Continuous Real Time Breath Gas Monitoring in the Clinical

Sep 18, 2013 - Beate Brock , Svend Kamysek , Josephine Silz , Phillip Trefz , Jochen K ... R Lindley , CL Paul Thomas , Matthew A Turner , James C Rey...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Continuous Real Time Breath Gas Monitoring in the Clinical Environment by Proton-Transfer-Reaction-Time-of-Flight-Mass Spectrometry Phillip Trefz,† Markus Schmidt,† Peter Oertel,† Juliane Obermeier,† Beate Brock,† Svend Kamysek,† Jürgen Dunkl,‡ Ralf Zimmermann,§ Jochen K. Schubert,† and Wolfram Miekisch*,† †

Department of Anaesthesia and Intensive Care, University Medical Center Rostock, Schillingallee 35, 18057 Rostock, Germany Ionicon Analytik GmbH, Eduard-Bodem-Gasse 3, A-6020 Innsbruck, Austria § Joint Mass Spectrometry Centre, Chair of Analytical Chemistry, University of Rostock, Dr. Lorenz Weg 1, 18059 Rostock, Germany and Joint Mass Spectrometry Centre, Cooperation Group “Comprehensive Molecular Analytics”, Helmholtz Zentrum München, D-85764 Neuherberg, Germany ‡

S Supporting Information *

ABSTRACT: Analysis of volatile organic compounds (VOCs) in breath holds great promise for noninvasive diagnostic applications. However, concentrations of VOCs in breath may change quickly, and actual and previous uptakes of exogenous substances, especially in the clinical environment, represent crucial issues. We therefore adapted proton-transfer-reaction-time-of-flight-mass spectrometry for real time breath analysis in the clinical environment. For reasons of medical safety, a 6 m long heated silcosteel transfer line connected to a sterile mouth piece was used for breath sampling from spontaneously breathing volunteers and mechanically ventilated patients. A time resolution of 200 ms was applied. Breath from mechanically ventilated patients was analyzed immediately after cardiac surgery. Breath from 32 members of staff was analyzed in the post anesthetic care unit (PACU). In parallel, room air was measured continuously over 7 days. Detection limits for breath-resolved real time measurements were in the high pptV/low ppbV range. Assignment of signals to alveolar or inspiratory phases was done automatically by a matlab-based algorithm. Quickly and abruptly occurring changes of patients’ clinical status could be monitored in terms of breath-to-breath variations of VOC (e.g. isoprene) concentrations. In the PACU, room air concentrations mirrored occupancy. Exhaled concentrations of sevoflurane strongly depended on background concentrations in all participants. In combination with an optimized inlet system, the high time and mass resolution of PTR-ToF-MS provides optimal conditions to trace quick changes of breath VOC profiles and to assess effects from the clinical environment.

B

preconcentration prior to analysis and long run times represent important disadvantages of chromatographic techniques. As a consequence, rapid changes of VOC profiles cannot be monitored by means of GC-MS. Direct mass spectrometric or laser spectroscopic methods may solve this problem as they do not require time-consuming sample preparation.22−26 Proton Transfer Reaction-Mass Spectrometry (PTR-MS) has already been applied successfully for online analysis of VOCs in human breath, since it offers fast response times and detection limits down to the pptV range.19,27,28 Physiologically induced changes of breath biomarker profiles can thus be monitored online. These results may aid in the understanding of biochemical pathways and metabolic processes.29−33 However, the poor mass resolution of PTR-Quadrupol-MS (QMS) is a crucial issue, as

reath gas analysis holds great promise as a potential new diagnostic tool. Not only the quest for endogenously generated biomarkers but also the monitoring of exogenous substrates or drugs administered to the patient has attracted attention of researchers and practitioners.1−7 Recent developments in analytical instrumentation provide identification of potential marker substances down to the ultra trace pptV−ppbV level.8−11 Mieth et al. reported on detection of endogenous biomarkers and confounding compounds by means of needle trap microextraction (NTME) in combination with twodimensional gas chromatography-time-of-flight (GCxGC-TOF).12 Potential application of these biomarkers include medical diagnosis,13−15 detection of smoking-related compounds in the breath,16,17 drugs,18,19 and environmental monitoring.20,21 As gas chromatography-mass spectrometry (GC-MS) provides accurate identification and quantification of potential biomarkers, it has become a standard method widely used for screening purposes and biomarker identification. Need for © 2013 American Chemical Society

Received: July 16, 2013 Accepted: September 16, 2013 Published: September 18, 2013 10321

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

and in real time? How do actual and previous exposures affect breath measurements in the clinical environment?

isobaric VOCs with the same nominal mass, such as furan and isoprene, cannot be separated. Untargeted screening can hence be problematic, and a precise identification of substances is difficult if not impossible in many cases. The high mass resolution of time-of-flight (ToF)-mass spectrometers can solve this problem. Herbig et al.33 reported on first breath analyses by means of PTR-ToF-MS. Kohl et al. applied PTR-MS and PTRToF-MS34 to determine breath markers of kidney function; others investigated patients with cystic fibrosis35 and liver cirrhosis.36 In these studies, exhaled air from spontaneously breathing patients or volunteers was analyzed. As buffered end tidal sampling instead of breath-resolved continuous analysis was applied in most cases, the potential of PTR-TOF for continuous breath-resolved analysis was not exploited. As concentrations of volatile substances in the breath may change quickly and abruptly, only breath-resolved real time monitoring will provide comprehensive information. In addition to these issues concerning time resolution, a number of even more general questions inevitably arise in breath research. The effects of actual or past intake of exogenous substances from the individual environment represent crucial issues in this context. Other open questions touch upon the impact of individual (patho)physiological conditions onto concentrations of potential marker substances in breath. First, investigations indicate that pathways for generation of marker substances may be much more complex than assumed and that breath levels of potential marker substances do not only depend on endogenous production but also on distribution into different compartments.29,32 These effects may easily obscure the true biochemical information of exhaled volatile substances and have to be taken into account, if breath markers are meant to move forward to diagnostic application. In the clinical environment, things become even more complicated. Volatile contaminations such as disinfectants, volatile anesthetics, or emissions from plastic materials may affect results from breath analysis. In addition, changes of vital parameters in seriously ill or mechanically ventilated patients are often much more pronounced and rapid than in healthy volunteers. Up to now, most studies in ventilated patients relied on collection of single point breath samples and subsequent analysis.15,37−42 This kind of (punctual) offline analysis bears the risk of missing important changes in breath biomarker concentrations induced through physiological or pathophysiological events. In addition, dilution, contamination of samples, and loss of analytes during sampling procedures have resulted in large variations of substance concentrations reported in different studies. These effects are even more pronounced in the clinical environment and under mechanical ventilation because of positive airway pressures. For each substance, it must be determined if its source was actually the patient or if it came from ambient air or previous exposure. To assess these issues fast, reliable, and highly sensitive analytical techniques have to be applied. For that purpose, we optimized a PTR-ToF-MS with a mass resolution of >4000 for continuous and breath-resolved analysis in the clinical environment. The following questions were addressed in detail: Can breath gas profiles determined by means of PTR-TOF be assigned to breath phases (e.g., alveolar or inspired air) without the need for further equipment? Is it possible to analyze background air and breath samples in a time-resolved way and continuously in the clinical environment? Can changes in breath VOCs in mechanically ventilated patients be monitored continuously

■

EXPERIMENTAL SECTION Development of a PTR-ToF-MS Setting for Clinical Measurements. Instrumentation. A PTR-TOF-MS 8000 (Ionicon Analytik GmbH, Innsbruck, Austria) was used. For data acquisition, TOF DAQ was applied. Application of the PTRMS technique for breath analysis has been described before.33 Briefly, the primary ions (H3O+) are produced in the ion source by a hollow cathode discharge, using water vapor as a reactant gas. The proton transfer reaction between the formed H3O+ and neutral analyte molecules (M) occurs in the drift tube (M + H3O+ → MH+ + H2O). A differentially pumped ion transfer unit couples the PTR-ion source with the TOF-MS. The mass analyzer used in this instrument is a high mass resolution, orthogonal acceleration, reflectron TOF-MS (Tofwerk AG). Optimizing and Evaluation of the Transfer Line for Breath Sampling. PTR-ToF transfer lines have to be connected directly to mouth pieces or to the respiratory circuit when breath is to be sampled online and in real time from spontaneously breathing individuals or from mechanically ventilated patients, respectively. For reasons of potential interferences with medical equipment, the PTR-ToF instrument cannot be placed directly at the patients’ bedside. Hence, transfer lines have to be long enough to allow for a safe distance between MS and patients’ bed. For that purpose, Harrison et al. used an unheated 4 m transfer line when they analyzed exhaled propofol during surgery.4 King et al. used a 3 m long, 1/4″ teflon tube with a temperature of 40°C, in combination with an additional flow-triggered valve to sample alveolar breath in a laboratory setting, resulting in a mean response time of 10 seconds and an average of approximately 3 breaths.29 Up to now, the effects of different lengths and materials of transfer lines onto results of breath analysis were not addressed at all. We, therefore, investigated the influence of material, temperature, length, and inlet flow of the transfer line onto response times of the MS analysis and onto applicability in breath analysis. Two different transfer line materials were chosen in the experiments: a polyether ether ketone (PEEK) transfer line, i.d. of 0.75 mm (VICO, LA) and a Silcosteel transfer line, i.d. of 0.75 mm (Restek, Bellafonte, PA). Transfer lines were heated by means of a custom-made heating hose and heating jacket (Kletti GmbH, Sandhausen, Germany). Transfer lines with a length of 1.2 and 6 m were compared. The volume of the 6 m transfer lines (0.75 mm) was approximately 2.65 mL. Transfer line temperatures from 26−80 °C and additional inlet flows up to 70 mL/min were investigated with respect to the response time of the system. Response times were analyzed for endogenous breath gas components (acetone and isoprene) as well as for the iv-anesthetic propofol. Five microliters of a commercially available propofol solution (Propofol-Lipuro 1%, 10 mg/mL) were transferred into a Tedlar bag with a volume of 1 L. Then the bag was filled with nitrogen and heated up to 60 °C for 10 min. Optimizing the Time Resolution of the PTR-ToF-MS for Real Time Breath Analysis. High time resolution and good signal-tonoise ratio are required for online analysis of VOCs in trace concentrations in human breath. The noise increases with shorter intervals between data acquisition points. The intervals have to realize a reasonable compromise between time resolution and signal-to-noise ratio. Since normal respiratory rates are in the range of 12−16 breaths/min in adults and in the range of 20−30 breaths/min in children, a time resolution of 100 to 200 ms is 10322

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

placed in a room adjacent to the patient’s bed in the intensive care unit. The heated 6 m long silcosteel transfer line of the PTRToF-MS was connected to the respiratory system by means of a sterilized T-piece located at the end of the endotracheal tube (ETT). After 20 min, a clinical intervention resulting in a change in ventilation/perfusion ratio took place for ten minutes. Counts per seconds (cps) measured for the compounds of interest were normalized to the primary ions, according to standard practice in PTR-MS data analysis. Based on the ion chemistry in the PTR drift tube, which has been studied for many years, estimation of concentrations based on the normalized counts and the k rate of each compound is possible. However, k rates are not available for all substances, and fragmentation can occur for some substances. To ensure precise quantitation of the compounds under investigation and in order to mimic high concentrations of some substances as occurring in the clinical environment, we performed additional external calibrations. Concentrations of sevoflurane, isoprene, and propofol were determined by means of calibrations with at least six concentration levels. Two replicates of each standard were analyzed. Water-saturated standards in nitrogen were prepared using a liquid calibration unit (Ionicon Analytik GmbH, Innsbruck, Austria). Calibration range was from 304 to 18.24 ppmV for sevoflurane, 0.94 to 470 ppbV for isoprene, and 4.82 to 227.26 ppbV for propofol. Detection and quantification limits were calculated using blank measurements. Ten water-saturated blanks were measured for that purpose, and LOD and LOQ were calculated by means of mean +3 SD and mean +10 SD. Samples were measured for one minute. The LOD for sevoflurane was 19.86 ppbV, and the LOQ was 27.58 ppbV when mass 198.999 (deprotonated sevoflurane) was used for calibration. The LOD of isoprene was 1.98 ppbV, and the LOQ was 2.39 ppbV. Detection limit of propofol was 0.70 ppbV, LOQ was 1.04 ppbV. Isopropanol concentrations were calculated using a k rate of 2.09 × 10−9cm3/s.44 Effect of Clinical Contaminations onto Breath Gas Profiles. The study had been approved by the local University Medical Centre Ethics Committee. In order to assess the effects of background contaminations from the clinical environment onto results of breath gas analysis, PTR-ToF-MS measurements were carried out in the post anesthesia care unit (PACU) of the University Hospital Rostock. Continuous measurements of room air as well as breath gas analyses from medical staff were done. Included in the study were 15 physicians, 10 nurses, and 7 volunteers not working in the clinical environment. Table S-1 of the Supporting Information shows a detailed overview on gender, age, BMI, and smoking status of the participants. Participants were asked to sit down and breathe through an exchangeable mouth piece that was connected to the heated silcosteel transfer line of the PTR-ToF-MS. Participants could breathe through the mouth piece with no additional resistance. Moreover, they were instructed to breathe through the mouth only and to hold their nose or use a nose clip during the measurement. Ten breaths were sampled each time and participants were asked to give breath samples on every working day at the beginning and at the end of their shift. The study went on for seven days. Furthermore, room air was analyzed continuously during a period of 7 days. Room air and breath gas measurements took place in the PACU. PTR-ToF-MS settings were the same as described in Pilot Measurements in Ventilated Patients (H3O+, NO+, and O2+),

required for breath-resolved measurements. A lower time resolution is only appropriate when respiratory rates are below 12 breaths/min. Evaluation of the Influence of the Ventilatory System onto in Vivo Measurements. Breath was sampled from a healthy volunteer who was ventilated via a face mask with a turbinedriven portable ventilator system (BREAS LTV 1000 , Moelnycke, Sweden). The effect of positive end expiratory pressures (PEEP) ranging from 5−15 mbar and inspiratory pressures in the range of 12−28 mbar was investigated for respiratory rates from 8−12 breaths/min. The primary ion count remained constant throughout the pressure variations induced by the ventilation via the face mask. The experiments showed that pressure fluctuations in ventilated patients did not affect the PTR-ToFMS analyses. Hence, this optimized setup could be applied in spontaneously breathing volunteers as well as in mechanically ventilated patients. Effect of Different Reactant Species (H3O+, NO+, and O2+). H3O+, NO+, or O2+ can be applied for chemical ionization (CI) in (P)TR-ToF-MS. For a given substance, mass spectra will change dependent on whether H3O+, NO+, or O2+ is used for CI. In a pilot experiment, all three ionization reagents were used to analyze breath from a spontaneously breathing volunteer. The respiratory rate was adjusted to 10 breaths/min by means of a metronome. Time resolution was 200 ms. The volunteer was asked to breathe into the instrument for three minutes in each ionization mode. Figure S-1 shows spectra averaged over these three minutes for all three ionization reagents. To minimize fragmentation of the VOCs and to exclude effects of more than one precursor ion, the H3O+ mode was used in all experiments. Despite the lower ionization energy (if compared to NO+ and O2+), fragmentation of larger molecules such as sevofluran can still take place. As previously described by Critchley et al.,43 sevofluran fragments with m/z ratios of 181.0071 and 198.999 could also be observed in our data with mass trace 181.0071 being the main fragment. Data Processing Algorithm. For automatic recognition of alveolar and inspiratory phases a matlab (version 7.12.0.635, R2011a)-based data processing algorithm (“breath tracker”) was applied. A detailed description of the data processing algorithm can be found in S-2 of the Supporting Information. Pilot Measurements in Ventilated Patients. The study had been approved by the local University Medical Centre Ethics Committee, and the subjects gave their written informed consent. Breath from mechanically ventilated patients who had just undergone heart surgery was analyzed continuously over 60 min. Measurements were carried out directly after surgery in the intensive care unit. Time resolution of the PTR-ToF was 200 ms. H2O flow was 6 sccm, and the additional sampling flow was set to 30 mL/min. Temperature of the drift tube and the transfer line were both at 75 °C. The drift voltage was 610 V, and the drift tube pressure was 2.3 mbar, resulting in an E/N ratio of 139 Td. The current of the ion source was 4 mA. Every minute a new data file was recorded. The mass scale was recalibrated after every run (60 s). Mass calibration was done using four masses that were abundant both in inspiratory and expiratory air. Masses used for that purpose were 21.0226 (H3O+-Isotope), 29.9980 (NO+), 57.0699 (C4H8H+), and 198.999 (sevoflurane fragment). Isoprene was used as tracker substance for the recognition of alveolar phases (see Data Processing Algorithm) in this experiment. Continuous measurements in mechanically ventilated patients were carried out for a period of one hour. The PTR-ToF-MS was 10323

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

Figure 1. Continuous breath monitoring in a mechanically ventilated patient on the intensive care unit. The upper diagram (a) shows alveolar concentrations (in ppbV) of isoprene, propofol, sevofluorane, and isopropanol over a time of 60 min. Sevoflurane concentration was divided by a factor of 200 for better visibility. Mean alveolar concentrations and variations refer to data averaged over 1 min. At minute 43, cardiac output increased from 3.8 to 8.0 L/min and fell back to 5.7 L/min within 3 min. The lower diagram (b) shows one of these data entries in a time-resolved profile of isoprene, CO2, O2, and sevoflurane over 60s without averaging (minute 44 of the experiment). High deviations in the upper diagram, therefore, indicate changes of marker concentrations within the averaging time (60 s).

except that a new data file was recorded every three minutes instead of every minute, and the additional sampling flow was 19 mL/min. Acetone was used as a tracker mass for the recognition of alveolar phases (see Data Processing Algorithm) in these experiments. Acetone and sevoflurane concentrations were determined by means of calibrations with at least 6 concentration levels. Two replicates of each standard were analyzed. Water saturated standards in nitrogen were prepared using a liquid calibration unit (Ionicon Analytik GmbH, Innsbruck, Austria). Two calibrations were used for the quantification of acetone. Concentrations up to 59.4 ppbV were quantified using a calibration in the range from 0.99 to 59.4 ppbV; higher concentrations were quantified using a

calibration in the range from 59.4 to 495 ppbV. The calibration range was from 30.4 to 912 ppbV for sevoflurane (mass 181.0071). Detection and quantification limits were calculated using blank measurements. Ten water saturated blanks were measured for that purpose, and LOD and LOQ were calculated by means of mean +3 SD and mean +10 SD. Time resolution was 200 ms and samples were measured for one minute. The detection limit of acetone was 0.90 ppbV, and LOQ was 1.03 ppbV. LOD for sevoflurane was 0.40 ppbV, and LOQ was 0.59 ppbV, when mass 181.0071 (main fragment) was used for calibration. 10324

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

Figure 2. VOC profiles determined in the PACU over 24 h. Concentrations were normalized to maximum values. (a) Measurements during a working day. (b) Measurements on a Sunday.

■

RESULTS Optimization of the Sampling System for Real Time Breath Analysis. The sampling system is of crucial importance when breath-resolved real time measurements in a clinical environment are to be realized. Response times have to be identical for small, highly volatile and larger, less volatile compounds to ensure unequivocal attribution of substance concentrations to different breath phases. Because of clinical safety requirements, the transfer line has to be long enough to place the PTR-ToF-MS outside the patient’s room. Temperature, sampling flow, and materials of transfer lines are crucial parameters to ensure phase resolution of breath cycles

and short response times. Table S-2 of the Supporting Information shows an overview on investigated transfer line materials, temperatures, flows, and related response times. Response times for “sticky” compounds such as propofol were higher than 120 s with the PEEK transfer line. Response times could be reduced to 40 s, when the transfer line was heated to 80 °C. In contrast, the silco steel transfer line, operated at room temperature, showed response times of 55 s for propofol. When the silco steel transfer line was heated, response times for propofol could be reduced to 5 s (80 °C). Response times could be reduced to 2.5 s when the additional sampling flow was increased. No differences in response times between highly 10325

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

volatile substances, such as acetone or isoprene, and “sticky” compounds of low volatility, such as propofol, could be observed within the adapted setting. For these reasons, silco steel transfer lines were used in all experiments, a transfer line temperature of 80 °C and an additional sampling flow of at least 19 mL/min was applied. In previous PTR settings, only a small number of preselected VOCs were analyzed. Harrison et al. determined exhaled concentrations of propofol and its metabolites in mixed expired air.4 King et al. monitored exhaled concentrations of acetone, ammonia, and isoprene by cutting out and averaging several alveolar phases.29 As can be seen from our data, transfer line temperatures of 40 °C or less result in different response times for different VOCs. As a consequence, these previously described sampling systems do not enable simultaneous breath phase-resolved analysis for substances of different physicochemical properties. Data Analysis Algorithm for the Identification of Breath Phases (Breath Tracker). PTR-ToF-MS realizes acquisition of large amounts of data in a continuous real time mode. Continuous measurements can be done for hours and even for days. For biomarker discovery and monitoring identification of alveolar and inspiratory phases is mandatory because exhalations consist of three phases. At first, dead space air from the mouth and the trachea is exhaled. The second phase contains a mixture of alveolar air and dead space air. Finally, alveolar air is exhaled. If blood born biomarkers are to be assessed by means of breath analysis, alveolar concentrations have to be determined, since gas exchange takes place in the alveoli of the lungs and only alveolar substance concentrations can be correlated to blood levels.15 Hence, an algorithm had to be developed to distinguish inspiratory from alveolar phases. Processing of large amounts of data and automatic recognition of respiratory phases were basic requirements for the algorithm. In principle, respiratory phases can be identified by means of the water cluster signal, since expiratory phases contain more water than inspiratory phases.28 Another possibility is to choose a tracker mass, which is supposed to have clearly higher signal intensity during alveolar phases than during inspired phases. Acetone or isoprene are endogenous blood borne substances that meet this requirement, since they are always abundant in human breath. Intensities of all mass traces could be assigned to alveolar or inspiratory phases automatically by means of the time frames determined through the breath tracker. In this way, endogenous compounds could easily be distinguished from contaminants from the clinical environment. Exhaled concentrations of endogenous substances such as acetone and CO2 had the highest intensities in the expiratory phases and lowest responses in the inspiratory phases. In contrast, signals of O2+ were highest in the inspiratory phases and lowest in the expiratory phases. Since oxygen is a marker for gas exchange, the lower concentrations in the expiratory phases confirm the reliability of the data processing algorithm. (cf. Figure 1) VOC Monitoring in Mechanically Ventilated Patients. Figure S-3 of the Supporting Information shows a mass spectrum from a mechanically ventilated patient, averaged over 10 breaths on a logarithmic scale. Table S-3 of the Supporting Information contains a list of tentatively identified substances in the mass spectrum. Figures 1a and S-4 of the Supporting Information show profiles of alveolar concentrations over a time of 60 min for isoprene, propofol, sevoflurane, and isopropanol. Concentrations of exogenous isopropanol and the anesthetic drugs

propofol and sevoflurane were not affected through changes of ventilation (lung recruitment) and increased cardiac output, respectively. Endogenous isoprene, in contrast, strongly depended on these parameters and showed an increase when cardiac output was increased (Figure 1b) and a decrease when ventilation was increased. In contrast to quad-MS systems, which are restricted to some target markers, PTR-TOF provides MS analysis of a broad spectrum of exhaled VOCs. The software algorithm described above can then help to process the large amounts of data, typically acquired within clinical studies. As we, in contrast to Herbig et al.,33 did not observe a lag time for CO2 in the breath (Figure 1b) in our optimized setting CO2 may also be used as a tracker mass for the identification of alveolar phases. Application of direct PTR-ToF-MS with a time resolution of 200 ms enables continuous breath-resolved monitoring of a broad range of breath biomarkers in real time. In contrast to previous settings,29,32 breath-resolved determination of alveolar concentrations as well as the parallel analysis of inspired concentrations without the need of any additional equipment was feasible. As changes in marker profiles are pronounced and may occur between single breath cycles or in the time range of seconds (Figure 1b), real time monitoring rather than averaging of several breath cycles is required. Real time PTR-ToF-MS measurements in ventilated patients can thus provide continuous and instantaneous information on breath biomarker or drug concentrations as well as knowledge on potential contaminants from the clinical environment. Effect of Clinical Contaminations onto Breath Gas Profiles. Figure 2 shows two heatmaps containing 75 selected m/q traces and their trends during a whole day. A typical workday (Friday, Figure 2a) and a Sunday (Figure 2b) are shown in comparison. Data were normalized onto maximum values to emphasize the relative changes over time. A list of VOCs related to some mass traces is given in Table S3. As could be expected, concentrations of most VOCs are comparably low during the night and in the early morning hours on a work day. Operating theaters were run during the daytime, except for emergencies at night. As a consequence, there were hardly any patients in the PACU during the nights, and overall VOC concentrations were low. Between 9:45 AM and 10:00 AM, the VOC profile started to change and increasing concentrations of several VOCs, including sevoflurane, could be observed. Concentrations decreased again after 30 min. This tendency repeated itself several times during working days. From 4:00 PM onward, concentrations steadily decreased and reached base levels again several hours later (e.g., at around 9:00 PM for sevoflurane).

Figure 3. Time course of sevofluorane (solid line) concentrations in the PACU during a normal working day. Red lines: patients in the PACU who had previously been anaesthetized with sevoflurane. 10326

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

Figure 4. Box plots of sevoflurane concentrations in alveolar breath of clinical staff (a) and in inspired air (b). Acetone concentrations in alveolar breath of clinical staff (c) and in inspired air (d) collected during breath sampling. The y scale was made logarithmic for better visibility.

the algorithm described above. Acetone was used as tracker substance and intensities were normalized onto primary ion counts. Figure 4 shows box plots of sevoflurane and acetone concentrations in the breath of the study participants and in inspiratory air collected during breath sampling. Exhaled substance concentrations in the different volunteer groups (physicians, nurses, and volunteers not working in the clinical environment) were compared before and at the end of the volunteers’ working shift. The nurses’ group was further divided into three subgroups with respect to the different shifts they were working (early, late, and night shifts). Figure 4a shows expiratory sevoflurane concentrations. Physicians and volunteers not working in the clinic had higher sevoflurane concentrations in their breath at the end of their work shift than at the beginning. The same applied for the nurses working the early shift. Nurses

Other workdays showed comparable profiles. On Sunday, however, VOCs stayed at the base level during the whole day. Figure 3 shows the time course of sevoflurane during the same working day that was already shown in Figure 2a. The red bars show the occupancy of the PACU with patients that had previously been anesthetized with sevoflurane. Sevoflurane concentrations started to increase within 15 min after the first patients entered the PACU and varied depending on the occupancy of the PACU with patients. At the end of the workday, sevoflurane concentrations decreased and reached base levels again at around 9:00 PM. In order to investigate the effects of background concentrations onto breath gas profiles, breath samples, and samples from inspired air were analyzed and compared with each other. Results from breath measurements were processed by means of 10327

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

concentrations or the impact of acute and previous environmental exposures onto breath profiles can be assessed. In contrast to previous settings, no additional equipment or sample collection devices were needed for data acquisition. Time resolutions of some hundred milliseconds in combination with high mass resolution enabled continuous monitoring of alveolar breath biomarker concentrations down to the pptV level. Requirements for simultaneous analysis of exhaled breath and determination of ambient substance concentrations were met. The optimized and flexible PTR-TOF setting was successfully applied in spontaneously breathing volunteers as well as in mechanically ventilated patients. Our results prove that concentrations of breath biomarkers, but also those of typical contaminants, may change rapidly and in a pronounced way in the clinical environment. Time and place of breath sampling can have a distinct impact onto exhaled substance concentrations and may easily override endogenous effects. Punctual breath sampling or simple subtraction of background air is not sufficient to account for this problem. Continuous and real time monitoring of substances from the clinical environment in combination with breath-resolved determination of exhaled VOC concentrations in patients and in volunteers can help to solve crucial problems in breath analysis, such as effects of previous and acute exposures onto human breath profiles. When optimized sample collection, intelligent data processing, and careful study planning are combined, the high mass and time resolution of the PTR-ToF-MS provides a powerful tool to address fundamental as well as practical questions in clinical breath analysis.

working the late shift showed the opposite behavior, while sevoflurane concentrations during the night shift were low before and after work. Concentrations of inspired air shown in Figure 4b showed the same tendencies. However, concentrations were even higher than in breath samples. In contrast, inspired concentrations of acetone (Figure 4d) were magnitudes lower than concentrations in breath samples (Figure 4c). Moreover, acetone concentrations did not show the same tendencies as sevoflurane concentrations with respect to the time of breath measurement. Rieder et al. also reported a sevoflurane burden dependent on patient turnover in the room air of the urologic PACU with a mean exposure of 15.9 to 9.5 ppbV.45 Summer et al. reported significantly increased sevoflurane concentrations in the breath of staff members after the end of their duty in the operating theater (0.8 ppbV) in comparison with values before their shift (0.26 ppbV).46 Cope et al. found significantly increased isoflurane levels in the breath of PACU nurses after their work shift on a Monday compared to isofluran levels determined before their shift. Moreover, the increase was significantly higher on a Monday compared to a Friday.47 Figure S-5 of the Supporting Information shows a comparison of room air concentrations with expiratory concentrations obtained by means of the breath tracker algorithm. On the y axis, room air concentrations are plotted versus expiratory and inspiratory concentrations on the x axis. The volunteer group not working in the clinical environment was chosen for that purpose, since they had not previously or chronically been exposed to sevoflurane as was the case for physicians and nurses. A good linearity could be observed (R2 = 0.94), proving the reliability of the breath tracker algorithm as well as the strong dependency of exhaled sevoflurane concentrations on actual background concentrations. In previous settings, only expired alveolar concentrations were monitored by PTR-QMS.29,32 As can be seen from our results, qualitative and quantitative aspects of contaminations in the clinical environment may change fast and in a pronounced way. Therefore, parallel monitoring of alveolar and inspiratory air is mandatory in a clinical setup. As sevoflurane represents an exogenous substance, its concentrations in the breath depended on previous exposure48 and background concentrations. When breath analysis was carried out in the PACU, even volunteers not working in the clinic exhaled sevoflurane concentrations as high as the exhaled concentrations observed in physicians and nurses who were exposed to sevoflurane on a regular basis. Since solubility of sevoflurane in blood is low (blood-gas coefficient = 0.6849,50) and solubility in fatty tissue is also low,50 no long-term uptake due to chronical exposure was observable in our data. However, sevoflurane concentrations in ambient air were immediately reflected in the breath of people exposed to it. These results emphasize the importance of taking background contaminations into account during breath measurements. Time and place of breath sampling can affect the results significantly and must not be ignored. Thus, breath-resolved time resolution, parallel determination of inspired concentrations, and analysis of a broad range of potential marker substances are important prerequisites for the application of real time mass spectrometry in the clinical environment.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] Tel: +49 381 494 5955. Fax: +49 381 494 5942. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The PTR-TOF instrument used in this study was entirely funded by the European fund for regional development (EFRE). The authors would like to thank the staff of the PACU and the ICU for participating and supporting this study.

■

REFERENCES

(1) Pauling, L.; Robinson, A. B.; Teranish., R.; Cary, P. Proc. Natl. Acad. Sci. U.S.A. 1971, 68, 2374−. (2) Pleil, J. D.; Stiegel, M. A.; Risby, T. H. J. Breath Res. 2013, 7, 017107. (3) Lindstrom, A. B.; Pleil, J. D.; Berkoff, D. C. Environ. Health Perspect. 1997, 105, 636−42. (4) Harrison, G. R.; Critchley, A. D. J.; Mayhew, C. A.; Thompson, J. M. Brit. J. Anaesth. 2003, 91, 797−799. (5) Mukhopadhyay, R. Anal. Chem. 2004, 76, 273A−276A. (6) Reynolds, J. C.; Blackburn, G. J.; Guallar-Hoyas, C.; Moll, V. H.; Bocos-Bintintan, V.; Kaur-Atwal, G.; Howdle, M. D.; Harry, E. L.; Brown, L. J.; Creaser, C. S.; Thomas, C. L. Anal. Chem. 2010, 82, 2139− 44.

■

CONCLUSION Continuous real time PTR-ToF-MS measurements provide a promising tool for clinical breath gas analysis. Crucial issues such as monitoring of fast changes of breath biomarker and drug 10328

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329

Analytical Chemistry

Article

(7) Mukhopadhyay, R. Anal. Chem. 2007, 79, 2610. (8) Grote, C.; Pawliszyn, J. Anal. Chem. 1997, 69, 587−596. (9) Sanchez, J. M.; Sacks, R. D. Anal. Chem. 2003, 75, 2231−2236. (10) Sanchez, J. M.; Sacks, R. D. Anal. Chem. 2006, 78, 3046−3054. (11) Ciaffoni, L.; Hancock, G.; Harrison, J. J.; van Helden, J. P.; Langley, C. E.; Peverall, R.; Ritchie, G. A.; Wood, S. Anal. Chem. 2013, 85, 846−50. (12) Mieth, M.; Schubert, J. K.; Groger, T.; Sabel, B.; Kischkel, S.; Fuchs, P.; Hein, D.; Zimmermann, R.; Miekisch, W. Anal. Chem. 2010, 82, 2541−2551. (13) Chambers, S. T.; Bhandari, S.; Scott-Thomas, A.; Syhre, M. Med. Mycol. 2011, 49 (Suppl 1), S54−61. (14) Fuchs, P.; Loeseken, C.; Schubert, J. K.; Miekisch, W. Int. J. Cancer 2010, 126, 2663−2670. (15) Miekisch, W.; Schubert, J. K.; Noeldge-Schomburg, G. F. Clin. Chim. Acta 2004, 347, 25−39. (16) Alonso, M.; Castellanos, M.; Sanchez, J. M. Anal. Bioanal. Chem. 2010, 396, 2987−95. (17) Buszewski, B.; Ulanowska, A.; Ligor, T.; Denderz, N.; Amann, A. Biomed. Chromatogr. 2009, 23, 551−556. (18) Miekisch, W.; Fuchs, P.; Kamysek, S.; Neumann, C.; Schubert, J. K. Clin. Chim. Acta 2008, 395, 32−7. (19) Kamysek, S.; Fuchs, P.; Schwoebel, H.; Roesner, J. P.; Kischkel, S.; Wolter, K.; Loeseken, C.; Schubert, J. K.; Miekisch, W. Anal. Bioanal. Chem. 2011, 401, 2093−2102. (20) Pleil, J. D.; Stiegel, M. A.; Sobus, J. R. J. Breath Res. 2011, 5, 046005. (21) Pleil, J. D.; Stiegel, M. A.; Sobus, J. R.; Liu, Q.; Madden, M. C. J. Breath Res. 2011, 5, 037104. (22) Sinues, P. M.; Kohler, M.; Zenobi, R. Anal. Chem. 2013, 85, 369− 373. (23) Martı ́nez-Lozano, P.; Fernandez de la Mora, J. Anal. Chem. 2008, 80, 8210−8215. (24) Mühlberger, F.; Streibel, T.; Wieser, J.; Ulrich, A.; Zimmermann, R. Anal. Chem. 2005, 77, 7408−7414. (25) Spanel, P.; Smith, D. Mass Spectrom. Rev. 2011, 30, 236−67. (26) Wang, C. J.; Sahay, P. Sensors (Basel) 2009, 9, 8230−8262. (27) Schwarz, K.; Pizzini, A.; Arendacka, B.; Zerlauth, K.; Filipiak, W.; Schmid, A.; Dzien, A.; Neuner, S.; Lechleitner, M.; Scholl-Burgi, S.; Miekisch, W.; Schubert, J.; Unterkofler, K.; Witkovsky, V.; Gastl, G.; Amann, A. J. Breath Res. 2009, 3, 027003. (28) Schwoebel, H.; Schubert, R.; Sklorz, M.; Kischkel, S.; Zimmermann, R.; Schubert, J. K.; Miekisch, W. Anal. Bioanal. Chem. 2011, 401, 2079−2091. (29) King, J.; Kupferthaler, A.; Unterkofler, K.; Koc, H.; Teschl, S.; Teschl, G.; Miekisch, W.; Schubert, J.; Hinterhuber, H.; Amann, A. J. Breath Res. 2009, 3, 027006. (30) King, J.; Mochalski, P.; Kupferthaler, A.; Unterkofler, K.; Koc, H.; Filipiak, W.; Teschl, S.; Hinterhuber, H.; Amann, A. Physiological Measurement 2010, 31, 1169−1184. (31) King, J.; Unterkofler, K.; Teschl, G.; Teschl, S.; Koc, H.; Hinterhuber, H.; Amann, A. J. Math Biol. 2011, 63, 959−99. (32) King, J.; Unterkofler, K.; Teschl, G.; Teschl, S.; Mochalski, P.; Koç, H.; Hinterhuber, H.; Amann, A. J. Breath Res. 2012, 6, 016005. (33) Herbig, J.; Müller, M.; Schallhart, S.; Titzmann, T.; Graus, M.; Hansel, A. J. Breath Res. 2009, 3, 027004. (34) Kohl, I.; Beauchamp, J.; Cakar-Beck, F.; Herbig, J.; Dunkl, J.; Tietje, O.; Tiefenthaler, M.; Boesmueller, C.; Wisthaler, A.; Breitenlechner, M.; Langebner, S.; Zabernigg, A.; Reinstaller, F.; Winkler, K.; Gutmann, R.; Hansel, A. J. Breath Res. 2013, 7, 017110. (35) White, I. R.; Willis, K. A.; Whyte, C.; Cordell, R.; Blake, R. S.; Wardlaw, A. J.; Rao, S.; Grigg, J.; Ellis, A. M.; Monks, P. S. J. Breath Res. 2013, 7, 017112. (36) Morisco, F.; Aprea, E.; Lembo, V.; Fogliano, V.; Vitaglione, P.; Mazzone, G.; Cappellin, L.; Gasperi, F.; Masone, S.; De Palma, G. D.; Marmo, R.; Caporaso, N.; Biasioli, F. PLoS One 2013, 8, e59658. (37) Schubert, J. K.; Miekisch, W.; Geiger, K.; Noldge-Schomburg, G. F. Expert Rev. Mol. Diagn. 2004, 4, 619−29.

(38) Schubert, J. K.; Muller, W. P.; Benzing, A.; Geiger, K. Intensive Care Med. 1998, 24, 415−21. (39) Schubert, J. K.; Spittler, K. H.; Braun, G.; Geiger, K.; Guttmann, J. J. Appl. Physiol. 2001, 90, 486−92. (40) Birken, T.; Schubert, J.; Miekisch, W.; Noldge-Schomburg, G. Technology Health Care Journal 2006, 14, 499−506. (41) Schubert, J. K.; Esteban-Loos, I.; Geiger, K.; Guttmann, J. Technology Health Care Journal 1999, 7, 29−37. (42) Schubert, J. K.; Miekisch, W.; Birken, T.; Geiger, K.; NoldgeSchomburg, G. F. Biomarkers 2005, 10, 138−52. (43) Critchley, A.; Elliott, T. S.; Harrison, G.; Mayhew, C. A.; Thompson, J. M.; Worthington, T. Int. J. Mass Spectrom. 2004, 239, 235−241. (44) Cappellin, L.; Karl, T.; Probst, M.; Ismailova, O.; Winkler, P. M.; Soukoulis, C.; Aprea, E.; Mark, T. D.; Gasperi, F.; Biasioli, F. Environ. Sci. Technol. 2012, 46, 2283−90. (45) Rieder, J.; Prazeller, P.; Boehler, M.; Lirk, P.; Lindinger, W.; Amann, A. Anesth. Analg. (Hagerstown, MD, U. S.) 2001, 92, 389−392. (46) Summer, G.; Lirk, P.; Hoerauf, K.; Riccabona, U.; Bodrogi, F.; Raifer, H.; Deibl, M.; Rieder, J.; Schobersberger, W. Anesth. Analg. (Hagerstown, MD, U. S.) 2003, 97, 1070−3. (47) Cope, K. A.; Merritt, W. T.; Krenzischek, D. A.; Schaefer, J.; Bukowski, J.; Foster, W. M.; Bernacki, E.; Dorman, T.; Risby, T. H. Journal of PeriAnesthesia Nursing 2002, 17, 240−50. (48) Ghimenti, S.; Di Francesco, F.; Onor, M.; Stiegel, M. A.; Trivella, M. G.; Comite, C.; Catania, N.; Fuoco, R.; Pleil, J. D. J. Breath Res. 2013, 7, 036001. (49) Dudziak, R.; Vettermann, J. Anaesthesist 1996, 45, S1−S9. (50) Clarke, K. W. Veterinary Clinics of North America: Small Animal Practice 1999, 29, 793−810.

10329

dx.doi.org/10.1021/ac402298v | Anal. Chem. 2013, 85, 10321−10329