High Accuracy Prostate Cancer Detection Using Human Blood Sera Metabolomic Profiling

XIAOLING ZANG; CHRISTINA JONES; TRAN QUOC LONG; MARÍA EUGENIA MONGE; MANSHUI ZHOU; L. DEETTE WALKER; ALEXANDER GRAY; JOHN F. MCDONALD; NIKHIL SHAH; RAJESH LAUNGANI; FACUNDO M. FERNANDEZ

Minneapolis

Conferencia; 61st ASMS Conference on Mass Spectrometry & Allied Topic; 2013

American Society for Mass Spectrometry

Introduction Prostate cancer is the sixth deadliest cancer among men. Although the Prostate-Specific Antigen (PSA) test has been widely used to screen for prostate cancer, certain advisory groups recommend against its use because it suffers from false positive results and over-treatment. These drawbacks have led to the increased interest of using metabolomic profiling to discover new differential metabolic biomarkers that could improve the specificity of prostate cancer diagnosis. However, global metabolomic profiling of prostate cancer patients is still at an early stage. In this study, Ultra Performance Liquid Chromatography (UPLC®) coupled to mass spectrometry and machine learning methods were used to profile and identify metabolites in blood sera samples from prostate cancer patients that discriminate them from healthy individuals. Methods Age-matched serum samples from prostate cancer patients (n=64, age range: 49-65, x ̅= 59 ± 4) and healthy individuals (n=53, age range: 45-76, x ̅= 58 ± 8) were profiled by UPLC®-MS, using a Waters ACQUITY H Class system fitted with a Waters ACQUITY UPLC® BEH C18 column (2.1 × 50 mm, 1.7 µm) and coupled to a Synapt G2 High Definition Mass Spectrometer (Waters Corporation). Metabolites were extracted using a mixture of acetone, acetonitrile and methanol (1:1:1 v/v). Dichloromethane was used to extract the lipid fraction. High resolution mass spectra were acquired for the non-lipid fraction in negative ionization mode for m/z 50-2000. Metabolomic features were obtained using MarkerLynx 4.1 software. Support vector machines (SVMs) were used for sample classification. Preliminary data Serum samples were successfully classified as cancerous or healthy based on their metabolic features (retention time, m/z pairs) with 92% accuracy, 98% sensitivity, and 88% specificity using SVMs. Of the 51 metabolic features found to best discriminate prostate cancer patients from healthy individuals, 40 were tentatively associated with metabolites in the Human Metabolome Database (HMDB). Several metabolites may be differentially produced/consumed in prostate cancer progression and proliferation, based on previous reported findings. Amino acid derivatives and dipeptides identified potentially result from alterations in amino acid metabolism associated with prostate cancer. Additionally, dipeptides might be produced by the proteolysis pathway of hepsin, a cell surface serine protease found to be markedly upregulated in human prostate cancer. Steroid derivatives such as 5α-dihydrotesterone sulfate were identified as well, similarly to published work suggesting the involvement of steroid metabolism in prostate cancer progression. Other differential metabolites include components of the cellular membrane, such as fatty acids and cholesterol sulfate, which may be related to increased membrane lipid synthesis and high energy demand in proliferating prostate cancer cells. Furthermore, the polyamine N1,N12-diacetylspermine, tentatively identified in our work, has been previously reported to detect early stage colorectal and breast cancers. UPLC®-MS/MS experiments will be performed to confirm the identities of the metabolites responsible for classification, providing further insights into the metabolic alterations and related biological pathways associated with prostate cancer. Novel aspect High classification accuracy suggests this approach is promising for accurate prostate cancer detection.