Esophageal cAncer Screening Trial

Overview

This multicenter study aims to include 15000 participants undergoing screening upper gastrointestinal endoscopy and establish a risk prediction model for esophageal squamous cell carcinoma and esophagogastric junctional (EGJ) adenocarcinoma in high-risk areas. The prediction model will be built based on epidemiological and cytological features, acquired from the esophageal sponge cytology test. The primary study outcome is the diagnostic performance of the model to detect high-grade lesions (including carcinoma and high-grade intraepithelial neoplasia) of the esophagus and EGJ. Secondary outcomes include the number needed to screen, and dignostic performance of cytologist under AI assistance and abnornal cell count.

Esophageal cancer is the seventh most common cancer and the sixth leading cause of cancer related death worldwide in 2020. The highest-incidence areas include Eastern and Central Asia, sub-Saharan Africa, and some South American countries, with esophageal squamous cell carcinoma (ESCC) being the dominant histologic subtype (nearly 90%). While esophageal adenocarcinoma, as a type of adenocarcinoma of the esophagogastric junction (AEG), was less common in these high-risk areas, but accounts for approximately two-thirds of esophageal cancer cases in Europe and North America. However, cardiac gastric cancer, another type of AEG, often shows high incidence in ESCC high-risk areas. Strikingly, 57.5% of global ESCC cases and 60.8% of cardiac gastric cancer cases were estimated to occur in China in 2018. Most esophageal cancer high-risk areas locate in developing countries with limited medical resources, high incidence and co-occurrence ESCC and AEG have posed significant challenges for disease control. Both ESCC and AEG demonstrate dismal prognosis, with an overall 5-year survival of less than 30%. This is largely due to the asymptomatic nature of early-stage lesions, which frequently leads to delayed diagnosis. Population-based studies in China have confirmed that upper gastrointestinal endoscopy with Lugol's staining could increase the early detection rate and decrease both the incidence and mortality of ESCC and the mortality of AEG in very high-risk areas. However, the invasiveness and resource intensiveness of endoscopic procedure and relative low prevalence of target lesions in the general screening population (0.8% to 1.6%) limit its feasibility and cost-effectiveness in massive screening. Therefore, developing less invasive, readily accessible methods with acceptable diagnostic accuracy is urgently needed for enrichment of high-risk individuals prior to endoscopy. To date, few non-endoscopic preliminary screening methods were available for ESCC and AEG. The esophageal sponge cytology is an emerging, minimally invasive, and convenient test for esophageal cancer as well as its precancerous lesions and conditions. The expandable sponge could harvest epithelial cells from the entire esophagus and the esophagogastric junction (EGJ) during withdrawn, and subsequent morphological or biomarker analysis could be performed. For Barrett esophagus (BE), the Cytosponge-trefoil factor 3 is a well-established non-endoscopic test, and could increase the detection of BE for 10.6-fold among patients with reflux symptoms in the primary care setting. For high-risk areas, where ESCC is the dominant subtype, several types of sponge cytology tests also gained promising yet preliminary results. In our previous study, we have developed an artificial intelligence (AI)-assisted sponge cytology test, which automatically indicated potentially abnormal cells for cytologist diagnosis, and found the sensitivity and specificity to be 90.0% and 93.7% in community-based ESCC screening in China. However, the inadequacy of experienced cytologists makes this method still suboptimal for population-based screening, especially for resource-limited areas. Furthermore, AEG, which is also within the detection range of the sponge cytology, has not been investigated in previous studies. Therefore, we conducted the Esophageal cAncer Screening Trial (EAST) to develop and validate a fully-automated machine learning model based on cytological and epidemiological features for ESCC and AEG screening in high-risk areas.