This paper outlines the design and performance of an observational study on the profiles of volatile organic compounds (VOCs) in the breath of 37 lung cancer patients and 23 healthy controls of similar age. The need to quantify each VOC considered as a potential disease marker on the basis of individual calibration is elaborated, and the quality control measures required to maintain reproducibility in breath sampling and subsequent instrumental trace VOC analysis using solid phase microextraction-gas chromatography-mass spectrometry over a study period of 14 months are described. Twenty-four VOCs were quantified on the basis of their previously suggested potential as cancer markers. The concentration of aromatic compounds in the breath was increased, as expected, in smokers, while lung cancer patients displayed significantly increased levels of oxygenated VOCs such as aldehydes, 2-butanone and 1-butanol. Although sets of selected oxygenated VOCs displayed sensitivities and specificities between 80% and 90% using linear discriminant analysis (LDA) with leave-one-out cross validation, the effective selectivity of the breath VOC approach with regard to cancer detection is clearly limited. Results are discussed against the background of the literature on volatile cancer marker investigations and the prospects of linking increased VOC levels in patients' breath with approaches that employ sniffer dogs. Experience from this study and the literature suggests that the currently available methodology is not able to use breath VOCs to reliably discriminate between cancer patients and healthy controls. Observational studies often tend to note significant differences in levels of certain oxygenated VOCs, but without the resolution required for practical application. Any step towards the exploitation of differences in VOC profiles for illness detection would have to solve current restrictions set by the low and variable VOC concentrations. Further challenges are the technical complexity of studies involving breath sampling and possibly the limited capability of current analytical procedures to detect unstable marker candidates.

No abstract

In this study, a standardized experimental set-up with various combinations of herbs as odor sources was designed. Two training approaches for sniffer dogs were compared; first, training with a pure reference odor, and second, training with a variety of odor mixtures with the target odor as a common denominator. The ability of the dogs to identify the target odor in a new context was tested. Six different herbs (basil, St. John's wort, dandelion, marjoram, parsley, ribwort) were chosen to produce reference materials in various mixtures with (positive) and without (negative) chamomile as the target odor source. The dogs were trained to show 1 of 2 different behaviors, 1 for the positive, and 1 for the negative sample as a yes/no task. Tests were double blind with one sample presented at a time. In both training approaches, dogs were able to detect chamomile as the target odor in any presented mixture with an average sensitivity of 72% and a specificity of 84%. Dogs trained with odor mixture containing the target odor had more correct indications in the transfer task.

In vitro cultured lung cancer cell lines were investigated regarding the possible identification of volatile organic compounds as potential biomarkers. Gas samples from the headspace of pure culture medium and from the cultures of human lung adenocarcinoma cell lines A549 and Lu7466 were exposed to polypropylene fleece in order to absorb odour components. Sniffer dogs were trained with loaded fleeces of both cell lines, and honey bees were trained with fleeces exposed to A549. Afterwards, their ability to distinguish between cell-free culture medium odour and lung cancer cell odour was tested. Neither bees nor dogs were able to discriminate between odours from the cancer cell cultures and the pure culture medium. Solid phase micro extraction followed by gas chromatography with mass selective detection produced profiles of volatiles from the headspace offered to the animals. The profiles from the cell lines were largely similar; distinct differences were based on the decrease of volatile culture medium components due to the cells' metabolic activity. In summary, cultured lung cancer cell lines do not produce any biomarkers recognizable by animals or gas chromatographic analysis.

No abstract