Article pp.193-206 du Vol.25 n°3 (2005)

(1)

ARTICLE ORIGINAL ORIGINAL PAPER

Evaluation of the solid phase microextraction (SPME) technique for the analysis

of human breath during eating

E. Pionnier

¹

, E. Sémon, C. Chabanet and C. Salles*

RÉSUMÉ

Évaluation de la technique de microextraction sur phase solide (SPME) pour l’analyse de l’air humain exhalé pendant la consommation d’aliments.

Une méthode de couplage microextraction sur phase solide/chromatographie en phase gazeuse/spectrométrie de masse (SPME-GC-MS) a été utilisée comme alternative à l’analyse par API-MS des effluves nasales pour suivre la libération de composés d’arômes présents dans l’air expiré pendant la dégustation d’un fromage modèle aro- matisé avec du diacétyle, de l’heptan-2-one, de l’hexanoate d’éthyle, de l’heptan-2-ol, de l’acide propanoïque et de l’acide butanoïque. Cinq types de fibres possédant un greffage différent ont été évalués pour leur sélectivité, sensibilité, stabilité et les phéno- mènes de compétition. Parmi ceux-ci, la fibre greffée avec un polymère de type polydi- méthylsiloxane (PDMS) était la plus appropriée. Une méthode basée sur un temps d’échantillonnage court (8 s) a montré son efficacité en suivant la cinétique de libéra- tion de ces composés pendant 3 minutes. Il a été demandé à huit juges de manger 5 g d’un fromage modèle sans contrainte particulière et des prélèvements d’effluves nasales ont été effectués pendant cette période. L’utilisation de la SPME-GC-MS a été vali- dée par comparaison avec la spectrométrie de masse à ionisation à pression atmosphérique (API-MS) car des corrélations entre les deux méthodologies ont été observées pour l’heptan-2-one et l’hexanoate d’éthyle. Des différences entre les juges ont pu être observées dans les cinétiques de libération des composés d’arômes, mais pour un juge donné, la même allure de profil de libération a été observée quel que soit le composé d’arôme étudié. Les relations linéaires entre la concentration de composés volatils dans le fromage modèle et les surfaces des pics respectifs obtenus par GC ont été étudiées. Cependant, un manque de sensibilité du détecteur de l’appareil GC-MS a été constaté pour de très faibles concentrations. Aucune compétition entre les composés d’arôme pour la fibre PDMS n’a été observée.

Mots clés

SPME, PDMS, flaveur, effluves nasales, API-MS, fromage modèle.

1. Institut National de Recherche Agronomique – Unité Mixte ENESAD-INRA de Recherche Flaveur – Vision – Comportement du consommateur – 17, rue Sully – BP 86510 – 21065 Dijon cedex – France.

* Correspondence : Tél. : 33 380 69 30 79, Fax : 33 380 69 32 27, E-mail : salles@dijon.inra.fr

(2)

SUMMARY

As an alternative to nosespace analysis using API-MS, a solid phase micro-extraction/

gas chromatography/mass spectrometry (SPME-GC-MS) method has been used in order to follow the release of aroma compounds present in exhaled air during eating a model cheese flavoured with diacetyl, heptan-2-one, ethyl hexanoate, heptan-2-ol, propanoic acid and butyric acid. Among five different fibre coatings evaluated for selectivity, sensitivity, stability, competition phenomena, polydimethylsiloxane coating (PDMS) was the most appropriate. A method based on short-time sampling (8 s) has been shown to be efficient in following the kinetics of release of these aromas over 3 min. Eight panellists were asked to eat 5 g of a model cheese in their own way and nosespace experiments were carried out during that period. The use of SPME-GC-MS has been validated in comparison with atmospheric pressure ionisation-mass spectrometry (API-MS) as correlations between the two methodologies were observed for heptan-2-one and ethyl hexanoate. Differences between panellists could be observed in the kinetics of aroma release but for a given panellist, the same pattern of release was observed whatever the aroma compound studied. The linear relationships between the concentration of volatile compounds in model cheese and the respective GC peak areas were studied. However, a lack of sensitivity of the GC-MS detector was observed at very low concentrations. No competition was observed between the aroma compounds for the PDMS fibre.

Keywords

SPME, PDMS, flavour, nosespace, API-MS, model cheese.

1 – INTRODUCTION

The release of aroma compounds from food during eating and the subsequent perception of the aroma compounds by the olfactory system are the factors that determine their final aroma quality (INGHAM et al., 1995a). For these reasons, the temporal aspect of flavour release and transport of flavours in food have been extensively studied during last recent years (TAYLOR et al., 2000).

The procedure used by LINFORTH et al. (1994) in order to measure volatile compounds present in the expired air during eating tomato slices consisted in trapping volumes of nosespace on Tenax^® traps followed by GC analysis. Expired air was globally collected for all the eating period. Consequently, nosespace samples represented only an averaged profile of aroma released during the chewing period. The same method was used to study the volatile changes in time-course nosespace profiles during the eating of strawberries (INGHAM et al., 1995b). It was necessary to collect breath samples from five panellists in order to compensate for the lack of sensitivity of the methodology and to reduce variations in results. Thus, they obtained the averaged time-course profile of the five panellists and could not observe inter-individual differences. Some studies in which the methods employed to sample breath volumes required complex devices such as bags or Tenax^® traps reported particular problems such as contamination (GROTE and PAWLISZYN, 1997). The sample preparation step for purge-and-trap analyses is frequently the most time-consuming and labor intensive step. It is also the primary point of analyte loss from the matrix (SONG et al., 1997). Therefore, liquid nitrogen is necessary to the purge and trap technique and water present in human breath can saturate the traps (GROTE and PAWLISZYN, 1997) and consequently be responsible for artefacts.

More recent technological developments permit monitoring of aroma compounds expired from the nose breath-by-breath during eating using Atmospheric Pressure Ionisa- tion-Mass Spectrometry (API-MS), allowing for successful real-time measurements of flavour release in vivo ( INGHAM et al., 1995c; TAYLOR et al., 2000). However, with API-MS, it could be difficult to distinguish between two compounds with the same molecular mass if

(3)

single stage mass spectrometry is used. Moreover this technology is rather expensive and many laboratories cannot afford such equipment.

Solid phase microextraction (SPME) (A^RTHUR et al., 1992) usually employed to determine volatile compounds in food (KATAOKA et al., 2000), has been also used for global breath volatile components studies (GROTE and PAWLISZYN, 1997). This technique, never used to follow time course flavour release in human breath, could be an alternative method to carry out nosespace experiments and to analyse all the aroma compounds of interest.

The aim of this study was to evaluate a SPME technique for sampling the aroma compounds present in expired air during the chewing of a flavoured model cheese and to study the time-course profiles of volatile compounds over a 3-min experimental chewing period.

2 – MATERIAL AND METHODS

2.1 Model cheese

A flavoured model cheese was made as described by PIONNIER et al. (2004a). As aroma compounds it contained propanoic acid, butyric acid, heptan-2-one, heptan-2-ol, ethyl hexanoate and diacetyl (Sigma Aldrich, Saint Quentin Fallavier, France). The propor- tions of volatile compounds were optimised to be consistent with literature data (MOLI- MARD and SPINNLER, 1996; ENGEL et al., 2001) and to give to the model cheese a good acceptability. The chromatographic purity of the six aroma compounds was in all cases greater than 95%.

2.2 SPME methodology for headspace measurement

2.2.1 Choice of the fibre coating, reproducibility in production of different fibres, storage of fibres after sorption

Various fibres were used throughout this study, all of them from Supelco (Bellefonte, PA): a 100 µm PDMS (polydimethylsiloxane), a 65 µm PDMS/DVB (divinylbenzene), a 75 µm CAR/PDMS (carboxen), a 65 µm CW/PDMS (carbowax) and an 85 µm PA (polyacr- ylate). They were conditioned as recommended by the manufacturer.

To compare the extraction efficiency of the five different fibres, 5 g of model cheese were placed in a 40 mL closed vial and allowed to equilibrate at 25°C for 1 hour. Each SPME fibre was inserted into the headspace through a Mininert^® valve (Supelco) and the extraction lasted for 5 s.

To compare the reproducibility in the production of 10 PDMS fibres (called L, O, B, P, I, Q, R, K, T and U) used during our experiments, and to study the loss of aromas during fibre storage after the sorption, 10 mL of a solution containing the six aromas studied (A) were placed in a 40 mL vial and allowed to equilibrate at 25°C for 1 hour. Solution A contained aroma compounds (in water) at levels that were in the linear response range of the fibres investigated: diacetyl (1.45 ppm), heptan-2-one (2.6 ppb), ethyl hexanoate (0.56 ppb), heptan-2-ol (5.1 ppb), propanoic acid (6.43 ppm), butyric acid (2.19 ppm). The aroma compounds were then extracted over 30 s. That headspace period provided approximately the same peak areas as those observed during nosespace experiments.

To study the loss of aroma compounds during the fibre storage, the fibres were analysed at successive periods. All these experiments were conducted in duplicate.

(4)

2.2.2 Competition for absorption on the fibre

Three 100 µm PDMS fibres (B, Q, U) were used to test the possibility of competition between aroma compounds for the fibre. Competition study focused on three aroma compounds out of the six. Heptan-2-one (2.1 ppb), ethyl hexanoate (2.3 ppb) and heptan- 2-ol (7.5 ppb) were each tested both separately and mixed with the five other aroma compounds (diacetyl (12.6 ppb), propanoic acid (1186 ppb) and butyric acid (501 ppb)). These solutions were prepared so that a sorption of 30 s in the headspace of these solutions gave approximately the same peak area as a sorption of 8 s in the nosespace of the panellists during chewing. Ten mL of solution containing either one or six aromas were placed in a 40 mL vial and allowed to equilibrate at 25°C for 30 min. The aroma compounds were then extracted from the headspace above the solutions over 30 s. They were present in these solutions at levels in the linear response range of the fibres investigated. The chromatographic peak areas obtained after analysis by GC-MS were compared. These experiments were conducted in triplicate.

2.2.3 Determination of linear response range

The linear response range of each fibre used for nosespace experiments was deter- mined on the concentration range observed during nosespace experiments by diluting aroma solutions containing the six aroma compounds: diacetyl (0.54 - 0.0013 ppm), heptan- 2-one (14.6 - 0.036 ppb), ethyl hexanoate (0.44-0.001 ppb), heptan-2-ol (22.6-0.03 ppb), propanoic acid (83.7-0.21 ppm), butyric acid (5.4-0.013 ppm). The SPME headspace analyses of these solutions were carried out in duplicate as previously described.

2.3 GC-MS analyses

After sampling, the fibre was placed into the injection port of a GC for 3 min at 250°C to desorb the volatile compounds into a GC-MS. After desorption, each fibre was then placed for 10 min in another injection port at 250°C in order to eliminate any compound from the fibre. A GC apparatus 6890 (Agilent Technologies, Palo Alto, CA) equipped with a splitless/split injector and coupled with a mass selective detector 5973 (Agilent Technolo- gies) was used. The column was a DBWAX (30 m; 0.25 mm i.d., 0.25 µm film thickness;

J&W Scientific, Agilent Technologies), with helium as carrier gas (35 cm/s). The oven temperature was initially raised from 50°C to 140°C at 6°C/min then raised from 140°C to 220°C at 15°C/min and finally kept at 220°C for 5 min. The mass spectrometer was used in selected ion monitoring (SIM) mode: m/z 43 and 86 for diacetyl; m/z 43, 58 and 114 for heptan-2-one; m/z 88, 99 and 101 for ethyl hexanoate; m/z 45, 55 and 83 for heptan-2-ol;

m/z 45, 73 and 74 for propanoic acid; m/z 60 and 73 for butyric acid.

2.4 SPME and API-MS methodology for nosespace sampling

A SPME method, carried out simultaneously with nosespace API-MS, was developed to follow, in a discontinuous way, the changes in the concentration of aroma compounds released in expired air during the eating of model cheese. A glass Y-junction was set up between the entry of the API-MS capillary, the entry for the SPME fibre and the nose of the subject. It was as short as possible and the fibre was introduced in this device as close as possible to the panellist’s nostril in order to limit losses due to aroma condensation phenomenon on the surfaces. API-MS allowed for continuity in the detection, during eating, of acetone, heptan-2-one and ethyl hexanoate.

2.4.1 SPME

The release of diacetyl, heptan-2-one, ethyl hexanoate, heptan-2-ol, propanoic acid and butyric acid was measured in a discontinuous way by inserting a SPME fibre at different moments (table 1) of the mastication in the Y-junction and sampling expired air for 8 s. Ten fibres were needed to perform these experiments and all along the experiments,

(5)

each fibre was attributed to the same period of each mastication. After nosespace sampling, the first fibre was immediately desorbed and the aroma compounds analysed as described above. The nine other fibres were stored just after the extraction phase until GC-MS analysis in a glass tube hermetically sealed, at room temperature, to avoid losses of aroma compounds due to exchanges between the fibre and the laboratory air.

Table 1

Nosespace sorption time, chewing period and PDMS fibres used for each sampling time.

2.4.2 API-MS

Nosespace experiments were performed using API-MS with gaseous sample introduction. Acetone, heptan-2-one and ethyl hexanoate release measurements were carried out using an Esquire-LC mass spectrometer (Bruker Daltonique, Wissembourg, France) fitted with a new probe design to allow a gaseous sampling introduction (GINIES et al., 2001). Air was sampled at a flow rate of 55 mL/min through deactivated stainless steel tubing (i.d. = 0.53 mm) (Silcosteel^®, Restek, Evry, France) heated to 150°C. The nitrogen auxiliary gas flow was 9 L/min, corresponding to a pressure of 10 psi. This inert heated transfer line directly from the ambient air to the API source prevented vapour condensation and minimised unwanted chemical reactions (DUNPHY et al., 2000). Analyses were carried out in positive mode.

2.4.3 Nosespace sampling

In order to respect principles of hygiene and minimum invasion, small disposable plastic tubes were inserted in one nostril so that assessors could breathe, eat and drink normally as mentioned by TAYLOR et al. (2000). The regularity of their respiratory rhythm was charted out by the follow up of acetone which is present permanently in human breath as it is produced from glucose metabolism.

Four sessions of 30 min over four consecutive days for each panellist were conducted to obtain three replicates of SPME profiles and eight replicates for API-MS. Two successive mastications in the same session were necessary to obtain one replicate with the SPME method because there was an overlapping of sorption times (table 1). At each session, the panellists ate 2 samples of 5 g of model cheese (placed for one hour at 20°C).

They were asked to eat in their own way with usual swallowing, mouth closed, and to breathe into the plastic tube. The total sampling time was 3 min. They cleansed their pal- ate with some bread, apple and water. They began the second mastication several minutes later, after verification with API-MS of the absence of heptan-2-one and ethyl hexanoate in their expired air.

2.5 Data analysis

API-MS data were smoothed after selecting all the peaks corresponding to exhalation events. To summarize the information contained in the curves, some parameters were extracted and studied by analysis of variance. These parameters included Cmax (greatest

Time (s) 0 5 10 20 30 40 60 90 120 180

Sorption time (s) blank 1-9 6-14 16-24 26-34 36-44 56-64 86-94 116-124 176-184

Chewing period^a 1 2 1 2 1 2 2 2 2 2

PDMS coating^b L O B P I Q R K T U

a Two successive mastications in the same session were necessary to obtain one replicate of aroma release profile with the SPME method because of the overlapping of sorption times.

b 10 PDMS fibres from the same batch.

(6)

chromatographic peak area for SPME and highest intensity for API-MS) corresponding to the greatest amount of aroma in nosespace; Tmax, the time when Cmax is reached; slope, the initial slope of the curve measured between 0 and 10 sec and finally, AUC, the Area Under the Curve corresponding to the total aroma release over a 3-min period. All the one- way analyses of variance, mean comparisons (using a Newman-Keuls test at 5%), student t- tests and correlations were performed with Statbox software version 3.0 (Grimmer Logiciel, Paris, France) or SAS software version 8.01 (SAS Institute Inc., Cary, NC).

3 – RESULTS AND DISCUSSION

Prior to nosespace analysis, optimisation of the SPME method was conducted.

Except for sorption time, optimisation was carried out with headspace analyses rather than nosespace analyses in order to reduce the variability at maximum and to simplify the procedure. However, those headspace analyses were performed with concentrations leading to peak areas of the same magnitude order as nosespace analyses.

3.1 Choice of the SPME fibre

The fibre coating is a factor which greatly influences the both qualitatively and quanti- tatively content of volatiles extracted by the fibre. The selection of the most appropriate fibre between PDMS, PDMS/DVB, CW/PDMS, CAR/PDMS and PA was undertaken by taking into account constraints due to nosespace sampling.

Despite its high affinity for carboxylic acids (ROBERTS et al., 2000), the CW-PDMS coating was discarded because of its inefficiency towards neutral compounds. Results obtained with the four other fibres are presented in figure 1. Though some differences were observed, the extraction capacities of PDMS, CAR-PDMS and PDMS-DVB seemed acceptable for nose-space sampling applications. As PDMS was highly stable (ROBERTS et al., 2000) and showed an excellent repeatability in sampling gas mixtures (MARTOS and PAWLISZYN, 1997), this fibre was selected for our study.

0 1 000 000 2 000 000 3 000 000

Chromatographic peak area

diacetyl*10 heptan-2-one ethyl

hexanoate heptan-2-ol propanoic

acid butyric acid

PDMS PDMS/DVB CAR/PDMS PA b a ab c

ab ab b a

c a b c

b a ab b a a b b

NS

Figure 1

Comparison of extraction efficiencies of the PDMS, PDMS/DVB, CAR/PDMS, and PA using a model cheese containing diacetyl, heptan-2-one, ethyl hexanoate,

heptan-2-ol, propanoic acid and butyric acid. Means with the same letter (a-c) are not significantly different at the level of 5% according to Newman-Keuls tests (NS: no significant difference). Standard deviation is drawn at the top of each bar.

Each measurement was done in duplicate.

(7)

Variability displayed in figure 1 is greater than in other headspace studies conducted on the same concentration range of aroma compounds. This is mainly due to the brief extraction time. Indeed, during these headspace experiments, we chose a brief extraction period in order to obtain chromatographic peak areas of the same magnitude range as those obtained during nosespace experiments. However, with a 5 s sorption time, an accurate measurement of sorption time is difficult to obtain and we evaluated a time error of more or less 1 s could involve a difference of 20% on the quantity extracted by the fibre.

3.2 Optimisation of SPME conditions and methodology

3.2.1 Sorption time

In the literature, most of the analyses using SPME involved considerable sorption time (from several minutes to one hour) (JIA et al., 1998; MARSILI, 1999; ROCHA et al., 2001) in order to reach equilibrium conditions between the matrix, the headspace and the poly- meric fibre coating phases allowing the best reproducibility (SOSTARIC et al., 2000).

Cheese aroma compounds present in the expired air were sampled at different moments of the mastication by SPME fibres. It was important to minimise absorption duration in order to ensure the maximum accuracy in measurement time. In this perspec- tive, nosespace experiments carried out by sorbing the expired air from one subject showed that a sorption time of 8 sec was sufficient to ensure enough aroma compounds on the fibre and to obtain the maximum accuracy in the measurement time, leading to an acceptable reproducibility.

3.2.2 Performance comparison between PDMS fibres

For each time-course profile of aroma compound present in expired air, one fibre was needed to extract aliquots at each of the ten sampling times during mastication. In order to avoid variability in comparative analyses, as observed by FABRE et al. (2002) using different fibres with the same coating, probably due to little differences in fibre length (GROTE and PAWLISZYN, 1997; MARTOS and PAWLISZYN, 1997), the 10 PDMS fibres used in that study from the same batch have been tested in duplicate with solution A. A one-way anova performed on log-transformed data for each compound and a mean comparison test showed significant differences among the 10 fibres only for propanoic acid extraction. So, a correction factor was applied for this compound.

3.2.3 Storage period of fibres after sampling

Since 35 min was needed to analyse by GC-MS the aroma compounds extracted by each fibre, the tenth fibre was analysed 5.15 h after the first one. No significant difference for any aroma compound was observed except only for ethyl hexanoate for 3 h of storage, probably due to an experimental error (results of the one-way anova: diacetyl F=1.12, p=0.44, heptan-2-one F=0.94, p=0.51, ethyl hexanoate F=5.05, p=0.04, heptan- 2-ol F=2.95, p=0.11, propanoic acid F=1.21, p=0.4, butyric acid F=1.79, p=0.24). These results were in accordance with those obtained by GROTE and PAWLISZYN (1997) for the CAR-DVB and PDMS-DVB fibres studying possible losses in acetone, ethanol and iso- prene.

3.3 Linear response range and competition study

3.3.1 Linear range

A determination coefficient less than 0.8 suggest an eventual non linearity of the response of the fibre. Then, the linearity was checked by the examination of the data. The

(8)

use of the fibres B, L, O, P, I, Q, R, and U demonstrated linear relationships between the concentrations of the 6 volatile compounds in expired air and the respective peak areas obtained. Only, three fibres provided non linear responses for several aroma compounds, mainly butyric acid and diacetyl. Moreover, the sensitivity of the mass spectrometer did not allow the correct detection of some weak peak areas, mainly corresponding to diacetyl, propanoic acid and butyric acid.

3.3.2 Competition

Competition phenomena on the fibre coating have been reported (ROBERTS et al., 2000; PINHO et al., 2002). Competition phenomena occur once the concentration exceeds the upper limit of the linear range (ROBERTS et al., 2000). Short sampling times can be used to reduce the possibility for fibre overloading and avoid the resulting distortions, especially when compounds of both low and high affinity for the fibre are analysed.

Although the probability to observe a saturation of the fibre was very low in this study due to very short sampling times (8 sec), we have verified that such a competition phenomenon did not occur. Due to the potential problem of linearity and possible chromatographic resolution difficulties, the competition phenomena were not studied with diacetyl, propanoic acid and butyric acid. The concentrations tested were in the linear range of the fibres investigated and were chosen to obtain approximately the same peak areas during these headspace experiments as those obtained with these three fibres during nosespace experiments. The results are shown in figure 2. No significant difference (Student’s t-test) was observed in peak areas between each aroma compound alone or mixed with the other 5 compounds indicating that no competition for the PDMS fibres among these aroma compounds occurred.

Our results agree with KOSIEL et al. (2000) who argued that the PDMS coating was not affected by competition between the absorbed compounds due to its mechanism of sorption. Indeed, the PDMS coating is a liquid phase and sorption of analytes is performed via absorption contrary to the other coatings for which sorption is performed by adsorption. This non polar coating is often tested as it has been used successfully for the analysis of both polar and non polar volatile components (STEFFEN and PAWLISZYN, 1996;

SOSTARIC et al., 2000).

3.4 Validation

API-MS or Proton Transfer Reaction Mass-Spectrometry (PTR-MS) are currently the most widely used methods for in-nose measurement of flavour release (GRAB and GFEL- LER, 2000; HARVEY et al., 2000; YERETZIAN et al., 2000). In order to validate the SPME methodology, the results obtained by SPME-GC-MS were compared to those obtained by API-MS using the parameters extracted from the curves. The comparisons were made with heptan-2-one and ethyl hexanoate which gave the best results in breath analysis by both SPME-GC-MS and API-MS. We observed positive and significant correlations between the two methodologies considering all the parameters investigated (table 2). The use of SPME method giving the same patterns of release for heptan-2-one and ethyl hexanoate as API-MS for all the subjects was thus validated in our experimental conditions.

(9)

0 (a)

(b)

(c) 4 000 8 000 12 000 16 000 20 000

heptan-2-one ethyl hexanoate heptan-2-ol

chromatographic peak area

1 compound mix of 6 compounds

0 4 000 8 000 12 000 16 000 20 000

Figure 2

Comparison of the chromatographic peak areas of heptan-2-one, ethyl hexanoate and heptan-2-ol when these aromas are alone or mixed

with the five other aroma compounds concerning fibre B (a), fibre Q (b) and fibre U (c). Standard deviation is drawn at the top of each bar.

Each measurement was done in triplicate.

(10)

Table 2

Correlation coefficients between SPME and API-MS tested

on the extracted parameters Tmax, Cmax, AUC and Slope concerning heptan-2-one and heptan-2-ol (24 observations).

3.5 Release of aroma compounds during mastication

Flavour release is subject to constant changes in quantity and quality during eating.

Figure 3 illustrates the results of release profiles obtained for heptan-2-ol and heptan-2- one for three subjects. The patterns of release differed according to the subject in terms of Cmax, Tmax, area under the curve of flavour release (AUC) and slope. Subject 1 showed the significantly greatest release for heptan-2-ol in terms of Cmax and AUC. Cmax was obtained at 60 s. In contrast with subject 1, the release profile obtained for subject 6 was very weak and Cmax seemed to be obtained at 30 s. For subject 8, heptan-2-ol seemed to be released very progressively during the 3-min release monitoring as Cmax was observed at 90 sec. The same observations can be done from heptan-2-one flavour release. The physico-chemical properties of these two aroma compounds may not be sufficiently different to observe differences in flavour release. The reproducibility observed through standard deviations seemed to be quite good for such in vivo experiments. These SPME profiles were similar to those previously observed by API-MS (HARVEY et al., 2000).

Moreover, even if each subject gave a different release profile for each aroma compound as shown in figure 3, we observed that for a given panelist, the shape of the curve was the same whatever the compound considered as observed from correlations between parameters (table 3). For instance, if a subject had the highest concentration of heptan-2-one in the exhaled air, he had also the highest concentration of heptan-2-ol and ethyl hexanoate as these three correlations are positive and significant. The same obser- vation was made by DELAHUNTY and GUILFOYLE (2000) using the Aroma Stimulus Index (ASI). Figure 3 showed also that the temporal aroma release curves slowly decreased after Cmax to tentatively reach the original value whereas the totality of the bolus was totally swallowed. That might be explained by possible retention and release phenomenon of aroma compounds by the oral cavity, as suggested by BUETTNER (2004) for several volatile compounds and could be responsible for persistence in flavour perception.

Correlation studies between flavour release parameters, temporal perception and oral parameters were already reported (PIONNIER et al., 2004a, b).

The preliminary experiments carried out on one subject from the laboratory showed that a sorption time of 8 sec was sufficient to extract enough volatile compounds present in the expired air and thus to follow the release of the six aromas during the chewing of the model cheese. However, the experiments with the panel of 8 subjects brought to the fore a potential lack of sensitivity on the part of the mass spectrometer for a few subjects.

Indeed, for some subjects, several very weak peak areas were observed for diacetyl, propanoic acid and butyric acid, and integration precision was very poor.

SPME API-MS

Heptan-2-one Ethyl hexanoate

Tmax Cmax AUC Slope Tmax Cmax AUC Slope

Tmax 0.43* 0.43*

Cmax 0.55** 0.68***

AUC 0.47* 0.63***

Slope 0.51** 0.72***

* p < 0.05, **p < 0.01, ***p < 0.001.

Tmax: time to reach the maximum intensity.

Cmax: greatest chromatographic peak area for SPME and highest intensity for API-MS.

AUC: area under the curve.

Slope: initial gradient of the curve measured between 0 and 10 s.

(11)

Table 3

Correlation coefficients between the release parameters of heptan-2-ol (Hol), heptan-2-one (Hon), ethyl hexanoate (EH).

Tmax Cmax AUC Slope

Hol Hone EH Hol Hone EH Hol Hone EH Hol Hone EH Hol

Hone 0.48** 0.87*** 0.80*** 0.32^t

EH 0.55** NS 0.85*** 0.91*** 0.78*** 0.78*** 0.6** 0.88***

tp < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001, NS stands for p > 0.1.

Tmax: time to reach the maximum concentration.

Cmax: maximum concentration.

AUC: area under the curve.

Slope: initial gradient of the curve measured between 0 and 10 s.

Heptan-2-ol

0 2 000 4 000 6 000 8 000 10 000

0 20 40 60 80 100 120 140 160 180 200

time (s)

peak area

Subject 1 Subject 6 Subject 8

Heptan-2-one

0 10 000 20 000 30 000 40 000

0 20 40 60 80 100 120 140 160 180 200

time (s)

peak area

Subject 1 Subject 6 Subject 8

Figure 3

Volatile release patterns of heptan-2-ol and heptan-2-one during the chewing of a model cheese for subjects 1, 6 and 8. Standard deviation is represented

for each point measurement. Each measurement was done in triplicate.

(12)

The quantity of aromas extracted by a SPME fibre could be influenced by the humidity rate in the headspace. MARTOS and PAWLISZYN (1997) have investigated this phenomenon and indicated that high humidity (greater than 90% of relative humidity) could interfere in the analyte mass uptake due to adsorption of water by the fibre, thus possibly changing the fibre’s characteristics. In our study, all the samples were taken in a high humidity atmosphere as all the aroma compounds extracted came directly from the nose. ZEHENTBAUER et al. (2000) have measured the relative humidity in the nose of three subjects and found it to be approximately 85%. Thus, if that parameter has caused a dis- tortion in the quantity of aromas extracted by the fibre, that bias would have the same impact for all the subjects, as for all subjects, the relative humidity in the expired air is approximately the same.

4 – CONCLUSION

The SPME-GC-MS methodology developed in this study has been shown to allow, with very short sampling times (8 sec), the measurement of aroma compound concentration changes in expired air over a chewing period of three minutes. No competition between compounds was observed. These experiments demonstrated the usefulness of this technique in studying flavour release during eating when no API-MS (or PTR-MS) is available. However this technique needs further developments to improve sensitivity and reduce the time required for the GC-MS analyses. A longer extraction time could be chosen as a good compromise between experimental sensitivity and accuracy (the extraction rate would be increased), the application of fast-GC-MS with the use of chromatographic microcolumns could reduce significantly the time of analysis, the use of the CW-PDMS fibre (which had a very strong affinity for propanoic and butyric acid) simultaneously with the PDMS fibre could heighten sensitivity in the analysis of acids.

5 – AKNOWLEDGMENTS

This research was financed by INRA and the Regional Council of Burgundy.

Moreover, we greatly acknowledge Janick Nicolas for technical support and Mary Bouley for reading this article.

REFERENCES

ARTHUR C.L., POTTER D.W., BUCHHOLZ K.D., MOTLAGH S., PAWLISZYN J., 1992.

Solid phase microextraction for the direct analysis of water: theory and practice. LC- GC Intl., 5, 8-14.

BUETTNER A. 2004. Investigation of potent odorants and afterodor development in two Chardonnay wines using the buccal odor screening system (BOSS). J. Agric.

Food Chem., 52, 2339-2346.

DELAHUNTY C.M., GUILFOYLE B., 2000.

Volatile compounds release during con-

sumption: a proposed aroma stimulus index. In: ROBERTS D.D., TAYLOR A.J.

(eds), Flavor release, 405-412, The Ameri- can Chemical Society, Washington.

DUNPHY P., BOUKOBZA F., CHENGAPPA S., LANOT A., WILKINS J., 2000. Wound responses in plants: an orchestrated flavor symphony. In: ROBERTS D.D., TAY- LOR A.J. (eds), Flavor release, 44-57, The American Chemical Society, Washington.

ENGEL E., SEPTIER C., LECONTE N., SAL- LES C., LE QUERE J.L., 2001, Determina-

(13)

tion of taste-active compounds in a bitter camembert cheese by omission tests. J.

Dairy Res., 68, 675-688.

FABRE M., AUBRY V., GUICHARD E., 2002, Comparison of different methods: Static and dynamic headspace and solid-phase microextraction for the measurement of interactions between milk proteins and flavor compounds with an application to emulsions. J. Agric. Food Chem., 50, 1497-1501.

GINIES C., PIOMBINO P., RODRIGUES V., RENAUD R., LE QUERE J.L., 2001. Analy- sis of aroma compounds by APCI/MS:

Interface for in vivo analysis of human breath volatile content. Poster communi- cation, 18^e Journées Françaises de Spec- trométrie de Masse 2001, September 11- 14, La Rochelle, France.

GRAB W., GFELLER H., 2000. A technology for the measurement of fast dynamic changes of flavour release during eating.

In: ROBERTS D.D., TAYLOR A.J. (eds), Flavor release, 33-43, The American Che- mical Society, Washington.

GROTE C., PAWLISZYN J., 1997. Solid-Phase Microextraction for the Analysis of Human Breath. Anal. Chem., 69, 587-596.

HARVEY B.A., BRAUSS M.S., LINFORTH R.S.T., 2000. Real time flavor release from foods during eating. In: ROBERTS D.D., TAYLOR A.J. (eds), Flavor release, 22-32, The American Chemical Society, Washington.

INGHAM K., LINFORTH R.S.T., TAYLOR A.J., 1995a. The effect on eating on aroma release from mint-flavoured sweets. Fla- vour Fragr. J., 10, 15-24.

INGHAM K.E., LINFORTH R.S.T., TAYLOR A.J., 1995b. The effect of eating on aroma release from strawberries. Food Chem., 54, 283-288.

INGHAM K.E., LINFORTH R.S.T., TAYLOR A.J., 1995c. The effect of eating on the rate of aroma release from mint-flavoured sweets. Food Sci. Technol. – Lebensm.

Wiss. Technol., 28, 105-110.

JIA M.Y., ZHANG Q.H., MIN D.B., 1998. Opti- mization of solid-phase microextraction analysis for headspace flavor compounds of orange juice. J. Agric. Food Chem., 46, 2744-2747.

KATAOKA H., LORD H.L., PAWLISZYN J., 2000. Applications of solid-phase

microextraction in food analysis. J. Chro- matogr. A, 880, 35-62.

KOSIEL J., JIA M., PAWLSZYN J., 2000. Air sampling with porous solid-phase microextraction fibers. Anal. Chem., 72, 5178-5186.

LINFORTH R.S.T., SAVARY I., PATTENDEN B., TAYLOR A.J., 1994. Volatile compounds found in expired air during eating of fresh tomatoes and in the headspace above tomatoes. J. Sci. Food Agric., 65, 241-247.

MARSILI R.T., 1999. SPME-MS-MVA as an electronic nose for the study of off-flavors in milk. J. Agric. Food Chem., 47, 648- 654.

MARTOS P.A., PAWLISZYN J., 1997. Calibra- tion of solid phase microextraction for air analyses based on physical chemical properties of the coating. Anal. Chem., 69, 206-215.

MOLIMARD P., SPINNLER E., 1996. Com- pounds involved in the flavor of surface mold-ripened cheeses: origins and properties. J. Dairy, Sci., 79, 169-184.

PINHO O., FERREIRA I.M.P.L.V.O., FERREIRA M.A., 2002. Solid-phase microextraction in combination with GC/MS for quantifica- tion of the major volatile free fatty acids in ewe cheese. Anal. Chem., 74, 5199-5204.

PIONNIER E., CHABANET C., MIOCHE L., LE QUÉRÉ J.L., SALLES C., 2004a. In-vivo aroma release during eating a model cheese: relations with oral parameters. J.

Agric. Food Chem., 52, 557-564.

PIONNIER E., NICKLAUS S., CHABANET C., MIOCHE L., TAYLOR A.J., LE QUÉRÉ J.L., SALLES C., 2004b. Flavor perception of a model cheese: relationships with oral and physico-chemical parameters. Food Qual. Pref., 15, 843-852.

ROBERTS D.D., POLLIEN P., MILO C., 2000.

Solid-phase microextraction method development for headspace analysis of volatile flavor compounds. J. Agric. Food Chem., 48, 2430-2437.

ROCHA S., RAMALHEIRA V., BARROS A., DELGADILLO I., COIMBRA M.A., 2001.

Headspace solid phase microextraction (SPME) analysis of flavor compounds in wines. Effect of the matrix volatile compo- sition in the relative response factors in a wine model. J. Agric. Food Chem., 49, 5142-5151.

(14)

SONG J., GARDNER B.D., HOLLAND J.F., BEAUDRY R.M., 1997. Rapid analysis of volatile flavor compounds in apple fruit using SPME and GC/time-of-flight mass spectrometry. J. Agric. Food Chem., 45, 1801-1807.

SOSTARIC T., BOYCE M., SPICKETT E., 2000. Analysis of the volatile components in vanilla extracts and flavorings by solid- phase microextraction and gas chromatography. J. Agric. Food Chem., 48, 5802- 5807.

STEFFEN A., PAWLISZYN J., 1996. Analysis of flavor volatiles using headspace solid- phase microextraction. J. Agric. Food Chem., 44, 2187-2193.

YERETZIAN C., JORDAN A., BREVARD H., LINDINGER W., 2000. Flavor release on- line monitoring of coffee roasting by Pro- ton-Transfer-Reaction-Mass-Spectrome- try. In: ROBERTS D.D., TAYLOR A.J.

(eds), Flavor release, 112-123, The Amer- ican Chemical Society, Washington.

TAYLOR A.J., LINFORTH R.S.T., HARVEY B.A., BLAKE B., 2000. Atmospheric pressure chemical ionisation mass spectrometry for in vivo analysis of volatile flavour release. Food Chem., 71, 327-338.

ZEHENTBAUER G., KRICK T., REINECCIUS G.A., 2000. Use of humidified air in optimi- zing APCI-MS response in breath analysis. J. Agric. Food Chem., 48, 5389-5395.