Celiac disease is an autoimmune-like systemic disorder in genetically susceptible individuals, perpetuated by daily ingested gluten cereals wheat, rye and barley and with manifestations in the small- intestinal mucosa and in organs outside the gut. The gold standard for celiac disease diagnosis is the finding of gluten-induced small-intestinal mucosal injury [1, 2]. The mucosa will heal upon introduction of a gluten-free diet and the mucosal damage will reappear if gluten is reintroduced [2]. Extensive time-course studies have provided evidence that during gluten challenge an inflammatory process with a dose-dependent accumulation of intraepithelial lymphocytes (IELs) is followed by mucosal villous atrophy with crypt hyperplasia [3]. In untreated celiac disease the degree of malabsorption is determined by the length of the functionally impaired bowel and the presence of symptoms is noway related to the histological features of a proximal biopsy [3–5]. More recently, it was concluded that only patients with extensive and severe enteropathy will give evidence of steatorrhea and increased intestinal permeability[6]. Biopsy has thus remained the golden standard in measuring gluten-induced ill health.

Previously a gluten challenge was part of the routine diagnostic procedure in children and a characteristic mucosal lesion reappeared in most patients within two years[2, 7]. In gluten challenge studies, the amount of gluten in normal food has been estimated to be as high as 10-20 g per day[8–10]. A reduced daily gluten dose (2.5-5 g per day) during a gluten-free diet has still enabled mucosal healing in half of the patients [11]. Also some adults challenged with 2.5-5 g of gluten daily for three to 14 months have shown no significant changes in mucosal morphology [12]. These findings are in marked conflict with those from a recent challenge study where it was shown that even a minimal dose of 50 mg gluten daily induced a significant mucosal deterioration [13]. Altogether, the patient groups in these studies have been rather small and no unequivocal conclusions can be drawn.

A gluten-free diet is socially troublesome and expensive and research has shown compliance to the diet often to be poor [14]. Gluten contamination in a gluten-free diet is also difficult to avoid [15, 16]. For these reasons many celiac patients have expressed a desire for other alternative or complementary treatments, less burdensome than a strict life-long gluten-free diet [17, 18]. Any drug under evaluation as clinically effective and trusted by celiac disease patients should significantly reduce or prevent gluten-induced mucosal injury. However, data are insufficient as to the amount of gluten to be used in a challenge study and for how long.

In this pilot study we challenged adult celiac disease patients with gluten aiming to gain a clearer conception of the gluten amount needed to cause some mucosal deterioration but without inducing excessive ill health with a resultant dropout of trial subjects. We challenged treated celiac disease patients with low to moderate amounts of gluten for up to 12 weeks to establish whether a morphometrically measurable mucosal deterioration could be detected. We especially measured separately mucosal morphological change as villous height crypt depth ratio (Vh/CrD) and the inflammatory aspect in terms of density of mucosal IELs, both known to be sensitive continuous functional parameters of gluten ingestion in celiac disease. Further, we sought to establish whether the gluten-dependent serum antibody markers show seroconversions.

Altogether four celiac disease patients belonging to the moderate gluten dose group discontinued the study due to abdominal symptoms shortly after the challenge was initiated (after three, eight, ten and 23 days), and they were excluded from the final analyses. Celiac serology remained negative in all. The first two drop-outs with extremely brief gluten challenges underwent the follow-up endoscopy, and no small- bowel mucosal deterioration was evident.

During the gluten challenge, a clinically significant decrease in Vh/CrD was found in 14 (67%) out of the 21 celiac patients; seven were in the moderate and seven in the low gluten dose group (Table 2, Figures 1 and 2). The Vh/CrD in the moderate-dose group deteriorated from a mean 3.0 (95%CI 2.8-3.2) to 1.7 (95%CI 1.0-2.4) and in the low-dose group from 2.9 (95%CI 2.5-3.4) to 1.9 (95%CI 1.2-2.6), both changes being statistically significant (p = 0.002 and p = 0.016, respectively). The changes in Vh/CrD did not correlate with daily gluten intake (r = -0.138, p = 0.552).

In line with the morphological findings, a clinically significant change in the density of CD3+ IELs was seen in altogether 14 (67%) subjects; eight in the moderate-dose and six in the low-dose of (Table 2, Figures 3 and 4). In the moderate gluten dose group the IEL count increased significantly from a mean 49 cell/mm (95%CI 38-60) to 88 cell/mm (95%CI 70-106, p = 0.001). In the low-dose group the increase did not reach statistical significance (from a mean 78 cell/mm [95%CI 56-100] to 98 cell/mm [95%CI 61-135], p = 0.188). All in all, daily gluten intake did not correlate with increased inflammation (gluten intake vs. ΔCD3+ IEL count, r = 0.251, p = 0.273).

Altogether nine (43%) celiac disease patients, five in the moderate- and four in the low-dose gluten challenge group, showed significant increases in TG2 antibody titers (Table 3). The subjects consuming moderate doses of gluten evinced clearly higher TG2 antibody titer responses than those consuming low doses; this was already seen after four weeks' challenge (Figure 5). Interestingly, after the highest positive titer during the gluten challenge, the serum TG2 antibody levels tended to decrease even if gluten consumption continued (Figure 5). Serum EMA and DGP antibody titers were in line with TG2 antibody results; furthermore, after the gluten challenge all these three antibodies appeared in the serum simultaneously (data not shown). Small-bowel mucosal TG2-targeted autoantibody deposits generally followed the gluten-induced serological response in sera and morphological changes seen in the biopsy specimens (Tables 2 and 3).

At the outset, all subjects were free of gastrointestinal symptoms. Altogether 15 (71%) out of the 21 celiac disease patients experienced mild to moderate abdominal symptoms upon gluten challenge; eight of them (including the four withdrawing prematurely, see above) were from the moderate-dose and seven from the low-dose group (Table 2). None developed severe symptoms. Thirteen (87%) out of the 15 symptomatic patients also developed small-intestinal mucosal changes (significant decrease in Vh/CrD and/or increase in the density of CD3+ IELs). Of note, symptoms did not reveal four celiac disease patients having clinically significant gluten-induced small-bowel mucosal damage.

Twelve weeks' gluten challenge with both moderate and low amounts of gluten sufficed to induce small-bowel mucosal morphological damage in the majority of volunteer treated celiac disease adults not, however, in all. An increase in the density of CD3+ IELs paralleled this phenomenon especially in the moderate-dose gluten challenge group. At the same time, moderate gluten doses triggered more gastrointestinal symptoms leading to premature withdrawals from the study. This suggests that in the 12-week time-frame smaller gluten doses of 1-3 g per day might be preferable to higher doses in order to obtain sufficient complete read-outs at the end of a challenge study.

In this study, small-bowel mucosal Vh/CrD measurement seemed to be the most sensitive parameter in both moderate and low gluten doses in assessing gluten-induced damage in celiac disease. This conception is also in line with previous findings, where the Vh/CrD index has been used to show harmful effects of gluten in both short-term gluten challenges and long-term follow-up studies [8, 13, 26, 27]. It is not surprising that the IEL count was a reliable marker for gluten toxicity only with moderate, but not with low gluten doses, as it has previously been shown that in celiac disease during a gluten-challenge, IELs accumulate in the surface and crypt epithelium in a dose-dependent manner [3]. It is also important to note that even if the majority of celiac disease patients after gluten challenge yielded positive read-outs both in the Vh/CrD index and in the IEL count, there were some discrepant cases responding only with one or other parameter (Table 2). Marsh classification entailing six diagnostic categories (type 0-type 3c) has been widely used in the diagnostic work-up and follow-up of celiac disease [3]. Marsh grades type 1-type 3c all indicate cursorily that small-bowel mucosal IEL density is increased (>40 IELs/100enterocytes), but different categories point more specifically to different morphological features (such as normal villous morphology, crypt hyperplasia or villous atrophy). Nonetheless the Marsh classification does not take into account minor gluten- induced mucosal changes (e.g. a significant decrease in Vh/CrD or increase in density of IELs within one Marsh grade) or discrepant results in Vh/CrD or IEL counts during a gluten challenge. It would thus be advisable in gluten challenge research projects, to use two separate continuous readout parameters in assessing gluten-dependent functional mucosal changes, i.e. the morphological Vh/CrD and the inflammatory IEL changes, as opposed to applying a simplified grouped classification.

Even if abdominal symptoms after the consumption of cereals do not generally constitute a specific marker of untreated celiac disease [28], gluten-induced gastrointestinal symptoms in our challenge study showed high positive predictive value (87%) for small-bowel mucosal injury in celiac disease patients. However, in 22% of the cases who developed significant small-intestinal damage, symptoms remained absent. It is widely known that many untreated celiac disease patients do not suffer from symptoms even if they have a manifest gluten-induced small-bowel mucosal lesion [29]. Furthermore, in some treated celiac disease patients, symptoms may disappear on a strict gluten-free diet despite continued ongoing mucosal villous atrophy [30, 31]. Such persistent small-intestinal villous atrophy in symptom-free patients clearly carries a risk of subsequent severe complications such as osteoporosis and malignancies [30]. This warrants caution in relying solely on symptoms or other patient-related outcomes without small-bowel biopsies in gluten challenge research projects.

In studies where a gluten challenge has been applied for diagnostic purposes in children on a normal daily gluten-containing diet (10-20 g daily), a small-bowel mucosal histological relapse has usually developed within three to twelve months in the majority of patients [8, 9]. Seroconversion of EMA and TG2 antibodies is known to be a reliable predictor of histological relapse and antibody testing has helped in the timing of the biopsy [9]. In this study with low to moderate gluten doses, only 43% of the challenged adult celiac disease patients evinced a significant serological response. In line with earlier challenge studies, here all those who had increased serum autoantibodies during the challenge also developed small-bowel villous atrophy. However, at the same time serology missed nine cases with significant small-bowel mucosal changes (Tables 2 and 3). These cases might have called for a longer gluten challenge period or higher gluten amounts. However, it is well-known that serology may fail to detect occasional slight dietary transgressions and incomplete mucosal healing [30, 32, 33], and therefore celiac serology is not always a reliable means of assessing gluten-induced harm. Furthermore, it is important to note that after the gluten challenge all three antibody - EMA, TG2 and DPG antibody - tests performed similar way, and recently introduced DPG antibody test was not superior to conventional serological assays. As small-bowel mucosal TG2-specific autoantibody deposits have also been found in seronegative celiac patients and patients with early-stage celiac disease [23, 24, 34], it has been hypothesized that these mucosal autoantibodies might also appear early during a gluten challenge. In our study, mucosal autoantibodies followed the morphological damage, and they added only little to the normal morphometrical measurements. It might be that after a short-term acute gluten challenge the distribution of mucosal autoantibody depositions is patchy, and single mucosal biopsies may not reveal the positivity. It remains to be seen whether they will be of use at earlier time-points in the gluten challenge, but due to the invasive nature of small-bowel biopsies we were not able to take specimens on several occasions during the challenge.

There were some celiac disease patients who did not respond in terms of any of the measured outcome variables during the 12 weeks' gluten challenge. It is well-known that celiac patients show varying sensitivity to gluten and the process of mucosal deterioration may take years or even decades [35–37]. Furthermore, it should be remembered that the phenotype of the disease may vary in the same person over time; celiac patients suffering originally from abdominal symptoms at diagnosis may later experience only extraintestinal symptoms such as dermatitis herpetiformis when gluten is re-introduced into the diet [37]. These issues need to be taken into account when sample sizes in study cohorts are determined.

At the baseline of the study, the majority of the treated celiac disease patients evinced increased densities of IELs and one patient even had villous atrophy with crypt hyperplasia despite adherent to a strict long-term gluten-free diet. According to the literature, such cases exist and it has been shown that even 80% of treated patients may have ongoing villous atrophy during a strict diet [38–40]. These persistent mucosal changes might be due to constant residual gluten contamination in the food, which again would indicate that alternative treatment options would be desirable.

All in all, further studies are still needed to establish the optimal gluten dose on the other hand causing some small-bowel mucosal deterioration but without inducing excessive ill health. Our results suggest that in drug trials, the duration of the gluten challenge might be shorter than 12 weeks and the gluten dose could be an average of those used in the current study (approximately 3g). As there are always some celiac disease patients who will not respond within these conditions, sample sizes must be sufficiently large to ensure adequate statistical power for the study. The Vh/CrD index seemed to be the most reliable outcome measurement in both moderate and low gluten doses. Furthermore, it must be remembered that during a gluten challenge, some celiac patients may remain asymptomatic even if they develop manifest gluten-induced mucosal damage - this might in the long-run predispose these cases to celiac disease-related complications.