Real-time versus thermal desorption selected ion flow tube mass spectrometry for quantification of breath volatiles

Rationale Selected ion flow tube mass spectrometry (SIFT-MS) is versatile, rapidly provides result output and determines a wide range of volatiles, making it suitable for biomedical applications. When direct sampling into the SIFT-MS instrument is impractical, combining thermal desorption (TD) and SIFT-MS might offer a solution as it allows sample storage on sorbent tubes for later analysis. This work compares off-line TD SIFT-MS and real-time SIFT-MS for the quantification of selected breath volatiles. Methods Ten healthy non-smoking individuals provided 60 breath samples per method. For off-line analysis, breath was collected onto sorbent tubes via a breath sampler provided with filtered inspiratory air. After TD, samples were re-collected in Tedlar bags which were then connected to the SIFT-MS instrument. For real-time analysis, breath was sampled directly into the instrument. In both cases the analytical method included a total of 155 product ions, and 14 selected volatiles were quantified. The agreement between the methods was assessed using Pearson correlation coefficients and Bland-Altman plots. Results Overall, correlations between real-time and off-line analysis were moderate to very strong (r = 0.43-0.92) depending on the volatile of interest, except for 2,3-butanedione and styrene. The difference between real-time and off-line measured breath concentrations (average bias) ranged between -14.57 and 20.48 ppbv. For acetone and isoprene, it was 251.53 and 31.9 ppbv, respectively. Conclusions Real-time SIFT-MS and off-line TD SIFT-MS for quantification of selected breath volatiles did not show optimal agreement. Analyzing a multitude of analytes in breath via direct exhalation into a SIFT-MS instrument for real-time analysis is challenging. On the other hand, off-line analysis using a breath collection device also has its issues such as possible sample losses due to selective absorption depending on the sorbent used or during desorption and transfer to the instrument. Despite these drawbacks, both methods were moderately well correlated.