Fast Detection of Recent Cannabis Sativa L. Consumption in Exhaled Breath Us- ing a Mobile Ion Mobility Spectrometer

The volatile metabolite pattern in exhaled breath after smoking Cannabis sativa L. was determined by ion mobility spectrometry coupled to gas-chromatographic pre-separation (GC-IMS). The aim was to identify typical biomarkers indicating a recent consumption of herbal Cannabis sativa L. In parallel, blood samples were taken and THC, 11-Hydroxyand 11-Carboxy-THC were analyzed by GC-MS. A high number of volatile constituents – most probably metabolized from the essential oil but not identified as cannabinoids – were detected in the breath. A significant correlation between blood and exhaled breath concentration of THC could not be observed. However, Cannabis sativa L. smoke contains distinct essential oil constituents which may act as indicative biomarkers providing proof for smoking herbal Cannabis sativa L. A characteristic pattern of compounds could be observed using GC-IMS over a detection period of up to 4h after last consumption. Method development was carried out successfully leading to a rapid total analysis time of 90 seconds combined with the high userfriendliness of the mobile equipment. In this manuscript, we present the scientific background, the technical implementation and the real-world operation of the commercial prototype.


Introduction
Cannabis (Cannabis sativa L) is a psychotropic plant that is frequently used by smoking or oral ingestion for relaxation. Cannabis sativa is the most reported illicit drug identified in impaired drivers in Europe and the US [1]. Driving under the influence of Cannabis causes severe accidents. Nearly 60% of injured drivers with trauma are tested positive for drugs or alcohol in the US. According to the annual report of the US National Driving Board [2], 11.4% of Americans and 10% of Canadians [3] drove in 2010 and 2012 under the influence of illicit drugs.
Herbal Cannabis is the most consumed illicit drug worldwide [4] and a strong increase in the number of future drug users can be expected with the on-going legalization. Thus, driving under the influence of Cannabis is an urgent and growing health and traffic security concern. Epidemiologic data [2,3] show that the risk of involvement in a motor vehicle accident is twofold higher after consuming the drug. The primary reason being the dosedependent effect and impairment of cognitive and psychomotoric functions [5,6]. As a consequence, drivers under the influence of Cannabis show a higher risk of collision and severe accidents combined with impaired driving skills.
Currently, no valid, rapid, and mobile screening methods are available which fulfil the needs as we know for alcohol checks. Police worldwide use detection devices like wipe tests for sweat or urine tests, based on thin-layer chromatography, which is easy to handle but show low sensitivity, hygienic discomfort and significant error rates in practical use [7][8][9]. Since cannabis is legalized in many countries for medicinal and recreational use, police are more confronted with drivers of all ages consuming cannabis on a regular basis. Driving under the influence of any drug is still illegal worldwide and countries are starting the debate about thresholds in the range of 1-7 ng/mL of THC in blood [10]. Rapid test devices will not meet this limit precisely due to the nature of the chemical reaction as a principle proof.
Consequently, there is a pressing need for a rapid field test that will enable law enforcement agents, customs officers, and other responsible personnel to obtain presumptive evidence on suspected sample identity.
Today, in 29 countries cannabis is legal for medicinal use and moreover, in Spain, Uruguay and some US states recreational use is legal as well. Similar to other medicinal plants, Cannabis is a phytochemical multi-component mixture with more than 120 cannabinoids including, tetrahydrocannabinolic acid, cannabidiolic acid, cannabigerol acid and cannabichromene acid being known as the most dominant constituents [11]. Over 500 metabolites have already been identified in this species, includ-ing terpenes, sugars, hydrocarbons, steroids, flavonoids, nitrogenous compounds, non-cannabinoid phenols and amino acids [11,12]. Moreover, it is clear that besides the dominant THC other constituents have been found to be medicinally active [13].
Cannabis Sativa L. is a typical aromatic plant with a significant concentration of essential oil (approx. 0.2-0.3%). These, mostly terpenophilic constituents and related human metabolites after application, are volatile and exhaled via the natural breath. Many of the constituents can be detected in the exhaled air which is a boon and a burden at the same time because essential oils consist of 300 up to 1000 compounds [14]. This is why analytics and distinct compound detection and elucidation is complicated. In Cannabis sativa L., about 90 primary constituents in the essential oil are known [15]. But, how to overcome the irresolvable question of compound identification or the best strategy for pattern recognition in the exhaled breath air is one of the key questions in this study.
Various methods for the determination of cannabinoids in plant material have been developed and are summarised by Klein [16]. In the past decades, thin layer chromatography (TLC) was used for cannabinoids separation, combined with colorimetric tests or UV detection under 254 nm for their identification [17,18]. Nowadays, commonly applied techniques are HPLC and GC, and very often coupled with tandem with the mass spectrometric detector. But so far, no one of these sophisticated techniques is mobile and can be used in the field, which would be an outstanding technical advantage.
Ion mobility spectrometry [19] is a powerful technique that fasts, sensitive and mobile and therefore, we investigated its potential concerning detection of drug consumption.
The method is based on ion generation and separation of ions in the gaseous phase under an electric field [20,21]. This technique separates ions in an electric field by size and shape due to collisions with a drift gas on their way to a detector. IMS is a very powerful and extremely sensitive technique to detect volatile molecules at ppt v concentration ranges. In combination with gas-chromatographic pre-separation (GC-), extremely complex mixtures such as exhaled breath can be comprehensively analyzed quantitatively. IMS and GC-IMS instruments are currently used in various technical fields like explosive detection at airport security checks [22], chemical warfare agents [23], food control [24], process control [25] and biomedical analysis [26][27][28], illicit drugs [29], and forensics [30].
This study aimed to investigate if cannabinoids and other characteristic volatile compounds of the psychotropic plant can be identified in exhaled breath by GC-IMS and to find a cor-relation between the blood concentrations and detected volatile THC or other relevant biomarkers. Therefore, additional blood sampling was conducted and analyzed for the presence of THC, 11-Hydroxy-and 11-Carboxy-THC.

Plant Material
Cannabis sativa L. variety Bedrocan ® was supplied by Bedrocan BV (The Netherlands). The plants were grown indoors under standardized conditions as explained in a previous report [31]. The saplings were initially generated from cuttings of stand-

Chemicals
All chemicals were obtained from Sigma-Aldrich in the highest available purity.

GC-IMS -gas-chromatography coupled to ion mobility spectrometry
A custom-made ion mobility spectrometer coupled to a multi-capillary gas-chromatographic column (GC-IMS, ISAS, Dortmund, Germany) was applied for the present investigations [26]. Ion mobility spectrometry is based on the characteristic mobility of ions in an electric field under a counter-current of a neutral drift gas [19]. The ions collide with the drift gas molecules, thus being separated based on size and shape. Exhaled breath sampling was carried out by exhalation through a mouthpiece with a flow sensor (differential pressure measurement). If the volunteer exhales, the sample flow is drawn from the main exhalation stream at a rate of 100 ml/min [32].

GC-MS Analysis
Qualitative

Ethics & legal situation
All volunteers inhaling herbal Cannabis sativa L. flos were legal users due to medical conditions. They gave written in-

Results and Discussion
The analysis of blood samples by GC-MS showed that the volunteers were not under the influence of cannabis at the beginning of the test. Maximum THC concentrations were reached shortly after consumption; 8 ng/mLand 10 ng/mL for the two volunteers and were detectable up to 5 hours post consumption.
The analysis of the headspace of herbal material found that different brands resulted in slightly different GC-IMS chromatograms (see Figure 1). The differences are not only in the overall intensity but also in the relation of the different signal intensities. However, most of the detected signals could be found in the headspace analysis data of all cannabis samples. In the current study, only medical cannabis was used by the volunteers due to the restrictions of the ethics committee and considering the legal situation in Germany. But in view of the similarities in the peak pattern from different types and brands of Cannabis, similar results from breath analyses can be expected as well.  In comparison with reference compounds, no THC or related human metabolites were detected. However, the GC-IMS data obtained showed clear peaks related to the uptake of essential oil constituents which were present even after 240 min p.a.
( Figure 2). Before administration of dried herbal Cannabis sativa L. flost he headspace and natural breath of both volunteers were recorded as a positive control for further in vivo methods.
Besides the many unknown peaks, a characteristic pattern of 3 peaks -peak 1, 2 and 4 as also detected in the headspace of the cannabis samples (see Figure 1) was indicative for smoking herbal medicinal cannabis and was significant over 240 min.
The slope of one of the relevant signals over 4 hours after smoking is presented in Figure 3. Other peaks from the pattern in Figure 1 was also detected in the first breath analysis, 10 min after administration, but disappeared over time.  IMS can be considered as a highly sensitive analytical technique for the detection of cannabis biomarkers after smoking, even if THC as an illicit compound cannot be detected directly. This is critical for drug enforcement because a direct proof is missing, but using 12 discriminating peaks from the recorded spectrum, a highly positive indication is obtainable.
Identification of some of the peaks is ongoing, but the presence of e.g. caryophyllene oxide as one peak is proven and it is important as a biomarker (Figure 4). Due to the presence of caryophyllene oxide in other plants or food products (e.g. carrots or hop) false-positive proofs are, in principle, possible. In a second experiment, V1 ate 1 kg fresh carrots and besides caryophyllene oxide (Figure 4) no significant peak pattern as found from cannabis consumption was detected. In our ongoing study, both volunteers ate, over three days, hemp-based food products like pasta, sauces, cannabis oil, sweets, etc, but no cannabinoids were detected by IMS before and after food uptake. No direct proof of THC was possible, but in headspace studies with a methanolic THC solution (5 mg/mL) the illicit drug was detected at concentrations of at least 1 ng/mL and below (data not presented). These data suggest that failure detection is not due to the malfunction of the IMS detector itself. A major reason for not detecting THC is the fact, that this compound is not volatile at the body temperature of 37 °C. Therefore, it can be detected in breath only in the first minutes after smoking, being still present as a product of the cannabis pyrolysis.
Furthermore, fast clearance in the body with a distribution time of 8-10 min and half lifetime for THC of T 50 =30 min gives an initial explanation as to why detection after 1 hour of smoking is critical and not easy [33]. Most of THC uptake (150 mg/cigarette) is fully absorbed and will not diffuse from the blood into the respiratory air. Currently used on-site rapid drug screening tests like wipe tests or urine tests have a detection limit of 2ng/mL. It must also be pointed out that there is a high false-positive/false negative error rate of on-site urine, oral fluid or sweat tests (10-15%). This is due to a failed detection at higher concentrations above 50 ng/mL [31]. Kintz et al. 2005 stated in their technical note [34] that some drug wipe on-site test is critical at high concentration and they are not safe at low It has to be kept in mind, that due to the restrictions from the ethics committee vote, only medical cannabis could be applied to the volunteers for the experiments described in this study. However, due to the fact, that the pattern detected in the headspace of medical and "street" cannabis is very similar to differences in the signal intensities of certain peaks only, it can be expected, that this will also result in similar patterns in exhaled breath after consumption. Certainly, further studies with more volunteers and different types of cannabis -e.g. in countries where medical and recreational consumption is legal -have to be carried out for validation of the method presented here.
Furthermore, other pathways for consuming Cannabis such as eating cannabis cookies should also be investigated in the future as a significantly different metabolic process will influence the possible detection in exhaled breath.
The instrument used includes an internal circuit for the operational gas and a battery, thus allowing autonomous operation for up to 4 hours (see Figure 5). This mobile GC-IMS named ION drug is currently commercially available. The operation does not require any special training. The user-interface informs about the instruments' status, e.g. for being ready for operation after a short warm-up phase. After the start of an analysis run via the touch screen of the instrument, the user is then guided through the procedure of controlled sampling, analysis after successful sampling and -if sampling was not successful -on a repetition of the procedure (see Figure 6). After the complete analysis, which takes 90 seconds, the operator is then given the information if recent cannabis consumption was determined or not (see Figure 6) using the analogy of a traffic light. However, the comprehensive raw data are stored internally for a possible detailed examination later on.