The safe threshold for gluten contamination in gluten-free products. Can trace amounts be accepted in the treatment of coeliac disease?


Dr P. Collin, Medical School, FIN-33014 University of Tampere, Tampere, Finland.


Background : Gluten contamination in gluten-free products cannot totally be avoided. The safe threshold for gluten remains obscure.

Aim : The purpose was to estimate a reasonable limit for residual gluten, based on current literature and measurement of gluten in gluten-free products on the market.

Methods : The gluten content of 59 naturally gluten-free and 24 wheat starch-based gluten-free products were analysed by enzyme-linked immunosorbent assay. The daily intake of flours was calculated in 76 adults on gluten-free diet, and the intake compared with mucosal histology.

Results : A number of both naturally gluten-free (13 of 59) and wheat starch-based gluten-free (11 of 24) products contained gluten from 20 to 200 ppm (=mg/kg). The median daily flour consumption was 80 g (range: 10–300). Within these limits, the long-term mucosal recovery was good.

Conclusions : The threshold for gluten-contamination can safely be set at 100 ppm. Provided that the daily flour intake is even 300 g, a level of 100 ppm results in 30 mg of gluten intake. This has been shown to be safe, when correlated to histology, in clinical and challenge studies. The level can be achieved by the industry, and does not make the diet too cumbersome.


Coeliac disease is characterized by small intestinal inflammation, villous atrophy and crypt hyperplasia after ingestion of gluten in genetically susceptible patients. The mucosal lesion recovers when the gluten-containing cereals wheat, rye and barley are withdrawn. A lifelong gluten-free diet is the only effective treatment of the disorder.

An unanimous view is that gluten-free dieting should be as strict as possible. However, a diet completely devoid of gluten would be difficult if not impossible to maintain. The products on the market may contain trace amounts contaminating gluten. The current standards of the Codex Alimentary Commission for a gluten-free diet go back to 1981.1 As at that time no sensitive and specific methods for gluten determination were available, a threshold value of 0.05 g nitrogen per 100 g dry matter was set for wheat starch. This limit is valid for wheat starch only, as many other gluten-free products contain endogenous proteins, which invalidate the nitrogen measurement. In the Codex Alimentarius Commission, there is an ongoing discussion regarding the safe threshold for residual gluten in gluten-free products, an issue which evidently needs to be considered.2, 3 One approach to the problem has been to carry out a gluten challenge and demonstrate how much gluten has to be taken before intestinal mucosal inflammation and villous shortening ensue. However, such mainly short-term studies have yielded somewhat divergent results,4–7 and they have not been randomized.

A second concern is that the histological responses to gluten withdrawal or challenge are extremely variable in patients with coeliac disease. Small intestinal mucosal recovery on gluten-free diet may take from a couple of months to many years.8, 9 Similarly, the intestinal mucosal deterioration after gluten challenge occurs in some patients within a few weeks, whereas in some more than 10 years’ normal diet is required for the mucosal relapse.10 Further, it is possible that the minor inflammatory response after inadvertent gluten ingestion such as inevitably occurs daily in most coeliac disease patients is not harmful at all.

The question of tolerable amounts of gluten in coeliac disease can also be settled in the context of everyday clinical practice. It is logical to assume that such gluten contamination, which has taken place for many years along with successful treatment of coeliac disease, would be safe enough. In other words, provided that we can show products containing a given trace amount of gluten contamination to be safe in long-term use, we can allow such contamination.

Wheat starch-derived gluten-free products may, after purification, still contain trace amounts of gluten. They have been allowed for decades for coeliac disease patients in Finland, Sweden and the United Kingdom, and complete intestinal mucosal recovery has been evident both in children and adults.9, 11 Dietary compliance has also been good9 and the rates of coeliac complications such as malignant development are low in patients adhering to such a diet.12 The first randomized study comparing wheat starch-based gluten-free to naturally gluten-free products showed that the morphological and clinical responses were equally good by both treatments.13

Recent research in the matter of determining trace amounts of gluten have focused on development of immunological methods with high sensitivity and specificity. Until recently, there has been no commercially available enzyme-linked immunosorbent assay (ELISA) for gluten determination which meets all the requirements for a suitable method.14 However, a promising new ELISA has been developed15 and made commercially available. The monoclonal antibody used recognizes the pentapeptide QQPFP, and homologous sequences that occur repetitively in the prolamins from wheat, rye and barley.15 Furthermore, this system has been tested in a collaborative trial,16 and is proposed by the ‘Working Group on Prolamin Analysis and Toxicity’ for further evaluation by the Codex Committee on Methods of Analysis and Sampling.2

In connection with the clinical data available, there is now evidence to provide recommendations as to the safe threshold for residual gluten in gluten-free products.

Patients and methods

Food samples

The gluten content was determined in the most widely used naturally gluten-free and wheat starch-based gluten-free flours and baked products. Twenty-seven naturally gluten-free and nine wheat starch-based flours available on the market were collected in the year 2000, and again 32 (13 baked) and 15 (seven) products in 2003.

Sandwich enzyme-linked immunosorbent assay

Gluten content was determined with a recently developed ELISA,15 available as a kit (RIDASCREEN Gliadin, Art. No. R7001) produced by R-Biopharm AG, Darmstadt, Germany. The monoclonal secalin antibody R5 used in the test detects gliadin fractions of wheat and corresponding prolamins from rye and barley. Prolamins from oats, maize and rice are not detected. Furthermore, the antibody has no cross-reactivity with glutenin. Heat-processed samples were extracted with a cocktail solution (provided with the kit) by the procedure described in the leaflet. All other samples were extracted with 60% ethanol solution. The detection limit used in this study was 10 ppm gluten; the interassay coefficient of variation was 9.8%. This coefficient of variation is close to what has been published by others (8.7%).15 A content of 10 ppm gluten corresponds to 10 mg gluten/1 kg product and roughly to 5 mg gliadin/1 kg.

The daily use of gluten-free flours and the effect of the diet

Apart from knowledge of the trace amounts of gluten in gluten-free products, the amount of these products consumed daily must be taken into account. Daily use was therefore estimated from 4-day food records in 76 consecutive adults and 16 children with coeliac disease adhering to a strict gluten-free diet for 1–10 years (median 2 years); 28 adults were taking naturally gluten-free, 48 adults and all children wheat starch-based gluten-free products; the diets did not include oats. The daily intake of flour was compared with the small intestinal mucosal histology and antiendomysial antibodies. Mucosal morphology was evaluated blindly from well-orientated small intestinal biopsy samples by measuring the villous height/crypt depth ratio (Vh/CrD). The density of intraepithelial CD3+ lymphocytes was determined as cells per millimetre epithelium, as previously described.9

Serum immunoglobulin A (IgA) class endomysial antibodies were determined by an indirect immunofluorescence method using human umbilical cord as antigen;17 a dilution of 1:≥5 was considered positive.

Statistical analysis

The correlation between the daily intake of flours and the small intestinal histology was estimated by two-tailed Pearson correlation test.


The study protocol was accepted by the Ethical Committee of Tampere University Hospital.


Residual gluten in gluten-free products

Trace amounts of gluten were found in both naturally gluten-free and wheat starch-derived flours, 42 of 59 and two of 24 were free of gluten contamination (<10 ppm), respectively. Five naturally gluten-free and two wheat starch-based products contained more than 100 ppm gluten; none exceeded 200 ppm (Figure 1).

Figure 1.

Gluten contamination in naturally gluten-free and wheat starch-based gluten-free products measured by enzyme-linked immunosorbent assay (ELISA).15 Gluten-free flours are indicated by closed and gluten-free baked products by open squares.

The daily use of flours and clinical outcome

The median daily use of flours was 80 g (range:10–300) in adults. The correlation between the use of flours and intestinal mucosal histology in adults is depicted in Figure 2. No significant correlation was seen. One patient adhering to a naturally gluten-free and one to a wheat starch-based gluten-free diet proved positive for endomysial antibodies after 1 year on a gluten-free diet, the titres being 1:5 and 1:50, respectively. In 16 children, the mean daily use of flours was 60 g (range: 20–140); they all had normal small bowel morphology at follow-up.

Figure 2.

The estimated daily ingestion of flours in 76 coeliac adults on gluten-free diet, and its correlation with small intestinal villous structure (a = wheat starch-based, b = naturally gluten-free diet) and intraepithelial CD3+ lymphocytosis (c = wheat starch-based, d = naturally gluten-free diet). No significant correlations were seen: correlation coefficient (a) −0.21, P = 0.15; (b) −0.20, P = 0.31; (c) −0.22, P = 0.13; (d) −0.03, P = 0.89. Short-term treated (1–1.2 years) patients are illustrated with open and long-term (2–10 years) treated with closed squares.


Evidently a gluten-free diet completely devoid of gluten is unrealistic. Coeliac disease patients are exposed to products containing trace amounts of gluten, even when the products are sold as naturally gluten-free (Figure 1). In order to estimate the safe and rational threshold for daily gluten intake, the amount of residual gluten in gluten-free products and the total intake of these products must be considered. Provided that we can demonstrate that the use of a variety of gluten-free products results in both clinical and histological recovery, we can assume that the gluten level in these products is acceptable, notwithstanding the fact that the sensitivity to gluten is individual.

Studies on possible toxic or harmful effects of dietary transgressions or gluten challenge are summarized in Table 1. The study designs and the duration and the amount of gluten load have been highly variable. Gluten challenge studies with <10 mg have evidently not shown any effect, whereas some effect on the histology can be observed when the daily intake has been 500 mg or more. Catassi et al.6 reported that challenge with 100 mg of daily gliadin (corresponding to 200 mg gluten) resulted in a decrease in mean Vh/CrD from 1.5 to 1.3 and an increase in intraepithelial lymphocytes from 11/100 enterocytes to 19/100, the respective values by 500 mg challenge being from 1.6 to 1.1, and from 10/100 to 25/100. The changes were altogether marginal and notably, the villous recovery had initially not been as good as in well-treated children in general (Vh/CrD: 3.5–4.5).9 Nevertheless, these earlier papers indicate that the safe threshold should clearly be set below 500 mg of daily gluten, but could well be more than 10 mg.

Table 1.  Small-bowel biopsy findings in coeliac patients taking small amounts of gluten
AuthornAge groupConsumption of glutenSymptomsAbnormal findings in small-bowel biopsy
  1. Vh/CrD, villous height and crypt depth ratio; PVA, partial villous atrophy, SVA, subtotal villous atrophy; IELs, intraepithelial lymphocytes.

  2. * Gliadin content estimated from the wheat starch-based gluten-free products.

Ciclitira et al.4 1Adults10 mg gliadin (4 h)No
  1 600 mg gliadin (4 h) SVA and increase in IELs
Ciclitira et al.24 7Adults1.2–2.4 mg gliadin daily (1 week)*No
Ciclitira et al.2510Adults1.2–2.4 mg gliadin daily (6 weeks)*Diarrhoea in fourNo
Montgomery et al.7 8Adults2.5–5 g gluten daily, median (6 months)Increase in IELs
Catassi et al.610Children100 mg gliadin daily (4 weeks)AsymptomaticDose-dependent decrease in Vh/CrD, increase in IELs
 10 500 mg gliadin daily (4 weeks)Anorexia and pale stools in three 
Sturgess et al.26 4Adults1 g gliadin (4 h)Decrease in Vh/CrD in 3/4, increase in IELs in all
Srinivasan et al.27 2Adults500 mg gluten daily (6 weeks)Decrease in Vh/CrD in two
Laurin et al.28 1Children0.2 g gluten (39 weeks)Symptoms in 74% within 4 weeksIncrease in IELs
  1 0.8 g gluten (5 weeks) No changes
 22 0.7–4.3 g gluten daily (5–51 weeks) Decrease in Vh/CrD and increase in IELs

The current study verifies that most wheat starch-based gluten-free products contain trace amounts of gluten (Figure 1). More than 90% of Finnish coeliac disease patients consume these products. Dietary compliance, even long-term, has been good, as more than 85% are adhering to a strict diet. Complete mucosal recovery has been evident in both children and adults,9 and even better18 than in some other series,19 probably because of the good compliance. The quality of life has been as good as in non-coeliac controls even after long-term treatment.20 Further, the incidence of small intestinal lymphoma has been lower12, 21 in patients on a long-standing strict gluten-free diet including wheat starch for a long period than in some other reports.22,23

In earlier studies, only occasional abdominal complaints but no intestinal mucosal deterioration have been reported after taking wheat starch-based gluten-free products in an open study design (Table 2). We may therefore conclude that these products are safe in clinical practice. This was also verified in a prospective controlled study where no differences in histology, serology or quality of life were seen between wheat starch-based and naturally gluten-free products.13

Table 2.  Studies on the effects of wheat starch-based gluten-free diet in coeliac disease and dermatitis herpetiformis
AuthorStudy designNumber of patients with wheat starch diet (all patients)Patient groupDuration of wheat starch intakeOutcome methodsUntoward effects of wheat starch-based gluten-free diet
  1. Cr-EDTA, crept-ethylenediaminetetraacetic acid; AGA, antigliadin antibodies; EmA, antiendomysial antibodies; ARA, antireticulin antibodies; BMD, bone mineral density.

Ciclitira et al.24Open challenge 7 (7)Adults1 weekSmall-bowel biopsyNo
Ciclitira et al.25Open challenge10 (10)Adults6 weeksSmall-bowel biopsy Cr-EDTA excretionAbdominal symptoms in some
Ejderhamn et al.11Cross-sectional11 (11)Adolescents10 years on averageSmall-bowel biopsyNo
Chartrand et al.29Open challenge17 (31)Children and adults0.5–10 monthsAGA, EmASymptoms in 11/17
Kaukinen et al.9Cross-sectional40 (52)Children and adults8 years on averageSmall-bowel biopsy, AGA, EmA, ARANo
Selby et al.30Cross-sectional39 (89)Adults0.6–29 yearsSmall-bowel biopsy, AGA, EmANo differences from naturally gluten-free diet
Lohiniemi et al.20Cross-sectional48 (53)Adults9–11 yearsQuality of life, small-bowel biopsy to 23No
Peräaho et al.13Prospective, randomized20 (40)Adults1 yearSmall-bowel biopsy, AGA, EmA, BMD, quality of lifeNo differences from naturally gluten-free diet

Table 3 shows how the daily amount of gluten load depends on the gluten level and the intake of flours. In the present study, we showed that coeliac disease patients have been using products containing 50–100 ppm gluten or even more. The daily amount of flours has mostly been <100 g (Figure 2). Provided that the threshold for gluten content is set at 100 ppm and the daily intake is 300 g, the gluten amount does not exceed 30 mg. Such a gluten load has been shown to be safe in both clinical practice and challenge studies (Tables 1 and 2). The assay used here is new, and recognizes only one epitope found in wheat, rye and barley, but not in oats and other cereals. This is a promising method to solve problems in gluten analysis of food products.16

Table 3.  Estimated amount of daily ingested gluten from different amounts of gluten-free foods with varying gluten contamination. Six slices of bread correspond to approximately 100 g baking mix
Gluten content in products, ppm (=mg/kg)Amount of daily gluten-free foods (g)
 Amount of daily gluten ingestion (mg)
100 5102030
 50 2.5 51015
 20 1 2 4 6

The results of this study, in accord with those in previous publications, suggest that the acceptable threshold for gluten content in gluten-free products can be from the clinical point of view set at 100 ppm (=mg/kg). This ensures a wide variety of gluten-free products and does not jeopardize the treatment of coeliac disease with inconvenient restrictions.


The present study and Coeliac Disease Study Group is supported by grants from the Medical Research Fund of Tampere University Hospital, the Foundation of the Friends of the University Children's Hospitals in Finland, the Päivikki and Sakari Sohlberg Foundation, the Finnish Medical Foundation and the Academy of Finland Research Council for Health.