Detection of Δ9 THC in oral fluid following vaporized cannabis with varied cannabidiol (CBD) content: An evaluation of two point‐of‐collection testing devices

Abstract Point‐of‐collection testing (POCT) for Δ9‐tetrahydrocannabinol (THC) in oral fluid is increasingly used to detect driving under the influence of cannabis (DUIC). However, previous studies have questioned the reliability and accuracy of two commonly used POCT devices, the Securetec DrugWipe® 5 s (DW5s) and Dräger DrugTest® 5000 (DT5000). In the current placebo controlled, double‐blind, crossover study we used liquid chromatography‐tandem mass spectrometry (LC–MS/MS) to accurately quantify cannabinoid concentrations in the oral fluid of 14 participants at various timepoints (10, 60, 120, and 180 minutes) following vaporization of 125 mg of THC‐dominant (11% THC; <1% CBD), THC/CBD equivalent (11% THC; 11% CBD) and placebo (<1% THC; <1% CBD) cannabis. At each timepoint, oral fluid was also screened using the DW5s (10 ng/mL THC cut‐off) and DT5000 (10 ng/mL THC cut‐off). LC–MS/MS analysis showed peak oral fluid THC concentrations at the 10 minute timepoint with a rapid decline thereafter. This trajectory did not differ with THC dominant and THC/CBD equivalent cannabis. With a 10 ng/mL confirmatory cut‐off, 5% of DW5s test results were false positives and 16% false negatives. For the DT5000, 10% of test results were false positives and 9% false negatives. Neither the DW5s nor the DT5000 demonstrated the recommended >80% sensitivity, specificity and accuracy. Accuracy was lowest at 60 minutes, when THC concentrations were often close to the screening cut‐off (10 ng/mL). POCT devices can be useful tools in detecting recent cannabis use; however, limitations should be noted, and confirmatory LC–MS/MS quantification of results is strongly advisable.


| INTRODUCTION
The ongoing amendment of medicinal and recreational cannabis laws worldwide has made driving under the influence of cannabis (DUIC) a key public safety concern. 1,2 There are two main approaches that are used to assess DUIC. The first is an effect-based approach whereby a police officer or drug recognition expert (DRE) must demonstrate behavioral impairment. Although widely used, there are concerns around the effectiveness of this approach and cases can be difficult to prosecute. 3,4 Many jurisdictions therefore enforce per se or zero tolerance policies for DUIC. Under such laws, a driver has committed an offence if delta-9-tetrahydrocannabinol (THC) is detected at or above a given concentration in a specified biological matrix, irrespective of actual impairment. Oral fluid is now increasingly relied upon as a matrix for DUIC detection as samples can be readily obtained in a non-invasive manner and rapidly analyzed at the roadside using point-of-collection testing (POCT) devices.
POCT devices are used by authorities to detect DUIC in a number of countries including Norway, 5 Germany, 6 Belgium, 7 and Australia, where this process is referred to as mobile drug testing (MDT). First introduced in 2004 in the state of Victoria, Australia, 8 MDT has since been adopted by all Australian states and territories. In the state of New South Wales (NSW), Australia, the procedure involves an initial test for oral fluid THC using the Securetec DrugWipe ® (DW) device.
If positive, this is followed by a secondary test using the Dräger DrugTest ® 5000 (DT5000) device. If the DT5000 test is also positive, confirmatory analysis is conducted to confirm the presence of THC.
Authorities have not revealed the THC screening cut-offs used for MDT: the DT5000 screening cut-off can be set by the operator to 5, 10, or 25 ng/mL, and the DW device variant which is used by NSW Police is not commercially available.
The primary aim of MDT is to improve road safety by detecting and therefore deterring DUIC. It is therefore essential that POCT devices accurately discriminate between drug-positive and drugnegative cases. False positives may lead to unjust punishment for drivers (eg, license disqualification, criminal conviction), while false negatives undermine the aims and integrity of the MDT program. To assess the real-world performance of the DW5s and DT5000 devices, controlled laboratory studies that compare POCT device results with confirmatory analysis using sensitive analytical methods are crucial.
Vaporization of cannabis is an increasingly common route of administration among both medicinal and recreational cannabis users; [21][22][23] however, only a small number of studies have described oral fluid concentrations 24,25 and POCT device performance 20 following vaporized cannabis. Moreover, these studies have been limited to THC-dominant cannabis. Cannabis chemovars ('strains') and medicinal cannabis products often contain significant levels of cannabidiol (CBD), a nonintoxicating cannabinoid with anxiolytic, anticonvulsant, and antipsychotic properties. [26][27][28] For example, the 'light cannabis' products that are legally available in a number of EU countries must contain less than 0.2% THC but may contain up to 40% CBD. 29 Medicinal cannabis products containing both THC and CBD include Nabiximols (Sativex), a buccal spray with a 1:1 ratio of THC and CBD, as well as commercially available cannabis botanicals and extracts 30 and homegrown illicit artisanal preparations. 31 It is currently unclear whether CBD content might influence the performance of POCT devices or influence the underlying pharmacokinetics of THC in oral fluid.
The current study therefore sought to evaluate the performance of the DW5s and DT5000 POCT devices relative to liquid chromatogra-

| Participants
Healthy adults (aged 18-65 years) with a history of infrequent cannabis use were recruited for this study. Inclusion criteria were selfreported cannabis consumption ≤2 times/week in the previous three months and ≥ 10 lifetime exposures (see Table 1 for details).
Exclusion criteria included current mood disorder; lifetime major psychiatric illness; history of clinically significant adverse response to previous cannabis exposure; any moderate or severe substance use disorder as assessed by an addiction medicine specialist; pregnant/nursing; interest in treatment to reduce cannabis use; current use of medications known to affect driving; active hypertension, cardiovascular disease, or chronic pulmonary disease. Volunteers were recruited through online advertisement, social media (e.g., Facebook) and word of mouth. All participants meeting inclusion/exclusion criteria underwent a comprehensive medical and psychiatric evaluation and provided written informed consent prior to study enrolment.

| Study design and procedures
This randomized, placebocontrolled, within-subjects, double-blind, crossover study included three experimental sessions that were scheduled at least seven days apart to avoid carryover effects. Participants were instructed to abstain from illicit drugs for the duration of the study (i.e., from the time of study enrolment until the final session) and from alcohol on the night before research sessions, to maintain any use of regular medications, and to consume no more than their regular caf-

| Oral fluid collection and POCT procedures
Oral fluid samples were collected using Quantisal™ collection devices Results are available after 8 minutes (negative, non-negative, or invalid) and can be printed using an attached printer. Test results where the indicator line did not turn blue were excluded. Test results for both devices were read and filed by an independent observer and only made available to the researchers upon completion of the study. Chromatographic separation was achieved using an Eclipse XDB-C18 column (50 mm x 2.1 mm i.d., particle size 3.5 μm; Agilent Technologies, Singapore) using gradient elution with mobile phases 0.1% formic acid in water and acetonitrile, at a flow rate of 0.3 mL/min. This was coupled to a Shimadzu LCMS-8030 mass spectrometer for analyte identification and quantification.

| Oral fluid analysis via LC-MS/MS
The LC-MS/MS analysis was validated for selectivity, linearity, accuracy, precision, bench-top and autosampler stability, dilution integrity, limit of detection (LOD), and limit of quantification (LOQ) ( Table 2), following Food and Drug Administration (FDA) validation guidelines. 33 Selectivity was verified by analyzing cannabinoid-free saliva samples for interferences. Linearity was assessed using calibrators at seven ascending concentration levels. Intra-assay accuracy and precision were determined using six replicate quality control (QC) samples at low, medium, and high concentrations relative to the concentration range on the same day. Inter-assay accuracy and precision were determined using similar QC samples three different days (three replicates per day). Repeat injections at 0-, 4-, and 8-hour timepoints were used to assess autosampler stability. Dilution integrity was assessed for 10x dilutions. The lower limit of quantification (LLOQ) was selected based on accuracy of calibrator samples (lowest calibrator within ±20% of the nominal value), while the LOD was set as the lowest calibrator concentration with signal-to-noise greater than 3. Samples that fell above the linear quantification range were diluted appropriately and re-analyzed.

| LC-MS/MS method
The LC-MS/MS method was accurate, precise, and had LODs of 1 ng/mL for both THC and CBD, and LLOQs of 2 and 6 ng/mL for THC and CBD, respectively. Although some matrix effect was apparent, this was accounted for with the use of deuterated internal standards for both analytes. We also verified that other common phytocannabinoids that could also be present in saliva (THCA, THCV, CBN, CBDA, CBG, CBGA, and CBC) were chromatographically separated and did not interfere with CBD or THC quantification (data not shown).   Medium QC precision (%RSD) 9.2 7.5 LOD = limit of detection. LLOQ = lower limit of quantification. N.B. For accuracy and precision, low = 10 ng/mL, medium = 100 ng/mL, and high = 400 ng/mL. were no significant differences between the THC and THC/CBD conditions at any timepoint.

| Oral fluid cannabinoid concentrations
Oral fluid CBD concentrations differed significantly between con-     TN, FP, FN) for the DW5s and DT5000 and overall device performance (sensitivity, specificity, and accuracy) at a 10 ng/mL confirmatory cut-off, while Table 5 describes these parameters when a 2 ng/mL and 1 ng/mL confirmatory cut-offs are applied. Figure 2 shows the LC-MS/MS quantified THC concentration corresponding to each test result. both false positives and false negatives was greatest at the 60-minute timepoint. As Table 5 shows, fewer false positives and more false negatives were observed with confirmatory cut-offs of 2 ng/mL and 1 ng/mL. Overall accuracy was greatest with a 10 ng/mL confirmatory cut-off applied.

| DrugTest 5000
A total of 163 DT5000 test results involving four different time points were evaluated relative to LC-MS/MS verified oral fluid THC concentrations. At a 10 ng/mL confirmatory cut-off (Table 4), overall sensitivity, specificity, and accuracy were calculated as 67%, 86%, and 80%.
Of the 47 test results that were positive, 17 false positives were detected with corresponding oral fluid THC concentrations ranging from 0 to 6.4 ng/mL. Of the 116 test results that were negative, 15 false negatives were detected, with corresponding oral fluid THC concentrations ranging from 10.1 to 203 ng/mL. As with the DW5s, the incidence of false positives and false negatives were greatest at the 60-minute timepoint. Applying a confirmatory cut-off of 2 ng/mL or 1 ng/mL decreased the number of false positives but substantially increased the number of false negatives (Table 5). Overall accuracy was highest with a 10 ng/mL confirmatory cut-off.

| DISCUSSION
The present study was designed to provide insights into the accuracy and reliability of two commonly used POCT devices. We assessed the performance of the DW5s and DT5000 devices by comparing observed test results against confirmatory LC-MS/MS quantified oral fluid THC and CBD concentrations at various timepoints following controlled laboratory vaporization of three different cannabis types (placebo, THC-dominant, and THC/CBD-equivalent) using a withinsubjects, crossover design.
Overall, our data confirm that oral fluid THC is a good indicator of very recent cannabis use. 20,24,[34][35][36] As with previous studies, 24,25,34,36,37 oral fluid cannabinoid concentrations were maximal at the time point closest to vaporization (10 minutes) and declined rapidly thereafter.
Sensitivity was highest at 10 minutes, when the incidence of positive test results was highest. Accuracy was highest for both devices at 180 minutes, when all samples were found to be true negatives, but this is something of an artefact given that sensitivity at this time point could not be computed. Accuracy was highest with a confirmatory cut-off of 10 ng/mL, and decreased progressively with lower cut-offs due to the substantial increase in false negatives.
These data are consistent with previous studies which have collectively reported DT5000 sensitivity, specificity and accuracy as 49.5%-100%, 55%-90%, and 55%-86.4%. 5,9,14,[16][17][18][19][20] One controlled laboratory study 20 reported considerably better DT5000 performance (>80% specificity, sensitivity, and accuracy) than we observed here. In that study, however, samples were collected up to 72 hours after cannabis administration and therefore included a much larger proportion of samples with very low THC concentrations that were classified as true negatives. Another study 19 reported DT5000 specificity and accuracy similar to that observed here, but with greater sensitivity than reported here (92.7% vs 67%). This discrepancy is due to the difference in the While the short detection window for THC is a key benefit of the POCT approach, the erratic distribution of THC in oral fluid, and the magnitude of intra-and inter-individual variability following standardized cannabis administration, may preclude its use as a meaningful marker of acute intoxication or impairment. We recently demonstrated impaired driving performance and reduced confidence in driving ability in these same 14 participants at both 30 and 210 minutes following vaporization. 32 However, at 180 minutes, there were no true positive test results in the present study and THC concentrations were typically below the LOQ. Consistent with this, a controlled laboratory study by Ramaekers et al 42 found only a weak relationship between oral fluid THC concentrations and magnitude of impairment on a range of driving-related cognitive tasks following smoked cannabis. Moreover, oral fluid THC concentrations can exceed 10 ng/mL (ie, the DW5s and DT5000 screening cut-offs) following passive exposure to cannabis smoke 43,44 or consumption of high CBD cannabis with negligible THC content. 29 It is therefore possible for an individual who has not actually consumed cannabis to test positive for cannabis with the two POCT devices examined here. These findings offer some support to concerns around the validity of the MDT program. 45 Given the widespread and increasing use of POCT as a method for detecting DUIC, it is essential that these limitations are considered.

| CONCLUSIONS
Here we have compared, for the first time, cannabinoid concentra-