Differentiation between genetic mutations of breast cancer by breath volatolomics

• Published in: Oncotarget Vol. 6; no. 42; pp. 44864 - 44876
• Main Authors: Barash, Orna, Zhang, Wei, Halpern, Jeffrey M, Hua, Qing-Ling, Pan, Yue-Yin, Kayal, Haneen, Khoury, Kayan, Liu, Hu, Davies, Michael P A, Haick, Hossam
• Format: Journal Article
• Language: English
• Published: United States Impact Journals LLC 29-12-2015
• Subjects: Adult; Aged; Artificial Intelligence; Biomarkers, Tumor - genetics; Biomarkers, Tumor - metabolism; Breast Neoplasms - genetics; Breast Neoplasms - metabolism; Breath Tests - methods; Carcinoma, Intraductal, Noninfiltrating - genetics; Carcinoma, Intraductal, Noninfiltrating - metabolism; Case-Control Studies; Diagnosis, Differential; Female; Gas Chromatography-Mass Spectrometry; Genetic Predisposition to Disease; Humans; Microarray Analysis; Middle Aged; Mutation; Nanotechnology; Phenotype; Predictive Value of Tests; Research Paper; Volatile Organic Compounds - metabolism; Young Adult; breast cancer; spectrometry; molecular; sensor; volatolomic
• ISSN: 1949-2553; 1949-2553
• DOI: 10.18632/oncotarget.6269

Mapping molecular sub-types in breast cancer (BC) tumours is a rapidly evolving area due to growing interest in, for example, targeted therapy and screening high-risk populations for early diagnosis. We report a new concept for profiling BC molecular sub-types based on volatile organic compounds (VOCs). For this purpose, breath samples were collected from 276 female volunteers, including healthy, benign conditions, ductal carcinoma in situ (DCIS) and malignant lesions. Breath samples were analysed by gas chromatography mass spectrometry (GC-MS) and artificially intelligent nanoarray technology. Applying the non-parametric Wilcoxon/Kruskal-Wallis test, GC-MS analysis found 23 compounds that were significantly different (p < 0.05) in breath samples of BC patients with different molecular sub-types. Discriminant function analysis (DFA) of the nanoarray identified unique volatolomic signatures between cancer and non-cancer cases (83% accuracy in blind testing), and for the different molecular sub-types with accuracies ranging from 82 to 87%, sensitivities of 81 to 88% and specificities of 76 to 96% in leave-one-out cross-validation. These results demonstrate the presence of detectable breath VOC patterns for accurately profiling molecular sub-types in BC, either through specific compound identification by GC-MS or by volatolomic signatures obtained through statistical analysis of the artificially intelligent nanoarray responses.