Exploring the potential of NTME/GC-MS, in the establishment of urinary volatomic profiles. Lung cancer patients as case study

The growing cancer incidence and mortality worldwide claims for the development of novel diagnostic strategies. In this study we aimed to explore the potential of an innovative methodology, based on a needle trap microextraction (NTME), combined with gas chromatography-mass spectrometry (GC-MS), as new approach to isolate and profile urinary volatile organic metabolites (VOMs) from lung cancer (LC) patients and healthy individuals (CTRL). In this context, different experimental parameters with influence of NTME extraction efficiency including, temperature, equilibration time, headspace volume, ionic strength, pH, effects of sample volume and stirring, were investigated and optimized. For the DVB/CarX/Car1000 needle trap device (NTD), the best results were obtained using 40 mL headspace of a 4-mL acidified (pH = 2) urine sample with 20% NaCl and an extraction temperature of 50 °C for 40 min of equilibration time. The stability of the isolated VOMs was investigated up to 72 h after extraction. From the VOMs identified, belonging namely to ketones, sulphur and benzene derivatives, 98 presented a frequency of occurrence above 90%. Data were processed by discriminant analysis, retrieving differentiated clusters for LC and CTRL groups. As far we are aware, this is the first study using NTME/GC-MS to establish urinary volatomic profiles. Preliminary results are very promising, as broad and comprehensive volatile profiles were obtained. Moreover, the extended storage stability of the NTD devices opens new opportunities for sampling other matrices in a wide range of applications.

. Configuration of the Needle Trap Device (NTD) used in this work. (A) Schematic representation of the triple bed DVB/CarX/Car1000 NTD, which allows the sequential retention from the big and polar to the small and less polar volatiles (DVB, divinylbenzene, is a week sorbent that retains mainly C7 to C20 compounds; CarX, Carboxen X, is a middle strong sorbent with higher affinity for C3 to C8 volatiles; Car1000, Carbopack 1000, is the strongest sorbent, retaining mainly C2 to C6 compounds). (B) The experimental layout used, (C) the DVB/CarX/Car1000 NTD loading the extracted volatiles in the GC-MS.
SCIenTIFIC REPoRTS | (2018) 8:13113 | DOI: 10.1038/s41598-018-31380-y Extraction Temperature. The extraction temperature affects the extraction efficiency as higher temperatures facilitate the mass transfer of the compounds from the bulk sample to the headspace vapor phase. This increases the diffusion coefficient and decreases the distribution constant, reaching faster the balance between the two phases 13,14 . By increasing the temperature, we therefore can reduce the extraction time. This, however, is only valid until thermal degradation and/or isomerization of the sample metabolites occurs 13 . We have assayed extractions at 30 °C, 40 °C, 50 °C and 60 °C and the results shows a considerable increase in terms of number of extracted compounds and total area from 30 °C up to 50 °C, but not a relevant gain from 50 °C to 60 °C (Fig. 2B). This result agrees with our previous work using other extraction methodologies for the characterization of urinary volatile profiles 5,6 . Based on the obtained results, 50 °C was selected as the most appropriate temperature for the extraction of urinary volatile metabolites.
Ionic Strength. The addition of salt (NaCl) to urine samples increases the ionic strength, enhancing the salting-out effect during the extractive process and therefore facilitating the diffusion of metabolites to the vapor phase 12,13 . On the other hand, excess of salts in solution do not favour the extraction 13 . In fact, we obtained a gain in the extraction efficiency with ionic strengths up to 20% NaCl (w /v). At higher NaCl concentrations (30% NaCl; w/v), a slight decrease in the total area was observed (Fig. 2C), although particular VOMs, such as dimethylsulfide and isoprene, have better responses under higher ionic strengths ( Supplementary Fig. 1). Considering the total area, a concentration of 20% NaCl (w/v) was selected for the extraction of urinary VOMs.
Sampling Headspace. Since NTME is an exhaustive technique, the sample headspace volume that is loaded through the sorbent is proportional to the response until the sorbent saturation is reached (breakpoint). Thus, by increasing the headspace volume of the sample, we will increase the extraction efficiency. In the evaluation of this parameter, the equilibration time and sample flow are also important parameters and should be considered. The first of these interfering factors was assayed in this work, while the second, the sample flow, was used as described by Trefz et al. 15 , that previously optimized this condition for the exhaled breath analysis using NTME 15 . Also in that report, the authors demonstrated that the extractive response of the sorbent was greater when larger sample headspace volumes were used, at least for the model metabolites isoprene, pentane, toluene and pentanal 15 . Furthermore, the breakthrough volume of this type of NTD was not reached up to a headspace volume of 60 mL 15 and a linear response was obtained in this range. Our results show a considerable increase in the number and total area of the VOMs extracted between 10 and 30 mL of headspace, whereas the gain is very moderate up to 40 mL of sample headspace and there is a decrease in the total area of the metabolites extracted (and less 2 VOMs identified) for 50 mL of headspace (Fig. 2D). Based on the obtained results, and in order to minimize the extraction time and increase the reusability of the sorbents, 30 mL was selected as the appropriate headspace sample volume. Equilibration Time. The equilibration time of the urine sample with its headspace affects the VOMs diffusion to the vapor phase. Analysis of the results obtained shows an incremental response of the total area and number of metabolites up to 40 min of equilibration (Fig. 2E). For a longer equilibration time of 60 min, the increase in the total area is not significant and the number of compounds decreases from 88 to 84. This suggests that some metabolites degradation may occurs as result of the prolonged time under heating and stirring conditions. Furthermore, the behaviour of specific VOMs analysed individually is not uniform, with better responses observed with 60 min of equilibration for (+) -4-carene and dimethylsulfide, the opposite for acetone and 2-butanone and irregular variations for toluene, phenol, isoprene and 2-pentanone ( Supplementary Fig. 1). An equilibration time of 40 min was therefore selected as the most favourable.
Sample Stirring. Constant and efficient sample stirring facilitates sample homogenization 16 and VOMs difusion from the liquid to the vapor phase, thus reaching equilibrium more quickly 13 . Our results show that a constant sample stirring at 800 rpm contributes to a considerable improvement in the extraction efficiency of the methodology by comparison with a no stirring approach. This is clearly expressed by the large increase (almost two times) in the number of extracted metabolites and total area (Fig. 2F). An equivalent result was obtained for several metabolites analysed individually (data not shown).

Sample Volume.
In the evaluation of the optimum sample volume, we assayed 2, 4 and 8 mL of urine. The results obtained reveals a small increase in the total area and number of VOMs identified from 2 to 4 mL, but no significant gain with 8 mL samples (Fig. 2G). When this analysis was detailed to specific VOMs, heterogeneous variations were obtained for the selected metabolites ( Supplementary Fig. 1). Considering these results, particularly the negligent gain in the extraction efficiency by doubling the sample volume from 4 mL to 8 mL, the lower sample volume was selected as the most suitable for the methodology being optimized. This selection also allows a significant reduction in the sample residues produced. Storage Time. The stability of the extracted VOMs in the sorbent as a function of the storage time (0-72 h) was also evaluated. This parameter is very relevant because samples are collected outside the laboratory, mostly in clinical facilities, where patients are observed. In this context, the possibility of performing in situ sampling and store the extracted analytes in the NTDs for later analysis, therefore avoiding the transport of biological samples from the collection point to the lab, would be a major advantage. The results obtained were very interesting as we verified that during the first 12 hours of storage there are no significant differences (p < 0.05) in terms of total areas, while after 24 hours, the decrease in the metabolite stability is marginal (decrease of less than 10% total area compared to the starting point -VOMs extracted and analysed after sample collection). Even after 48 and 72 hours of extraction, the loss of extracted analytes is about 15% over the initial period, revelling therefore a high stability and reliability in the use of the NTDs as a sampling and storage device for the VOMs extracted at least up to 3 days after extraction (Fig. 3). This study was further extended to the analysis of a set of representative VOMs. As can be observed in Fig. 3, the decrease in total areas obtained for the selected metabolites is not appreciable, confirming the high stability of the selected metabolites in the sorbent during the period considered.
VOMs Profile of Lung Cancer Patients. The urinary VOMs profile of LC was established using NTME followed by high-resolution GC-MS to process 30 urine samples from healthy subjects (CTRL) and 17 urine samples from LC patients. The typical chromatograms of the volatile urine profile of subjects from each study group is presented in the Supplementary Fig. 2.
The analysis of the chromatograms obtained allowed the identification of 98 VOMs with a frequency of occurrence (FO) higher than 90% in the volatile composition of the urine of all samples used in this study. Table 1 summarizes the metabolites identified for each group, the chromatographic data and the possible origin 17 .
The metabolites identified in CTRL and LC urine samples span diverse chemical families, being the most representative the ketones, in addition to sulfuric, benzenic, furanic and terpenic compounds. Figure 4 shows the distribution of the metabolites identified in the groups under study according to their chemical families. As can be easily observed, the urinary profiles of both groups are dominated by sulphur compounds and ketones (each class representing nearly 30% of the total area), with the remaining classes representing less than 10% of the total area obtained in each group. The only exception are benzenes, which are highly abundant in the control samples (around 30% of the total area obtained), while in the patients group, only a third of the equivalent representativeness was found. In the same direction, terpenics and alcohols are significantly more abundant in the control group. Phenolics and aldehydes, in turn, are more abundant in the LC than in the control group. These volatomic differences are eventually related with the altered metabolic activity that cancers cells have in comparison with the normal cells. Nevertheless, external factors contributing with exogenous VOMs must be also considered as important interferent factors 5,6 . And this problem cannot be overcome just by discarding such metabolites, because for many of them it is not possible to associate a unique origin, being very difficult to completely elucidate their metabolic pathway 6,18,19 . Statistical Analysis. Following the characterization of the urinary volatile profiles of LC and CTRL subjects, the data obtained was processed with different statistical tools to assess the existence of discriminative models SCIenTIFIC REPoRTS | (2018) 8:13113 | DOI:10.1038/s41598-018-31380-y able to distinguish LC from CTRL samples. In addition to the procedures described in the experimental section, and to try to obtain more robust and meaningful models, only the VOMs identified with a FO of at least 90% were considered in the statistical analysis. Moreover, the VOMs unambiguously reported in the literature 17 as exogenous were also discarded. Firstly, a variable normalization was applied to the data matrix to obtain a homogeneous distribution. This procedure reduced the size of the data matrix by eliminating redundant variables that do not contribute to differentiate the groups being analysed. Following this, a multivariate analysis using supervised partial least square discriminant analysis (PLS-DA) models 20,21 was performed with the MetaboAnalyst 3.0 web based tool 22 . The statistical model obtained segregates LC an CTRL samples in two clusters corresponding to the cancer and non-cancer samples analysed (Fig. 5). Additionally, top 29 VOMs differentially expressed were found with variable importance in projection (VIP) > 1 ( Table 2). Among these, several VOMs have been previously involved in the discrimination of different clinical conditions. Hexanal, as well as other aldehydes, for instance, has been already reported by several authors as discriminant metabolites for LC [23][24][25] . In turn, Rudnicka et al. showed that, although carbon disulfide was present in the volatile profile of healthy individuals and smokers, it has a higher significance in LC, with a considerably higher response in the patients 26 . A similar result was reported by Buszewski et al. 27 when comparing the volatile composition of normal and stomach cancer tissue. Regarding 2-heptanone, Hanai et al. 28 identified this ketone in the urine and culture of rats LC cell lines, with a higher abundance than the respective control groups. Finally, to check the robustness of statistic model obtained, a random permutation test with 100 permutations was performed with PLS-DA model. The permutation test yielded a R2 (Goodness of fit) of 0.915 and a Q2 (Predictive ability) of 0.760, indicating that the model is not over fitted and has good predictive ability to distinguish between the studied groups. It should be highlighted, however, that this is just a proof of concept study involving a limited number of samples. To be able to extrapolate the discrimination power for LC diagnosis, broader and extensive studies with a statistical significant number of samples and independent patient cohorts are mandatory. The methodology reported here constitutes a breakthrough given the superior sampling and storage abilities of the NTDs and the broader coverage of the volatomic profiles obtained and may therefore help to reach our ambitious aims.

Conclusions
To the best of our knowledge, this is the first study reporting the high throughput potential of NTME combined to GC-MS to characterize the volatomic profile of urine samples without complicated pre-treatments, so could be a versatile tool for future application, providing a good technical advance. In this context, different key parameters of the extraction procedure were evaluated and optimized according to the nature of the sample and its volatile composition. The optimized experimental layout involves a DVB/CarX/Car1000 NTD to extract 40 mL headspace of a 4-mL acidified urine added NaCl 20% (m/v), during 40 min at temperature of 50 ± 1 °C.

Materials and Methods
Chemicals and Materials. Hydrochloric acid (HCl), sodium chloride (NaCl), all the standards used for VOMs confirmation (purity higher than 98.5%) and the n-alkanes mixture containing C 8 -C 20 straight-chain alkanes in hexane were obtained from Sigma-Aldrich (St. Louis, MO, USA). Helium, ultra-pure grade (Air Liquide, Portugal) was used as carrier gas in the GC system. Clear glass screw cap vials for extraction with PTFE/silica septa were purchased from Supelco (Bellefonte, PA, USA). The NTDs used in this work, "NeedleEx", were custom manufactured by Shinwa Ltd., Japan (60 mm × 0:41 mm id, 0.72 mm od, triple bed configuration Divinylbenzene/Carboxen X/Carbopack 1000 -DVB/CarX/Car1000) and purchased from PAS Technology (Magdala, Germany). Prior to their use, NTDs were conditioned in a heating device (PAS Technology, Magdala, Germany) at 250 °C, under permanent helium flow for at least 20 h to eliminate any contaminations from the manufacturing process or shipping. Afterwards, both ends of the needles were sealed with Teflon caps and stored. Before being used, the NTDs were conditioned again for 30 min in the heating device.    (Table 3). Participants were given sterile and disposable glass bottle collectors and instructed to collect the first urine in the morning after the rejection of the first urine stream. All participants were fully informed of the objectives of the study and signed the informed consent before donating the urine samples. After collection, each sample was individually homogenized, aliquoted in 5 mL ambar glass vials and stored at −80 °C until analysis. This study was previously approved by the Ethics Committee of HCF.
Optimization of NTME extraction. To increase the NTME efficiency, different key parameters were optimized according to the characteristics of the sample and extractive methodology 6,29 . This included (i) the pH of the sample (acid -pH 2, basic -pH 11, and neutral -pH 7), (ii) the extraction temperature (30 °C, 40 °C, 50 °C to 60 °C), (iii) the ionic strength (0, 10%, 20% and 30% NaCl, w/v), (iv) sample headspace volume (10,25,30,40 and 50 mL), (v) the equilibration time (5,10,20,40, 50 and 60 min), (vi) sample stirring (0 and 800 rpm) and (vii) sample volume (2, 4 and 8 mL) were tested using an univariate procedure to achieve extraction conditions suitable for this application (lowest sample volume, preferentially near room temperature and reasonable equilibration time). In this study, we also evaluated the storage time of the VOMs in the NTD (2,4,6,12,24,48 and 72 hours after the NTME extraction). All extractions were performed in duplicate.  NTME extraction. Following the optimization step, 4 mL of urine sample was insert in 20 mL tubes (extraction tubes -ETs) and added 0.5 mL HCl 5 M to acidify the sample (pH = 2). The pH was verified with the aid of a pH meter (Hanna Instruments) with a clean electrode. After the measurements, the electrode was carefully cleaned to avoid contamination between samples. To promote the salting-out effect, facilitating the transference of the urinary VOMs for the headspace, 20% NaCl (w/v) were added, the ET sealed, and the system equilibrated for 40 min at 50 ± 1 °C. Then, the NTDs pre-attached to a disposable 1 mL syringe were insert into the headspace of the ETs and 40 mL of the gas phase were manually loaded through the sorbent (40 withdraw-loading cycles, average speed 10 ± 2 mL/min). After the extraction, the syringe was discarded and the NTD was sealed in both ends with PTFE caps. Finally, the NTD was injected in the GC-MS at 250 °C for 30 seconds for thermal desorption of the extracted VOMs. Before the next extraction, the sorbent was reactivated by placing the NTDs in a conditioner at 250 °C under constant flow of helium (purity 5.0, Air Liquid, Portugal) at a constant pressure of 1 bar for 30 min.
Gas chromatography-mass spectrometry analysis (GC-MS). The analysis was carried out with an Agilent 6890N gas chromatograph system (Agilent Technologies, Palo Alto, CA, USA) coupled with an Agilent 5975 quadrupole inert mass selective detector. The separation of the extracted compounds was performed on a BP-20 fused silica capillary column (60 m × 0.25 mm I.D. × 0.25 µm film thickness). Splitless injection was employed using helium as carrier gas at a constant flow rate of 1.0 mL/min. Oven temperature conditions were: 45 °C (held for 2 min), followed by a gradient temperature ramp from 45 °C to 60 °C, held for 1 min at a rate of 0.7 °C/min, followed by a flow rate of 1.0 °C/min until 110 °C (held for 1 min); then a flow rate of 3.0 °C/ min until 150 °C (held for 1 min) and finally from 150 °C to 220 °C held for 10 min at a rate of 15 °C/min. The injection and ion source temperatures were 250 °C and 230 °C, respectively. The mass spectra of the compounds were acquired in electron-impact (EI) mode at 70 eV. The electron multiplier was set by the auto tune procedure. Data acquisition was performed in scanning mode (mass range m/z = 35-300 amu; six scans per second). Chromatograms and spectra were recorded and processed using the Enhanced ChemStation software for GC-MS (Agilent Technologies, Palo Alto, CA, USA). VOMs identification was based on the comparison between the GC retention times (RT) of the chromatographic peaks with those, when available, of authentic standards run under the same conditions. MS fragmentation patterns were compared with those of pure compounds, and mass spectrum database search was performed using the National Institute of Standards and Technology (NIST) MS 05 spectral database. Finally, confirmation also involved the determination of the RI of each peak of C 8 -C 20 n-alkanes series. Once again, the values were compared, when available, with values reported in the literature for similar chromatographic columns. Chromatographic peak areas, expressed in arbitrary units (a.u.) of area were determined using the FullScan chromatogram, and were used as an indirect approach to estimate the relative content of each volatile metabolite. For semi-quantification purposes, each sample was injected in triplicate, and the chromatographic peak areas (as kcounts amounts) were determined by a reconstructed full-scan chromatogram using for each compound some specific quantification ions: these corresponded to base ion (m/z 100% intensity), molecular ion (M + ), and another characteristic ion for each molecule.
Multivariate statistical analysis (MVSA). In order to identify potential discriminative features for the selected groups (LC and CTRL), the statistical data analysis was carried out using the web based application, Metaboanalyst 3.0 22 . The data was normalized by the median, auto scaled and subjected to multivariate statistical analysis. Partial least squares-discriminant analysis (PLS-DA) was then used to visualize the separation between LC and CTRL groups and to identify set of VOMs able to discriminate between the two groups.
Ethical approval. The study was approved by ethics committee of Hospital do Funchal, Madeira, Portugal.
Informed consent. Prior informed consent was obtained from all the participants in the study with institutional review approval.
Research involving human participants and/or animals. All protocols and procedures were adhered to institutional ethical standards and/or research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.