The objective of this study was to detect and evaluate reliable metabolite markers for screening and monitoring treatment of patients with esophageal adenocarcinoma (EAC) by studying metabolomics. The sensitivity and specificity of the study were evaluated not only for EAC but also for Barrett esophagus and high-grade dysplasia, which are widely regarded as precursors of EAC.

Profiles of metabolites in blood serum were constructed using nuclear magnetic resonance spectroscopy and statistical analysis methods. The metabolite biomarkers discovered were selected to build a predictive model that was then used to test the classifications accuracies.

Eight metabolites showed significant differences in their levels in patients with cancer and in the control group on the basis of Student t test. A partial least-squares discriminant analysis model built on these metabolites provided excellent classifications of patients with cancer and the control group, with the area under the receiver operating in a characteristic curve of > 0.85 for both training and validation sample sets. Evaluated by the same model, the Barrett esophagus samples were of mixed classification, and the high-grade dysplasia samples were classified primarily as cancer samples. A pathway study indicated that altered energy metabolism and changes in the trochloroacetic acid cycle were the dominant factors in the biochemistry of EAC.

¹H nuclear magnetic resonance–based metabolite profiling analysis was shown to be an effective approach to differentiating between patients with EAC and healthy subjects. Good sensitivity and selectivity were shown by using the 8 metabolite markers discovered to predict the classification of samples from the healthy control group and the patients with the disease. Serum metabolic profiling may have potential for early diagnosis of EAC and may enhance our understanding of its mechanisms.