Prediction of Breast Cancer Risk with Volatile Biomarkers in Breath

BACKGROUND: Human breath contains volatile organic compounds (VOCs) that are biomarkers of breast cancer. We investigated the positive and negative predictive values (PPV and NPV) of breath VOC biomarkers as indicators of breast cancer risk. METHODS: We employed ultra-clean breath collection balloons to collect breath samples from 54 women with biopsy-proven breast cancer and 124 cancer-free controls. Breath VOCs were analyzed with gas chromatography (GC) combined with either mass spectrometry (GC MS) or surface acoustic wave detection (GC SAW). Chromatograms were randomly assigned to a training set or a validation set. Monte Carlo analysis identified significant breath VOC biomarkers of breast cancer in the training set, and these biomarkers were incorporated into a multivariate algorithm to predict disease in the validation set. In the unsplit dataset, the predictive algorithms generated discriminant function (DF) values that varied with sensitivity, specificity, PPV and NPV. RESULTS: Using GC MS, test accuracy = 90% (area under curve of receiver operating characteristic in unsplit dataset) and cross-validated accuracy = 77%. Using GC SAW, test accuracy = 86% and cross-validated accuracy = 74%. With both assays, a low DF value was associated with a low risk of breast cancer (NPV \u3e 99.9%). A high DF value was associated with a high risk of breast cancer and PPV rising to 100%. CONCLUSION: Analysis of breath VOC samples collected with ultra-clean balloons detected biomarkers that accurately predicted risk of breast cancer