Society: AGA

Background: Tools predicting incident esophageal adenocarcinoma (EAC) and gastric cardia adenocarcinoma (GCA) that can be automated in electronic health records (EHRs) to guide screening decisions are needed. However, EHRs often have missing data for smoking, body mass index (BMI), and gastroesophageal reflux disease (GERD) which are important factors used by currently validated tools (Trøndelag Health [HUNT] and Kunzmann). We aimed to accurately predict EAC/GCA even with missing data using machine learning.
Methods: We performed retrospective analyses in the Veterans Health Administration (VHA) Corporate Data Warehouse among Veterans with ≥1 encounter between 2005 and 2018. Cases diagnosed with EAC/GCA were identified in the VHA Central Cancer Registry. The index date was the date of diagnosis for cases and randomly selected for controls. We collected prescriptions, laboratory results, and International Classification of Diseases diagnoses 1 to 5 years prior to index. We randomly divided the cohort into training (50%), preliminary validation (25%), and testing (25%). In the preliminary validation set, simple random sampling imputation and extreme gradient boosting machine learning were most accurate. In the test set, we compared the final model, the Kettles Esophageal and Cardia Adenocarcinoma predictioN (K-ECAN) Tool, to HUNT, Kunzmann, and published guidelines. To simulate a non-VHA population, we randomly under-sampled males. We ranked the proportion of the total gain in the loss function and the mean Shapley Additive Explanations for each variable.
Results: We identified 8,430 cases of EAC, 2,965 of GCA, and 10,256,887 controls. The mean age was 59.6 years, 92% were male, 80% white, and mean BMI 29.3 kg/m². In the test set, K-ECAN was well calibrated (Figure 1) and had better discrimination (area under the receiver operating characteristics curve [AUC] = 0.77) than HUNT (AUC = 0.68), Kunzmann (AUC = 0.64), or guidelines (Figure 2). Using only data from 4-5 years prior to index slightly diminished its accuracy (AUC = 0.75). Under-sampling men to simulate a non-VHA population, the AUCs of HUNT and Kunzmann improved, but K-ECAN was still most accurate (AUC = 0.85, Figure 2). The most important variables influencing K-ECAN included 4 known risk factors (age, race, sex, BMI) and 9 novel (COPD, greater Hct, lower HDL, greater LDL, lower serum CO₂, lower Na, lower BUN, lower ALT, and greater WBC). While GERD was strongly associated with EAC, it only contributed a small proportion of gain in information.
Conclusions: We developed and internally validated a novel prediction tool for incident EAC/GCA using EHR data. K-ECAN identifies individuals who are at increased risk for EAC/GCA ≥ 3 years in advance and is more accurate than published guidelines. Further work is needed to validate K-ECAN outside VHA and to assess how best to implement it within EHRs.

<b>Figure 1. Calibration Plot</b><br /> Predicted and observed risks are the cumulative incidences per 100,000 individuals in the testing set over the 14 years of ascertainment. Each dot represents 2% of the testing set (51,398 individuals).

Figure 1. Calibration Plot
Predicted and observed risks are the cumulative incidences per 100,000 individuals in the testing set over the 14 years of ascertainment. Each dot represents 2% of the testing set (51,398 individuals).

<b>Figure 2. Receiver Operating Characteristic Curves.</b><br /> <b>Panel A: Entire Testing Set. </b>AUCs [95% CIs] are displayed in parentheses.<br /> <b>Panel B: Simulated non-VHA Population</b><br /> Because Kunzmann and HUNT were both developed in populations that were roughly 50% male and both rely heavily on sex for classifying risk, this analysis simulated a non-VHA population by using all available female controls and cancer cases and under-sampled men by including a random selection of an equal number of male controls as female controls and a random sample of male cancer cases to match the expected odds ratio of male sex of 8.33 in the US population.<br /> ACG: American College of Gastroenterology, ACP: American College of Physicians, AGA: American Gastroenterological Association, ASGE: American Society for Gastrointestinal Endoscopy, BSG: British Society of Gastroenterology, ESGE: European Society for Gastrointestinal Endoscopy, HUNT: Trøndelag Health, VHA: Veterans Health Administration

Figure 2. Receiver Operating Characteristic Curves.
Panel A: Entire Testing Set. AUCs [95% CIs] are displayed in parentheses.
Panel B: Simulated non-VHA Population
Because Kunzmann and HUNT were both developed in populations that were roughly 50% male and both rely heavily on sex for classifying risk, this analysis simulated a non-VHA population by using all available female controls and cancer cases and under-sampled men by including a random selection of an equal number of male controls as female controls and a random sample of male cancer cases to match the expected odds ratio of male sex of 8.33 in the US population.
ACG: American College of Gastroenterology, ACP: American College of Physicians, AGA: American Gastroenterological Association, ASGE: American Society for Gastrointestinal Endoscopy, BSG: British Society of Gastroenterology, ESGE: European Society for Gastrointestinal Endoscopy, HUNT: Trøndelag Health, VHA: Veterans Health Administration

Introduction: Eosinophilic esophagitis (EoE) is a rare, chronic disease characterized by infiltration of eosinophils (eos) into the esophagus and symptoms related to esophageal dysfunction, such as dysphagia. Benralizumab, an eos-depleting anti-IL-5Rα monoclonal antibody, reduced eos in the esophagus in smaller studies and may be effective in EoE.
Methods: MESSINA (NCT04543409) was a phase 3, randomized (1:1), double-blind, placebo-controlled study evaluating benralizumab 30 mg subcutaneously every 4 weeks. Patients were 12-65 years with symptomatic (i.e., dysphagia) and histologically active EoE (≥15 eos/high-power field [hpf] on ≥2 esophageal levels); stable standard-of-care therapies were allowed. Dual primary endpoints were proportion of patients with histologic remission (≤6 eos/hpf) and change from baseline (CFB) in Dysphagia Symptom Questionnaire (DSQ) score at Week 24. Key secondary endpoints included percentage CFB in peak tissue eos and absolute CFB in Eosinophilic Esophagitis Histology Scoring System (EoE-HSS) grade and stage scores and Endoscopic Reference Score (EREFS). Symptoms were also assessed using the Eosinophilic Esophagitis Daily Dysphagia Diary (EoE-3D) and the Patient Global Impression of Severity (PGI-S). Selected patients underwent a sub-study with additional endoscopies and biopsies at Weeks 4 & 12 to understand the time course and impact of tissue eos depletion.
Results: A total of 210 patients (103 benralizumab; 107 placebo) were randomized and received treatment. Baseline characteristics were balanced between groups. At Week 24, a significantly higher proportion of patients on benralizumab vs placebo achieved histologic remission (87.4% vs 6.5%; p<0.0001); however, there was no difference in CFB in DSQ score between groups (–12.1 vs –15.1; p=0.1770) (Figure 1). Nominal statistical significance was observed for benralizumab vs placebo across the histology-based secondary endpoints (percentage CFB in peak tissue eos and EoE-HSS grade and stage scores) but not in CFB in EREFS at Week 24 (Figure 2). There were no differences between treatment groups for symptoms as measured by the EoE-3D and PGI-S. The rate of adverse events was 64.1% for patients on benralizumab and 61.7% for placebo. Benralizumab was well tolerated with no new safety signals. Additionally, of the 25 sub-study patients (11 benralizumab; 14 placebo), a higher proportion of patients on benralizumab vs placebo achieved histologic remission, and they had a higher percentage CFB in peak tissue eos at Weeks 4 & 12 than placebo, which was maintained to Week 24 (Figure 2).
Conclusions: Benralizumab reduced esophageal eos at 4 weeks and maintained this reduction to 24 weeks without a significant improvement in symptoms for patients with EoE vs placebo, suggesting that eos are not the primary driver of disease activity and symptomology.

Figure 1

Figure 2

Harald Fjallbrant

Appearances