An evaluation of Fourier transform infrared (FTIR) spectroscopy for detecting organic solvents in expired breath.

The aim of this study was to test the performance of gas-phase FTIR analysis on human breath samples. Ten volatile organic compounds (VOC) were examined for applicability to FTIR spectroscopy (ethanol, ethylbenzene, n-hexane, methyl ethyl ketone, methyl tert-butyl ether, m-xylene, 1,1,1-trichloroethane, trichloroethylene, tetrachloroethylene, and toluene). Three sets of detection limits (LOD) were determined for comparison. LOD(1) were generated from partial least squares (PLS) calibration methods using spectroscopic software, LOD(2) from spiked breath samples, and LOD(3) from blank breath samples. Mixed expired breath samples from four subjects were spiked at varying levels with four different VOC (hexane, methyl ethyl ketone, m-xylene and 1,1,1-trichloroethane) to validate spectral data and test overall accuracy. Breath samples spiked with m-xylene also were validated by GC/FID analysis. PLS-derived LOD(1) ranged from 0.06-2.47 ppm. Spiked breath sample LOD(2) ranged from 0.52-1.21 ppm. Blank breath LOD(3) measurements ranged from 0.17-1.70 ppm, except for ethanol, which had an LOD of 11.2 ppm. Predicted concentrations for carbon dioxide (slope = 1.06), m-xylene (slopes = 1.19, 1.21), and methyl ethyl ketone (slope = 0.93) were fairly accurate, while concentrations were underpredicted for n-hexane (slope = 0.69) and 1,1,1-trichloroethane (slopes = 0.58-0.66).