Celiac disease (CD) is an immune-mediated systemic disorder triggered by ingested gluten and related prolamines in genetically susceptible individuals (Husby et al., 2012; Green et al., 2015). CD has progressively increased in prevalence and is associated with significant morbidity, impacting quality of life and health care costs (Biagetti et al., 2015; Burden et al., 2015; Catassi et al., 2015). Overall prevalence is nearly 1%, ranging from 0.3 to 5% in different regions (Catassi et al., 2015; Green et al., 2015). In Chile, a National Health Survey performed in 2009 assessed a representative sample of individuals older than 15 years of age using anti-transglutaminase (tTG) antibodies detection and estimated a prevalence of 0.8% (Minsal Chile, 2010). Initially considered a gluten dependent enteropathy, CD is currently recognized as a systemic autoimmune disease with clinical manifestations ranging from asymptomatic to malabsorptive chronic diarrhea associated with severe malnutrition (Husby et al., 2012; Agardh et al., 2015; Green et al., 2015; Schøsler et al., 2016). Likewise, the intensity of inflammation and damage of the intestinal mucosa in untreated patients varies widely, from the absence of any detectable lesions to total mucosal atrophy (Donaldson et al., 2008; Husby et al., 2012). Mechanisms underlying this wide clinical and histological spectrum remain elusive.

The intestinal microbiota is currently under intense evaluation as a potentially relevant pathogenic factor in CD. Differential microbiota patterns in duodenal biopsies and stool samples have been described for CD patients as compared to healthy controls (Collado et al., 2008, 2009; Di Cagno et al., 2011; Cenit et al., 2015). Helicobacter pylori (H. pylori), a Gram-negative, spiral bacterium capable of infecting human gastric mucosa, has also been isolated from the duodenum (Csendes et al., 1995; Kim et al., 2004; Nagasawa et al., 2009). There is evidence that H. pylori is associated with an increased regulatory T cell (T-reg) response in blood and gastric mucosa, mainly in children, and specifically with higher number of Foxp3+ T cells and higher expression of Foxp3 and TGF-β mRNA/protein among other markers (Lundgren et al., 2005; Gil et al., 2014; Yang et al., 2014). In particular, CagA, a pleomorphic virulence factor of H. pylori, involved in oncogenesis and epithelial barrier disfunction, has also been involved in T-reg pathway activation (Kido et al., 2011) (reviewed in Stein et al., 2013). Meta-analysis presented in recent systematic reviews suggest a protective role of H. pylori against atopy, an immune mediated disorder (Lionetti, 2014; Taye et al., 2015), and some studies have proposed a modulator role in CD. Lebwohl et al. reported a strong inverse association between H. pylori presence in gastric mucosa and CD in a large cross-sectional study (Lebwohl et al., 2013). However, these findings have not been reproduced elsewhere (Jozefczuk et al., 2015; Simondi et al., 2015). Although, Simondi et al. did not find a difference in the prevalence of H. pylori between celiac and non-celiac individuals, they described an association of the presence of this bacterium in the gastric mucosa of celiac patients with milder duodenal lesions, suggesting H. pylori plays an immune regulator role (Simondi et al., 2015). Konturek et al. described lower seroprevalence of anti-cagA antibodies in celiac patients compared to healthy controls, suggesting a possible protective role of H. pylori harboring specific virulence factors (Konturek et al., 2000).

We aimed to determine, using a case-control design, whether H. pylori gastric infection, specifically with cagA+ strains, is associated with reduced rates of duodenal atrophy in celiac patients. Our secondary aimwas to compare duodenal expression of representative T-reg markers among CD individuals with and without infection by cagA+ H. pylori and different histological severity. Our hypothesis was that infection with cagA+ H. pylori would be more common among individuals with milder forms of CD and associated with an increase in expression of T-reg markers.

In this case-control study including predominantly young patients, overall H. pylori infection rates were similar between subjects with and without CD, and were independent of the severity of duodenal atrophy in the latter. However, in individuals infected by H. pylori, detection of cagA+ strains was significantly more common among individuals with potential-CD compared to active-CD and non-CD patients. A similar H. pylori prevalence between CD and non-CD subjects have been previously reported. Aydogdu et al. found a 22% of H. pylori prevalence in 96 Turkish CD children compared to 24% in 235 non-CD children (mean age of 8 years old; Aydogdu et al., 2008). Similar prevalence rates were reported by Luzza et al in Italian children with and without CD (Luzza et al., 1999). On the other hand, Simondi et al. determined a higher, but also similar prevalence of H. pylori in CD and non-CD adults (36 and 41%, respectively; mean age of 41 years old; Simondi et al., 2015). In contrast to these findings, Lebwohl et al. reported a strong inverse association between the presence of H. pylori in gastric mucosa and CD in a large cross-sectional study that included 136,179 patients undergoing upper endoscopy in the USA (OR = 0.48; 95%; CI: 0.40–0.58; Lebwohl et al., 2013). This study was performed in a country with a low-prevalence of H. pylori (prevalence 4.4 and 8.8% in CD and non-CD subjects, respectively) and included older patients compared to our clinical series (mean age 51 years old vs. 13 years old), which could in part explain the discrepancy.

Milder histological damage among CD patients infected with H. pylori was previously reported by Villanacci et al. (in a group of 80 CD adults H. pylori was detected in 7/9 individuals Marsh 1–2 and 23/71 Marsh 3; Villanacci et al., 2006), Aydogdu et al. (in a group of 96 CD children H. pylori was detected in 56% of Marsh 1, 38% of Marsh 2 and 5% of Marsh 3, p < 0.05; Aydogdu et al., 2008), and Simondi et al. (in a group of 73 CD adults H. pylori was detected in 50% of Marsh 1–2 and 33% of Marsh 3; p > 0.05; Simondi et al., 2015). Our study differs from the above mentioned because the association between H. pylori and CD related histological damage was the primary and not secondary outcome and we additionally explored the possible effect of a virulence factor (cagA). Overall H. pylori prevalence did not differ between potential- and active-CD patients (both nearly 30–40%); however among infected subjects, the presence of cagA+ strains was significantly associated with milder histological damage (Marsh 0–1). Prevalence of cagA+ H. pylori strains vary between populations between 40 and 80% (Feliciano et al., 2015; O'Ryan et al., 2015; Sayehmiri et al., 2015; Scarpulla et al., 2015) and could explain at least in part the difference observed between our study and previous findings. The presence/absence of the cagA+ gene among infecting H. pylori strains is a factor that may be modulating the severity of duodenal lesions. Konturek et al. reported that anti-CagA antibodies were detected significantly less frequently in H. pylori+ celiac subjects than in healthy H. pylori+ controls (Konturek et al., 2000), which is in line with our findings.

H. pylori infections, which are acquired mostly early in life in Chilean children (O'Ryan et al., 2015), may have a limited role in overall development of CD, possibly modulating disease in subsets of individuals. The fact that for potential-CD subjects infected with H. pylori most are infected with cagA+ strains, in contrast to subjects with active disease where most are infected with cagA- strains, suggests that infection with the former pathotype may be modulating disease to less severe forms of presentation.

Conversely, it is possible that gastric/duodenal tissue of these potential-CD individuals is more prone to be colonized with cagA+ strains. However, counter evidence to this hypothesis was reported in a previous Chilean cohort study in which acquisition of persistent H. pylori infection occurred mainly during the first 3 years of life (O'Ryan et al., 2015), an age that is 10 years younger than the median age of our series of patients. It is reasonable to extrapolate that in most of our cases, H. pylori infection preceded the development of CD. Whether persistently infected children change strains over time, from cagA+ to cagA- (or vice versa), remains as yet an open question requiring a longitudinal study design.

The number of positive stained Foxp3 cells in duodenal lamina propria and the expression of TGF-β1 were higher in active than in potential CD patients, consistent with findings by Tiittanen et al. (2008), Brazowski et al. (2010), and Borrelli et al. (2013). In active-CD, the T-reg pathway is fully activated downregulating the ongoing Th1/Th17 inflammatory process (Tiittanen et al., 2008). However, in vitro evidence suggests that effector T cells become resistant to suppression by T-regs in this setting (Hmida et al., 2012). According to our results, this increase in Foxp3 cells was even more evident in active-CD patients infected with cagA+ H. pylori strains. Although duodenal inflammatory profiles have not been previously explored in patients infected with H. pylori, this agent has been associated with a local gastric and systemic response characterized by Th1/Th17 inflammatory processes as well as activation of T-reg pathways contributing to infection persistence (Gil et al., 2014; Hussain et al., 2016). The T-reg pathway is especially enhanced in children compared to adults (Freire de Melo et al., 2012; Serrano et al., 2013; Gil et al., 2014) and in individuals infected with cagA+ H. pylori strains (Kido et al., 2011; Hussain et al., 2016). We hypothesize that priming with cagA+ H. pylori at a young age, may protect against chronic inflammatory processes and loss of food/self antigen tolerance through activation of the T-reg pathway.

Further studies of cytokine expression in cultured duodenal biopsies of CD patients exposed in vitro to H. pylori with and without cagA gene and cohort studies of high risk CD patients infected and non-infected with cagA+ H. pylori strains are needed to confirm this proposal.

Animal models fully reproducing CD are not currently available but comparing potential-CD with active-CD individuals is an interesting “in vivo” model to advance knowledge on mechanisms underlying this condition. The identification of factors that could modulate histological damage and inflammation in CD can potentially open new avenues for prevention and treatment aiming to improve the quality of life of these patients.

In conclusion, H. pylori infection rates were similar among individuals with and without CD, but infection with cagA+ strains was significantly associated with milder histological damage in celiac patients infected by H. pylori and in active-CD cases with higher expression of T-reg markers. These results suggest that infection by cagA+ H. pylori may be protective for CD progression, or conversely, that these strains are prone to colonize when intestinal damage is less severe. Future research should focus on clarifying mechanisms involved in this association.

Substantial contributions to the conception or design of the work: YL, AO, MO, RQ, NE, RV, and MA. Contribution to the acquisition, analysis, or interpretation of data for the work: YL, AO, MO, RQ, NE, RV, DS, EA, CA, MF, AV, IG, JM, JC, GE, MG, JO, EM, FP, MC, and MA. Drafting the work or revising it critically for important intellectual content: YL, AO, MO, RQ, NE, RV, DS, EA, CA, MF, AV, IG, JM, JC, GE, MG, JO, EM, FP, MC, and MA. Final approval of the version to be published: YL, AO, MO, RQ, NE, RV, DS, EA, CA, MF, AV, IG, JM, JC, GE, MG, JO, EM, FP, MC, and MA. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: YL, AO, MO, RQ, NE, RV, DS, EA, CA, MF, AV, IG, JM, JC, GE, MG, JO, EM, FP, MC, and MA.