Statistical and bioinformatical methods to differentiate chronic obstructive pulmonary disease (COPD) including lung cancer from healthy control by breath analysis using ion mobility spectrometry

Abstract

Human breath analysis is a powerful and especially a non-invasive technique for the monitoring and hopefully also for the diagnosis of respiratory diseases, including chronic obstructive pulmonary disease (COPD). The exhaled breath of 95 patients suffering COPD and of 35 healthy controls was investigated using an Ion Mobility Spectrometer (IMS) coupled to a Multi-Capillary Column (MCC) without any pre-separation or pre-enrichment. Starting with the results from a Mann–Whitney-Wilcoxon rank sum test to find analytes with the highest potential with respect to differentiation, box and whisker plots, metabolic maps and probability charts were introduced and compared. In addition, the sensitivity, specificity, positive and negative predictive values and the accuracy of the relation were also summarized. The findings were compared to the results of a principal component analysis. Finally, decision trees were introduced to visualize the interdependencies between the analytes and the classifications. The application of these biostatistical methods with simultaneous inclusion of several VOCs for disease classification by ion mobility spectrometry of human breath will provide much more information than using single peaks and single concentration dependencies for disease classification and discrimination of various groups. Towards the future application of potential biomarkers for clinical diagnostic procedures, complex analytical methods, such as ion mobility spectrometry, need statistical and bioinformatical tools which are simple in application, visualize the results and support decisions on the basis of the data obtained from measurements of analytes in exhaled human breath.