Prospective tuberculosis diagnostic biomarkers from human skin volatile organic compounds

The data cohort consisted of 38 individuals ( 15 patients & 23 controls) who gave written consent prior to taking part in the study. The control group was clinically proven to be TB negative, similarly, TB patients were clinically provent to be TB positive and patients were sampled prior to comencing with the TB treatment. Ethical approval for the study was received from the UP medical health faculty ethical commitee, the NAS faculty ethical committe, Steve Biko Academic hospital (SBAH) ethical committee and Tshwane district hospital (TDH) ethical committee. Both hospitals gave permission for the study to be conducted on their premises. Non-invasive PDMS samplers were used to collect human skin volatile organic compound samples on the skin above the wrist. 70% isopropanol swabs were used to clean the sampling site, Mylar was used to cover the sampler and Micropore tape was used to attach the sampler and mylar to the skin. Sampling was conducted for a duration of one hour, after which the PDMS sampler was removed and placed in a glass container and transported to the University of Pretoria Hatfield campus Chemistry department. Samples were stores in a Freezer at minus 18 degress celsius, samples were allowed to reach room temperature on a work bench prior to instrumental analysis with the Gerstel thermal desorption (TDS 3) unit coupled to the LECO Pegusus 4D Time-of-flight Mass Spectrometer. The data was processed using the ChromaTof software (version 4.50.8.0), the minimum signal-to-noise ratio required for the sub peak to be retained was 20 with a baseline offset of 1.0, which is defined by the instrument. A 750 (of 1000) match was required to combine sub peaks while a 75% match was required for mass spectral library identification using the NIST14 library. Tentative identification of the skin VOCs was done by mass spectral library matching with a 75% match criterion and by calculating the linear retention indices using a C8-C28 n-alkanes standard (Sigma-Aldrich®, South Africa) according to the Van den Dool and Kratz method. Experimental linear retention indices were compared to those reported in the literature. The Mass Spectrometry Total Useful Signal (MSTUS) normalization method was used.Statistical analysis was conducted using two approaches namely; targeted analysis where compounds found in this study were compared to those in the literature, and non-targeted analysis which focused on finding novel compounds that may have biomarker capabilities. All calculations were carried out using SPSS® 25 statistical software (IBM® Corp. Armonk, New York, USA) except for the High Performance variable selection calculation which was done on the SAS 9.4 (Cary, North Carolina, USA).