Genetic Markers May Predict PPI Response in Pediatric EoE

A new multicenter study published in Frontiers in Pharmacology provides exploratory evidence that genetic variability in drug metabolism genes may help predict which children with eosinophilic esophagitis (EoE) will respond to proton pump inhibitor (PPI) therapy. The research, led by Mariamena Arbitrio and Licia Pensabene, investigated 58 pediatric patients to identify pharmacogenomic determinants of treatment response, moving beyond the well-established CYP2C19 pathway to explore a broader set of absorption, distribution, metabolism, and excretion (ADME) genes.

Eosinophilic esophagitis is a chronic immune-mediated condition characterized by esophageal dysfunction and inflammation. It has become the leading cause of dysphagia in children and young adults, with incidence and prevalence increasing over recent decades. PPIs are currently recognized as a first-line treatment option due to their low cost, favorable safety profile, and ease of administration. However, response rates in pediatric populations vary widely, ranging from 30% to 71%, and identifying responders currently relies on an empirical trial of the drug followed by repeat endoscopy and biopsies.

This reliance on empirical trials means that patients who do not respond are exposed to unnecessary invasive procedures, prolonged symptoms, and an increased risk of complications such as food impaction and fibrostenotic disease. Consequently, there is a critical need for reliable non-invasive or minimally invasive predictive markers that can stratify patients before treatment begins. The study aimed to determine whether genetic variants in 231 ADME genes are associated with histologic and clinical response to PPI treatment, potentially offering a pharmacogenomic biomarker for patient stratification.

The study included 58 pediatric patients diagnosed with EoE at two Italian centers: the Pediatrics Unit of Magna Graecia University in Catanzaro and the Gastroenterology and Allergy Units of Bambino Gesù Children’s Hospital in Rome. The cohort consisted of 52 males and 6 females, with a mean age of 15.4 years. All patients received double-dose PPI therapy, primarily lansoprazole, for 8 to 12 weeks. Patients were classified as responders if they achieved both symptomatic improvement, defined as a 30% reduction in the Pediatric Eosinophilic Esophagitis Symptom Score, and histological remission, defined as fewer than 15 eosinophils per high-power field on biopsy.

Genomic DNA was extracted from peripheral blood and genotyped using the Drug Metabolizing Enzymes and Transporters (DMET) Plus microarray. This platform covers 231 ADME-related genes, including 1,931 single nucleotide polymorphisms (SNPs) and 5 copy number variations. After quality-control filtering, 348 informative markers were retained for the final association analysis. The researchers integrated these genetic data with clinical variables using a machine learning approach, specifically a decision-tree classifier, to identify combinatorial patterns that might predict treatment response.

The strongest nominal association with PPI response was observed for the CHST3 rs4148944 G/G genotype. This genotype was present in all 17 responders (100%) and in 28 of the 41 non-responders (68.3%). The unadjusted p-value for this association was 0.0064. However, the study explicitly notes that this association did not meet the pre-specified significance threshold after multiple-testing correction, with an FDR-adjusted q-value of 0.0643. Therefore, the authors emphasize that this finding is exploratory and hypothesis-generating rather than a statistically confirmed biomarker.

Integrative analysis revealed that the predictive signal was not isolated to CHST3 alone. The machine learning model identified combinatorial patterns involving CHST3, CHST11 (rs7847), SULT1A2/A3 (rs11859842), and NR1I3 (rs55802895). Notably, the presence of food allergy emerged as a relevant clinical feature that improved patient stratification when combined with these genetic variants. The decision tree achieved an accuracy of 0.8333 on an independent test set of 12 patients, with an area under the receiver operating characteristic curve (ROC-AUC) of 0.7812. These metrics suggest moderate discriminatory ability, but the authors caution that the small test set limits the precision of these estimates.

The biological plausibility of these findings lies in the role of these genes in sulfation pathways. CHST3 and CHST11 encode carbohydrate sulfotransferases involved in glycosaminoglycan sulfation, a key process in extracellular matrix organization. Alterations in sulfation patterns may affect cytokine binding, growth factor signaling, and mucosal barrier integrity, potentially influencing the inflammatory response and tissue remodeling in EoE. SULT1A2/A3 and NR1I3 are involved in phase II metabolism and nuclear receptor-mediated regulation, respectively, suggesting that drug response variability may be influenced by interactions between metabolic, regulatory, and inflammatory pathways.

Importantly, the study found no significant association between PPI response and CYP2C19 variants, which have previously been linked to PPI metabolism. Genotype distributions for CYP2C19 were comparable between responders and non-responders, and no metabolizer phenotype was significantly associated with response status. This finding extends the pharmacogenomic framework for EoE beyond the well-established CYP2C19 and STAT6 pathways, although STAT6 could not be evaluated because it is not represented on the DMET Plus platform.

The study has several limitations that must be considered. The cohort size was relatively small, with only 17 responders, which increases the risk of overfitting and limits the generalizability of the findings. The independent test set consisted of only 12 patients, making the performance estimates inherently imprecise. Additionally, the DMET platform provides targeted coverage of ADME genes but does not capture the full spectrum of genomic variability, including variants identified in genome-wide association studies or genes involved in immune regulation. The retrospective assessment of clinical data also meant that a minority of patients received different PPIs or treatment durations, although all received guideline-recommended doses.

Despite these limitations, the study provides proof-of-concept evidence that integrating pharmacogenomic and clinical data can help identify EoE patient subgroups with distinct PPI responsiveness. The identification of CHST3 and related pathways suggests a potential link between sulfation processes, the tissue microenvironment, and pharmacological response. The machine learning approach enabled the detection of interaction patterns that might not be readily apparent using conventional univariate statistical methods.

The authors conclude that these findings should be regarded as exploratory and require validation in larger, independent cohorts before clinical application can be considered. Future studies integrating additional omics layers, such as transcriptomic and proteomic data, will be essential to further elucidate the mechanisms underlying PPI responsiveness. If validated, such pharmacogenomic approaches could support more personalized therapeutic strategies, potentially preventing delays in achieving remission and reducing the burden of unnecessary invasive procedures for children with EoE.

This research contributes to the growing body of evidence supporting the role of precision medicine in eosinophilic esophagitis. By identifying candidate genetic markers and clinical features that may predict treatment response, the study paves the way for more tailored management strategies. The integration of genetic profiling with clinical allergy data represents a promising opportunity to enhance the accuracy of patient stratification and improve clinical outcomes in pediatric EoE.

Subject of Research: Pharmacogenomic predictors of proton pump inhibitor response in pediatric eosinophilic esophagitis

Article Title: Pharmacogenomic predictors of proton pump inhibitor response in pediatric eosinophilic esophagitis

Article References: Arbitrio, M., Tambucci, R., Tallarico, V., Gnasso, I., Scionti, F., Agapito, G., Altomare, E., Cannataro, M., D'Arrigo, G., Tripepi, G., Concolino, D., De Masi, F., Puoti, M. G., Rea, F., Malamisura, M., Ricci, A. M., Artesani, M. C., Urbani, S., Calandrelli, V., … Pensabene, L. (2026). Pharmacogenomic predictors of proton pump inhibitor response in pediatric eosinophilic esophagitis. Frontiers in Pharmacology, 17, Article 1913372. https://doi.org/10.3389/fphar.2026.1913372

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1913372

Keywords: eosinophilic esophagitis, proton pump inhibitors, pharmacogenomics, CHST3, pediatric medicine, drug response, machine learning, ADME genes, sulfation pathways, clinical trials, Pharmacogenomic, predictors