Background: Pancreatic cancer (PDAC) has been associated with development of glucose intolerance and dysregulation. Based on our previous report of strong potential for existence of metabolite biomarkers for PDAC detection, in this study, we have pursued initial characterization of plasma metabolites for discriminating PDAC cases among diabetic patients as a means to delineate further subset of patients who may benefit from screening protocol.

Methods: IRB-approved prospectively collected database and corresponding plasma from UC Davis Pancreas Registry was searched for appropriate cases: PDAC patients with recent-onset diabetes (≤3yrs prior to the index PDAC diagnosis) (Grp 1) and non-PDAC patients with long history of diabetes mellitus (>3 yrs) (Grp 2).

The plasma was processed and non-targeted GC/MS and targeted LC/MS were performed to assay for identifiable plasma metabolites. The data was processed and the peak intensity levels were normalized, scaled and identifiable metabolites were analyzed via metabolomics specific multivariate & bootstrap methods as well as ROC evaluation (Metaboanalyst & ROCCET) for visualization and biomarker development.

Results: We have identified 36 patients in Grp 1 and 22 patients for Grp 2. The metabolite specific multivariate and ROC analyses revealed specific features such as elaidic acid, uric acid, 2,3-propanediol, arachidonic acid, docosahexanoic acid, 5-oxo-EET, lysine, LysoPC(18:2), 9(10)-EpOME, LysoPC(16:0), sphingosine-1-phosphate and others as strong discriminators for recent-onset diabetic patients with pancreatic cancer. Such set yielded AUC of 0.964 with 15 features [CI=0.876-1].

Conclusion: We have demonstrated that an identifiable set of plasma metabolite biomarkers exists for discriminating diabetic patients with developing pancreatic cancer. These features could be utilized in a larger set of new-onset diabetic population for further validation in near-future study.

Citation Format: Shiro Urayama, Vladimir Tolstikov. Discriminant, identifiable plasma metabolites in pancreatic cancer-associated diabetes - candidate early-detection biomarkers. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 895. doi:10.1158/1538-7445.AM2014-895

©2014 American Association for Cancer Research.
2014