EASD 2019), 16–20 September 2019, Barcelona, Spain.