Multi-Component Profiling of Trace Volatiles in Blood by Gas Chromatography/Mass Spectrometry with Dynamic Headspace Extraction

Abstract

A dynamic headspace extraction method (DHS) with high-pressure injection is described. This dynamic extraction method has superior sensitivity to solid phase micro extraction, SPME and is capable of extracting the entire gas phase by purging the headspace of a vial. Optimization of the DHS parameters resulted in a highly sensitive volatile profiling system with the ability to detect various volatile components including alcohols at nanogram levels. The average LOD for a standard volatile mixture was 0.50 ng mL⁻¹, and the average LOD for alcohols was 0.66 ng mL⁻¹. This method was used for the analysis of volatile components from biological samples and compared with acute and chronic inflammation models. The method permitted the identification of volatiles with the same profile pattern as in vitro oxidized lipid-derived volatiles. In addition, the concentration of alcohols and aldehydes from the acute inflammation model samples were significantly higher than that for the chronic inflammation model samples. The different profiles between these samples could also be identified by this method. Finally, it was possible to analyze alcohols and low-molecular-weight volatiles that are difficult to analyze by SPME in high sensitivity and to show volatile profiling based on multi-volatile simultaneous analysis.

INTRODUCTION

The relation between human odor and disease is an important field of investigation. Odor is one of factors that reflects biological information, and in ancient times, doctors differentiated between such diseases based on smell.¹⁾ Oxidative stress related to various inflammations or diseases enhance the production of certain types of active species, which react with biological metabolites such as lipids and proteins. Several reports have linked oxidative stress with lifestyle- or ageing-related sicknesses.^2–7)

Recently, oxidized lipids produced by oxidative stress were correlated as marker compounds for various diseases. For example, the levels of oxidized lipid were reported to increase at different stages during the development of arteriosclerosis.⁸⁾ Lipid oxidization in the body occurs via a variety of pathways and various secondary metabolites are produced from these oxidized lipids.⁹⁾ Short-chain compounds such as aldehydes and carboxylic acids are known to be released from these compounds.¹⁰⁾ Three categories of volatile compounds are produced by such oxidation reactions: 2-alkenals, 4-hydroxy-2-nonenals, and ketoaldehydes. Some volatile compounds such as acrolein, 4-hydroxy-2-nonenal, 4-hydroxy-2-hexenal, malondialdehyde, and 4-oxo-2-nonenal have also been reported.^11–16) These volatiles are the main compounds produced from lipid oxidation reactions, and are subjects of significant interest because of their production rate, high reactivity, or ease of analysis.^13,14) Biologically-derived lipids are composed of various fatty acid side-chains with different carbon chain lengths and degrees of unsaturation. The oxidation of these lipids in the body produces various volatile compounds.¹⁷⁾ In a previous study, we reported on the analysis of volatiles released from oxidized lipids by the in vitro oxidation of various phospholipids having different fatty acid or polyunsaturated fatty acid groups. As a result, we were able to detect alcohols, ketones, and aldehydes.

Recent developments in metabolomics research have enabled the analysis of various phenotypes based on comprehensive metabolite analysis. Researchers have attempted to identify diseases such as colorectal and lung cancers by multi-compound analyses of blood.^18,19) By applying metabolomics techniques to volatile profiling studies, new knowledge concerning the production and cleavage processes of oxidized lipids can be obtained. Some volatiles released by the cleavage of oxidized lipids in the body were reported to have biological activities along with oxidative stress. One of the cytotoxic lipid-derived aldehydes, 4-hydroxy-2-nonenal, is efficiently reduced to an alcohol by aldo-keto reductase, AR.²⁰⁾ AR also regulates inflammation signals induced by oxidants such as cytokine.²¹⁾ Recently, 4-hydroxy alkenals were reported to have signaling activities in diabetes by aldehyde stresses.²²⁾ Therefore, various volatiles that are produced in the body in addition to aldehydes and alcohols clearly exist, and significant research is focused on identifying such volatiles as possible biomarker because of their various biological activities.²³⁾

Since these volatiles are only present in the body at very low concentrations, approximately ppb or ppt levels, a headspace micro extraction method (HS-SPME) is typically applied as the pretreatment process.²⁴⁾ This technique consists of an extraction and concentration method by exposing the headspace to fibers. This pretreatment process was used to identify significantly increased concentrations of hexanal and heptanal in the blood of lung cancer patients, as compared with those of volunteers.²⁵⁾ It is believed that HS-SPME is a suitable method for the analysis for high concentrations or relatively non-polar volatiles.

However, this method has some disadvantages. For example, the absorbent fiber is fragile and some fiber-derived artifacts are coeluted with volatiles when the fiber is exposed to the high temperature used in the thermal desorption process.²⁶⁾ The sensitivity also differs from compound to compound, since the limit of detection of hexanal and heptanal were reported to be 0.026 and 0.032 nmol L⁻¹, respectively,²⁵⁾ while for ethanol and methanol the value was 0.17 and 0.28 mmol L⁻¹, respectively. Sensitivity for alcohols tends to be poor for HS-SPME.^27,28) In addition, this method uses static extraction; thus, a long fiber exposure time is required to improve the sensitivity. However, interanalyte displacement occurs in complex samples, and profiles of volatiles with a higher affinity for the fiber tended to enhance specific compounds.²⁹⁾ Therefore, a profiling method that is capable of extracting and analyzing a wide variety of volatiles is required.³⁰⁾

In this study, we developed a dynamic head-space extraction method (DHS), which consists of dynamic extraction by purging the head-space of a vial into a sorbent. DHS can extract the entire gas phase compared to the equilibrium driven process of static extraction. Using this method, various volatiles can be extracted with improved sensitivity. In addition, it is able to trap the sorbent with improved extraction efficiency for trace volatiles by combining techniques including cooling the sorbent and adding some salt to produce a salting out effect.^31,32) The combination of these techniques would be expected to improve the sample volume loaded on the column and the sensitivity for the individual volatile components.

Therefore, we developed a volatile profiling system for biological samples by DHS with high-pressure injection, and performed the volatile profiling of both in vivo acute and chronic inflammation models by this analytical system. A detailed comparison and discussion is presented based on these results.

MATERIALS AND METHODS

Materials

Acetone (HPLC grade) was purchased from Kishida Chemical Co., Ltd. (Osaka, Japan). Sodium phosphate, monobasic and dihydrogenphosphate dodecahydrate were purchased from Nacalai Tesque, Inc. (Kyoto, Japan). Sodium chloride (pesticide residue-PCB analysis grade), potassium carbonate, distilled water (HPLC grade), and 2,2′-azobis(2-amidinopropane) dihydrochloride (AAPH) were purchased from Wako Pure Chemical Industries, Ltd. Fifty volatile standard compounds were purchased from Wako Pure Chemical Industries, Ltd., Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), Kishida Chemical Co., Ltd., Nacalai Tesque, Inc., and Sigma-Aldrich (St. Louis, MO, USA) (Table 1). These volatiles were previously reported as oxidized lipid-derived volatiles.¹⁷⁾

Table 1. Target volatile metabolites.

Functional group	CAS	Analyte name	RT (min)	MW (g/mol)	Target ion (m/z)
Alkane	142-82-5	Heptane	10.33	100.21	57
	19549-87-2	2,4-Dimethyl-1-heptene	15.64	126.24	55
Aldehyde	78-84-2	2-Methylpropanal	6.76	72.11	57
	123-72-8	Butanal	7.56	74.12	44
	590-86-3	3-Methylbutanal	9.23	86.13	58
	96-17-3	2-Methylbutanal	9.51	86.13	58
	4170-30-3	2-Butenal	9.60	70.09	69
	110-62-3	Pentanal	10.61	86.13	44
	497-03-0	(E)-2-Methyl-2-butenal	12.57	84.12	84
	1576-87-0	(E)-2-Pentenal	13.05	84.12	83
	66-25-1	Hexanal	14.33	100.16	41
	6728-26-3	(E)-2-Hexenal	16.68	98.14	41
	111-71-7	Heptanal	18.20	114.19	43
	18829-55-5	(E)-2-Heptenal	20.51	112.17	41
	124-13-0	Octanal	21.87	128.21	41
	100-52-7	Benzaldehyde	21.95	106.12	105
	4313-03-5	2,4-Heptadienal	22.94	110.15	81
	2548-87-0	(E)-2-Octenal	24.08	126.20	55
	124-19-6	Nonanal	25.33	142.24	41
	30361-28-5	(E,E)-2,4-Octadienal	26.22	124.18	81
	18829-56-6	(E)-2-Nonenal	27.42	140.22	41
	112-31-2	Decanal	28.57	156.27	67
	5910-87-2	(E,E)-2,4-Nonadienal	29.51	138.21	81
	25152-84-5	2,4-Decadienal	32.50	152.23	81
Alcohol	71-23-8	1-Propanol	7.63	60.10	59
	78-92-2	2-Butanol	8.24	74.12	59
	78-83-1	2-Methyl-1-propanol	9.22	74.12	73
	71-36-3	1-Butanol	10.42	74.12	56
	616-25-1	1-Penten-3-ol	11.02	86.13	57
	123-51-3	3-Methyl-1-butanol	12.71	88.15	42
	108-11-2	4-Methyl-2-pentanol	13.15	102.17	45
	71-41-0	1-Pentanol	13.92	88.15	55
	1576-95-0	(Z)-2-Penten-1-ol	14.65	86.13	57
	111-27-3	1-Hexanol	17.60	102.17	56
	111-70-6	1-Heptanol	21.20	116.12	55
	50999-79-6	1-Octen-3-ol	21.49	128.21	57
	104-76-7	2-Ethyl-1-hexanol	23.15	130.23	57
	35192-73-5	1-Nonen-4-ol	24.44	142.24	55
	111-87-5	1-Octanol	24.62	130.23	55
	18409-17-1	(E)-2-Octen-1-ol	24.85	128.21	57
	100-51-6	Benzyl alcohol	26.58	108.14	79
Ketone	107-87-9	2-Pentanone	10.29	86.13	58
	4312-99-6	1-Octen-3-one	21.07	126.20	55
	111-13-7	2-Octanone	21.43	128.21	58
	1669-44-9	3-Octen-2-one	23.35	126.20	55
Cyclic compound	625-86-5	2,5-Dimethylfuran	10.91	96.13	96
	108-88-3	Toluene	13.32	92.14	91
	4466-24-4	2-n-Butylfuran	17.79	124.18	81
	100-42-5	Styrene	18.33	104.15	104
	3777-69-3	2-Pentylfuran	21.43	138.21	81
Internal standard	920-66-1	1,1,1,3,3,3-Hexafluoro-2-propanol	12.87	168.04	99

Each volatile standard was diluted with acetone to a concentration of 100 ng μL⁻¹. Mouse plasma was purchased from Kohjin Bio Co., Ltd. (Saitama, Japan). This pooled plasma originated from a mixture of male and female mice that were 8–12 weeks old and sexually mature.

In vitro oxidized mouse plasma sample

The oxidative treatment was based on the protocol described by Niki.³³⁾ Mouse plasma (100 μL) was added to 250 μL of phosphate buffered saline (PBS). AAPH (50 μL) was added 1 h after the start of incubation at 37°C and the sample was agitated at 1400 rpm agitation under aerated conditions. The final concentration of AAPH was 0.3 and 30 mmol L⁻¹ and was used to monitor the effect of oxidative stress from a mild level to a high level that is technically impractical. The mixture (total volume=400 μL) was then incubated under the same conditions for 4 h. This mixture was used as the in vitro oxidized mouse plasma sample.

In vivo mouse plasma sample

We used two in vivo mouse model samples: the first was zymosan-A treated mice and the second was IL-10 knockout mice. In the former, the acute inflammation sample, peritonitis was induced in C57BL/6J male mice, n=5, by an intraperitoneal injection of zymosan-A (1 or 10 mg/mouse) while the control group was only injected with vehicle.³⁴⁾ Plasma samples were taken from mice that were sacrificed after 72 h. The latter chronic inflammation sample is an age-related spontaneous model of chronic enterocolitis (Crohn’s disease). IL-10 knockout mouse plasma, n=6, was prepared at 8 and 16 weeks to compare the difference between slight and severe inflammation conditions. The control sample was C57BL/6J wild mouse at 8 weeks.

Sample extraction and GC/MS analysis

Mouse plasma (100 μL) or oxidized plasma (400 μL) samples were diluted with PBS to a final volume of 1 mL. Plasma from zymosan-A treated mice (50 μL) was diluted with PBS to a final volume of 1 mL. The mixtures were transferred to 20 mL vials and potassium carbonate (1.4 g) was added to salt out the volatile components. The internal standard (1 μL of a 100 ng μL⁻¹ solution) was added to the vials, which were then sealed with crimp caps (silicone/PTFE high-temperature seals, diam. 20 mm).

HS-SPME analysis condition was based on as below. The vial was incubated at 50°C in water bath for 10 min. After incubating, a DVB/CAR/PDMS fiber (2 cm–50/30 μm, Supelco) was exposed in the headspace for 5 min to extract volatile compounds. The fiber was then exposed in the inlet for 2 min for thermal desorption under the splitless mode. The GC inlet temperature was 280°C.

Dynamic headspace extraction was performed with HS20 (Shimadzu, Kyoto, Japan). The extraction process was carried out in the trap mode using helium gas with the HS20 side sample and transfer line temperatures of 150°C. The cooling trap and waiting temperature were −10°C and 25°C, respectively. Vial agitation was set at three. Before purging with inert gas, the vial was agitated at 50°C for 10 min followed by purging at 100 kPa for 2 min. The purge equilibration time was 0.1 min, and the purge gas was loaded onto the sorbent (50 mg, 60/80 mesh Tenax GR) for a period of 1 min. The loading equilibration time was also set at 0.1 min. This procedure was repeated five times. After the samples were loaded, the sorbent was purged to dryness at 60 kPa for 10 min. The injection temperature was 280°C for 2 min, and the needle flush time was 5 min.

Gas chromatography mass spectrometry analysis was performed with a GCMS-TQ8030 instrument (Shimadzu, Kyoto, Japan). For separating the volatile mixture, an InertCap 5MS/NP column (0.25 μm×30 m, 1 μm, GL Sciences Inc., Tokyo, Japan) and an InertCap WAX-HT column (0.25 μm×30 m, 0.25 μm, GL Sciences Inc., Tokyo, Japan) were connected by an inner-seal connector (GL Sciences Inc., Tokyo, Japan). The sample was injected in the pulsed split injection mode with a split ratio of 1 : 10 (v/v), and an injection pressure of 360 kPa for 2 min after injection. Argon was used as the CID gas at a pressure of 200 kPa. Helium was used as the carrier gas with a linear velocity of 30 cm/sec. The column temperature was kept at 50°C for 3 min, then increased by 5°C/min to 230°C, and held at this temperature for 1 min. The mass spectrometer side transfer line and ion source temperature were 230°C. Spectra were recorded in the scan/SIM mode. Scan time and SIM loop time were 0.1 and 0.2 s, respectively. The mass range was 35 to 300 m/z in scan mode. The minimum event time was 0.012 s and the dwell time was 4.5 ms.

Identification of volatiles

The results were checked using the GCMSsolution software package (Shimadzu, Kyoto, Japan). The detected peaks were identified based on target m/z and retention time (Table 1). Unknown peaks were compared with the NIST library, NIST/EPA/NIH Mass spectral Library (NIST 11). Peak areas were calculated for each target m/z in the extracted ion chromatogram.

Statistical analysis

Each peak area was divided by the peak area for the internal standard. Relative quantitative data are presented as the mean±the standard deviation (S.D.). Significance test was performed using Microsoft Excel 2010.

RESULTS AND DISCUSSION

Optimization of high-pressure injection conditions

We focused on the dynamic head-space extraction method (DHS) because it permits various volatiles including alcohols to be simultaneously analyzed, which show poor sensitivities with HS-SPME. DHS uses a carrier gas to purge the head-space through a sorbent, which extracts the volatile components. The volatiles were then extracted and concentrated in a trap, which was rapidly heated for thermal desorption and loaded onto a column for separation.

For the analysis of trace volatiles, to maintain a sufficient carrier gas flow for the thermal desorption process, which is required for DHS, the split injection mode was applied. However, loading samples at any split ratio resulted in them being ejected from the inlet. Thus, some sample loss occurred with the DHS method even though the extraction efficiency is higher. Therefore, we focused on reducing sample loss by using a high-pressure injection method. We envisioned that increasing the carrier gas flow rate would result in sufficient total flow by pressuring the inlet during injection. This would result in a decreased split ratio at the same flow rate and a higher coverage of volatiles with improved sensitivity.

We first examined the influence of injection pressure at 300, 330, 360, and 400 kPa with 100 ng of the volatile standard mixture sample. The volatile standard was selected based on analysis of the in vitro oxidized lipid standard samples taken from our previous study.¹⁷⁾ Tenax TA, with an injection time of 3 min was used for the sorbent trap. The results for selected volatiles from each category are shown in Fig. 1, and all results for volatile compounds are shown in the supplemental data (S. Fig. 1).

Fig. 1. Area value for different injection pressures. Closed bars: 300 kPa, Open bars: 330 kPa, Dotted bars: 360 kPa, Hatched bars: 400 kPa.

All trace volatiles, including alcohols at the ng mL⁻¹ level, were detected when this extraction method was used. Compound peak areas increased with increasing inlet pressure and the split ratio was the lowest for the highest pressure tested (400 kPa). However, the peak area of some volatile components decreased despite having the highest loaded sample volume. This can be attributed to a faster flow rate and a lower split ratio. During injection at a higher pressure, the flow rate was increased and the volatiles eluted earlier, resulting in a poor separation. Moreover, the split ratio was also lowered due to the high pressure injection, which, within a certain range, would be expected to improve the peak height. However, when the split ratio became too low, instead of higher peaks, the peaks turned broad, which directly led to decreased peak areas. The average RSD value for alcohols at 400 kPa was 7%, while the average RSD value for other compounds was 15%. In contrast, the average RSD value of alcohols at 330 kPa and 360 kPa were 4% and 6%, respectively, which were the lowest of the four pressures examined. The average RSD value for aldehydes and furan compounds was also the lowest at 6% for an injection pressure of 360 kPa. However, the RSD value for alkanes was higher because of measurement error and was improved to 7% by excluding an outlier. A pressure of 360 kPa gave the highest peak area values for all target volatiles. Therefore, an injection pressure of 360 kPa was determined to be optimal because of the improved sensitivity for low-molecular-weight compounds and alcohols which are difficult to analyze by SPME.

Next, we examined the injection time for a pressure of 360 kPa (Fig. 2). The injection time indicated here refers to both the value set for the thermal desorption time and that of high pressure split injection, which are technically the same. While the sample loading volume improved for longer injection time, the peak separation and shape were affected because the flow rate was faster than the normal injection mode. In addition, a shorter injection time would contribute to measurement errors because of inadequate thermal desorption. Therefore, injection times of 1, 2, 3, and 5 min were examined for the volatile standard mixture, and the area value differences were compared. The results for selected volatiles from each category are shown in Fig. 2, and results for all volatile are shown in the supplemental data (S. Fig. 2).

Fig. 2. Area value for different injection times. Closed bars: 1 min, Open bars: 2 min, Dotted bars: 3 min, Hatched bars: 5 min.

While the area values were slightly different for the 2 min injection time, this was a sufficient period of time to achieve thermal desorption. A shorter injection time of 1 min afforded poor RSD values for 1-propanol (25%) and other compounds (10%). This can be attributed to an insufficient thermal desorption time, which resulted in completely different ratios for each volatile component. In contrast, a 2 min injection time produced RSD values lower than 10% for all volatile including 1-propanol (7%) with a concomitant increase in its area value. A 3 min injection time afforded constant area values although the area values for low-molecular-weight compounds such as 1-propanol, 2-butanol, 2-methyl-1-butanol, butanal, and 2-methyl-1-propanol were decreased. This was because these compounds were not retained in the column for sufficient time due to the long duration of the fast flow rate, thus making the peak shape lower. In addition, several peaks were overlapped with the injection shock peak because they were eluted too rapidly from the column. Therefore, the optimized conditions for the DHS method with high-pressure injection were an injection pressure of 360 kPa and an injection time of 2 min.

Investigation of the trapping sorbent

With optimized injection conditions for the DHS method determined, we now focused on factors affecting extraction efficiency including the trapping material. The purged gas phase from the DHS was trapped and extracted. While silica gel was used as a sorbent for polar compounds, its high affinity for water and variable characteristics at high temperatures posed significant limitations. To circumvent this problem, two different types of sorbents were evaluated; the first was a synthetic resin and the second was a carbon-based material. A 2,6-diphenyl-p-phenylene oxide polymer is mainly used as a synthetic resin, because it can extract various compounds such as low-molecular-weight and polar compounds. In contrast, graphite carbon black (nonporous) and carbon molecular sieve (porous) are used as a carbon-based sorbent. These sorbents are efficient for extracting intermediate and high-boiling-point volatiles.

In this study we tested three different sorbents: the synthetic resin Tenax TA, the carbon sorbent Carboxen 1000/Carbopack, and the synthetic resin with a graphite carbon Tenax GR. The peak area values for the standard volatile mixture were compared using these sorbents; the results for selected volatiles from each category are shown in Fig. 3, and the results for all volatiles are shown in the supplemental data (S. Fig. 3).

Fig. 3. Area value for different sorbents. Closed bars: Tenax TA, Open bars: Tenax GR, Dotted bars: Carboxen 1000/Carbopack.

Tenax GR had an improved breakthrough point compared to Tenax TA because the sorbent is mixed with carbon material and has a higher affinity for low-molecular-weight volatiles.³⁵⁾ The RSD value for each sorbent was calculated and Tenax GR was found to be superior with an average RSD of 4%. In contrast, Tenax TA had an average RSD of 7% while the RSD for some volatiles such as 2-butanol, 1-octanol, 2-ocnten-1-ol, and benzyl alcohol were approximately 10%. Moreover, the area values for Tenax GR were double that of Tenax TA. The area value for Carboxen 1000/Carbopack was the worst of the three sorbents. Compared with Tenax, this sorbent strongly held water and required a longer drying purge time. The low area value likely results from the loss of volatile compounds from the sorbent during the purge process. Therefore, Tenax GR was selected as the sorbent for volatile profiling.

Comparison of headspace micro extraction method

We next compared the difference between the volatile profile with the DHS system and that for HS-SPME. HS-SPME is a static extraction method, which characteristically emphasizes volatile compounds with either a higher concentration or that interact strongly with the fibers. Therefore, this method affords lower recovery rates for low-molecular-weight compounds or compounds that have a poor affinity for the fiber, thus resulting in poor sensitivity. In contrast, the DHS method is a dynamic extraction method in which the entire headspace is purged, resulting in an increased sensitivity for volatile components. We previously showed that the DHS method could be used to analyze alcohols and low-molecular-weight volatiles with a high sensitivity, which is difficult for HS-SPME. Therefore, we further compared the two methods to identify the differences in their respective volatile profiles.

The results for some selected volatiles from each category using both methods are shown in Fig. 4, and all volatile results are shown in the supplemental data (S. Fig. 4). In the HS-SPME method, low-molecular-weight compounds such as 1-propanol, 2-butanol, 2-methyl-1-propanol, 1-penten-3-ol, 2-pentanone, 2-methylpropanal, 3-methylbutanal, 2-methylbutanal, and 2-pentenal were undetectable. In addition, the area values for compounds such as 3-methyl-1-butanol, 4-methyl-2-pentanol, and 2-penten-1-ol were one order of magnitude lower than the corresponding values for DHS. However, high-molecular-weight compounds such as 2-ethyl-1-hexanol and 1-octen-3-one gave similar results. In a previous study, the parameter of fiber-sample partition coefficient, K_fs, was found to be important in estimating the sensitivity of SPME extraction. Volatiles present at high concentrations with high K_fs values tend to displace those with smaller values during the extraction process.²⁹⁾ This is because HS-SPME is based on a mobile equilibrium between the water phase and gas phase, as well as that between the gas phase and the fiber. Some volatiles with hydroxyl groups tend to interact with water, leading to decreased recovery rates. To solve this problem, we added salt to the samples to create a salting out effect. By applying the salting out effect to HS-SPME, the signal intensity of the alcohols was improved.³¹⁾ However, the simultaneous volatile mixture extraction and the limited fiber coating volume affected the SPME extraction and volatiles having low interactions with the fibers might be released from fiber due to being more readily displaced.^36,37) Therefore, low-molecular-weight volatiles tend to have lower area values while higher-molecular-weight volatiles have improved area values (Fig. 4). In contrast, the DHS method is less influenced by temperature because of the separate purging and sorbent extraction steps. Furthermore, the sorbent temperature could be set at −10°C during the extraction process thereby improving the breakthrough volume and increasing the extraction efficiency for low-molecular-weight volatiles.^32,38)

Fig. 4. Comparison of DHS extraction and HS-SPME extraction. Closed bars: DHS method, Open bars: HS-SPME method.

The average RSD value of all volatiles was 7% for DHS and 12% for HS-SPME. In addition, the average RSD for alcohols was 10% for DHS and 15% for HS-SPME. As a result, the RSD for the DHS method was slightly better than that for HS-SPME, however the profiles of these two methods were completely different, especially in the case of low-molecular-weight volatiles. Temperature is considered to be one of the main factors affecting extraction efficiency.

The use of heated vials in the HS-SPME extraction procedure caused the fiber in the head space to become heated, which resulted in the desorption of some compounds from the fiber in the extraction process.³⁷⁾ In contrast, the DHS method is less influenced by temperature because of the separate purging and sorbent extraction steps. Furthermore, the sorbent temperature could be set at −10°C during the extraction process thereby improving the breakthrough point and increasing the extraction efficiency for low-molecular-weight volatiles.³²⁾

Therefore, an efficient volatile profiling system using DHS and high-pressure injection was developed. By applying these methods, sample losses from the split mode were reduced and various volatiles such as alcohols and low-molecular-weight volatiles could be detected.

Validation of the DHS method

The DHS volatile profiling system was validated under optimized conditions. Validation results including linearity, R², LOD, RSD at a concentration near the LOD, intra- and inter-day variation, and recovery rate are shown in Table 2. The LOD was calculated based on the formula for the standard deviation and slope of the calibration curve defined by IUPAC (LOD=3.3×S.D./slope).³⁹⁾ RSD values from results at a concentration near the LOD were also calculated. The intra- and inter-day variations were calculated by analyzing the standard mixture three times per day for three days and calculating the RSD values based on these data. Finally, we prepared 100 μL pooled mouse plasma samples and spiked the standard for the recovery rate experiment.

Table 2. Validation result for each volatile component.

Compound	Slope	Intercept	R2	Linear (ng mL−1)	LOD (ng mL−1)	RSD (%)*		RSD (50 ng mL−1)		Recovery (25 ng mL−1, n=3)
								Intraday (n=3)	Interday (n=9)	Average (%)	RSD (%)
Alkane
Heptane	7353	2437	0.9997	0–10	0.14	3.0	b	3.1	12	105	5.9
2,4-Dimethyl-1-heptene	14292	8590	0.9991	0–100	0.08	14	a	1.6	13	98	1.4
Alcohol
1-Propanol	2612	1743	0.9960	0–50	0.32	8.4	b	9.3	12	85	13
2-Butanol	3836	1531	0.9998	0–100	0.07	14	a	7.4	9.9	95	11
2-Methyl-1-propanol	115	34	0.9989	1–100	4.20	11	d	6.9	9.3	147	15
1-Butanol	13499	33574	0.9950	0–100	0.31	5.5	b	6.3	8.1	96	7.9
1-Penten-3-ol	41632	42049	0.9984	0–100	0.08	15	a	6.8	9.2	94	11
3-Methyl-1-butanol	16560	18299	0.9981	0–100	0.08	21	a	5.8	8.8	93	10
4-Methyl-2-pentanol	48338	67615	0.9969	0–100	0.06	12	a	5.5	7.9	98	9.4
1-Pentanol	28655	30427	0.9986	0–100	0.09	15	a	6.0	8.8	92	9.3
2-Penten-1-ol	12086	6758	0.9996	0–100	0.09	25	a	5.7	8.6	84	8.6
1-Octen-3-ol	28041	19704	0.9992	0–100	0.20	6.8	b	3.6	7.6	90	7.4
1-Hexanol	14910	14651	0.9990	0–100	0.15	5.0	b	4.5	7.9	90	8.6
1-Heptanol	14543	8829	0.9997	0–100	0.28	11	b	3.6	8.0	88	8.5
2-Ethyl-1-hexanol	21992	260285	0.9996	0–100	3.71	11	d	3.8	8.8	84	2.2
1-Nonen-4-ol	36095	8761	0.9997	0–100	0.08	19	a	3.7	7.8	87	8.3
1-Octanol	14541	7333	0.9997	0–100	0.51	20	b	3.5	8.2	89	9.1
2-Octen-1-ol	11441	281	0.9994	0–100	0.33	20	b	3.6	8.0	82	8.6
Benzyl alcohol	20360	62381	0.9994	0–100	0.58	8.0	b	3.8	6.9	77	6.0
Ketone
2-Pentanone	6122	3925	0.9976	0–100	0.11	7.0	b	3.7	13	100	2.7
1-Octen-3-one	12077	−3209	0.9993	0–50	0.49	20	b	14	11	71	2.4
2-Octanone	24930	39841	0.9968	0–100	0.02	6.0	a	3.1	12	101	2.2
3-Octen-2-one	31667	14012	0.9980	0–100	0.02	1.5	a	3.3	12	93	2.0
Aldehyde
2-Methylpropanal	172	200	0.9872	0–100	0.90	18	c	4.3	25	102	11
Butanal	4710	3421	0.9930	0–50	0.24	2.6	b	1.0	14	95	2.7
3-Methylbutanal	3182	3425	0.9980	0–100	0.07	15	a	2.4	10	112	4.1
2-Methylbutanal	5538	4103	0.9987	0–100	0.06	20	a	2.6	9.8	112	4.2
2-Butenal	6393	2700	0.9995	0–100	0.10	23	a	4.9	10	126	16
Pentanal	5352	13476	0.9958	0–100	1.03	10	c	0.7	12	128	3.1
2-Methyl-2-butenal	13361	18808	0.9963	0–100	0.04	12	a	2.0	11	119	2.9
2-Pentenal	10972	1341	0.9997	0–100	0.04	9.0	a	3.3	10	69	13
Hexanal	6864	7373	0.9984	0–100	0.20	13	b	3.2	12	121	4.0
2-Hexenal	6892	1113	0.9997	0–100	0.25	9.3	b	3.4	11	89	4.2
Heptanal	2025	3809	0.9976	0–100	0.86	3.1	c	4.2	9.9	116	5.3
2-Heptenal	6947	−5171	0.9983	0–100	0.11	17	b	5.4	9.0	81	7.0
Octanal	5533	−589	0.9968	0–100	0.27	5.4	b	13	8.9	112	5.3
2,4-Heptadienal	28487	−494	0.9994	0–100	0.04	10	a	7.3	7.6	91	9.0
2-Octenal	6626	−16090	0.9852	0–100	0.12	22	b	13	7.7	75	9.1
Nonanal	3877	−2753	0.9865	0–100	1.26	22	d	23	15	108	5.8
2,4-Octadienal	24419	−19659	0.9956	0–100	0.05	13	a	14	9.1	92	5.9
2-Nonenal	1671	−3695	0.9997	0.5–50	0.99	10	c	26	21	66	16
Decanal	2089	−1445	0.9683	0–100	1.97	24	d	35	27	94	9.2
2,4-Nonadienal	41974	−41813	0.9903	0–100	0.08	15	a	24	20	92	9.0
2,4-Decadienal	8695	−14893	0.9781	0–100	0.12	11	b	37	39	92	12
Benzaldehyde	34279	64892	0.9984	0–100	0.41	8.7	b	5.5	8.8	144	10
Cyclic compound
2,5-Dimethylfuran	19228	18172	0.9981	0–100	0.12	7.0	b	4.9	12	94	2.7
Toluene	103199	3000000	0.9980	0.5–100	5.64	4.4	d	3.7	11	95	3.2
2-n-Butylfuran	41227	46372	0.9973	0–100	0.01	6.6	a	2.9	12	101	2.4
2-Pentylfuran	29035	40791	0.9971	0–100	0.01	5.2	a	2.8	12	99	2.3
Styrene	35586	35963	0.9974	0–100	0.05	16	a	2.9	13	99	2.2

* RSD values of the standard mixture with the concentrations closed to their respective LODs. a: 0.1 ng mL⁻¹, b: 0.5 ng mL⁻¹, c: 1 ng mL⁻¹, d: 10 ng mL⁻¹.

When the DHS method was applied, the various volatiles were detected at ng mL⁻¹ levels. As shown in Fig. 5, it was possible to detect volatiles even in trace concentrations (1 ng mL⁻¹). These peaks were also detected in the SIM mode, so it was possible to separate and identify each volatile compound by its specific target m/z (Table 1). Furthermore, alcohols were detected with high sensitivity as well as other volatiles. While it has been reported that, in the analysis of alcohols by HS-SPME, the analytical range for polar volatiles is limited.⁴⁰⁾ For example, the LOD values for ethanol and methanol in human blood by HS-SPME were reported to be 5.6 μg mL⁻¹ and 12.8 μg mL⁻¹, respectively.²⁷⁾ Alcohol detection sensitivities also tend to be poor compared with other compounds. Because alcohols contain hydroxyl groups and are hydrophilic compounds, they tend to interact with the water phase, which leads to low recovery rates in HS-SPME. A previous study by Zuba et al. reported on an improved performance of alcohol analysis by employing a salting out effect.³¹⁾ Sensitivities for volatile compounds varied with their solubility. As a high soluble salt, potassium carbonate has been shown to be highly effective in driving volatiles, especially alcohols into the gas phase. However, the LODs were only improved to μg L⁻¹ or mg L⁻¹ levels due to the static extraction process of HS-SPME.

Fig. 5. Total ion current chromatograms of different concentrations of volatile standards. (A) retention time: 7–16 min, (B) retention time: 16–25 min, (C) retention time: 25–34 min.

In contrast, this DHS method can detect alcohols at lower concentrations because it relies on purging the entire gas phase into the sorbent instead of equilibrium extraction. Thus, low-molecular-weight volatiles are trapped in the sorbent resulting in improved sensitivities. RSD values calculated for concentrations near the LOD were different for each volatile compound. This result indicates that some volatile compounds were analyzed with good or poor reproducibilities, even though they had a lower LOD value. In this way, we were able to gain knowledge regarding the features of each volatile compound by this method. Moreover, in the intra- and inter-day variation analysis, the RSD values for 2-nonenal, decanal, 2,4-nonadienal, and 2,4-decadienal were in excess of 20%. This is possibly caused by their tendency to adsorb to the inner wall of the vial due to their low solubilities in the water phase, and consequently this unstable condition adds some level of uncertainty to the extraction efficiency and the following analysis.²⁹⁾

Finally, we examined the recovery rate for the volatiles. The RSD values for most of the volatiles were less than 10%, with the exception of 2-methyl-1-propanol (15%). The average recovery rate also did not show significant changes in sensitivity because of matrix effects. Since the HS-SPME extraction process is based on an equilibrium between two phases, the recovery rate for HS-SPME tends to be lower in the case of an analysis of multiple compound samples.⁴¹⁾ As shown in a previous report, the recovery rates of volatiles were different between SPME and DHS. For example, in contrast to the low recovery rate for toluene for HS-SPME, 46.2%,⁴²⁾ the recovery rate for toluene was as high as 95% for DHS. Therefore, the above findings demonstrated that various volatiles can be analyzed simultaneously at nanogram levels by improving the sample loading volume of the DHS method.

Biological sample analysis

We carried out analyses of biological samples by the DHS method. Since various lipids are present in the body, we envisioned that various volatile compounds would be produced by oxidation. Therefore, we prepared in vitro oxidized biological samples to analyze the volatile components and determine how the volatile profiles changed. In this study, pooled mouse plasma samples were oxidized based on Niki’s report with oxidant concentrations of 0, 0.3, and 30 mmol L⁻¹.³³⁾

Three different profiles were obtained from the analysis of the oxidized mouse plasma. The first profile showed an increase in volatiles with a low oxidant concentration and a decrease with a high oxidant concentration. Regarding volatiles, the second profile showed a decrease while the last profile showed an increase in according to the oxidant concentration (Table 3).

Table 3. Volatile profiles for in vitro oxidized mouse plasma.

Increased in low concentration; Decreased in high concentration	Increased according to concentration		Decreased according to concentration
Alkane	Alcohol	Nonanal	Alcohol
Styrene	1-Hexanol	2,4-Octadienal	1-Propanol
Heptane	1-Heptanol	2,4-Heptadienal	2-Octen-1-ol
2,4-Dimethyl-1-heptene	2-Ethyl-1-hexanol	Ketone
Alcohol	1-Octanol	2-Pentanone
1-Butanol	1-Penten-3-ol	2-Octanone
2-Butanol	2-Penten-1-ol	1-Octen-3-one
3-Methyl-1-butanol	1-Pentanol	Cyclic compound
Benzyl alcohol	1-Octen-3-ol	2-n-Butylfuran
Aldehyde	Aldehyde	2-Pentylfuran
2-Methyl-2-butenal	2-Methylpropanal
Octanal	Butanal
2-Octenal	Pentanal
Decanal	Hexanal
Cyclic compound	Heptanal
Toluene	Benzaldehyde

In previous studies with in vitro oxidized samples, mild oxidation involved the use of a low concentration of oxidant, while a hemolytic reaction was observed at an oxidant concentration of 200 mmol L⁻¹. However, concentrations of antioxidants such as α-tocopherol and ascorbic acid were reported to decrease and lipid peroxides were produced after adding a low concentration of oxidant.³³⁾ In this study, the relative concentrations of volatiles also changed under different oxidation conditions, and some volatiles compounds increased at low oxidant concentration and decreased under enhanced oxidative stress. This result was caused by the further reaction of the volatiles produced under low oxidative stress at higher oxidant levels resulting in oxidative cleavage to afford different volatile components. However, the concentration of some volatiles increased with increasing oxidant concentration, which tends to be produced during the progression of certain types of diseases. For example, significantly increased concentrations of hexanal and heptanal in lung cancer patients and 1-octen-3-ol in liver cancer patients have been reported.^43,44) 1-Propanol and 2-octen-1-ol also decreased with oxidant concentration because of subsequent oxidative cleavage under increased conditions of oxidative stress. However, 1-propanol was reported to be significantly increased in breath samples of lung cancer patients.⁴⁵⁾ It has been shown that cytochrome p450 enzymes are induced in cancer patients and this enzyme catalyzes the hydroxylation of some volatiles.⁴⁵⁾ In addition, aldose reductase can also catalyze the conversion of aldehydes to alcohols in the human body.²⁰⁾ Therefore, the volatile profiles of in vitro oxidized samples using DHS were different from those in previous reports.

We next analyzed the volatile profiles of zymosan-treated mouse plasma and IL-10 knockout mouse plasma samples. These models show different disease states, with the former model representing acute inflammation while the latter is a chronic inflammation model. As shown in Table 4, 39 compounds were detected in zymosan-treated mouse plasma, and the concentrations of 20 of these compounds were significantly different for each sample.

The profile pattern for eight of the 20 volatiles produced in the in vitro oxidized mouse plasma samples were the same, while ten volatiles have not been previously reported in blood. These results may be due to dynamic metabolic changes in the acute inflammation model. In a previous report, inflammation was enhanced after the injection of 10 mg/body zymosan and acute exacerbation was observed 72 h after the treatment.³⁴⁾ Applying this method, the volatile profile for the acute inflammation sample was determined, and various volatiles derived from oxidized lipids were detected in the blood.

Table 4. List of compounds with significantly different concentrations in zymosan-treated mouse plasma.

Not reported compounds		Reported compounds
Control vs. 1 mg/body	1 mg/body vs. 10 mg/body	Control vs. 1 mg/body	3-Methyl-1-butanol
1-Hexanol	2-Butenala	Acetoneb	2,4-Dimethyl-1-heptene
1-Octen-3-one	1-Octen-3-one	1-Propanol	1 mg/body vs. 10 mg/body
2-Ethyl-1-hexanola	2-Octanone	2-Butanola	Acetoneb
1-Octanol	2-Butyl-1-octanolb	3-methylbutanala	1-Propanol
Control vs. 10 mg/body	1-Decanolb	Heptanea	1-Pentanol
2-Butenala		1-Butanola	1-Octen-3-ol
2-n-Butylfurana		Control vs. 10 mg/body	Nonanal
2-Octen-1-ola		1-Propanol
		2-Pentanone

^a: Profile pattern similar to that of in vitro oxidized mouse plasma. ^b: Annotated by NIST library.

For the analysis of IL-10 knockout mouse plasma, we used plasma samples derived from mice with heavy bowel inflammation and either none or very slight inflammation in the dissection view. As shown in Table 5, 40 volatiles were detected and 14 volatiles had significantly different concentrations.

Table 5. List of compounds with significantly different concentrations in IL-10 KO mouse plasma.

Not reported compounds		Reported compounds
Control vs. 8 weeks	8 weeks vs. 16 weeks	Control vs. 8 weeks	2,4-Dimethyl-1-heptenea
3-Methyl-1-butanola	2-Pentanone	Hexanala	Acetonea,b
Control vs. 16 weeks	2-n-Butylfurana	2,4-Dimethyl-1-heptenea	2-Propanola,b
Butanala	2-Methyl-2-butanolc	Control vs. 16 weeks	8 weeks vs. 16 weeks
2-n-Butylfurana		2-Butanola	Toluenea
2-Butyl-1-octanola		Heptane	2-Propanola,b
3-Methyl-2-butanolc		Toluenea
2-Methyl-2-butanolc		Hexanala

^a: Profile pattern similar to that of in vitro oxidized mouse plasma. ^b: Annotated by NIST library. ^c: Specific compounds from IL-10 KO mouse.

Ten of the 14 volatiles profile patterns were same as that in the in vitro oxidized mouse plasma, while seven volatiles have not been previously reported in blood. Moreover, 3-methyl-2-butanol and 2-methyl-2-butanol were detected only in IL-10 knockout mice samples. Analysis of the volatile compounds present in the two different inflammation models showed completely different volatile profiles.

Moreover, the acute and chronic inflammation models exhibited very different profiles even in the case of volatiles that were found in both models (S. Table 1). Increased concentrations of alcohols and aldehydes were detected in the acute inflammation samples, 6 of which were found to be significantly different between the acute and the chronic inflammation samples. In the acute inflammation model, the progression of inflammation is rapid from exacerbation to remission, during which, large amounts of oxidized lipid-derived volatiles are produced. In contrast, the chronic inflammation samples showed higher levels of other volatiles such as toluene and 2,4-dimethyl-1-heptene.

While many studies have attempted to distinguish diseases based on changes in a specific volatile component, the sensitivity of current detection methods has always been a limiting factor. In this study, a new method with a high sensitivity was developed for multi-volatile profiling of biological samples including alcohols.

CONCLUSION

In this study, we developed a DHS method with high-pressure injection that constitutes a highly sensitive multi-volatile profiling system. This DHS method has superior sensitivity to SPME, and is able to extract the entire gas phase by purging the headspace of a vial. As a result, we were able to detect fifty compounds including alcohols at nanogram levels. This method was successfully applied to biological samples for trace volatile profiling. In addition, different volatile profile patterns could be obtained as well as an increase with increasing oxidant concentration. This analysis model was applied to two in vivo models, acute and chronic inflammation samples, and the volatile profiles of the two samples were found to be different. In addition, the concentration of alcohols and aldehydes in samples from the acute inflammation model were significantly higher than the chronic inflammation model samples.

Various volatiles are produced via different metabolic pathways in the body. However, current studies have focused on identifying specific volatiles and the use of multi-volatile profiling has not been examined. The development of the DHS system in this study allowed us to carry out multi-volatile profiling analyses, and demonstrate the volatile profile associated with inflammation, which is a preliminary stage of various diseases. Further development of this method may permit volatile biomarker candidates related to inflammation or volatile profile changes produced by different disease types to be identified, instead of analyzing single marker changes with previous methods.

Acknowledgements

We thank Dr. Kawana for providing the sorbent trap, Tenax GR, and Carboxen 1000/Carbopack sorbent and Dr. Fujieda for providing zymosan-A treated mouse plasma samples. And this work was partially supported by a Grant-in-Aid for Young Scientists (A) (23686120).

REFERENCES

• 1) D. Penn, W. K. Potts. Chemical signals and parasite-mediated sexual selection. Trends Ecol. Evol. 13: 391–395, 1998.
• 2) Y. Christen. Oxidative stress and Alzheimer’s disease. Am. J. Clin. Nutr. 71 (suppl): 621S–629S, 2000.
• 3) P. Jenner. Oxidative stress in Parkinson’s disease. Ann. Neurol. 53 (suppl3): S26–S38, discussion, S36–S38, 2003.
• 4) J. E. Repine, A. Bast, I. Lankhorst, The Oxidative Stress Study Group. Oxidative stress in chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med. 156: 341–357, 1997.
• 5) N. R. Madamanchi, A. Vendrov, M. S. Runge. Oxidative stress and vascular disease. Arterioscler. Thromb. Vasc. Biol. 25: 29–38, 2005.
• 6) J. W. Baynes. Role of oxidative stress in development of complications in diabetes. Diabetes 40: 405–412, 1991.
• 7) T. Finkel, N. J. Holbrook. Oxidants, oxidative stress and the biology of ageing. Nature 408: 239–247, 2000.
• 8) J. M. Upston, J. Neuzil, R. Stocker. Oxidation of LDL by recombinant human 15-lipoxygenase: Evidence for α-tocopherol-dependent oxidation of esterified core and surface lipids. J. Lipid Res. 37: 2650–2661, 2002.
• 9) M. Rosario, M. Domingues, A. Reis, P. Domingues. Mass spectrometry analysis of oxidized phospholipids. Chem. Phys. Lipids 156: 1–12, 2008.
• 10) E. N. Frankel. Lipid oxidation: Mechanisms, products and biological significance. J. Am. Oil Chem. Soc. 61: 1908–1917, 1984.
• 11) K. Uchida. 4-Hydroxy-2-nonenal: A product and mediator of oxidative stress. Prog. Lipid Res. 42: 318–343, 2003.
• 12) K. Warner, W. E. Neff, W. C. Byrdwell, H. W. Gardner. Effect of oleic and linoleic acids on the production of deep-fried odor in heated triolein and trilinolein. J. Agric. Food Chem. 49: 899–905, 2001.
• 13) T. Hayashi, K. Uchida, G. Takebe, K. Takahashi. Rapid formation of 4-hydroxy-2-nonenal, malondialdehyde and phosphatidylcholine aldehyde from phospholipid hydroperoxide by hemoproteins. Free Radic. Biol. Med. 36: 1025–1033, 2004.
• 14) E. K. Long, M. J. Picklo Sr. Trans-4-hydroxy-2-nonenal, a product of n-3 fatty acid peroxidation: Make some room HNE. Free Radic. Biol. Med. 49: 1–8, 2010.
• 15) L. J. Marnett. Lipid peroxidation—DNA damage by malondialdehyde. Mutat. Res. 424: 83–95, 1999.
• 16) S. H. Lee, I. A. Blair. Characterization of 4-oxo-2-nonenal as a novel product of lipid peroxidation. Chem. Res. Toxicol. 13: 698–702, 2000.
• 17) S. Kakuta, Y. Bando, S. Nishiumi, M. Yoshida, E. Fukusaki, T. Bamba. Metabolic profiling of oxidized lipid-derived volatiles in blood by gas chromatography/mass spectrometry with in-tube extraction. Mass Spectrom. 2: A0018, 2013.
• 18) S. Nishiumi, T. Kobayashi, A. Ikeda, T. Yoshie, M. Kibi, Y. Izumi, T. Okuno, N. Hayashi, S. Kawano, T. Takenawa, T. Azuma, M. Yoshida. A novel serum metabolomics-based diagnostic approach for colorectal cancer. PLoS ONE 7: e40459, 2012.
• 19) S. Hori, S. Nishiumi, K. Kobayashi, M. Shinohara, Y. Hatakeyama, Y. Kotani, N. Hatano, Y. Maniwa, W. Nishio, T. Bamba, E. Fukusaki, T. Azuma, T. Takenawa, Y. Nishimura, M. Yoshida. A metabolomics approach to lung cancer. Lung Cancer 74: 284–292, 2011.
• 20) S. Srivastava, A. Chandra, A. Bhatnagar, S. K. Srivastava, N. H. Ansari. Lipid peroxidation product, 4-hydroxynonenal and its conjugate with GSH are excellent substrates of bovine lens aldose reductase. Biochem. Biophys. Res. Commun. 217: 741–746, 1995.
• 21) S. Pandey, S. K. Sricastaca, K. V. Ramana. A potential therapeutic role for aldose reductase inhibitors in the treatment of endotoxin-related inflammatory diseases. Expert Opin. Investig. Drugs 21: 329–339, 2012.
• 22) G. Cohen, Y. Riahi, V. Sunda, S. Deplaso, C. Chatgilialoglu, C. Ferreri, N. Kaiser, S. Sasson. Signaling properties of 4-hydroxyalkenals formed by lipid peroxidation in diabetes. Free Radic. Biol. Med. 65: 978–987, 2013.
• 23) I. Dalle-Donne, R. Rossi, R. Colombo, D. Giustarini, A. Milzani. Biomarker of oxidative damage in human disease. Clin. Chem. 52: 601–623, 2006.
• 24) C. L. Arthur, J. Pawliszyn. Solid phase microextraction with thermal desorption using fused silica optical fibers. Anal. Chem. 62: 2145–2148, 1990.
• 25) C. Deng, N. Li, X. Zhang. Development of headspace solid-phase microextraction with on-fiber dericatization for determination of hexanal and heptanal in human blood. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 813: 47–52, 2004.
• 26) T. Hyotylainen. Critical evaluation of sample pretreatment techniques. Anal. Bioanal. Chem. 394: 743–758, 2009.
• 27) R. Maleki, K. Farhadi, A. A. Matin. Analysis of ethanol and methanol in human body fluids by headspace solid phase microextraction coupled with capillary gas chromatography. Anal. Sci. 22: 1253–1255, 2006.
• 28) T. Kumazawa, H. Seno, X. P. Lee, A. Ishii, O. Suzuki, K. Sato. Detection of ethanol in human body fluids by headspace solid-phase micro extraction (SPME)/capillary gas chromatography. Chromatographia 43: 393–397, 1996.
• 29) S. Risticevic, J. Pawliszyn. Solid-phase microextraction in targeted and nontargeted analysis: Displacement and desorption effects. Anal. Chem. 85: 8987–8995, 2013.
• 30) A. Zlatkis, R. S. Brazell, C. F. Poole. The role of organic volatile profiles in clinical diagnosis. Clin. Chem. 27: 789–797, 1981.
• 31) D. Zuba, A. Parczewski, M. Reichenbacher. Optimization of solid-phase microextraction conditions for gas chromatographic determination of ethanol and other volatile compounds in blood. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 773: 75–82, 2002.
• 32) A. Kroupa, J. Dewulf, H. V. Langenhove, I. Viden. Breakthrough characteristics of volatile organic compounds in the −10 to +170°C temperature range on Tenax TA determined by microtrap technology. J. Chromatogr. A 1038: 215–223, 2004.
• 33) E. Niki, Y. Yamamoto, M. Takahashi, K. Yamamoto, Y. Yamamoto, E. Komuro, M. Miki, H. Yasuda, M. Mino. Free radical-mediated damage of blood and its inhibition by antioxidants. J. Nutr. Sci. Vitaminol. 34: 507–512, 1988.
• 34) Y. Fujieda, A. Manno, Y. Hayashi, N. Phodes, L. Guo, M. Arita, T. Bamba, E. Fukusaki. Inflammation and resolution are associated with upregulation of fatty acid β-oxidation in zymosan-induced peritonitis. PLoS ONE 8: e66270, 2013.
• 35) K. Ventura, M. Dostal, J. Churacek. Retention characteristics of some volatile compounds on Tenax GR. J. Chromatogr. A 642: 379–382, 1993.
• 36) F. Pelusio, T. Nilsson, L. Montanarella, R. Tilio, B. Larsen, S. Facchetti, J. Ø. Madsen. Headspace solid-phase microextraction analysis of volatile organic sulfer compounds in black and white truffle aroma. J. Agric. Food Chem. 43: 2138–2143, 1995.
• 37) M. Tankiewicz, C. Morrison, M. Biziuk. Application and optimization of headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography-flame-ionization detector (GC-FID) to determine products of the petroleum industry in aqueous samples. Microchem. J. 108: 117–123, 2013.
• 38) X. L. Cao, C. N. Hewitt. Evaluation of Tenax-GR adsorbent for the passive sampling of volatile organic compounds at low concentrations. Atmos. Environ. 27: 1865–1872, 1993.
• 39) J. Mocak, A. M. Bond, S. Mitchell, G. Scollary. A statistical overview of standard (IUPAC and ACS) and new procedures for determining the limits of detection and quantification: Application to voltammetric and stripping techniques. Pure Appl. Chem. 69: 297–328, 1997.
• 40) I. Rasanen, J. Viinamaki, E. Yuori, I. Ojanpera. Headspace in-tube extraction gas chromatography-mass spectrometry for the analysis of hydroxylic methyl-derivatized and volatile organic compounds in blood and urine. J. Anal. Toxicol. 34: 113–121, 2010.
• 41) S. de Koning, H. G. Janssen, U. A. Th. Brinkman. Modern methods of sample preparation for GC analysis. Chromatographia 69(S1): S33–S78, 2009.
• 42) A. K. Gottzein, F. Musshoff, B. Madea. Qualitative screening for volatile organic compounds in human blood using solid-phase microextraction and gas chromatography-mass spectrometry. J. Mass Spectrom. 45: 391–397, 2010.
• 43) R. Guadagni, N. Miraglia, A. Simonelli, A. Silvestre, M. Lamberti, D. Feola, A. Acampora, N. Sannolo. Solid-phase microextraction-gas chromatography-mass spectrometry method validation for the determination of endogenous substances: Urinary hexanal and heptanal as lung tumor biomarkers. Anal. Chim. Acta 701: 29–36, 2011.
• 44) R. Xue, L. Dong, S. Zhang, C. Deng, T. Liu, J. Wang, X. Shen. Investigation of volatile biomarkers in liver cancer blood using solid-phase microextraction and gas chromatography/mass spectrometry. Rapid Commun. Mass Spectrom. 22: 1181–1186, 2008.
• 45) M. Hakim, Y. Y. Broza, O. Barash, N. Peled, M. Phillips, A. Amann, H. Haick. Volatile organic compounds of lung cancer and possible biochemical pathways. Chem. Rev. 112: 5949–5966, 2012.