From the Journals

Algorithm predicts villous atrophy in children with potential celiac disease

View on the News

Evidence-based prediction with a grain of salt

While the simplification of the diagnostic process for celiac disease (CD), now heavily reliant on CD-specific autoantibodies, has made the life of clinicians easier in many respects, new scenarios also have emerged that are posing new challenges. One of them is that a substantial, growing portion of subjects (who may or may not have symptoms) present with positive CD autoantibodies but a normal duodenal mucosa (“potential celiac patient”). If left on gluten, with time some will develop villous atrophy, but some won’t. What is the clinician supposed to do with them?

Dr. Stefano Guandalini

The paper by Auricchio et al. addresses this issue in a rigorous, well-structured way by closely prospectively monitoring a large series of pediatric patients. Their conclusions have very useful implications for the clinician. In fact taking into consideration several criteria, they found valuable after a long observation period – such as age of the child, HLA status, persistence of elevated CD-specific autoantibodies, and presence or absence of intraepithelial lymphocytes in the initial biopsy – they concluded that one can correctly identify at the beginning four out of five potential celiac patients who will not develop villous atrophy, and thus do not need to follow a gluten-free diet.

Ultimately, however, let’s not forget that we are still dealing with percentages of risk to develop full-blown CD, not with definitive certainties. Hence, the decision of starting a gluten-free diet or not (and of how often and in which way to monitor those who remain on gluten) remains a mutually agreed upon plan sealed by two actors: on one side the patient (or the patient’s family); and on the other, an experienced health care provider who has clearly explained the facts. In other words, evidence-based criteria, good old medicine, and a grain of salt!

Stefano Guandalini, MD, is a pediatric gastroenterologist at the University of Chicago Medical Center. He has no conflicts of interest.



A new algorithm may be able to predict which children with potential celiac disease will go on to develop villous atrophy, according to investigators writing in Gastroenterology.

The risk model was developed from the largest cohort of its kind, with the longest follow-up to date, reported lead author Renata Auricchio, MD, PhD, of University Federico II in Naples, Italy, and colleagues. Using the algorithm, which relies most heavily on a baseline number of intraepithelial lymphocytes (IELs) in mucosa, followed by age at diagnosis and genetic profile, clinicians may now consider prescribing gluten-free diets to only the highest-risk patients, instead of all suspected cases, noting that more than half of potential cases do not develop flat mucosa within 12 years.

Development of the algorithm began with enrollment of 340 children aged 2-18 years who were positive for endomysial antibodies immunoglobulin A antibodies and had tested positive twice consecutively for antitissue transglutaminase antibodies. Additionally, children were required to possess HLA DQ2- or DQ8-positive haplotypes and have normal duodenal architecture in five biopsy samples. Because of symptoms suggestive of celiac disease or parental discretion, 60 patients were started on a gluten-free diet and excluded from the study, leaving 280 patients in the final cohort. These patients were kept on a gluten-containing diet and followed for up to 12 years. Every 6 months, the investigators checked antibodies and clinical status, and every 2 years, small bowel biopsy was performed, if symptoms had not necessitated this earlier.

After a median follow-up of 60 months, ranging from 18 months to 12 years, 39 patients (13.9%) developed symptoms of celiac disease and were placed on a gluten-free diet, although they declined confirmatory biopsy, disallowing classification of celiac disease. Another 33 patients (11.7%) were lost to follow-up and 89 (32%) stopped producing antibodies, with none going on to develop villous atrophy. In total, 42 patients (15%) developed flat mucosa during the follow-up period, with an estimated cumulative incidence of 43% at 12 years. The investigators noted that patients most frequently progressed within two time frames; at 24-48 months after enrollment, or at 96-120 months.

To develop the algorithm, the investigators performed multivariable analysis with several potential risk factors, including age, sex, genetic profile, mucosal characteristics, and concomitant autoimmune diseases. Of these, a high number of IELs upon first biopsy was most highly correlated with progression to celiac disease. Patients who developed villous atrophy had a mean value of 11.9 IELs at first biopsy, compared with 6.44 among those who remained potential (P = .05). The next strongest predictive factors were age and genetic profile. Just 7% of children less than 3 years developed flat mucosa, compared with 51% of patients aged 3-10 years and 55% of those older than 10 years (P = .007). HLA status was predictive in the group aged 3-10 years but not significant in the youngest or oldest patients. Therefore, HLA haplotype was included in the final algorithm, but with smaller contribution than five non-HLA genes, namely, IL12a, SH2B3, RGS1, CCR, and IL2/IL21.

“Combining these risk factors, we set up a model to predict the probability for a patient to evolve from potential celiac disease to villous atrophy,” the investigators wrote. “Overall, the discriminant analysis model allows us to correctly classify, at entry, 80% of the children who will not develop a flat mucosa over follow-up, while approximately 69% of those who will develop flat mucosa are correctly classified by the parameters we analyzed. This system is then more accurate to predict a child who will not develop flat mucosa and then can be monitored on a gluten-containing diet than a child who will become celiac.”

The investigators noted that IEL count may be an uncommon diagnostic; however, they recommended the test, even if it necessitates referral. “The [IEL] count turned out to be crucial for the prediction power of the discriminant analysis,” the investigators wrote.

“The long-term risks of potential celiac disease have never been accurately evaluated. Thus, before adopting a wait-and-see strategy on a gluten-containing diet, a final decision should always be shared with the family.”

Still, the investigators concluded that gluten-free diet “should not be prescribed indistinctly to all patients” with potential celiac disease, as it is a “very heterogenic condition and is not necessarily the first step of overt disease.”

The investigators disclosed no funding or conflicts of interest.

SOURCE: Auricchio R et al. Gastroenterology. 2019 Apr 9. doi: 10.1053/j.gastro.2019.04.004.

Next Article: