Identification of highly sensitive biomarkers that can aid the early detection of pancreatic cancer using GC/MS/MS-based targeted metabolomics.

ABSTRACT

BACKGROUND: To improve prognosis of pancreatic cancer (PC) patients, the discovery of more reliable biomarkers for the early detection is desired. METHODS: Blood samples were collected by 2 independent groups. The 1st set was included 55 early PC and 58 healthy volunteers (HV), and the 2nd set was included 16 PC and 16HV. The 16 targeted metabolites were quantitatively analyzed by gas chromatography/tandem mass spectrometry together with their corresponding stable isotopes. In the 1st set, the levels of these metabolites were evaluated, and diagnostic models were constructed via multivariate logistic regression analysis, leading to validation using the 2nd set. RESULTS: In the 1st set, model X consisting of 4 candidates based on our previous report possessed higher sensitivity (74.1%) than carbohydrate antigen 19-9 (CA19-9). Model Y, consisting of 2 metabolites newly selected from 16 metabolites via stepwise method possessed higher sensitivity (70.4%) than CA19-9. Furthermore, combining model Y with CA19-9 increased its sensitivity (90.7%) and specificity (89.5%). In the 2nd set, combining model Y with CA19-9 displayed high sensitivity (81.3%) and specificity (93.8%). In particular, it displayed very high sensitivity (100%) for resectable PC. CONCLUSIONS: Quantitative analysis confirmed that metabolomics-based diagnostic methods are useful for detecting PC early.