Saliva's unrevealed stories: characterisation of the human saliva volatilome in remote clinical settings by Thermal Desorption Gas Chromatography Mass Spectrometry

Saliva is a complex, rich matrix containing significant biochemical information, which can potentially be used to monitor an individual’s health, well-being and toxicological state. Saliva samples can be taken non-invasively and at low cost under challenging conditions such as heavily contaminated environments, for example from casualties in a Hot Zone of a CBRN incident, in primary care settings with no clinical infrastructure, from unconscious patients, as well as anxious patients or children. However, little is known about the volatile organic compound (VOC) biochemistry of human saliva and use of VOC saliva analysis in clinical settings has yet to be described. This research set up saliva sampling in two clinical centres for analysis with thermal desorption gas chromatography mass spectrometry to evaluate the saliva volatilome.

Analytical workflows for saliva VOC analysis were developed and tested for clinical use. Recruitment included 16 participants (6 female and 10 male) in the age range 22 to 29 years who participated in a “Peppermint Test” that enabled preliminary comparisons between breath and saliva volatilome. The peppermint test also showed a well-defined eucalyptol washout, indicative of reliable and reproducible sampling and analysis. Clinical research assistants were trained to take VOC saliva samples following an end-to-end quality control and quality assurance protocol that included a proof of concept of digital global tracing system of samples. Forty-seven adults receiving treatment for acute pesticide poisoning (5 female and 42 male) in the age range 16 to 78 years were recruited at the Peradeniya Teaching Hospital Sri Lanka. Four samples were collected from each participant post treatment at 0, 1, 3 and 6 hr after recruitment. Samples were shipped to the Centre for Analytical Science Loughborough University Chemistry Department for analysis and subsequent data processing.

A total of 131 samples were obtained over a period of 23 months. 38257 VOC features were isolated and were then classified into 1937 separate clusters out of which 1619 were present in only less than 20% of the samples. They were statistically sorted by VOCCluster based on similarity in retention index values and the intensities of the five most abundant fragment ions. 318 features were commonly found in a saliva matrix. 13 known siloxane products were detected from which 7 are by-products originating from the polydimethylsiloxane sampling media used for sample collection and 6 siloxane products were identified as ingredients of herbal medicine and cooking oil. The comparison of breath and saliva volatilome demonstrated similarities in biochemistry mechanisms between the two matrices from the behavioural trends of commonly endogenous VOCs. The replication of exhaled peppermint oil volatiles elimination behaviour in data from saliva samples supports the proposition that exposure to, and elimination of VOC insults may be monitored using saliva samples and that the Peppermint Test may be used to determine the efficacy of a saliva volatilome workflow.

Targeted analysis from post-hoc samples confirmed that detectable levels of OP pesticides and their breakdown products were not recovered from the saliva samples as opposed to formulation solvents xylene and ethylbenzene that were detected and found to reliably indicate pesticide ingestion. A non-targeted (compound agnostic) multi-variate analysis workflow based on SIMCA-P consisting of data reduction with orthogonal partial least squares followed by principal component analysis and the proposal of a market score identified four candidate markers of pesticide ingestion, two of which were solvents identified through the targeted approach. Statistical scoring at 95% confidence intervals gave an AUROC of 0.959 with sensitivity of 0.714 and specificity of 1 suggesting a statistical viable model for identifying individuals exposed to pesticides.

This research demonstrated the feasibility for saliva to be used as a diagnostic tool in primary care settings. To explicitly unfold the richness of the saliva matrix, further studies are required with the use of more sensitive mass analysers such as time of flight mass spectrometers to accurately identify features of interest. Saliva VOC sampling is a platform which can further be used for the investigation of ethanol and radiation toxicology and its development and validation may allow rapid triage in acute care settings.