Article pubs.acs.org/ac
Sensitivity Enhancement in the Determination of Volatile Biomarkers in Saliva Using a Mass Spectrometry-Based Electronic Nose with a Programmed Temperature Vaporizer Miguel del Nogal Sánchez,† Pedro Á ngel Callejo Gómez,† José Luis Pérez Pavón,*,† Bernardo Moreno Cordero,† Á ngel Pedro Crisolino Pozas,‡ and Á ngel Sánchez Rodríguez‡ †
Departamento de Química Analítica, Nutrición y Bromatología, Facultad de Ciencias Químicas, Universidad de Salamanca, 37008 Salamanca, Spain ‡ Servicio de Medicina Interna, Hospital Virgen de la Vega, Complejo Asistencial Universitario de Salamanca, 37007 Salamanca, Spain S Supporting Information *
ABSTRACT: With a view to improving the sensitivity of direct coupling of a headspace sampler (HS) with a mass spectrometer (MS), here we propose the use of a programmed temperature vaporizer (PTV) in solvent−vent injection mode before the sample is introduced into the MS. This preconcentration scheme has been used for some time in many methods based on gas chromatography (GC), but to the best of our knowledge it has not yet been used in an electronic nose based on MS. The increase in the S/N ratio with the proposed instrumental configuration (HS-PTV/MS) lies between 6.9- and 22-fold. The main advantage of using this injector lies in the fact that it does not involve time-consuming steps. To check the possibilities of this methodology, saliva samples from healthy volunteers and patients with different types of illnesses (including some types of cancer) were analyzed. None of the compounds studied was detected in the samples corresponding to the healthy volunteers. One or more biomarkers, at levels ranging from 13 to 500 μg/L, were found in five of the samples from the patients. Additionally, separative analysis by HS-PTV-GC/MS was performed for confirmatory purposes and both methods provided similar results. The main advantage of the proposed methodology is that no prior chromatographic separation and no sample manipulation are required.
L
analyses, offer an alternative solution to invasive, timeconsuming, complicated, and expensive diagnostic methods using specimens such as blood, serum, and feces.12 Saliva sampling requires minimal training and costs, and no specialist equipment is required in comparison to breath sampling. Enormous efforts have been made to develop artificial noses for biomedical applications13−16 owing to their potential in rapid detection and odor characterization. Electronic nose technology has developed significantly, along with advances in different sensor platforms and complex microarray devices. Electronic noses based on MS17−19 have some advantages in comparison with conventional ones, and they can be considered an “open sensor array” in which each m/z ratio acts as a “‘sensor’” that detects any ion fragment with that ratio. Besides electronic noses, other sensitive and nonseparative methods20−23 such as proton transfer reaction mass spectrometry (PTR-MS), selected ion flow tube mass spectrometry (SIFT-MS), and ion mobility spectrometry (IMS) are often used for the online monitoring of target compounds present in exhaled breath.
ung cancer is the leading cause of cancer mortality, with more than one million deaths worldwide every year.1,2 Early diagnosis is limited by the fact that the disease usually develops asymptomatically. The 88% 5-year mortality associated with lung cancer is partly related to late clinical diagnosis as well as suboptimal therapeutic alternatives for patients with advanced disease.3,4 To improve overall survival, many screening methods, including chest radiography, sputum cytology, low-dose spiral computer tomography, fluorescence bronchoscopy, and positron emission tomography, have been used.5−8 However, these procedures are complicated, expensive, and time-consuming, and in many medical facilities they are unavailable. The detection of volatile organic compounds (VOCs) in human breath proposed by Pauling et al.9 in 1971 opened a new and challenging scientific field in the analysis of VOCs for clinical diagnosis. Presence of VOCs in body fluids can provide valuable information about the state of health of an individual because certain basic cellular functions, including the maintenance of cell membrane integrity, energy metabolism and, especially, oxidative stress, are linked to their formation.10 A desirable screening method should be noninvasive, painless, inexpensive, easily accessible to a large number of patients, and reliable. and it should facilitate the diagnosis of an early stage cancer.11 Saliva analyses, together with urine © 2014 American Chemical Society
Received: May 15, 2014 Accepted: July 3, 2014 Published: July 3, 2014 7890
dx.doi.org/10.1021/ac501917a | Anal. Chem. 2014, 86, 7890−7898
Analytical Chemistry
Article
In a previous work,24 we reported the use of direct coupling of a headspace sampler (HS) with a mass spectrometer (MS) for the rapid determination of five biomarkers in saliva samples, two of them related to lung cancer and liver disease.With a view to improving the sensitivity of the method (with which the detection limits achieved were in the ppm range), here we propose the use of a programmed temperature vaporizer (PTV) in solvent−vent injection mode before the sample is introduced into the MS. The gas phase generated in the HS is introduced into the PTV, the analytes of interest are retained in the liner by cold trapping, while other more volatile compounds and the solvent are eliminated through the split line. The PTV injector is equipped with an efficient heating and cooling system and by using liners packed with a selective adsorption material, the range of components that can be trapped in the liner can be significantly extended toward the more volatile components. The later application of a rapid temperature ramp allows the analytes to be introduced into the mass spectrometer. This preconcentration scheme has been used for some years in many GC methods applied to the environmental,25 food,26 and pharmaceutical fields;27 among others but to our knowledge a PTV inlet, in solvent vent injection mode, has not yet been used in an electronic nose based on MS. Some of the preconcentration techniques most widely used for the determination of biomarkers with GC-based methods are solid phase microextraction (SPME)22,28−30 and sorption on solid sorbents followed by thermal desorption (TD).31 The main advantage of a PTV injector over others is that its use does not involve time-consuming steps. Here we propose a rapid, simple, sensitive, and nonseparative method (HS-PTV/MS) that can be applied to a large number of patients with good precision and accuracy for the detection of lung cancer biomarkers in saliva samples. For positive samples, gas chromatography/mass spectrometry analysis was performed for confirmatory purposes. Finally, the results are compared with those obtained at a hospital after applying the corresponding medical routine. Eight biomarkers mainly related to lung cancer1,11,20 (benzene, 3-methyl-1-butanol, toluene, styrene, o-xylene, propylbenzene, 1,2,4-trimethylbenzene, and 2-ethyl-1-hexanol) were selected to check the applicability of the proposed methodology.
sample was analyzed by triplicate (three vials, one injection per vial). Although the addition of salts increased the analytical signal in the case of two of the eight compounds (3-methyl-1butanol and 2-ethyl-1-hexanol), it was decided not to use them owing to their progressive accumulation in the PTV liner and column head. Unstimulated saliva samples were obtained from 30 adults of both sexes and placed directly into a 10.0 mL vial sealed with Teflon/silicone septum caps. The saliva samples were collected and analyzed on the same day. They were maintained at 4 °C until analysis. Samples 1−18 were from healthy volunteers apparently unaffected by diseases (11 women and 7 men); samples 19−30 were from patients at the Internal Medicine Unit of the Virgen de la Vega Hospital in Salamanca (7 women and 5 men). To perform the measurements of these samples, 500 μL of saliva were placed in a 10.0 mL vial and 50 μL of ultrapure water was added to the vial. Each sample was analyzed by triplicate (three vials, one injection per vial). The study was authorized by the Ethics Committee Hospital. HS-PTV/MS Measurements. HS sampling was performed with a PAL autosampler (CTC Analytics AG, Zwingen, Switzerland). This sampler is equipped with a tray for 32 consecutive samples and an oven with six positions to accommodate sample vials. The oven temperature was kept at 70 °C and the equilibration time was set at 10 min. During this time, the vials were shaken at 750 rpm in the oven. A 2.5 mL syringe at 120 °C was used. The experiments were carried out with a programmed temperature vaporizer (PTV) inlet (CIS-4: Gerstel, Baltimore, MD), using the solvent−vent injection mode. A liner (71 mm × 2 mm) packed with Tenax-TA was used. In the first step, the sample from the headspace was injected into the PTV injector, which was at 25 °C, such that the analytes were retained in the liner while the split valve was open, allowing solvent elimination (0.10 min). The vent flow was adjusted to 50 mL/min and a venting pressure to 5.0 psi (34 474 Pa) was imposed. In the second step, involving the transfer of the sample to the column, the split valve was closed and the PTV was heated rapidly (12 °C/s) until it reached 250 °C. Thus, the analytes were desorbed and transferred to the column (1.5 min). Finally, the split valve was opened again to clean the system (purge flow 150 mL/min) and the liner temperature was held at 250 °C. Cooling was accomplished with liquid CO2. The interface between the PTV and the MS was a lowpolarity DB-VRX capillary column (20 m × 0.18 mm × 1 μm) from Agilent J&W, which was maintained at 240 °C in an Agilent 6890 GC device during the time of analysis. In this way, the separation capacity of the column was removed and it behaved as a simple transfer line from the PTV to the mass detector. This instrumental configuration meant that it was not necessary to change the interface when chromatographic analyses were performed. The detector was a quadrupole mass spectrometer (HP 5973 N) equipped with an inert ion source. It was operated in the electron−ionization mode using an ionization voltage of 70 eV. The ion source temperature was 230 °C, and the quadrupole was set at 150 °C. The analyses were performed in scan mode (3.46 scan/s). The m/z range was 25−125 amu. The signal-recording time was 2.5 min. An interval of 6.0 min between sample injections was chosen in order to allow the PTV to cool from the final (250 °C) to the initial temperature (25 °C).
■
EXPERIMENTAL SECTION Chemicals. Benzene, styrene, and 2-ethyl-1-hexanol were supplied by Sigma-Aldrich (Steinheim, Germany). 3-Methyl-1butanol, toluene, o-xylene, propylbenzene, and 1,2,4-trimethylbenzene were purchased from Acros Organics (Geel, Belgium). Methanol was supplied by Merck (Darmstadt, Germany). The purities of the compounds were at least 98%. Standard Solutions and Samples. A set of stock solutions (5000 mg/L) of these compounds in methanol was prepared and stored at 4 °C. Eight solutions ranging from 5.0 to 60 mg/L were also prepared by appropriate dilutions of the stock solutions in ultrapure water (each solution contains one analyte). In total, 31 working solutions containing the eight compounds studied were prepared in ultrapure water by taking different volumes from the solutions described above. To perform the measurements of the calibration and validation samples, 500 μL of ultrapure water was placed in a 10.0 mL vial (Agilent Technologies, DE, Germany) and 50 μL of the above solution was added to the vials, which were sealed with Teflon/ silicone septa (Agilent Technologies, DE, Germany). Each 7891
dx.doi.org/10.1021/ac501917a | Anal. Chem. 2014, 86, 7890−7898
Analytical Chemistry
Article
Figure 1. Schematic diagram of the instrumental configuration and the methods used.
Table 1. Relative Standard Deviation (RSD) of the Profile Signals (HS-PTV/MS) of a Set of 5 samples when Different Venting Temperatures Were Used. S/N Ratios for the Profile Signals when the HS/MS and HS-PTV/MS (Venting Temperatures 15, 25 and 35 °C) Nonseparative Methods Were Used Are Also Reported. RSD (HS-PTV/MS)
S/N HS-PTV/MS
extracted ion profile signal m/z m/z m/z m/z m/z m/z m/z m/z
78 55 91 104 106 120 105 57
5 °C
15 °C
25 °C
35 °C
HS/MS
15 °C
25 °C
35 °C
1.4 5.0 2.1 2.2 1.9 2.2 1.8 6.9
1.8 6.7 0.57 0.58 0.74 0.67 0.52 2.1
0.54 2.8 2.9 3.3 0.59 1.3 0.45 1.7
0.60 2.9 2.7 3.2 0.60 1.4 0.40 1.7
57 22 175 90 36 70 39 48
345 95 1209 1327 665 628 740 430
396 217 1249 1391 801 807 729 775
315 182 1081 1186 659 755 585 474
Figure 2. Extracted ion profile signal for m/z 91 (a) and 105 (b) when hot-split injection (HS/MS) and solvent−vent injection (HS-PTV/MS) were used.
HS/MS Measurements. The experimental conditions for the HS autosampler and the mass spectrometer were the same as those used for the methodology based on HS-PTV/MS. The experiments were carried out using the PTV inlet in hot-split injection mode (split ratio 1:5, 250 °C). The chromatographic column was maintained at 240 °C along the time of analysis (2.5 min), and an interval of 3.0 min between sample injections was chosen. HS-PTV-GC/MS Measurements. The experimental conditions for the HS autosampler and the PTV inlet were the
same as those used for the methodology based on HS-PTV/ MS. The experiments were carried out using the PTV inlet in the solvent−vent injection mode described above. The initial oven temperature was 45 °C for 1.5 min; this was increased at a rate of 60 °C/min to 175 °C and then further increased at 45 °C/min to 240 °C and held for 0.50 min. The carrier gas was helium (99.999% pure; Air Liquide) and the flow rate was 2.0 mL/min. The total chromatographic run time was 5.61 min. Additionally, about 7 min were necessary before the next sample could be measured, since the column had to be 7892
dx.doi.org/10.1021/ac501917a | Anal. Chem. 2014, 86, 7890−7898
Analytical Chemistry
Article
Table 2. Characteristics of the PLS Models and Relative Prediction Error in the Cross-Validation and External Validation Steps for the Methods Based on HS-PTV/MS and HS-PTV-GC/MS ∗a benzene
3-methyl-1-butanol
toluene
styrene
o-xylene
propylbenzene
1,2,4-trimethylbenzene
2-ethyl-1-hexanol
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
selected m/z variables
E (%)
PLS components
50,51,52,74,76,77,78,79,80,102 103 104
8.4 7.4 1.7 4.2 19 9.7 8.3 13 7.1 8.1 13 7.5 11 13 3.2 14 21 11 14 18 6.9 5.9 16 10
2
78 46,53,55,57,58,60,70,72,81,83,95,97,98
2
55 50,62,63,64,65,66,92,93,119
5
91 51,74,76,77,78,80,102,103,104
2
104 106,107
1
91 50,62,63,64,78,91,92,104,105,106,107,120,121
6
91 65,91,92,105,106,107,115,116,117,119,120,121
4
105 55,57,68,83,98,99
3
57
∗: 1, cross-validation results using the method based on HS-PTV/MS; 2, external validation results using the method based on HS-PTV/MS; and 3, external validation results using the method based on HS-PTV-GC/MS. a
cooled down from the final temperature (240 °C) to the initial conditions of 45 °C. An interval between sample injections of 13 min was chosen. The analyses were performed in synchronous SIM/scan mode, which allowed the collection of both SIM and full scan data in a single run. Full scan (25−125) was used for compound identification by comparison of the experimental spectra with those of the NIST’08 database (NIST/EPA/NIH Mass Spectral Library, version 2.0). Selected ion-monitoring (SIM) was used for quantification, choosing the characteristic ions in each case (Table S-1, Supporting Information), with a dwell time of 10 ms. The number of scans per second in synchronous SIM/scan mode varied between 8 and 11, depending on the number of ions in each SIM window. A schematic diagram of the methods used is shown in Figure 1. Data Analysis. Data collection was performed with Enhanced ChemStation32 from Agilent Technologies. Partial least-squares (PLS) multivariate calibration was performed using the Unscrambler, v10.2 statistical package.33
Table 3. Detection Limits in Saliva Samples for the Methods Based on HS-PTV/MS and HS-PTV-GC/MS detection limits (μg/L) HS-PTV/MS benzene 3-methyl-1-butanol toluene styrene o-xylene propylbenzene 1,2,4-trimethylbenzene 2-ethyl-1-hexanol
MDL1*a
MDL2*b
HS-PTV-GC/MS
7.9 22 3.8 5.2 6.9 4.4 4.9 11
9.4 89 5.6 5.9 20 8.4 12 18
0.27 1.3 0.19 0.15 0.11 0.19 0.17 0.20
a
MDL1*: Strategy developed by Faber and Bro. bMDL2*: Strategy based on prediction uncertainty provided by The Unscrambler.
25, and 35 °C). In all cases, temperature was kept stable throughout the venting process. Table 1 shows the relative standard deviation (n = 5) of the area of the profile signals for the different venting temperatures assayed. The most abundant m/z ratios of all the compounds were chosen with the exception of o-xylene and propylbenzene, for which the second most abundant ratio was selected because the base peak of both coincided with that of toluene (m/z 91). In all cases, the values were similar and no specific trend in the data was observed. The optimum venting temperature in this application was 25 °C, since with higher values (35 °C) a percentage of analytes ranging between 9 and 29% was eliminated through the split valve with respect to 25 °C and hence the S/N ratio decreased (Table 1, columns 8 and 9). With values below 25 °C, the
■
RESULTS AND DISCUSSION Study of the Signals Obtained. Here we studied the possibilities of the use of a PTV in solvent−vent injection mode as a preconcentration step in a nonseparative analytical scheme based on the introduction of the sample from the headspace to an MS through a chromatographic column kept at 240 °C. The critical point of this coupling (HS-PTV/MS) is in the PTV, located inside the oven of the chromatograph (240 °C), which had to reach the venting temperature and maintain it stably during the venting time. In order to check the possibilities of this coupling, samples of water spiked with the biomarkers studied were analyzed at different venting temperatures (5, 15, 7893
dx.doi.org/10.1021/ac501917a | Anal. Chem. 2014, 86, 7890−7898
Analytical Chemistry
7894
3-methyl-1-butanol
a
2-ethyl-1-hexanol
Toluene (1.0 ± 0.4 μg/L), propylbenzene (0.3 ± 0.2 μg/L), styrene (0.5 ± 0.3 μg/L), and 1,2,4-trimethylbenzene (0.2 ± 0.1 μg/L) were also quantified in sample no. 9 with the method based on HSPTV-GC/MS.
no. 12 no. 10
(2 ± 1) × 10 1.0 ± 0.4