Stock solutions
Sodium chloride (NaCL, ≥99.5 %) and hydrochloric acid (HCl, >37 %) were obtained from Sigma Aldrich, Dorset, UK. Sodium hydroxide (NaOH, >97 %) was obtained from BDH limited, Poole, UK. A 100 ml stock solution of saturated sodium chloride (salt solution) was prepared with distilled water and 38 g of NaCl; a 100 ml stock solution of 5M sodium hydroxide (base solution) was prepared with distilled water and 20 g of NaOH; and a 100 ml stock solution of 5M hydrochloric acid (acid solution) was prepared by diluting concentrated hydrochloric acid with distilled water.
Volunteer recruitment
The study presented here was performed in accordance with the Declaration of Helsinki and ethical approval was obtained from the North Wales Research Ethics Committee—West (REC reference number 13/WA/0266) with the Royal Liverpool and Broadgreen University Hospitals as research sponsor. Three healthy volunteers were recruited after obtaining informed consent. Volunteers were healthy male subjects of age 29 ± 2.5 years old (mean ± standard deviation) who were taking no medication for at least 4 months prior to sample collection.
Urine samples
A 200 ml sample of first pass urine was collected from each volunteer and quickly divided into 30 aliquots of 4 ml stored in 10 ml vials for SPME-GC-MS analysis (Sigma-Aldrich, Dorset, UK). From these, twenty-five aliquots were stored at −80 °C while five aliquots were kept at room temperature to be analysed within 15 h of sample collection.
Freeze-drying
An Edwards EF4 Modulyo freeze-dryer (Edwards High Vacuum, UK) operated at −35 °C and eight mbar was used to freeze-dry for 18 h five aliquots of 4 ml urine samples from each volunteer previously frozen at −80 °C.
Headspace-SPME-GC-MS analysis
A Perkin Elmer Clarus 500 GC/MS quadruple bench top system (Beaconsfield, UK) was used in combination with a Combi PAL auto-sampler (CTC Analytics, Switzerland) for the analysis of all samples. The GC column used was a Zebron ZB-624 with inner diameter 0.25 mm, length 60 m, film thickness 1.4 \(\mu\)m (Phenomenex, Maccles field, UK). The carrier gas used was helium of 99.996 % purity (BOC, Sheffield, UK). A CAR-PDMS 85 \(\mu\)m fibre was used to extract VOCs from the headspace air above the samples for 20 minutes (Sigma-Aldrich, Dorset, UK). The fibre was pre-conditioned before use, in accordance with the manufacturer manual. Urine samples were placed in an incubation chamber at 60 °C for 30 min before fibre adsorption. The fibre desorption conditions were 5 min at 220 °C. The initial temperature of the GC oven was set at 40 °C and held for 2 min before increasing to 220 °C at a rate of 5 °C/min and held for 4 min with a total run time of 42 min. A solvent delay was set for the first 4 min and the MS was operated in electron impact ionization EI+ mode, scanning from ion mass fragments 10–300 m/z with an inter scan delay of 0.1 s and a resolution of 1000 at FWHM (Full Width at Half Maximum). The helium gas flow rate was set at 1 ml/min. Urine samples were randomly analysed by headspace-SPME-GC-MS within 14 h following their treatment.
Experimental conditions
The following treatments or sample preparation methods were applied to the urine samples collected from each volunteer prior to analysis by headspace-SPME-GC-MS: Fresh, where five aliquots were kept at room temperature and quickly analysed after collection; Frozen, where five aliquots were frozen at −80 °C and defrosted; Freeze-dry, where five aliquots were frozen at −80 °C and freeze-dried for 18 h; NaCl, where five aliquots were frozen at −80 °C, defrosted and treated with 1 ml of salt solution (i.e. saturated NaCl); HCl, where five aliquots were frozen at −80 °C, defrosted and treated with 1 ml of acid solution (i.e. HCl 5M); and NaOH, where five aliquots were frozen at −80 °C, defrosted and treated with 1 ml of base solution (i.e. NaOH 5M). In total, 15 samples of each treatment, five samples from each volunteer, were analysed by headspace-SMPE-GC-MS. The dry residue of each freeze-dried urine was directly analysed by headspace-SMPE-GC-MS with no addition of water or any other substance. We compared the number of VOCs identified, the classes of compounds identified and the variability in metabolite quantification when using each treatment. In addition, we compared the GC column degradation promoted by each treatment.
Reference solution
Although the urine samples were randomly analysed, we assessed the stability of the headspace-SPME-GC-MS analysis method throughout the study by preparing a stock reference solution containing four compounds dissolved in water: 2-pentanone (CAS 107-87-9), pyridine (CAS 110-86-1), benzaldehyde (CAS 100-52-7) and indole (CAS 120-72-9). A 2 ml aliquot of this reference solution was then analysed on the days the urine samples were analysed. These compounds were selected as reference compounds as their retention times were considerably spread across the GC-MS run. A single stock of reference solution was prepared and used throughout the experiment. The stock solution was stored at room temperature.
Laboratory air
In order to correct the results for potential air contaminants, samples of the laboratory air were analysed among urine samples. A total of 22 laboratory air samples were analysed throughout the study. Compounds found in more than 50 % of the air samples (Additional file 1) were considered contaminants and were removed from the data analysis.
Mass spectral library
Two mass spectral libraries were built for this study (Additional files 2 and 3), one for processing samples from the reference solution and another for processing urine samples. They were both built using the Automated Mass Spectral Deconvolution System (AMDIS-version 2.71, 2012) in conjunction with the NIST mass spectral library (version 2.0, 2011). The AMDIS configuration used is available through Additional file 4.
Data analysis
The GC-MS data were processed using AMDIS in conjunction with the R package Metab [22]. All statistics were performed using R software [23]. A total of 90 urine samples were analysed by headspace-SPME-GC-MS, 30 samples per volunteer (Additional file 5). Outlier samples were those found to contain considerably fewer metabolites in comparison to the rest of the technical replicates. Principal component analysis was used to support the identification of outliers. These were removed from the analysis and comprised of seven samples from volunteer one (one frozen sample, two samples with NaCl, two samples with HCl and two freeze-dried samples) and two samples from volunteer three (one fresh sample and one sample with HCl). Every treatment tested is represented by a minimum of three urine samples. The p values lower than 0.05 were considered as significant.
Metabolite identification
Initially, the number of compounds identified per sample was determined for each volunteer. For an overall comparison across treatments, the number of compounds identified for a volunteer for all the different treatments tested was divided by the average number of compounds identified for this specific volunteer when using fresh urine. Compounds present in less than 30 % of the samples within a particular condition tested were considered false positives and, thus, were removed from the analysis. In this case, the levels detected for this specific compound were replaced by NA within samples of this particular condition. A Shapiro test indicated that the number of compounds identified per condition was not normally distributed. Thus, the Mann–Whitney U test was applied for comparing the number of compounds identified across treatments. The identified compounds were divided in chemical classes according to their functional groups. A single compound may have multiple functional groups, thus, it may be part of multiple chemical classes.
Metabolite quantification
The coefficient of variation (CV) (i.e. standard deviation of abundance divided by the mean abundance and multiplied by 100) was calculated per volunteer and per treatment for each compound identified. The CVs associated with each volunteer were then combined per treatment and the proportion of compounds showing CVs lower than 30 % was calculated. In addition, the proportion of compounds showing a CV of less than 30 % across treatments was compared statistically using 2-sample test for equality of proportions with continuity correction through the R function prop.test.
Polysiloxane or column degradation
Polysiloxanes, as part of the silicone septa, part of the SPME fibre or stationary phase of the GC column can degrade, resulting in its de-polymerisation and production of volatile siloxanes. In order to identify if the different sample treatments tested may promote degradation, we compared the abundances of compounds originating from column degradation across treatments. Siloxanes containing the ion fragment 73 were defined as compounds originating from decomposition [24]. Student t test and one-way analysis of variance (ANOVA) were applied on the log transformed abundances of compounds in order to assess statistical differences between treatments. For this analysis, samples that received the same treatment were considered as belonging to the same data class or experimental condition, disregarding the volunteer id. Compounds present in less than 30 % of the samples of all the treatments tested were considered as false positives and, thus, were removed from the analysis (Additional file 6).