Patients with pancreatic cancer (PC) are usually diagnosed at late stages, when the disease is nearly incurable. Sensitive and specific markers are critical for supporting diagnostic and therapeutic strategies. The aim of this study was to use a metabonomics approach to identify potential plasma biomarkers that can be further developed for early detection of PC. In this study, plasma metabolites of newly diagnosed PC patients (n = 100) and age- and gender-matched controls (n = 100) from Connecticut (CT), USA, and the same number of cases and controls from Shanghai (SH), China, were profiled using combined gas and liquid chromatography mass spectrometry. The metabolites consistently expressed in both CT and SH samples were used to identify potential markers, and the diagnostic performance of the candidate markers was tested in two sample sets. A diagnostic model was constructed using a panel of five metabolites including glutamate, choline, 1,5-anhydro-d-glucitol, betaine, and methylguanidine, which robustly distinguished PC patients in CT from controls with high sensitivity (97.7%) and specificity (83.1%) (area under the receiver operating characteristic curve [AUC] = 0.943, 95% confidence interval [CI] = 0.908-0.977). This panel of metabolites was then tested with the SH data set, yielding satisfactory accuracy (AUC = 0.835; 95% CI = 0.777-0.893), with a sensitivity of 77.4% and specificity of 75.8%. This model achieved a sensitivity of 84.8% in the PC patients at stages 0, 1, and 2 in CT and 77.4% in the PC patients at stages 1 and 2 in SH. Plasma metabolic signatures show promise as biomarkers for early detection of PC.
- Pancreatic cancer
- logistic regression
- multivariate statistical analysis
ASJC Scopus subject areas