Metabolomic profile in pancreatic cancer patients: A consensusbased approach to identify highly discriminating metabolites

Iole Maria Di Gangi, Tommaso Mazza, Andrea Fontana, Massimiliano Copetti, Caterina Fusilli, Antonio Ippolito, Fulvio Mattivi, Anna Latiano, Angelo Andriulli, Urska Vrhovsek, Valerio Pazienza

Research output: Contribution to journalArticle

Abstract

Purpose: pancreatic adenocarcinoma is the fourth leading cause of cancer related deaths due to its aggressive behavior and poor clinical outcome. There is a considerable variability in the frequency of serum tumor markers in cancer' patients. We performed a metabolomics screening in patients diagnosed with pancreatic cancer. Experimental Design: Two targeted metabolomic assays were conducted on 40 serum samples of patients diagnosed with pancreatic cancer and 40 healthy controls. Multivariate methods and classification trees were performed. Materials and Methods: Sparse partial least squares discriminant analysis (SPLS-DA) was used to reduce the high dimensionality of a pancreatic cancer metabolomic dataset, differentiating between pancreatic cancer (PC) patients and healthy subjects. Using Random Forest analysis palmitic acid, 1,2-dioleoylsn- glycero-3-phospho-rac-glycerol, lanosterol, lignoceric acid, 1-monooleoylrac- glycerol, cholesterol 5',6' epoxide, erucic acid and taurolithocholic acid (T-LCA), oleoyl-L-carnitine, oleanolic acid were identified among 206 metabolites as highly discriminating between disease states. Comparison between Receiver Operating Characteristic (ROC) curves for palmitic acid and CA 19-9 showed that the area under the ROC curve (AUC) of palmitic acid (AUC=1.000; 95% confidence interval) is significantly higher than CA 19-9 (AUC=0.963; 95% confidence interval: 0.896-1.000). Conclusion: Mass spectrometry-based metabolomic profiling of sera from pancreatic cancer patients and normal subjects showed significant alterations in the profiles of the metabolome of PC patients as compared to controls. These findings offer an information-rich matrix for discovering novel candidate biomarkers with diagnostic or prognostic potentials.

Original languageEnglish
Pages (from-to)5815-5829
Number of pages15
JournalOncotarget
Volume7
Issue number5
DOIs
Publication statusPublished - 2016

Fingerprint

Metabolomics
Pancreatic Neoplasms
Area Under Curve
Confidence Intervals
ROC Curve
Glycerol
Biomarkers
Lanosterol
Metabolome
Carnitine
Discriminant Analysis
Tumor Biomarkers
Least-Squares Analysis
Serum
Mass Spectrometry
Neoplasms
Healthy Volunteers
Adenocarcinoma
Research Design

Keywords

  • Metabolomic
  • Pancreatic cancer

ASJC Scopus subject areas

  • Oncology

Cite this

Metabolomic profile in pancreatic cancer patients : A consensusbased approach to identify highly discriminating metabolites. / Di Gangi, Iole Maria; Mazza, Tommaso; Fontana, Andrea; Copetti, Massimiliano; Fusilli, Caterina; Ippolito, Antonio; Mattivi, Fulvio; Latiano, Anna; Andriulli, Angelo; Vrhovsek, Urska; Pazienza, Valerio.

In: Oncotarget, Vol. 7, No. 5, 2016, p. 5815-5829.

Research output: Contribution to journalArticle

@article{859fd3c3888e4e9cadac66178b66877a,
title = "Metabolomic profile in pancreatic cancer patients: A consensusbased approach to identify highly discriminating metabolites",
abstract = "Purpose: pancreatic adenocarcinoma is the fourth leading cause of cancer related deaths due to its aggressive behavior and poor clinical outcome. There is a considerable variability in the frequency of serum tumor markers in cancer' patients. We performed a metabolomics screening in patients diagnosed with pancreatic cancer. Experimental Design: Two targeted metabolomic assays were conducted on 40 serum samples of patients diagnosed with pancreatic cancer and 40 healthy controls. Multivariate methods and classification trees were performed. Materials and Methods: Sparse partial least squares discriminant analysis (SPLS-DA) was used to reduce the high dimensionality of a pancreatic cancer metabolomic dataset, differentiating between pancreatic cancer (PC) patients and healthy subjects. Using Random Forest analysis palmitic acid, 1,2-dioleoylsn- glycero-3-phospho-rac-glycerol, lanosterol, lignoceric acid, 1-monooleoylrac- glycerol, cholesterol 5',6' epoxide, erucic acid and taurolithocholic acid (T-LCA), oleoyl-L-carnitine, oleanolic acid were identified among 206 metabolites as highly discriminating between disease states. Comparison between Receiver Operating Characteristic (ROC) curves for palmitic acid and CA 19-9 showed that the area under the ROC curve (AUC) of palmitic acid (AUC=1.000; 95{\%} confidence interval) is significantly higher than CA 19-9 (AUC=0.963; 95{\%} confidence interval: 0.896-1.000). Conclusion: Mass spectrometry-based metabolomic profiling of sera from pancreatic cancer patients and normal subjects showed significant alterations in the profiles of the metabolome of PC patients as compared to controls. These findings offer an information-rich matrix for discovering novel candidate biomarkers with diagnostic or prognostic potentials.",
keywords = "Metabolomic, Pancreatic cancer",
author = "{Di Gangi}, {Iole Maria} and Tommaso Mazza and Andrea Fontana and Massimiliano Copetti and Caterina Fusilli and Antonio Ippolito and Fulvio Mattivi and Anna Latiano and Angelo Andriulli and Urska Vrhovsek and Valerio Pazienza",
year = "2016",
doi = "10.18632/oncotarget.6808",
language = "English",
volume = "7",
pages = "5815--5829",
journal = "Oncotarget",
issn = "1949-2553",
publisher = "Impact Journals LLC",
number = "5",

}

TY - JOUR

T1 - Metabolomic profile in pancreatic cancer patients

T2 - A consensusbased approach to identify highly discriminating metabolites

AU - Di Gangi, Iole Maria

AU - Mazza, Tommaso

AU - Fontana, Andrea

AU - Copetti, Massimiliano

AU - Fusilli, Caterina

AU - Ippolito, Antonio

AU - Mattivi, Fulvio

AU - Latiano, Anna

AU - Andriulli, Angelo

AU - Vrhovsek, Urska

AU - Pazienza, Valerio

PY - 2016

Y1 - 2016

N2 - Purpose: pancreatic adenocarcinoma is the fourth leading cause of cancer related deaths due to its aggressive behavior and poor clinical outcome. There is a considerable variability in the frequency of serum tumor markers in cancer' patients. We performed a metabolomics screening in patients diagnosed with pancreatic cancer. Experimental Design: Two targeted metabolomic assays were conducted on 40 serum samples of patients diagnosed with pancreatic cancer and 40 healthy controls. Multivariate methods and classification trees were performed. Materials and Methods: Sparse partial least squares discriminant analysis (SPLS-DA) was used to reduce the high dimensionality of a pancreatic cancer metabolomic dataset, differentiating between pancreatic cancer (PC) patients and healthy subjects. Using Random Forest analysis palmitic acid, 1,2-dioleoylsn- glycero-3-phospho-rac-glycerol, lanosterol, lignoceric acid, 1-monooleoylrac- glycerol, cholesterol 5',6' epoxide, erucic acid and taurolithocholic acid (T-LCA), oleoyl-L-carnitine, oleanolic acid were identified among 206 metabolites as highly discriminating between disease states. Comparison between Receiver Operating Characteristic (ROC) curves for palmitic acid and CA 19-9 showed that the area under the ROC curve (AUC) of palmitic acid (AUC=1.000; 95% confidence interval) is significantly higher than CA 19-9 (AUC=0.963; 95% confidence interval: 0.896-1.000). Conclusion: Mass spectrometry-based metabolomic profiling of sera from pancreatic cancer patients and normal subjects showed significant alterations in the profiles of the metabolome of PC patients as compared to controls. These findings offer an information-rich matrix for discovering novel candidate biomarkers with diagnostic or prognostic potentials.

AB - Purpose: pancreatic adenocarcinoma is the fourth leading cause of cancer related deaths due to its aggressive behavior and poor clinical outcome. There is a considerable variability in the frequency of serum tumor markers in cancer' patients. We performed a metabolomics screening in patients diagnosed with pancreatic cancer. Experimental Design: Two targeted metabolomic assays were conducted on 40 serum samples of patients diagnosed with pancreatic cancer and 40 healthy controls. Multivariate methods and classification trees were performed. Materials and Methods: Sparse partial least squares discriminant analysis (SPLS-DA) was used to reduce the high dimensionality of a pancreatic cancer metabolomic dataset, differentiating between pancreatic cancer (PC) patients and healthy subjects. Using Random Forest analysis palmitic acid, 1,2-dioleoylsn- glycero-3-phospho-rac-glycerol, lanosterol, lignoceric acid, 1-monooleoylrac- glycerol, cholesterol 5',6' epoxide, erucic acid and taurolithocholic acid (T-LCA), oleoyl-L-carnitine, oleanolic acid were identified among 206 metabolites as highly discriminating between disease states. Comparison between Receiver Operating Characteristic (ROC) curves for palmitic acid and CA 19-9 showed that the area under the ROC curve (AUC) of palmitic acid (AUC=1.000; 95% confidence interval) is significantly higher than CA 19-9 (AUC=0.963; 95% confidence interval: 0.896-1.000). Conclusion: Mass spectrometry-based metabolomic profiling of sera from pancreatic cancer patients and normal subjects showed significant alterations in the profiles of the metabolome of PC patients as compared to controls. These findings offer an information-rich matrix for discovering novel candidate biomarkers with diagnostic or prognostic potentials.

KW - Metabolomic

KW - Pancreatic cancer

UR - http://www.scopus.com/inward/record.url?scp=84958063989&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84958063989&partnerID=8YFLogxK

U2 - 10.18632/oncotarget.6808

DO - 10.18632/oncotarget.6808

M3 - Article

AN - SCOPUS:84958063989

VL - 7

SP - 5815

EP - 5829

JO - Oncotarget

JF - Oncotarget

SN - 1949-2553

IS - 5

ER -