Exposure Sources, Amounts and Time Course of Gluten Ingestion and Excretion in Patients with Coeliac Disease on a Gluten-free diet

Exposure sources, amounts and time course of gluten ingestion and excretion in patients with coeliac disease on a gluten-free diet Jocelyn A Silvester, MD, PhD1,2,3,4, Isabel Comino, PhD5, Lisa N Rigaux, BSc6, Veronica Segura, BSc5, Kathy H Green, MSc, RD6, Angel Cebolla, PhD7, Dayna Weiten, RD8, Remedios Dominguez, PhD7, Daniel A Leffler, MD, MSc1,3, Francisco Leon, MD, PhD9, Charles N Bernstein, MD4, Lesley A Graff, PhD4, Ciaran P Kelly, MD1,3, Carolina Sousa, PhD5, Donald R Duerksen, MD4,6 1.Harvard Medical School Celiac Research Program, Boston, USA 2.Boston Children’s Hospital, Boston, USA 3.Beth Israel Deaconess Medical Center, Boston, USA 4.Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, Canada 5.Faculty of Pharmacy, University of Seville, Seville, Spain 6.St Boniface Hospital, Winnipeg, Canada 7.Biomedal S.L., Seville, Spain 8.Grace General Hospital, Winnipeg, Canada 9.Glutenostics LLC, USA Corresponding Author: Jocelyn A Silvester, Boston Children’s Hospital Gastroenterology Hunnewell Ground, 300 Longwood Avenue, Boston, MA 02115; Telephone +1 617-355-6058; Fax: +1 617-730-0495; jocelyn.silv[email protected]. Specific Author Contributions: Study concept and design – AC, DRD, CPK, DAL, FL, CS, JAS; acquisition of data – IC, LR, VS; analysis and interpretation of data – CNB, AC, IC, KHG, LAG, CPK, JR, CS, JAS, VS, DW; drafting of the manuscript - JAS; critical revision of the manuscript for important intellectual content – all authors; statistical analysis - JAS; administrative, technical, or material support – AC, RD, FL, LR. All authors approved the final version of the manuscript. Statement of Interests Potential Competing Interests: CNB has served on advisory boards of Abbvie Canada, Ferring Canada, Janssen Canada, Pfizer Canada, Roche Canada,Takeda Canada, and has consulted to Mylan Pharmaceuticals and Takeda. He has received unrestricted educational grants from Abbvie Canada, Janssen Canada, Pfizer Canada, Shire Canada, and Takeda Canada. He has been on speaker’s bureau of Ferring Canada, Janssen Canada, Takeda Canada, and Medtronic Canada. AC is the CEO and shareholder of Biomedal S.L. He is also partner of Glutenostics LLC. Biomedal was the licensor of G12/A1 monoclonal antibodies for detection in food, urine and stool at the time of the study. DRD has received research funding from Biomedal S.L. and has served as a consultant for Takeda Pharmaceuticals and Shire Canada and is Chair of the Professional Advisory Council of the Canadian Celiac Association. RD was an employee of Biomedal S.L. at the time of the study. CPK has acted as a scientific advisor to Cour Pharma, Glutenostics, Innovate, ImmunogenX, Innovate and Takeda. He also acts as Principal Investigator on a research grant on coeliac disease supported by Aptalis. DAL is a Medical Director for Takeda Pharmaceuticals. FL was an employee of Celimmune at the time of the study, and is currently an employee of Provention Bio. He also serves as a partner in Biomedal and Glutenostics. JAS has served on an advisory board of Takeda Pharmaceuticals and received research support from Cour Pharma, Glutenostics and the Celiac Disease Foundation. IC, KHG, LAG, LNR, CS, VS and DW have no disclosures. HHS Public Access Author manuscript Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Published in final edited form as: Aliment Pharmacol Ther . 2020 November ; 52(9): 1469–1479. doi:10.1111/apt.16075. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Structured Summary Background: A major deficit in understanding and improving treatment in coeliac disease (CD) is the lack of empiric data on real world gluten exposure. Aims: To estimate gluten exposure on a gluten-free diet (GFD) using immunoassays for gluten immunogenic peptides (GIP). Relationships among GIP detection, symptoms and suspected gluten exposures were examined. Methods: Adults with biopsy-confirmed CD on a GFD for 24 months were recruited from a population-based inception cohort. Participants kept a diary and collected urine samples for 10 days and stools on days 4–10. “Doggie bags” containing ¼ portions of foods consumed were saved during the first 7 days. Gluten in food, stool and urine was quantified using A1/G12 ELISA. Results: Eighteen participants with CD (12 female; age 21–70 years) and 3 participants on a gluten-containing diet enrolled and completed the study. 12/18 CD participants had a median 2.1 mg gluten per exposure (range 0.2 to >80 mg). Most exposures were asymptomatic and unsuspected. There was high intraindividual variability in the interval between gluten ingestion and excretion. Participants were generally unable to identify the food. Conclusions: Gluten exposure on a GFD is common, intermittent, usually silent, and excretion kinetics are highly variable between individuals. The amount of gluten varied widely, but was typically in the milligram range which was 10–100 times less than was consumed by those on an unrestricted diet. These findings suggest that a strictly gluten free diet is difficult to attain and specific exposures are difficult to detect due to variable time course of excretion. Keywords Coeliac disease; gluten immunogenic peptide; gluten-free diet; treatment adherence; immunoassay; stool test; urine test Introduction Coeliac disease is a chronic gluten-responsive immune mediated enteropathy1 that is treated with a gluten-free diet (GFD)2. Patients with coeliac disease report high rates of adherence to a strict GFD3; however, unintentional gluten exposures may be more common than realized and are distinct from lapses in an otherwise intentionally strict GFD4,5. Compared to other chronic health conditions, the treatment burden of coeliac disease is high6. The limitations and socio-emotional toll of a GFD are increasingly recognized7, which along with high rates of ongoing symptoms and enteropathy on a GFD are driving efforts to develop adjunctive or alternative therapies to a GFD. These include immunomodulatory therapies as well as gluten-digesting enzymes that hydrolyze prolyl-peptide bonds that are resistant to human gastrointestinal luminal proteases8. Recently, tests for gluten in food9–11, and gluten immunogenic peptides (GIP) in stool12,13 and urine14 have been developed. These immunoassays rely upon the G12 and/or A1 antibodies, which bind GIP - gluten peptides resistant to intraluminal and serum proteases that are recognized by T cells of patients with coeliac disease9–12. Using these methods, GIP were detected in feces from 1 in 4 Spanish patients with coeliac disease who reported strict Silvester et al. Page 2 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript adherence to a GFD and had no sources of gluten ingestion identified on a detailed 3-day dietary questionnaire12. These assays quantify gluten excretion, not ingestion. Syage et al have estimated that adults with coeliac disease on a GFD consume a median 141 mg gluten per day through extrapolation from stool GIP concentrations16; however, many factors affecting kinetics of GIP excretion (e.g., food matrices, individual variations, water ingestion) were not accounted for. Previously, in a sample of community-dwelling adults with biopsy-confirmed coeliac disease who were aiming to adhere to a strictly gluten-free diet, we demonstrated that low concentrations of gluten can be found in the food, urine and stool17. In this study we quantified the amount and frequency of gluten exposures; characterized the timing of gluten absorption and excretion; and examined the relationship of gluten exposure to acute symptoms and suspected gluten exposures. Methods As described previously17, participants were recruited from the Manitoba Celiac Disease Inception Cohort study. This prospective longitudinal study enrolled adults (16 years or older) with biopsy-confirmed coeliac disease. Diagnosis of coeliac disease was based upon villous atrophy and intraepithelial lymphocytosis (Marsh 3 lesions) on intestinal biopsy and elevated serum tissue transglutaminase (TTG IgA) and/or endomysial (EMA IgA) antibody levels17. In Manitoba, a central laboratory performs all serum TTG IgA testing which facilitated a population-based approach to reduce recruitment bias. Specifically, a list of physicians with patients with elevated serum TTG was generated weekly throughout the enrolment period. These physicians were contacted to inform them of the study, and provided with recruitment materials for their patients. Secondary recruitment methods included referral by the endoscopist, and advertisements at retailers of gluten-free products and in the newsletter of the Manitoba Chapter of the Canadian Celiac Association. Participants were recruited within 6 weeks of GFD initiation, with additional study visits at 6, 12, and 24 months. Per usual practice, participants were referred at diagnosis to a Registered Dietitian with GFD expertise and given information about the Canadian Celiac Association patient support group. For the DOGGIE BAG sub-study, coeliac disease participants on a strict GFD completed an additional diary, food record, and food, urine, and stool sample collection during the 10 days immediately prior to per-protocol follow-up biopsy 24 months after initiation of a GFD (Figure 1). Healthy persons without coeliac disease or a dietary gluten restriction were studied as positive controls for gluten ingestion. Exclusion criteria (applicable to both groups): 1) inability to complete diary and/or collect samples; 2) baseline stool frequency <3/week; or 3) concurrent digestive or malabsorptive disorder other than disaccharide intolerance (e.g., viral gastroenteritis, small intestinal bacterial overgrowth, inflammatory bowel disease). Participants on a “strict GFD” were trying to follow a strict gluten-free diet and did not report intentional gluten exposure. All DOGGIE BAG study participants were instructed to follow their usual dietary patterns including any packaged/processed and restaurant or take-out meals. To reduce participant Silvester et al. Page 3 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript burden and account for intestinal transit, food samples were collected for the first 7 days and stools were collected on days 4–10. Participants were instructed to collect their first morning urine (50 mL) plus two additional samples daily (preferably midday and at bedtime) throughout the study period. The participants provided with specific instructions to prevent contamination with gluten during sample collection. Sterile collection containers were provided as well as a −20°C freezer (Whynter LLC, Brea, CA) in which to store all food and body waste samples. GIP are stable for >72 hours at room temperature in urine and stool; nevertheless, a thermometer was also provided to monitor freezer temperature. The study protocol was reviewed by the University of Manitoba Health Research Ethics Board. Participant diary and standardized measures of coeliac disease activity and gluten-free diet adherence Participants logged sample collection times and kept a food record. Each day, participants reported whether they thought that they may have ingested gluten, and described any symptoms of suspected gluten exposure. Responses were independent, so participants could suspect gluten exposure without symptoms or vice versa. If gluten exposure was suspected, then participants indicated which food(s) they suspected contained gluten. At the end of the study, participants completed standardized coeliac disease specific self-report measures, including: Celiac Symptom Index (CSI)18, a 16-item measure with possible scores ranging from 16 to 80 on which scores of 30 are less are associated with better quality of life and GFD adherence; Celiac Diet Assessment Tool (CDAT)19, a seven-item self-report measure with questions about symptoms and personal traits, with possible scores ranging from 7 to 35 on which scores of 12 or less are associated with adequate adherence; and Gluten-Free Eating Assessment Tool Short (GF-EATs), a single-item self-reported adherence scale with 5 categorical responses ranging from frequent gluten (more than once per week) to no gluten20. In addition, GFD adherence was assessed by a Registered Dietitian with expertise in GFD based using the Dietitian Integrated Evaluation Tool for Gluten-Free Diets (GFD) which incorporates assessment of: 3-day food records; food frequency questionnaire including gluten-free and gluten-containing items; assessment of risk behaviors and precautions to avoid gluten at home, work/school, restaurants and other people’s homes; and label reading skills21. This was reported using a 10-point ordinal descriptive scale on which higher scores correspond to greater adherence with a score of 10 reserved for individuals living in gluten-free homes who eat out only a few times per year. Scores for the Manitoba Celiac Disease Inception Cohort at 24 months range from 1 to 10 with a median of 8 (IQR 6–8) Food collection During the first 7 days, participants provided portions of food consumed, including sauces, salad dressings, flavoured coffees and beer. Unprocessed, inherently gluten-free foods were not collected (e.g., raw or steamed fresh fruits or vegetables without dressing or sauce; water, wine, 100% fruit juice or unflavoured milk, tea, filtered coffee made at home from whole beans). A representative ¼ portion was selected as it was more accurate than a 1/3 portion (based upon weight of food sampled as a proportion of weight of the total meal) in a Silvester et al. Page 4 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript pilot study of 10 persons (data not shown). A digital scale (Ozeri ZK14-S, Ozeri USA, San Diego, CA) was provided to facilitate accurate sampling. Detailed in-person and written instructions were provided regarding collection of “representative” sample of all foods on their plate to account for complex meals with multiple food components. Thawed food samples were batch processed for each participant and pooled based upon time consumed (04.01–10.00, 10.10–16.00, 16.00–04.00). Each pooled sample was homogenized using a 1000W triple blade stainless steel blender (Breville, Saint Laurent, QC) with deionized water added when required to facilitate mixing. The blender was thoroughly cleaned and rinsed between each sample. Quantification of gluten in food samples Gluten content of food (ppm) was determined using gliadin as a standard according to the recommendations of the Working Group on Prolamine Analysis and Toxicity22 using an A1- G12 antibody sandwich enzyme-linked immunosorbent assay9 (ELISA; GlutenTox ELISA Sandwich A1-G12, Hygiena Diagnostica, Seville, Spain) that is specific and sensitive for the most gluten immunogenic peptides in food23. Gluten concentration in food samples was determined according to the following formula to account for sample dilution: gluten content (ppm) = ppm gluten in food sample ×weigℎtof samplewitℎwateradded sampleweigℎtascollected . The amount of gluten consumed was estimated using the following formula to convert from ppm to mg and account for sampling of ¼ of food as plated (the participant ate the other ¾): mg gluten consumed =ppm gluten in sample xsample weigℎt(kg)x3 . Following sample dilution, the working limit of quantification of this method is 1.6 ppm gluten. Quantification of GIP in stool samples Stool samples were thawed and the entire collection was manually homogenized. Stool GIP concentration was determined by sandwich ELISA (iVYLISA GIP Stool kit, iVYDAL In Vitro Diagnostics®, Biomedal S.L., Seville, Spain) according to the manufacturer’s protocol. Each sample was run in duplicate with ≥4 different aliquots tested on different days using αgliadin 33-mer as GIP reference material. The accuracy of this method in detecting gluten in stool has been reported previously: analytical sensitivity having limits of detection and quantification of 60 and 160 ng GIP per gram stool, respectively; and diagnostic sensitivity and specificity of 98.5% and 100%, respectively12. GIP are stable in stool stored between −20 and −80°C for more than 12 months, and through multiple freeze-thaw cycles. Quantification of GIP in urine samples Urine samples were processed for testing by the iVYCHECK GIP Urine (iVYDAL In Vitro Diagnostics®, Biomedal S.L., Seville, Spain) according to the manufacturer’s recommendations. This immunochromatographic test uses G12/A1 antibodies and provides Silvester et al. Page 5 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript a visual result as a red line in the test zone14. Samples with positive qualitative results were quantified using the iVYCHECK Reader. The validity of this method in detecting GFD transgressions was determined by the analytical sensitivity using α-gliadin 33mer peptide (limits of detection and quantification 2.2 and 6.3 ng per mL urine, respectively). Each sample was run in duplicate and at least two different aliquots of each sample were tested on different days. Tissue transglutaminase antibody levels Serum TTG IgA levels were determined by the Immunology Lab at St Boniface Hospital, Winnipeg, Canada. In May 2015, the assay was changed from ImmuLisa (Immco Diagnostics Inc, Buffalo NY) to Bioplex 2200 (Bio-Rad Laboratories (Canada) Inc, Montreal QC). All results are reported as multiples of the upper limit of normal (ULN) for the assay used (ImmuLisa 20, Bioplex 15). Results All recruited participants, which included 18 persons with coeliac disease on a GFD (12 female; median 41 years, range 21–77) and 3 healthy persons who consumed gluten regularly (1 female; ages 20, 36, 43 years) completed the study protocol (Table 1). As reported previously, participants with coeliac disease were generally asymptomatic although some reported mild symptoms; Celiac Symptom Index (CSI) scores ranged from 19 to 4017. Celiac diet adherence test (CDAT) scores were ≤14 suggesting adequate adherence; however, self-reported GFD adherence on the Gluten-Free Eating Assessment Tool short (GF-EATs) suggested higher rates of gluten exposure. Median serum TTG IgA at 24 months was 0.6 ULN (range 0.07 to 2.67), a more than ten-fold decrease from diagnosis, providing further evidence that gluten consumption had been reduced. The mean number of pooled food samples was 17.4 (range 12–21). Those with fewer samples tended not to have a sample for the 04:01–10:00 time interval because they ate fruit and/or drank coffee/tea at breakfast. Three food samples were not collected (2 forgot; 1 participant travelled to the United States and was unable to return with the sample). Participants collected a mean 29.0 (range 22–30) urine samples and 4.2 (range 3–6) stool samples. Urine sample collections were missed due to urinary incontinence, working conditions and forgetfulness. There were no freezer failures. Gluten in food and GIP in urine and stool: Healthy participants on a gluten-containing diet The 3 participants on a gluten-containing diet collected samples during 62 of 63 possible time intervals. Nine (14%) food samples did not contain detectable gluten and 1 contained only 4 ppm gluten. Most (75%) contained very high amounts of gluten that exceeded the upper limit of the dynamic range of the assay (~600 ppm gluten, depending on sample dilution). Consistent with these findings, 67 of 88 (76%) urine samples contained between 2.2 and 58.5 ng GIP/mL, and all 12 (100%) stool samples contained detectable GIP. Silvester et al. Page 6 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Gluten in food: Coeliac disease participants on a GFD Coeliac disease participants following a GFD ate food prepared in their own home, at others’ homes and at restaurants, and several participants travelled away from home. Overall, 25/313 (8%) pooled food samples tested positive. Based upon sample weights, median estimated gluten exposure was 2.1 mg per ingestion (range 0.2 to >80 mg). Foods that tested positive for gluten were consumed throughout the day and included baked goods, sausages, snacks and restaurant meals that were allegedly gluten-free (Table 3). GIP in urine and stool: Coeliac disease participants on a GFD Overall, 30/519 (6%) urine samples tested positive for gluten. Positive samples from 8 participants contained 2.2 to 26.8 ng GIP per mL. For stool, 8/75 (11%) tested positive, containing 140 to 910 ng GIP per gram stool (Table 2). Five participants had at least one positive stool sample. Association between suspected gluten exposure and detection of GIP For the coeliac disease participants, there were 20 days when gluten exposure was suspected, 14 of which were during the first seven days when food samples were collected. Twenty-seven foods were implicated related to concerns about cross-contact with gluten in restaurants or at home, specific ingredients (e.g., Worcestershire sauce), specific foods (e.g., cookie prepared by co-worker) and suspicions based upon symptoms. Most participants (73%) who suspected a gluten exposure had at least one positive food, stool or urine sample; however, only two participants correctly identified the culprit food (2 samples). There was only case in which gluten was detected when symptoms of gluten exposure were experienced. Four of the seven participants who did not suspect a gluten exposure had at least one sample that tested positive for GIP. Correlation and time course of documented gluten consumption and detection of GIP in urine or stool There were 25 documented gluten exposures from food, 8 of which were associated with a positive urine sample within 36 hours and 9 of which were associated with a positive stool sample 2 to 4 days later (Figure 2). The interval between consumption of a food sample that tested positive for gluten and GIP detection in urine ranged from 2 to 49 hours. Discussion This study confirms that gluten ingestion occurs frequently despite efforts to follow a strictly gluten free diet. Our findings likely represent a “best case” scenario as participants were highly engaged having participated in a research study since diagnosis and had access to highly skilled dietitians with GFD expertise. Gluten exposure was often unsuspected and occurred at home as well as when eating out. Twelve of eighteen persons with coeliac disease on an intentional GFD, based upon standardized self-report measures, were exposed to gluten within the 10 day study period. The estimated amount of gluten consumed ranged from 0.2 mg to more than 80 mg. This is orders of magnitude less than the 3 to 10 grams typically utilized in gluten challenge studies, which may partially explain why most gluten exposures were silent (no symptoms reported) and unsuspected by participants. The clinical Silvester et al. Page 7 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript consequences of chronic intermittent milligram range gluten exposures observed in our study are unknown. As previously reported, there was discordance between gluten exposures and villous atrophy in our cohort, which may be related to the timing of follow-up biopsy, duration of the sample collection period, and/or high sensitivity assay detecting milligramrange gluten ingestions17. We observed considerable inter-individual variability in the interval between gluten consumption and GIP excretion in urine, a phenomenon previously reported in healthy individuals14,24. Some participants absorbed and excreted GIP relatively rapidly, with GIP detectable in urine collected within hours of ingestion. Others displayed a more protracted time course with GIP appearing in urine collected 24 hours after gluten consumption. Renal dysfunction was an exclusion criterion but we did not study gastric emptying or small intestinal transit. Participants were instructed to collect three urine samples per day (preferably first morning, during the day and at bedtime), so we likely did not capture all urine that contained GIP. This complicates evaluation of the relationship between estimated gluten consumption and detection of GIP in urine. However, among individuals with multiple gluten exposures, the interval between gluten consumption and collection of urine with detectable GIP was generally consistent intra-individually (i.e., for a given participant) and ranged from <4 to >24 hours. There was also variation in the amount of gluten ingestion that was associated with detection of gluten in urine. The smallest estimated gluten exposure that was detected was 0.2 mg, which is two orders of magnitude less than what was determined to be the limit of detection for healthy individuals14. This participant reported that they may have been exposed to gluten transferred from toast prepared in a 4-slice toaster with a shared crumb tray, but dedicated gluten-free toasting slots. Situations in which gluten may be introduced through cross-contact highlight the challenges associated with estimating gluten content through sampling of food, as gluten may not be homogenously distributed and the sampled portion is not consumed. The kinetics of gluten absorption and excretion may be affected by multiple factors in addition to the amount consumed. Intact dietary proteins, including ovalbumin25, lactoglobulin26 and gliadin24,27, may be detected in serum within minutes of ingestion and the concentration tends to peak within hours. Numerous studies have demonstrated transcellular passage of proteins across the small intestinal epithelium which is likely the predominant mechanism in healthy persons. In a case series of 5 patients with CD, Husby et al26 demonstrated that absorption of ovalbumin and lactoglobulin was more rapid and more extensive in untreated patients with villous atrophy than in those who were following a GFD. Absorbed dietary proteins are not bound to immune complexes and serum protein concentration is not related to antigen-specific IgG, IgA or IgM25. The phenomenon of wheat dependent exercise-induced anaphylaxis (WDEIA) is instructive. Affected individuals are prone to anaphylaxis if they exercise after wheat consumption, but not with wheat consumption alone or exercise alone. Exercise is associated with increased serum gliadin levels28, and detectable serum gliadin is associated with positive response to wheat plus exercise challenge27. Non-steroidal anti-inflammatory drugs (NSAIDs) also increase antigen uptake24 and may induce wheat dependent exercise-induced anaphylaxis. Silvester et al. Page 8 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Detection of GIP in urine is dependent upon intestinal absorption as well as renal excretion. Circulating gluten is filtered by the kidneys which reabsorb most proteins. Urinary peptides are derived from proteolysis of non-reabsorbed proteins. Although poorly correlated, renal dysfunction is associated with a decrease in the urine peptide-to-protein ratio29. Urinary peptide excretion is increased in children with coeliac disease compared to age- and sexmatched controls30 and decreases following treatment with a GFD31. Many circumstances may contribute to gluten ingestion. Introduction of gluten to inherently gluten-free grains can occur during any stage from planting through harvest and processing32. Commercial seed may legally contain off-types (seeds other than the stated variety or strain). Accordingly, commercial “naturally” gluten-free grain products often contain >20 ppm gluten in Canada [61/640 (9.5%)]33 and the United States [7/22 (32%)]34, including grain products with a “gluten-free” label35. Food products labeled as gluten-free have also been tested. In Europe, there was negligible gluten content in the 205 commercially available gluten-free foods tested36, while in a US study approximately 5% of the 158 gluten-free products tested above levels of 20 ppm37. Further, these studies might be underestimating gluten content because they typically utilized the R5 antibody sandwich ELISA method which produced high variation of results in interlaboratory proficiency studies38 and does not detect small hydrolyzed gluten fragments or some of the most immunogenic gluten peptides39,40. In a systematic review of 24 cross-sectional studies, gluten contamination of both processed foods (13.2%) and raw ingredients (41.3%)41 was common. Gluten ingestion may also occur as a result of household contamination, such as through use of shared cooking utensils42. Finally, patients may inadvertently ingest gluten due to incomplete information about ingredients or misinterpretation of food labels due to misconceptions regarding the gluten-free status of an ingredient. All of these situations occurred during the study, and many participants suspected gluten ingestion based upon how food was prepared in a shared kitchen or unclear product label information. Our study was not designed to assess whether gluten exposure is associated with where a food is prepared. This question is difficult to answer definitively because food preparation is complicated and “self-prepared” meals are themselves sourced using ingredients from outside the home. Several participants commented upon this in the margin of their food records (e.g., “told it was gluten-free” in reference to products purchased at a farmer’s market). We did not collect sufficiently granular data on the source of all components of meals, or indeed gluten testing data, to discern the extent to which they were derived from pure ingredients vs assembled from partially prepared foods. Strengths of this study include analysis of food samples in addition to analysis of temporally relevant urine and stool using an assay for the immunodominant gluten peptide over a multiday period. As well, participants were recruited from a larger longitudinal research study with comprehensive regular follow-up that included review of practices to avoid crosscontact as well as assessments of label-reading skills. These participants were recruited within weeks of initiating a GFD so were all at a similar point in their disease trajectory when they participated in the study. While this may limit the generalizability of the findings, one might expect that the general population of patients with coeliac disease, many of whom do not receive regular disease- Silvester et al. Page 9 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript collected on the morning of day #8 contained detectable GIP. Participant 14 had consumed >20 mg gluten (>100 ppm) on day #2 with GIP detected in stool collected 48 hours later. Urine samples tested negative; however, no samples were obtained between 6 and 24 hours after gluten exposure. Participants 2, 5, 7 and 8 each had at least one exposure to up to 2.1 mg gluten with no GIP detected in any contemporaneous urine or stool samples. Participant 16 had repeated gluten exposure with 10 (59%) of food samples collected containing detectable levels of GIP (no samples tested positive on day#6). Stool samples collected on days 4, 5, 8 and 9 all contained GIP. GIP were not detected in any urine sample collected day #1 through day #4, but at least one urine sample contained GIP on day #5 through day #10. Silvester et al. Page 16 Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Author Manuscript Author Manuscript Author Manuscript Author Manuscript Silvester et al. Page 17 Table 1 – Coeliac Disease Participant Characteristics at Baseline and after 24 months on a Gluten-free Diet According to Gluten Ingestion and Excretion (N=18) Diagnosis 24 months Gluten detected (N = 12) No gluten detected (N=6) Age [years (IQR)] 41 (21–77) 48 (28–63) 39 (36–45) Female [% (n)] 56 (66%) 8 (66%) 4 (66%) HLA Genotype [% (n)] DQ2 15 (83%) 11 (92%) 4 (66%) DQ8 1 (6%) -- 1 (17%) DQ2/DQ8 2 (11%) 1 (8%) 1 (17%) TTG IgA serology (reported as multiples of upper limit of normal) [median (IQR)] 9 (3.6 – > 10) 0.73 (0.28–1.3) 0.50 (0.32–1.2) Celiac symptom index (CSI) [median (IQR)] 39.5 (39.3–47.8) 29 (26–33) 30 (27–35) 1 -- -- 2 (33%) Dietitian Global Assessment Score (DIET-GFD) 5 --- 1 (8%) -- 6 --- 2 (16%) 1 (17%) 7 --- 3 (25%) 2 (33%) 8 --- 5 (42%) 1 (17%) 9 --- 1 (8%) 2 (33%) Marsh classification [% (n)] 0 -- 4 (33%) 2 (33%) 1 -- -- 2 (33%) 2 -- -- -- 3a 30 (35%) 8 (66%) 2 (33%) 3b 36 (43%) -- -- 3c 19 (22%) -- -- DIET-GFD – Dietitian Integrated Evaluation Tool for Gluten-Free Diets Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Silvester et al. Page 18 Table 2 – Detection of Gluten in Food, and GIP in Urine and Stool Collected Over a 10 day Period From Participants with Coeliac Disease on a GFD for 24 months (N=18) Sample type Number of samples Number of Positive samples Number of Participants with positive sample Food 313 25 (8%) 9 (50%) Urine 519 30 (6%) 8 (44%) Stool 75 8 (11%) 5 (28%) All samples 907 63 (7%) 12 (67%) Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Silvester et al. Page 19 Table 3 – Details of pooled food samples that tested positive for gluten immunogenic peptides Estimated mg gluten Study ID Day eaten Time eaten (24 hr clock) Foods in positive sample Notes Marsh grade at 24 months 0.2 11 5 17.00 Cheese flavour corn chips, hard caramel candies Chips labelled gluten-free 3a 1 5 3 11.00 13.00 Caramel candy Muffin with cheese No gluten-containing ingredients 3a 1.1 2 2 10.30 Eggs, waffles Waffles made at home from mix labelled GF 3a 1.4 5 4 12–15.00 15.00 Smoothie Cookie, marshmallow Purchased from smoothie shop Suspected gluten in cookie 3a 1.5 16 3 7.00 2 slices GF bread, fried eggs, ham, cheese. 3a 1.6 17 1 13.45 Steak, baked beans, fried potatoes Restaurant meal 0 2.6 12 4 7.15 Toast (GF labeled bread), jam, peanut butter Toast in GF side of shared 4- slice toaster. Dedicated GF condiments. 3a 2.7 7 1 19.45 22.45 Ham/quinoa casserole Granola bar (certified GF oats) At a house party 0 3 16 3 18.00 21.00 Fish, duck, French fries Cornflakes with milk 3a 3 17 3 17.38 19.20 Pepperoni pizza Dark chocolate GF pizza from restaurant with shared kitchen. Suspected. Brand may contain barley. 0 3.2 8 7 9.00 Ham and tomato omelet 3a 4 16 7 12.00 Pork and potato 3a 5.4 16 2 19.00 Chicken, duck, fried rice Symptomatic, not suspected. 3a 6 17 5 9.20 Donut Labeled GF. Attributed suspected exposure to leftover pizza eaten later in day 0 7.5 16 1 18.00 Chicken, ham, potatoes Symptomatic, not suspected. 3a 9.5 11 1 12.30 Rice grits, potato chips Grits wheat-free, no glutencontaining ingredients 3a 10.1 12 3 16.30 19.15 21.00 PB&J sandwich bar GF spaghetti, parmesan cheese Cheesecake GF labeled oat product GF labeled pasta GF label 3a 19.5 12 5 19.30 Perogies, sausage, salad Perogies local GF baker. Sausage farmer’s market (told was GF). 3a 25 12 2 18.30 GF naan bread, vindaloo, rice Restaurant with GF menu, identified as coeliac needing GF 3a 27 16 5 11.30 Beef, potato, peas, carrots 3a 36 16 3 11.30 15.00 Sausage, potato Apple pie 3a 39 14 2 16.00 18.00 Sausage Flavoured corn chips Butcher told was GF 0 40 16 4 11.30 Chicken, borscht, carrot 3a Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01. Author Manuscript Author Manuscript Author Manuscript Author Manuscript Silvester et al. Page 20 Estimated mg gluten Study ID Day eaten Time eaten (24 hr clock) Foods in positive sample Notes Marsh grade at 24 months 57 16 7 21.30 Beer, hamburger, vegetables Intentional gluten ingestion? 3a 80 16 1 11.30 Hamburger, potatoes Symptomatic, not suspected. 3a Bold text indicates items that participants identified as potentially containing gluten; GF – gluten-free. Aliment Pharmacol Ther . Author manuscript; available in PMC 2021 November 01.