Serum Metabolite Profiling for the Detection of Pancreatic Cancer

Hirofumi Akita, Shawn A. Ritchie, Ichiro Takemasa, Hidetoshi Eguchi, Elodie Pastural, Wei Jin, Yasuyo Yamazaki, Dayan B. Goodenowe, Hiroaki Nagano, Morito Monden, Masaki Mori, Yuichiro Doki

Research output: Contribution to journalArticlepeer-review

17 Citations (Scopus)


Objectives To improve the detection of pancreatic cancer (PC), a robust diagnostic biomarker is essential. We have previously discovered 4 serum metabolites (PC-594, lysophosphatidylcholine, phosphatidylcholine, and sphingomyelin) in distinguishing patients with PC from healthy controls. Here, we report the results of our validation phase by using larger numbers of independent and blinded samples. Methods We collected 3 mL of serum from 116 patients with PC and 138 healthy controls. Samples were blinded and expression of the 4 candidate metabolites in each sample was determined by triple quadrupole tandem mass spectrometry. We then used cutoffs established in the discovery phase to predict the disease state of each of the validation samples. Results All 4 metabolites showed significantly lower expression in patients with PC compared with healthy controls. PC-594 showed 73.3% sensitivity and 92.0% specificity, whereas the other 3 metabolites showed 58.6% and 92.0%, 76.7% and 69.6%, and 58.6% and 81.9% sensitivity and specificity, respectively. Area under the receiver operating characteristic curve for PC-594 alone was 0.92, whereas a combination method using all 4 metabolites showed 86.2% sensitivity and 84.8% specificity. Conclusions Our validation results confirmed that a reduction in PC-594, along with 3 other serum-based choline metabolites, is highly associated with PC.

Original languageEnglish
Pages (from-to)1418-1423
Number of pages6
Issue number10
Publication statusPublished - Nov 1 2016

All Science Journal Classification (ASJC) codes

  • Internal Medicine
  • Endocrinology, Diabetes and Metabolism
  • Hepatology
  • Endocrinology


Dive into the research topics of 'Serum Metabolite Profiling for the Detection of Pancreatic Cancer'. Together they form a unique fingerprint.

Cite this