eCite Digital Repository

Discovery of serum biomarkers for Tasmanian Devil cancer (DFTD)

Citation

Karu, N and Wilson, RR and Hamede Ross, RK and Jones, Menna and Woods, GM and Hilder, EF and Shellie, RA, Discovery of serum biomarkers for Tasmanian Devil cancer (DFTD), Proceedings of The Australian & New Zealand Metabolomics Conference, 31st March - 1st April, 2016, Melbourne, Australia (2016) [Conference Extract]

Abstract

The Tasmanian Devil is endangered due to Devil Facial Tumour Disease (DFTD), a transmissible and fatal cancer. Currently there is no test to diagnose DFTD in the latent stage, prior to overt signs of the disease. This work describes the first moderate-scale metabolomics study of Tasmanian devil. The aim was to identify DFTD biomarkers in serum that could be used in a pre-DFTD diagnostic model. RPLC-MS (Thermo LTQ-Orbitrap) was utilised for metabolic fingerprinting of 140 serum samples from wild Tasmanian devils trapped near Cradle Mountain, Tasmania (2013-2015). DFTD prediction was modelled by SVM, random forests and logistic regression, based on 35 DFTD and 35 healthy samples as training set, and 70 samples as test set, including 35 latent devils (3-12 months prior to signs of DFTD). Models were assessed by cross-validations and permutations, and ROC curves plotted to examine specificity and sensitivity. Feature identification was by mass search in libraries, by chemical standards and MS/MS spectra. The best-performing test was a random forests model of 13 biomarkers (5 elevated in DFTD), mostly amino acids and derivatives. For the training set, the area under ROC curve was 0.97 (95% CI 0.88-1), average accuracy of prediction 89% (100 cross-validations) and p<0.001 after 500 permutations. The DFTD prediction rate in the test set was 62% (16/26 devils) 3-6 months before DFTD, and 44% (4/9 samples) up to 12 months before DFTD. The false positive rate was 10% (1/10) in both sets. Independent biomarker discovery for biological interpretation utilising age-matched subset (n=22x2) as training set, found and further assessed 40 metabolites with significant change in DFTD (p<0.02, FDR<0.05; FC≥1.3). These included common endogenous metabolites and also 10 small peptides >5-fold higher in DFTD. The gathered data provide the basis for a pre-DFTD diagnostic test by targeted quantitative analysis, which is the next step.

Item Details

Item Type:Conference Extract
Keywords:DFTD, metabolomics, biomarkers
Research Division:Biological Sciences
Research Group:Biochemistry and Cell Biology
Research Field:Cell Metabolism
Objective Division:Health
Objective Group:Health and Support Services
Objective Field:Diagnostic Methods
Author:Karu, N (Dr Naama Karu)
Author:Wilson, RR (Dr Richard Wilson)
Author:Hamede Ross, RK (Dr Rodrigo Hamede Ross)
Author:Jones, Menna (Associate Professor Menna Jones)
Author:Woods, GM (Professor Gregory Woods)
Author:Hilder, EF (Professor Emily Hilder)
Author:Shellie, RA (Associate Professor Robert Shellie)
ID Code:108640
Year Published:2016
Deposited By:Central Science Laboratory
Deposited On:2016-04-29
Last Modified:2017-03-20
Downloads:0

Repository Staff Only: item control page