Potential Celiac Children: 9-Year Follow-up on a Gluten-containing Diet

Renata Auricchio MD, PhD; Antonella Tosco MD; Emanuela Piccolo MD; Martina Galatola PhD; Valentina Izzo PhD; Mariantonia Maglio PhD; Francesco Paparo PhD; Riccardo Troncone MD, PhD; Luigi Greco MD, PhD


Am J Gastroenterol. 2014;109(6):913-921. 

In This Article


A sizeable proportion of potential celiac patients without clinical symptoms show fluctuating or negative antibody production, of those persistently positive for anti-TG2 the majority did not develop mucosal damage after 9 years of follow-up. Similar findings were recently described by Biagi et al.[22] in adult patients.

Despite having a similar prevalence in very different populations and being associated with the same genes in those populations, CD shows a multifaceted phenotype, ranging from no symptoms to severe, life-threatening conditions. The degree of gluten-induced mucosal damage is also quite variable, independently of the type of symptoms, ranging from light intraepithelial infiltration to total mucosal atrophy. Before the discovery of anti-TG2 antibodies, which bring very high sensitivity and specificity, the diagnosis was often based on the clinical phenotype associated with severe mucosal damage. Only recently we could identify individuals with no clinical symptoms, production of anti-TG2, and light or absent mucosal damage. These are the potential celiac patients, named after the assumption that they will, or may, develop over time a gluten-induced mucosal damage peculiar to the celiac status. The easy management of these patients is just to put them on a gluten-free diet on the basis of the production of anti-TG2, without considering the lack of damage in the target tissue. However, it has been often observed[23] that anti-TG2 production may not persist forever: indeed, many potential CD patients show a persistent serum negativity for anti-TG2 antibodies and no symptoms while on a gluten-containing diet. Many of them would not accept a gluten-free diet for life without a conclusive proof of a permanent gluten-induced disease. The persistence of autoantibodies in the serum is, on the other hand, a possible threat in the long run this risk has indeed to be considered in the management of potential celiac cases.[10,11] Unfortunately, no long-term longitudinal study of these individuals is actually available.

We previously suggested that potential celiacs might not only have a peculiar phenotype but might also have some different genetic profile resulting in different gene expression in the target mucosa.[13] In our cohort, HLA was frequently of the low-risk classes, and IL-21 expression was suppressed in the mucosa, suggesting the lack of activation of the mechanism of cellular killing and mucosal damage typical of classical CD. We may speculate that potential celiacs are a live model of the progression of the gluten-induced immune response, worth exploring to increase our understanding of the factors related to the progression of the disease. We included in the study only potential CD patients who had no clinical symptoms, neither growth impairment nor subtle laboratory findings such as anemia or low blood iron. A gluten-free diet was prescribed to all those who preferred early treatment. In those on a gluten-containing diet, we observed a slow progression of the mucosal damage: after 3 years, few individuals had progressed (13%), but afterward there was a significant change, with 26% developing mucosal damage by 6 years and 32% progressing to villous atrophy by 9 years.

It should be noted that boys, a minority with respect to girls, progress significantly more frequently than girls: a Carter effect should be considered, wherein boys have a higher load of risk factors to become celiacs.[24] The level of anti-TG2 at entry was a minor, but significant, predictor of the outcome, whereas mucosal inflammation, as estimated by the numbers of CD25+ cells and γδ+ lymphocytes, was significantly associated with the outcome. Intestinal deposits of anti-TG2 did not contribute to the prediction of cases who progress to mucosal atrophy because of the presence, in a multivariate function, of powerful competitors, but this does not diminish the relevance to clinical practice of this finding, as shown by the survival curve.

It is interesting to note that three SNPs of three genes associated with CD contributed significantly to the identification of individuals who eventually develop mucosal damage. These genes were strongly implicated in the pathogenesis of the gluten-induced damage and were consistently associated with an increased risk of familial recurrence.[20] It is interesting that the IL2/IL21 gene marker on chromosome 4 was selected as a good predictor also. We previously showed that the potential CD phenotype is associated with marked downregulation of the expression of this gene in intestinal mucosa, revealing a possible functional role of the associated polymorphism. Recent studies suggested a role of interleukin 12A in mediating the immune response in different diseases.[25,26] It is indeed possible that the genetic profile associated with the progression of the disease in our cohort confirms the involvement of IL12A in generating the mucosal damage in potential CD.[27,28]

The pattern of anti-TG2 antibody production over time is associated with the final outcome, as reported elsewhere.[22] Seventeen percent of the potentials stopped producing anti-TG2 antibodies over time, whereas 35% showed a fluctuating antibody production, as observed in other autoimmune conditions.[29] Forty percent of the potentials had a stable elevation of antibody production: they started from a mean level at entry significantly higher than the other two groups and had a higher rate of progression to mucosal damage.

By combining, at entry, different risk factors associated with the outcome in a multivariate discriminant function, it was possible to predict the outcome of 74% of the potentials. This may appear to be an overoptimistic estimate, but the cross-validation, through the sequential exclusion of each patient from the equation by which he was classified, indicates the robustness of this classification: 73% of cross-validated cases were correctly classified. Male sex, light mucosal inflammation at time 0, and a certain genetic profile (in IL12A, OLIG3, and IL2/IL21) may well start to delineate a cohort of individuals who react differently to the immunologically recognized gluten peptides. It should be noted that the genotypes of the associated genes are not independent of each other: only a slight difference in the power to discriminate may include one gene and exclude another one, merely because the previous one already explained enough variance. The variance ratio at step 1 of the analysis shows that several genes produced a similar contribution to the discrimination (Supplementary Table S3 online). Thus, the genotypes of OLIG3, RGS1, IL2/IL21, c-REL, and SCHIP1 had substantially similar contributions, but the selection of OLIG3 and IL2/IL21 also explained the residual variance of the other genes.

The very reason to develop a multivariate analysis model is to explore which variable or risk factor is likely to differentiate the patients who progress to flat mucosa from those who do not. Each variable is entered in the model after considering the contribution of all others. To evaluate the reliability of the model, a blinded prediction of the classification of cases is performed. Indeed, a 73% prediction is not a "safe" prediction, but this performance makes a substantial theoretical contribution to the understanding of the extensive polymorphism of the gluten-induced response in different individuals. We do not intend, at this phase, to keep any individual patient on a gluten-free or gluten-containing diet but rather to tailor the level of care to the peculiar characteristics of each individual.

The major limitation of this study is time: mucosal damage may develop over decades, and we were able to follow-up this cohort for less than one decade. However, the surveillance of the natural history of potential celiac children is the best option to protect their health. To date, there was no study of similar duration. The final enrolled population was 133/210 (63%), a substantial and representative portion of the original cohort, not very common in long-term survival studies.[29]

There is no doubt that it may not be possible to explore the phenotype of potential celiacs any further: as soon as we are aware of any risk to the patients, we advise them to start a gluten-free diet immediately. By the end of follow-up, a sizeable proportion of them showed fluctuating production of anti-TG2 antibodies, often ending in a complete cessation of their production. The majority of them had not developed mucosal damage after 9 years. If we ignore these findings, we are likely to prescribe a permanent gluten-free diet for individuals who might not benefit from such a major disruption of their daily life.

In conclusion, the relevance of this study consists both in highlighting the natural history of potential CD and in the identification of risk factors for the development of villous atrophy.