Detection of Metabolites of Trapped Humans Using Ion Mobility

Article pubs.acs.org/ac

Detection of Metabolites of Trapped Humans Using Ion Mobility Spectrometry Coupled with Gas Chromatography Wolfgang Vautz,*,† Rafael Slodzynski,† Chandrasekhara Hariharan,† Luzia Seifert,† Jürgen Nolte,† Rita Fobbe,† Stefanie Sielemann,‡ Bolan C. Lao,‡ and Lars Hildebrand§ †

Leibniz-Institut für Analytische Wissenschaften−ISAS−e.V., Bunsen-Kirchhoff-Straße 11, 44139 Dortmund, Germany Gesellschaft für analytische Sensorsysteme mbH (G.A.S.), Otto-Hahn-Straße 15, 44227 Dortmund, Germany § Department of Computer Science, University of Dortmund, Otto-Hahn Street 16, 44227 Dortmund, Germany ‡

S Supporting Information *

ABSTRACT: For the first time, ion mobility spectrometry coupled with rapid gas chromatography, using multicapillary columns, was applied for the development of a pattern of signs of life for the localization of entrapped victims after disaster events (e.g., earthquake, terroristic attack). During a simulation experiment with entrapped volunteers, 12 human metabolites could be detected in the air of the void with sufficient sensitivity to enable a valid decision on the presence of a living person. Using a basic normalized summation of the measured concentrations, all volunteers involved in the particular experiments could be recognized only few minutes after they entered the simulation void and after less than 3 min of analysis time. An additional independent validation experiment enabled the recognition of a person in a room of ∼25 m3 after ∼30 min with sufficiently high sensitivity to detect even a person briefly leaving the room. Undoubtedly, additional work must be done on analysis time and weight of the equipment, as well as on validation during real disaster events. However, the enormous potential of the method as a significantly helpful tool for searchand-rescue operations, in addition to trained canines, could be demonstrated.

S

of essential importance to localize trapped victims as early as possible, to avoid any decline in their vital status, even if they are not directly injured. Furthermore, immediate and reliable information on their health status would be of significant relevance to enable prompt inception of the particular operations and to initiate focused medical support. Therefore, USaR may be divided in four subsequent steps, and during all of these steps, the safety of the rescuers must be guaranteed continuously:6 (1) Localization of trapped bodies (2) Assessment to determine if the trapped victim is still alive (3) Monitoring of their vital statistics. (4) Rescue and first aid For the first step, the localization of trapped victims after a disaster event, dogs still are the method of choice, because of the extreme mobility and sensitivity of trained canines. However, dogs need a break quite frequently and they have problems in extremely difficult situations (e.g., with many dead victims in the surroundings). Therefore, within the framework of a joint research project funded by the European Union, we

earch and rescue of trapped humans is a permanent challenge for teams specialized in this field. Geological events such as earthquakes, volcanic eruptions, or landslides happen everyday.1 Most of these events are not perceived, since their impact is too small. But the situation changes dramatically if such phenomena strike urban areas.2 Every year, significant disasters occur, such as the recent earthquakes in Haiti3 and Japan in 2011, sometimes followed by a severe tsunami. As a consequence, buildings are destroyed and many people are trapped in the ruins.4 Similar scenarios may happen after tropical storms as well as events induced by terroristic attacks.4 In all cases, it is of essential importance to localize and rescue the trapped victims as soon as possible to save their lives. Many of them are severely injured and urgently need medical first aid. Although the equipment available for the search and rescue teams is developing continuously, there is always a need for further improvement.5 Time is the critical parameter during search-and-rescue operations. In most instances, initial search-and-rescue procedures are carried out by survivors.5 Then, fire and police departments and other local authorities will converge on the affected area and assume control and coordination. In a largescale disaster that overwhelms the resilience of the affected region, professional urban search and rescue (USaR) teams are mobilized at the national and eventually international level. It is © XXXX American Chemical Society

Received: September 23, 2012 Accepted: December 18, 2012

A

dx.doi.org/10.1021/ac302752f | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

intend to develop a “Second Generation Locator for Urban Search and Rescue”, to improve the technical equipment for search-and-rescue operations. In particular, all available sensors, such as audio and video (thermal, ultraviolet (UV), and visual cameras), should be combined with innovative instruments such as ion mobility spectrometers (IMS) or chemical sensor arrays. Furthermore, all data obtained should be evaluated in a comprehensive way, thus providing data of higher significance related to the detection of “signs of life” and even regarding the health status of a trapped person. The innovative sensors include portable ion mobility spectrometers for the detection of relevant volatile organic compounds (VOCs) in the field as available in a commercial design7 but also laboratory instruments previously used for the analysis of medical and biological applications,8 as well as for food characterization.9 Those instruments combine high sensitivity with rapid results in the range of secondsor minutes, if additional gas-chromatographic preseparation is applied.10 The work presented here is focused on the detection of VOCs, which could enable the localization of trapped living victims using ion mobility spectrometry with rapid gaschromatographic preseparation (GC/IMS). After optimization and calibration procedures, two GC/IMS systems were applied during the “Trapped Human Experiment” (abbreviated hereafter as THE),7 which was designed to simulate the situation of victims trapped under collapsed buildings. A pattern of relevant detected analytes was developed and verified during an independent validation experiment.

conditions, the only source of CO2 is the particular volunteer in the void. Ion Mobility Spectrometry Coupled with Gas Chromatography. Ion mobility spectrometry is a sensitive and rapid analytical method for the detection of trace substances in the gas phase.11 In the present study, analyte molecules are ionized by proton transfer from reactant ions that are produced by a β-radiation source. In general, protonated monomer ions are formed. However, when the concentration increases, dimers or even higher polymer ions also may be formed. The ions are accelerated by a weak electric field toward the detector, which is a Faraday plate. An ion gate is opened and an ion cloud is allowed to enter the drift region, thus starting to travel toward the detector. A drift gas flow in the opposite direction (nitrogen or clean and dry air) causes collisions of the ions with the present gas molecules. Hence, the ions reach a constant resulting drift velocity, depending on their charge, mass, and shape. The drift velocity is characteristic for every analyte molecule and can be calculated from the measured drift time at a known drift length. Normalizing the drift velocity to the electric field provides the ion mobility (K). A further normalization to pressure and temperature leads to the socalled “reduced ion mobility” (K0), which is a unique characteristic of the analyte and independent of the experimental conditions. Method development first started in the 1960s with initial military application for the detection of chemical warfare agents and, later, for security purposes (e.g., at airports, for the detection of explosives and drugs of abuse).11 During the past decade, the method was adopted to various additional fields (e.g., for air quality monitoring, process control, food quality and safety, and for biological and medical purposes). In those cases, samples are getting more and more complex and, in particular, during the analysis of human breath, very humid. As a consequence, the various analyte ions together with the water molecules present at the same time in the ionization region tend to form clusters, hence making their identification difficult or even impossible. Therefore, ion mobility spectrometry coupled with gas-chromatographic preseparation (GC/IMS) has become more and more important.10,11 The instruments applied during the experiments described above were both of that type: a custom-made research instrument8 (ISAS, used for GC/IMS) and a commercial instrument9 (BreathSpec, made by Gesellschaft für analytische Sensorsysteme mbH (G.A.S.), Dortmund, Germany). Both instruments were equipped with a multicapillary column for rapid gas-chromatographic preseparation (the ISAS system has minimal polar capabilities, whereas G.A.S. does have polar capabilities). They have already been applied successfully for various applications (e.g., for the analysis of human breath,12−14 as well as that of animal breath,15,16 and for the detection of the metabolites of micro-organisms17,18 but also for food quality and safety control19−21 and for process control22,23). For sampling, the room airparticularly, the air from the void to be investigatedwas drawn for 20 s through the sample loop to avoid memory effects. The volume of the sample loop then was introduced into the preseparation column for analysis using a six-way-valve. The instruments were operated with a slightly different preseparation to enhance the information obtained from the experiments. The full experimental setup, including the additional CO2 measurements, is described in Table 1.

■

EXPERIMENTAL SECTION The “Trapped Human Experiment”. The “Trapped Human Experiment” (THE) has previously been described in detail.7 A continuously replenished reservoir of fresh air with a reserve of almost 40 min was fed to a volunteer inside the simulator representing a void in a collapsed building. The volatiles released by the volunteer accumulated in the void and were subsequently passed through a model of the debris: a glass column containing cassettes of typical building materials. Each cassette consisted of a set of disks made from materials commonly used to construct a single story of a glass-clad, reinforced-concrete building. Altogether, 10 volunteers entered the void simulator for 6 h each.7 All of them gave informed consent. The experiments that we monitored for the safety of the volunteers for critical gas concentrations and the health status of the volunteers were recorded using a vital-signs monitor. Additional information on the body-mass-index (BMI), the diet, medication, recent digestion and on personal hygiene was collected directly from the volunteers prior to the particular experiments. However, only 7 of the particular experiments (Nos. 4−10) have been completely investigated by GC/IMS and additionally TD-GC/ MS, starting in the last hour of experiment 3. GC/IMS and TDGC/MS samples were drawn directly from the void simulator using 1-m Teflon tubes. Note that none of the volunteers urinated into the void, but volunteer 9 temporarily left the void for this purpose. During the THE, the environment in the void was controlled for the safety of the volunteers. One of the variables that were monitored continuously was the carbon dioxide concentration, which was measured with an electrochemical sensor (Figaro TSG 4161). The CO2 concentration profile is used to visualize the development of the particular experiments. It is the ideal parameter for this purpose: because of the controlled B



Article

mode was performed with 70 eV, and a mass range of m/z 33− 450 was detected. To validate those proposals, the proposed analyte was introduced in a HovaCAL (IAS, Frankfurt, Germany) calibration gas generator26 for approval of the ion mobility and retention time and for providing calibration curves for the quantification. Ethics. All volunteers during the THE, as well as those involved in the studies at ISAS and G.A.S. laboratories, gave written informed consent. The data was recorded, stored, and evaluated anonymously. During all experiments, the volunteers were medically controlled.

Table 1. Experimental Setup of Both GC/IMS Systems: ISAS and G.A.S.a parameter flowMCC lengthMCC temperatureMCC typeMCC ion source lengthdrift flowdrift temperaturedrift electric field volumeloop temperatureloop

ISAS GC/IMS Preseparation 150 mL/min 20 cm 40 °C OV-5, min polar IMS 63 Ni, 550 MBq 12 cm 100 mL/min ambient 320 V/cm Sampling 8 mL 40 °C

G.A.S. BreathSpec 100 mL/min 20 cm 40 °C OV-1701, polar 3

H, 100 MBq 5 cm 500 mL/min 40 °C 400 V/cm

■

RESULTS AND DISCUSSION In the GC/IMS spectra obtained during the experiment, altogether, 46 signals were investigated, all of them following the development of the experiments and the CO2 slope in particular, as discussed in the following in detail. Octanal, as a first example, was detected during all experiments but with a higher concentration during one particular experiment (No. 6) compared to the others. Octanal has already been found in samples of human breath,27 which explains the elevated concentration during the particular experiment (see Figure 2). On the other hand, it is also used as a solvent, which could be the reason for the elevated concentration in experiment 6, where the volunteer declared that a specific perfume was recently used. Not all signals appeared during all experiments. From the additional data obtained from the volunteers, one could assume that those signals are related to individual digestion, diet, or personal hygiene. Those signals cannot be expected for any trapped person, and, therefore, those peaks were not considered for further evaluation. The signal of 4-methyl-2-pentanone was used for validation of the correlation of the GC/MS analysis of the samples to the GC/IMS data. This analyte is supposed to be present in paint and glue and might be related to the construction of the void. Therefore, a significant increase could be observed during the experimentsmost probably due to accumulation in the closed voidwith good agreement with the GC/MS analysis. To obtain information on the concentration range of the detected analytes, the GC/IMS was calibrated with reference

5 mL 40 °C

a

CO2 reference measurements were made using the following equipment and conditions: Figaro TSG 4161 sensor, operating range = 0−10.000 ppm.

Identification and Quantification. For the identification of the signals in the GC/IMS, their ion mobility24 and retention time were compared to the values in the ISAS analyte database. If no match could be found, related TD-GC/MS analyses of samples from the void (1 L drawn by a pump via a Teflon tube) on TENAX adsorption tubes were consulted and, by using appropriate alignment procedures25 for the GC/MS and GC/ IMS retention times, proposals on the origin of unknown GC/ IMS signals could be given. For GC/MS analysis, the adsorbed analytes were desorbed thermally and the sample was injected splitless at 250 °C into an Agilent Technologies Model 6890N GC system connected with an Agilent Technologies Model 5973 mass-selective detector (MSD, Gerstel, Mühlheim, Germany). The initial oven temperature of 35 °C was maintained for 2 min and then increased to 250 °C at a rate of 7 °C/min and finally held for 27 min. A HP-5MS capillary column (60 m, 0.25 mm, 0.25 m film thickness; Wicom, Heppenheim, Germany) was used for compound separation with helium as the carrier gas at a constant flow rate of 1.0 mL/min. The electron ionization

Table 2. Reduced Ion Mobility and Retention Time of the Analytes Considered as Signs of Life (Some Exemplary References Are Indicated) Detection Limit (ppb) compound

data source

reduced ion mobility, K0 (cm2 V−1 s−1)

retention time, tR (s)

dry sample

humid samplea

2-ethyl-1-hexanol 2,2,4,6,6-pentamethylheptane acetone acetophenone ammonia benzaldehyde cumol (benzene, 1-methylethyl) decanal hexanal (capronaldehyde) limonene octanal pelargonaldehyde (nonanal)

ref 28 ref 30 refs 28, 30 ref 28 ref 30 ref 28 ref 31 ref 27 ref 27 ref 29 ref 27 ref 27

1.441 1.511 2.020 1.724 2.195 1.783 1.664 1.295 1.565 1.684 1.408 1.350

32.5 27.4 2.4 39.3 4.0 18.5 14.6 130.0 7.6 27.4 25.0 55.1

0.05 18 2 0.10 b 0.02 4 0.10 0.30 0.50 0.10 0.30

0.05 10 30 0.02 b 0.02 350c 0.1 10.00 1 0.1 0.3

a c

100% relative humidity (RH). bAccurate calibration was not possible in the present setup. However, the concentrations are in the lower ppbV range. 40% RH (for higher humidity, the LOD is too high to provide reliable concentrations with the available calibration gas generator). C



Article

analytes using the HovaCAL calibration gas generator.26 In general, all relevant analytes are detected in the lower ppbV range, down to the pptV level, with varying effect of humidity. However, accurate quantification of those analytes in a real case may be difficult, since, in many cases, the GC/IMS response is dependent on the humidity of the sample. Therefore, the limits of detection (LODs) are given in Table 2 for dry and for humid (100% relative humidity (RH)) samples, to demonstrate the variance of the values. However, by optimizing the experimental setup, the limits of detection may be improved significantly for a particular or distinct pattern of analytes. Furthermore, in complex and humid samples, peaks may overlap in retention time and, therefore, are competing for the protons provided by the reactant ions, thus further impeding quantification. Therefore, in the following, we do not display the concentration but rather the concentration values normalized to the maximum value during the experiments, since the focus is on the comparison of the slope of those signs of life. Moreover, during real disaster events, the background concentration may vary significantly. Under such conditions, it is more important to detect concentration changes than the accurate absolute value. In the following, we present the values obtained from the ISAS GC/IMS only to avoid negative influence on the correlation to the experiments by the slightly different performance of both ISAS and G.A.S. instruments. However, both instruments obtained comparable results with only minor deviations, most probably due to different experimental setups. Finally, 12 analytes that are already reported to be available in human breath in the literature (see Table 2, some exemplary references are indicated)independent of whether studies compared cancer/noncancer or smoker/nonsmoker groups were chosen to represent a human presence. Furthermore, those analytes have already been detected in human breath during many investigations of the authors’ group. All 12 substances have been observed during all experiments with a concentration slope related to the experiments or to CO2. The selected analytes (see Table 2) are giving the characteristic pattern of signs of life, as presented in Figure 1. The ion mobility and retention time values given in Table 2 cover a relatively wide range and the position of the particular peaks is isolated. Therefore, their accurate quantification is possible, because they do not overlap. The isolated analyte peaks are presented in Figure 1; some of them even formed

dimer ions at the particular concentration level. This was validated not only by calibrating the particular analytes separately but also by measuring them together in a reference mixture. However, in a real case, overlapping of the relevant signs of life with signals from other trace substances available in the ambient air on-site must be expected and must be considered when conclusions are drawn from a measurement on the probability of the presence of a living trapped victim. This must be investigated in the future by applying the proposed method during real disaster events. During the THE, not all 12 analytes could be evaluated properly, due to weak performance of the instrument after recent transportation and high humidity in the void. Therefore, acetone was not evaluated at allthe signal appears in the shoulder of the reactant ion peak, which was broadened during the experiment. Quantification of the ammonia peak was also difficult, because the instrument was contaminated, to a certain degree, with humidity, which overlaps the ammonia signal. Nevertheless, the results obtained during the experiments are compelling, as presented in the following discussion. Figure 2 presents the slope of the normalized concentration of the selected human metabolites during experiments 4−10. At least seven of the selected signs of life detected with the GC/IMS show a clear correlation with the progress of the experiment (see Figure 2A). Increasing concentrations can be observed during the particular experiments, while a rapid drop of concentration can be observed when the volunteers left the void at the end of the particular experiment (the breaks are indicated by the gray bars). The high fluctuations during experiment 4the first one that was investigated completely may be explained by the quite poor performance of the IMS device shortly after transportation. Overall, it can be observed that the slope of the different analytes show different concentration maxima for each experiment. This is due to the individual metabolism of the particular volunteers, including influencing parameters, such as diet, digestion, or personal hygiene. For validation of the conclusions drawn from the detection of particular analytes identified as signs of life, all data were compared to the slope of the CO2 concentration, as presented in Figure 2D. After a steep increase of ∼60 min after the start of a particular experiment, the CO2 concentration reaches a plateau of ∼4000 ppm, which rapidly decreases again after the volunteers have left the void 6 h later. During experiment 8, the gas sensor system including CO2 was disconnected for changing the sampling points, thus causing a temporary decrease of concentration while both GC/IMS were still connected to the void. During experiment 9, the volunteer temporarily left the void to urinate, which can also be detected by a breakdown of concentration in the CO2 profile. For the controlled experiments of the THE, the trapped volunteers were the only source of CO2; therefore, the CO2 slope can be considered as the ultimate sign of life for the simulation setup. In real disaster events, this is not necessarily true, because CO2 could also be derived from combustion or fires. However, all seven analytes clearly follow the slope of CO2 (see Figures 2A and 2D) and, hence, can be considered as reliable signs of life. Figure 2B shows the slope of three additional analytes initially considered as signs of life but with a less clear correlation to CO2. This might be due to the performance of the instrument as already mentioned previously for ammonia and acetone. However, they are verified human metabolites

Figure 1. Signs of life pattern (12 human metabolites) as detected using ion mobility spectrometry coupled with gas chromatography (GC/IMS) for the localization of trapped victims. Some of the analytes even formed dimer ions at the particular concentration level. D



Article

Figure 2. Slope of the human metabolites selected as signs of life, as detected during the “Trapped Human Experiment” (THE) with GC/IMS (A, B, C), and the slope of CO2 concentration, together with the normalized summation of the signs of life (D). The particular experiments and the breaks are indicated.

and, therefore, will be included in the pattern of signs of life in the following.

The slope of hexanal, as presented in Figure 2C, shows a completely different behavior than the other analytes E



Article

Figure 3. General slope of concentration for a low initial concentration combined with a decreasing (Case I) and an increasing emission rate (Case II).

Figure 4. The increase of the summation parameter, interpreted as a “sign of life” during the validation experiment (bold black line), is in excellent agreement with the presence of the volunteer (bold gray line), including two occasions, at 14:50 and 15:40, when the volunteer briefly left the room.

mentioned above. Instead of a steep increase as observed previously, the slope starts with a very slight increase with time. Under the controlled condition of the human experiment, a modeling of the slopes can be carried out quite simply. The air in the void was exchanged with a continuous flow of fresh air, for the safety of the volunteers. As a consequence, the development of the concentration in the void depends on the temporal development of the analyte emission rate or, in our case, the exhalation of the entrapped volunteers, respectively. Two different cases presented in Figure 3 show the general behavior of the concentration slope when starting with a low value for a decreasing emission rate (Case I) and an increasing emission rate (Case II). For example, for CO2, it can be expected that the volunteer will relax after a while in the void some of them even fell asleepthus reducing the turnover of inhaled and exhaled air, followed by a decreasing emission rate. This is obviously true for all of the selected metabolites, except

for hexanal. In this case, the emission rates evidently increased during the particular experiments. For many years, hexanal has been known as a metabolite of lipid peroxidase and is related closely to oxidative stress.32,33 An intensification of this process caused by the special situation of the volunteer could explain the significant concentration slope during all particular experiments. For a realistic application of the proposed pattern of signs of life for the search for victims in real disaster events, an evaluation of each particular analyte included in the pattern is not suitable. Therefore, a bulk parameter must be used as output of such measurements, giving the information to the USaR teams if there is a living victim in an investigated void or at least to what probability this could be the case. In a first and simple approach, we used the mean value of the normalized concentrations for this purpose. Figure 2D shows the slope of this normalized summation parameter, together with the CO2 F



Article

algorithm be developed. This does not necessarily have to be the presented normalized summation parameter: the different analytes under consideration may have to be weighted. Potentially, the result given from such an analysis must be transferred to a probability of the presence of a living victim. However, after these steps, the application of GC/IMS will establish a significantly helpful tool for search-and-rescue operations, in addition to trained canines. Furthermore, since the potential of IMS for medical diagnosis was recently demonstrated in the literature, the instrument used for localization of trapped victims might be used afterward for an instant overview of the health status of the victim by breath analysis.

concentration. It is evident that the presence of each individual volunteer in the void of the THE would have been recognized at least after ∼30 min from the increase of this parameter. For example, using a threshold value of 0.4, all volunteers would have been identified in the void without false alarms. Hence, in real cases, an increase in concentration, compared to the background level, will be the measure for an alarm. However, the background concentration levels will undergo significant changes, especially when switching from a controlled laboratory entrapment simulation to real disaster events. Therefore, we carried out first validation experiments at ISAS, which is still not realistic but independent from the THE, with respect to the void, the volunteers, and the equipment. The room air was monitored during the day and the pattern of signs of life was supposed to indicate when somebody was in the room with a volume of ∼25 m3 for a while. Again, the room air was drawn through the sample loop of the IMS (8 mL) to avoid contamination and then was introduced into the multicapillary column for preseparation. Indeed, after closing the door, a significant increase of the particular analytes included in the signs of life pattern could be observed (see Figure 4); some of them are indicated exemplarily as fine lines. The summation parameter, which is calculated as the mean value of the normalized concentration of all selected analytes, perfectly describes the presence of the volunteer. Even two periods, when the person left the room for a few minutes, are mirrored by the slope of the summation parameter at 14:50 and 15:40.

■

ASSOCIATED CONTENT

S Supporting Information *

This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: ++49-231-1392169. Fax: ++49-231-1392120. E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

■

CONCLUSIONS From the simulations of the “Trapped Human Experiment” (THE) and an additional independent validation experiment, it could be demonstrated that, when using ion mobility spectrometry coupled with gas chromatography (GC/IMS), relevant human metabolites could be detected with sufficient sensitivity and selectivity to indicate the presence of living trapped persons for the first time. Altogether, 12 analytes, which have already been verified as human metabolites in the literature, could be used for the identification of trapped humans after pattern recognition in the GC/IMS dataset onsite and immediately after the analysis of a sample drawn from a void. A short-term increase in the concentration of a normalized summation parameterin a real case, this is determined by comparison of background concentrations with those measured in a voidenables the identification of living trapped humans. Because of the rapid preseparation when using multicapillary columns, the analysis time with the present experimental setup is

Detection of Metabolites of Trapped Humans Using Ion Mobility

Recommend Documents