Professional Documents
Culture Documents
h i g h l i g h t s g r a p h i c a l a b s t r a c t
a r t i c l e i n f o a b s t r a c t
Article history: In response to the Canadian federal government's Cannabis Tracking and Licensing System compliance
Received 13 June 2019 standards, a quantitative method was created for cannabis analysis, and validated using Eurachem V.2
Received in revised form (2014) guidelines. Cannabinol, cannabidiol, cannabigerol, cannabichromene, cannabidiolic acid, canna-
12 August 2019
bigerolic acid, D-9-tetrahydrocannabinol, and D-9-tetrahydrocannabinolic acid A were all analysed by
Accepted 19 August 2019
Available online 20 August 2019
scheduled multiple reaction monitoring (MRM) via LC-MS/MS and isotope dilution. In addition, afla-
toxins B1, B2, G1, and G2 were also analysed by scheduled MRM via LC-MS/MS and matrix matched
calibration curves in order to achieve the reporting limits (2 mg kg1) set out by the European Phar-
Keywords:
Cannabis
macopoeia. The LODs/LOQs were 0.50/1.7, 2.0/6.7, 0.59/2.0, and 0.53/1.8 mg kg1, for B1, B2, G1, and G2
Terpenes respectively. Thirty one terpenes were analysed by selected reaction monitoring via GC-MS/MS and
Aflatoxins isotope dilution using b-myrcene-d6 as a surrogate. All quantitative analyses can be accomplished using
Quantitative method less than 1 g of material, with minimal solvent and consumable use, on low resolution instruments in less
Chromatography than 30 min of instrument time. Of important note is this method's power of selectivity, working ranges,
Mass spectrometry
* Corresponding author.
E-mail address: Gregg.tomy@umanitoba.ca (G. Tomy).
https://doi.org/10.1016/j.aca.2019.08.042
0003-2670/© 2019 Elsevier B.V. All rights reserved.
80 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88
and lack of need for extraction consumables such as SPE or QuEChERS, thereby minimising analytical
costs and time.
© 2019 Elsevier B.V. All rights reserved.
1. Introduction given the potential for variances in instrument response [18]. Thus,
a scheduled (or dynamic) multiple reaction monitoring (MRM)
The United States federal government currently categorises approach that uses a specific time window spanning calculated
cannabis as a Schedule 1 illegal drug under the Controlled Sub- retention times is essential to isolate ion transitions that are diag-
stances Act (1970) and thus there are no federal guidelines for nostic of the target cannabinoid.
contaminant restriction and analysis of the medicinal phyto- Cannabis is susceptible to a number of fungal endophytes and
chemicals in cannabis products [1]. However, since cannabis has pathogens, including Aspergillus species that produce the potent
become recognised as a valuable commercial crop, legislation class of carcinogens known as the mycotoxins. Health Canada and
should be forthcoming and standardised analytical testing will the federal Cannabis Act (2018) do not specifically outline regula-
become mandatory. In October 2018, the federal government of tory limits for the types or levels of mycotoxins in cannabis prod-
Canada legalised the recreational use of cannabis, with regulations ucts; however it is recommended that in the pursuit of complete
stipulating strict production safety and consumer protection regulatory compliance that a standardised benchmark (e.g. the
guidelines under the Cannabis Act and Regulations (2018), which in European Pharmacopoeia) be followed comprehensively. Specif-
turn are facilitated through the Cannabis Tracking and Licensing ically, the Canadian Food and Drug Act states that it is up to the
System [2]. For federally-regulated licenced producers of cannabis discretion of the licensed producer to decide which publication
to remain compliant with these regulations, it is essential to listed within Schedule B it should adhere to for analytical testing of
develop quantitative analytical methods to easily and accurately mycotoxins in cannabis products. However, any document chosen
quantify both the medicinal phytochemicals produced by this plant must be fully adhered to in its entirety in the Food and Drug Act
(collectively termed the cannabinoids), and potential fungal toxin [19]. Contaminant challenges are non-trivial with a plant such as
contaminants in cannabis products [2]. Previous analytical cannabis, which has the potential to present microbial contami-
methods were used to characterise medical cannabis for years prior nation in addition to organics [5]. Analyses for fungal mycotoxins,
to legalisation [3]. Cumulatively, the total suite of analytes neces- pesticides, and metals are necessary for Canadian federal compli-
sary for Canadian regulatory compliance at the federal level is ance [2]. Mycotoxins such as aflatoxins are typically found in the
cannabinoids, fungal toxins, pesticide residues, and metals content. high pg mL1 to low ng mL1 range; with EU (EU 2006) regulatory
However, there presently exists fundamental analytical challenges, levels of aflatoxins B1 of 2e12 mg kg1, and combined total B1, B2,
as the methods used were unable to unequivocally address G1, and G2 of 4e15 mg kg1 [20]. Thus, effective sample extraction
cannabis potency (due to identical molecular masses of many of the methods are necessary to isolate the target analytes, to minimise
cannabinoids), insufficient reporting limits for fungal contaminants background contamination, to maximise sensitivity of mass anal-
[4,5] and poor chromatography for many of the terpenes [6]. ysis, and to minimise LC ionisation matrix effects from this complex
Cannabis produces the cannabinoids as acidic molecules, and sample material. Pesticides (beyond the scope of this study) pose a
quantitatively the major cannabinoids include D-9 tetrahy- difficult challenge as there are currently 95 different pesticides that
drocannabinolic acid (THCA), cannabidiolic acid (CBDA), cannabi- require analysis as part of the regulatory control of cannabis [4],
gerolic acid (CBGA), and cannabichromenic acid (CBCA) [7]. Upon some of which are not analysable by LC-MS/MS due to inefficient
heating, these molecules spontaneously decarboxylate to the ionisation [21].
‘neutral’ cannabinoids D-9 tetrahydrocannabinol (THC), cannabi- Terpenes are volatile aromatic compounds that are responsible
diol (CBD), cannabigerol (CBG), and cannabichromene (CBC). The for the flavour profile of cannabis, and have also been implicated in
ratio of the acidic cannabinoids varies amongst different cannabis the medicinal effects of cannabis either explicitly (e.g. anti-
varieties, with THCA being more abundant in marijuana cannabis convulsant, anti-inflammatory, anti-depressant, anticancer) [22],
varieties and CBDA being more abundant in hemp cannabis vari- or in concert with cannabinoids (e.g. increasing the biological/
eties [8]. Dried flowers from all cannabis plants typically contain physiological effects) which is commonly referred to as the
these cannabinoids in the mass percent range [9]. ‘entourage effect’ [23]. A major analytical challenge is the co-
Cannabinoids were typically quantified using liquid chroma- elution of various unknown terpenes along with targeted com-
tography (LC) in conjunction with either UV-DAD [7,10e13] and MS pounds given the presence of hundreds of ancillary components
[14e17], and/or by gas chromatography (GC-MS) [6,8,10]. However, found in the cannabis plant (e.g. terpenes, flavonoids, fatty acids,
it became evident that potency levels could be problematic as a phenols, pigments) [9]. Thus, it is important from a quantitative
consequence of inaccuracies due to the isomeric nature of many standpoint to know what proportion of the identified MS signal is
cannabinoids. The acidic cannabinoids THCA, CBDA, and CBCA all due to the targeted terpene of interest, and the possible qualitative
have an identical molecular mass of 358.5 g mol1, whereas the elucidation of new/unidentified terpenes within cannabis strains
neutral cannabinoids THC, CBD, and CBC have an identical molec- [6]. Additionally, the lack of available commercial labelled stan-
ular mass of 315.3 g mol1. Thus, these molecules possess similar dards for terpenes makes unambiguous quantitation problematic,
LC-electrospray positive ion mass spectra. Furthermore, these in that isotopically labelled surrogates become necessary, thereby
cannabinoids are only a few of >90 that have been identified in possibly affecting values due to unequal responses during mass
cannabis [9]. The heat-labile nature of the acidic cannabinoids analysis (i.e. LC-ionisation matrix effects).
highlights the importance of using a non-destructive, low-tem- With all of these analytical challenges taken into consideration
perature method in order to achieve the necessary quantitative and using the methods contained herein, the objective of the pre-
accuracy of these molecules. Moreover, careful thought regarding sent study is to provide a validated quantitative analytical method
internal standards with which to quantify also becomes important for cannabinoids, aflatoxins, and terpenes from an ISO 17 025
A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88 81
accredited laboratory perspective to address major regulatory coat and submerge the bud with solvent, then sonicated for 5 min.
[1,2,4], biological [3,22,23], and environmental concerns [24]. The tubes were rolled vertically by hand once again for 30 s, and the
Regulatory compliance in a public safety context for licensed pro- solvent pipetted directly into an LC vial and subjected to LC-MS/MS
ducers notwithstanding, biological questions such as monitoring protocol described below.
for batch and plant strain consistency, novel medicinal properties Terpenes. 0.020 g of homogenized material was added to a
[22,23] and information relating to growth stage development and 1.7 mL microcentrifuge tube along with 1.00 mL of hexane. Tubes
plant components (e.g. inflorescence compared to stem, or roots in were capped and vortexed for 2 min, sonicated for 10 min, centri-
comparison to higher stem) become important [14]. Additionally, fuged for 5 min at 4500 g. The supernatant was transferred to a GC
with an estimated increase in the cannabis consumption by the vial using an ashed pipette. Another 1 mL hexane was added to the
Canadian population, questions surrounding the environmental microcentrifuge tube; and extraction was repeated once again. The
risk assessment of various cannabis and/or contaminants on non- two aliquots of extracts were combined and vortexed. A 50 mL
target organisms in the aquatic environment are increasingly aliquot of the combined extracts was diluted to final volume of 1 mL
becoming a concern [24]. with toluene (40 fold dilution).
each analyte (1 ng mL1, 1 mL injection). Once the most diagnostic Accuracy and matrix effects. Aflatoxins are most likely found in
and abundant precursor ion was determined it was then purposely the ng g1 range. Thus, in order to assess accuracy of aflatoxins,
selected and fragmented in our collision cell (Q2) to generate 10 mL of a 1000 pg mL1 solution containing all 4 native aflatoxins
product ions. To improve detection limits, pseudo-MRM transitions was spiked onto 0.10 g of whole bud cannabis, and extracted
were used for some compounds. through the complete method. Cannabinoid and terpenes assessed
in this study were pre-determined to be found in the mg g1 range.
2.5. Method performance characteristics Therefore, only isotopically labelled cannabinoid standards were
quantified throughout method development for 4 main reasons: 1)
The Eurachem Guide-The Fitness for the Purpose of Analytical Available unlabelled standards were in the 1 mg mL1 (which is still
Methods (2014) 2nd edition outlines the various performance in the 1% (w/v) range of cannabis). 2) The cost associated with
characteristics necessary for full validation [25], namely: selectivity, spiking numerous aliquots of entire standard ampules per extrac-
accuracy, LOD/LOQ, working range, precision, and ruggedness. tion was prohibitive. 3) Each standard was only available in
Selectivity. Selectivity for cannabinoids and aflatoxins was methanol, which was not the extraction solvent of choice (aceto-
determined through a scheduled (dynamic) MRM via LC-MS/MS nitrile). Thus, if the entire standard solution was spiked onto the
with quantitation achieved by selecting the m/z ion transition of cannabis material, the extraction process would have commenced,
greatest abundance for quantification and the second greatest and undesirable matrix components would have leached from the
abundant ion transition used for qualification. Of utmost impor- material into solution. 4) Cannabis contains hundreds of matrix
tance to this method was the necessity to create a binary solvent components including fatty acids, pigments, sterols, and sugars.
gradient that baseline separated the isomers comprising the Using another plant devoid of any cannabinoid or terpene in order
various individual cannabinoids. Individual unlabelled standards of to conduct recovery studies would not accurately reflect the impact
cannabinoids were subjected to the same gradient and retention of unique matrix effects on extraction capabilities or the ionisation
times isolated by selective reaction monitoring (SRM). The MRM process using ESI. To elucidate the extraction efficiency and
was then constructed in light of these specific SRMs. For terpenes, response of labelled cannabinoids, smaller aliquots of deuterated
selectivity was determined by comparing the selected ion moni- cannabinoid internal standard mixture were spiked onto the plant
toring (SIM) transitions of each individual terpene and constructing and the signal areas assessed. External and matrix-matched cali-
a dynamic MRM on GC-MS/MS. To isolate the various co-eluting bration curves of the labelled cannabinoids were made at 34, 67,
isomers the temperature gradient was effectively doubled in time 125, 250, and 500 ng mL1, and ion signal suppression assessed at
(i.e. 15 mine30.6 min) to allow for best separation for each point. First, external calibration curves in 50% acetonitrile: 50%
quantification. deionised water. Second, ground cannabis was spiked with labelled
Limits of Detection and quantitation. LOD/LOQs for cannabi- cannabinoids, extracted with 1 mL acetonitrile, and diluted 100 fold
noids were calculated by spiking very small amounts of deuterated in 50% acetonitrile: 50% deionised water (called 100 fold dilution).
labelled standards into cannabis plant material (n¼10) and stan- Third, ground cannabis was extracted with 1 mL acetonitrile, and
dard deviations of that signal calculated. The s0’ was calculated by diluted 100 fold in 50% acetonitrile: 50% deionised water, then
the ratio s0’ ¼ s0/√n. LOD was determined as 3* s0’, and LOQ as 10* spiked with labelled cannabinoids (called post 100 fold dilution).
s0’. Aflatoxins were calculated in a similar way with the exception of Then 7 separate spikes of 100 ng mL1 were added to ground
using unlabelled standards instead of labelled. Working range cannabis and extracted; then interpolated on the corresponding
(linearity) was determined by linear regression (R2 values) of the “post 100 fold dilution” matrix-matched calibration curve under
calibration curves associated with each method. Cannabinoids and identical conditions. This was used as the assessment for accuracy/
terpenes were determined using external calibration curves, and recovery of CBD-d3, CBN-d3, and THC-d3. Matrix effects were
aflatoxins were determined using unlabelled B1, B2, G1, G2 stan- assessed by comparing the signal area ratio between the external
dards and matrix-matched calibration curves. LOD/LOQs for ter- solvent calibration curve points and the post 100 fold dilution
penes were calculated by determining the s0 of native terpenes in points, and calculated proportionality ratio. Additionally, signal
cannabis plant material (n¼10). For the terpenes not detected in area ratios were also calculated between the 100 fold dilution and
plant material, analytical detection limits (ADLs) was used for the post 100 fold dilution spikes to see if the signal area increased
calculating LOD, which is calculated by determining the standard 100 fold as well, and dilution matrix effects calculated as well.
deviation (s0) of 50 ng mL1 native terpene solution (n¼10). The s0’ Similar to cannabinoids, for terpenes, small aliquots of b-myrcene-
was calculated as above for cannabinoids and aflatoxins. Labelled b- d6 internal standard were spiked onto plant material prior to
myrcene-d6 was used as instrumental performance internal stan- extraction and the signal areas assessed. External calibration curves
dard (IPIS) to account for the variability among different injections. of b-myrcene-d6 were made at 34, 67, 125, and 250 ng mL1, and
Calibration range. For cannabinoids an eleven point external signal suppression also assessed at each point. Replicate measure-
calibration curve was constructed from 0.1 to 9000 ng mL1 with ments (n ¼ 10) of plant extracts were analysed and the b-myrcene-
50 ng mL1 of internal standard mixture added to each point. Thus, d6 was quantified by the above external calibration method.
concentration of each point was calculated by signal proportion- Precision. Precision was calculated by extracting and quanti-
ality via relative response factors. For aflatoxins a six point matrix- fying each analyte from real matrix using the whole methods in
matched calibration curve was constructed from 0.1 to 20 ng mL1. triplicate over a 24 h period (intraday) and over 3 consecutive days
For terpenes a six point external calibration curve was constructed (interday). For cannabinoids and aflatoxins the same 100 ng mL1
from 50 to 1000 ng mL1. The concentration of the IPIS, b-myrcene- spikes of internal standard mixture used to assess accuracy and
d6, was added at 125 ng mL1 for each calibration point. Calibration precision. For terpenes, 125 ng mL1 spikes of the internal standard
standards were run randomly and in triplicate along with a solvent b-myrcene-d6 used in accuracy were used for precision assessment.
blank injection. The signal response of each analyte was normalised Ruggedness. The ruggedness of each analytical method was
to that of b-myrcene-d6 and plotted as a function of concentration. determined by purposefully changing either the mass of cannabis
Linearity was evaluated by the magnitude of R2 (correlation coef- extracted or the volume of solvent used in extraction by a specific
ficient) and the level of significance (i.e., p-value) for each of the nominal amount, in triplicate. Results were then normalised to the
analytes. Residual plots were also examined to ensure that data was unit mass of cannabis (i.e. w/w). The variations in the methods
randomly distributed about zero. were conducted both above and below the optimised values to
A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88 83
Fig. 2. An acquisition chromatogram resulting from liquid chromatography-electrospray(þ)-tandem MS analysis of a 40 ng mL1 mixture of aflatoxins B1, B2, G1, and G2 in
acetonitrile; the quantifier ions here are highlighted. The chromatogram outputs of 0e2.0 min, and 7.0e10.0 min were omitted to provide graphical clarity.
Fig. 3. Gas chromatography-electron impact-tandem MS analysis and mass chromatograms for terpene standards highlighting the modification in temperature gradient (top: initial
temperature gradient; bottom: final temperature gradient) in order increase sensitivity and separation of closely associated peaks.
caryophyllene, a-humulene, and cis- and trans-nerolidol) using the quantify terpenes will introduce a bias relative to MRM; specif-
two approaches. Not surprisingly, the chromatogram obtained us- ically, for monoterpenes the bias is greater for SIM than MRM while
ing the SIM method resulted in an extracted ion chromatogram that for all other terpenes the bias is smaller.
had more interferences (Fig. 4, bottom panel, and Fig. S1) and The identities of the species that appear in the ion chromato-
generally contained more background signal especially between 17 gram (peaks labelled A-Y) remain unknown. While beyond the
and 18 min making the quantification of cis-nerolidol (peak 27) scope of this study, high resolution mass spectrometry will un-
very challenging. Results of our terpene measurements comparing doubtedly help to elucidate the nature of these species.
SIM and MRM supports this and implies that using a SIM method to The LODs and LOQs of each analyte can be found in Table 1. The
A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88 85
Fig. 4. Gas chromatography-electron impact-tandem MS analysis and mass chromatograms for the multiple reaction monitoring method (top panel) and selected reaction
monitoring method (bottom panel) for plant extract. Elution order can be found in Table S1 (AeY: unidentified peaks in plant extracts).
European Pharmacopoeia has implemented stricter limits for the due to cannabis matrix, it was critical to use a matrix-matched
presence of aflatoxins in herbal drugs, 2 mg kg1 for B1 and calibration curve [18]. Fundamental to the compensation of ma-
4 mg kg1 for total aflatoxins. The same limit is set for the presence trix effects was the understanding that fungal mycotoxins are sur-
of aflatoxins by the British Pharmacopoeia; and there is no mention ficial contaminants, in that ground cannabis is not necessary for
of ochratoxins for mandatory regulation. Health Canada has no extraction. This was the key to obtaining satisfactory recoveries,
standards itself in the regulation of cannabis but strongly “sug- and was easily the most difficult and time consuming part of
gests” that regardless of which document is followed, that the method development for all analytes reported in this document.
document be adhered to in its entirety [4]. As can be seen in Table 1, Numerous different methods were tried before settling on the
LODs/LOQs obtained in this method were all 2 mg kg1; they were simplest approach, and the following were not successful: extrac-
0.50/1.7, 2.0/6.7, 0.59/2.0, and 0.53/1.8 mg kg1, for B1, B2, G1, and G2 tion as usual and passage through 3 cc, 60 mg C18 solid phase
respectively, and thus would be equal to or lesser than the extraction (SPE) cartridges, approximately 50% of aflatoxins were
reporting limits outlined by the European Pharmacopoeia above. scavenged; extraction as usual and Quick, Easy, Cheap, Effective,
Cannabinoids were assessed using an external calibration curve Rugged and Safe (QuEChERS) using AOAC 2007.01 rated 2 mL Agi-
composed solely by a mixture CBD-d3, CBN-d3, and THC-d3 spiked lent Bond-Elut™ tubes comprised of 50 mg primary secondary
onto ground cannabis as outlined in 2.9.4 above, from 34 to amine (PSA), 50 mg graphite carbon black (GCB), 150 mg MgSO4,
500 ng mL1. Then 7 replicates of 100 ng mL1 were spiked onto scavenged 100% of aflatoxins; 1 g cannabis extracted as above but
ground cannabis, extracted, and diluted 100 fold. Using this using 10 mL of acetonitrile and 6 g of the same QuEChERS sorbent
100 ng mL1 spike, the standard error (s0’) was calculated as in 2.9.2 as above but in dSPE loose salts package, scavenged 100% of afla-
above. Resultant LODs/LOQs were 610/2100 ng g1 (CBD-d3), 3100/ toxins; extraction as usual and aflatoxins spiked afterwards, then
10 300 ng g1 (CBN-d3), 7200/24 000 ng g1 (THC-d3). QuEChERS using Bond-Elut™ tubes, scavenged 100% of aflatoxins.
For terpenes with exception of caryophyllene, all terpenes had In addition various solvents were tested: water, 50% acetonitrile:
LODs/LOQs 0.25/0.084 mg g1. The LODs/LOQs ranged from 50% water, toluene, ethyl acetate, and methanol alone, all yielded
0.00089/0.0030 mg g1 to 0.82/2.7 mg g1 for g-terpinene and no positive results. In support of these findings, the comprehensive
caryophyllene respectively. mycotoxin analysis review by Zhang et al. (2018) [18] noted that
Calibration standards for cannabinoids (solvent), aflatoxins GCB is prone to adsorption of planar structures such as aflatoxins.
(matrix-matched), and terpenes (solvent) were injected in tripli- The effect of plastic centrifuge tubes or glass test tubes used in
cate starting from lowest concentration to greatest with lab blanks the extraction process on recovery of aflatoxins was assessed and
injected between each level to prevent the potential for carry-over found no statistical difference. Therefore, the main consideration in
in each subsequent set of injections. The correlation coefficients extraction of aflatoxins was the shape of the bottom of the test tube
(R2) were all >0.99 for cannabinoids and aflatoxins, and >0.90 for used. Larger round bottom test tubes volume (e.g. 30 mL vs. 10 mL)
terpenes, with p-values <0.0001. Residual plots demonstrate data were more effective at completely submerging the cannabis bud in
normality scattered around zero. Thus, the linear regression models acetonitrile than smaller ones (and also conical bottoms), due to the
for each analyte are acceptable. increased diameter, thus maximising submersion during extrac-
Accuracy was conducted using three different methods as noted tion. Seven replicates over 3 days of 10 mL of 1000 ng mL1 unla-
above: aflatoxins were calculated using the nominal spike and re- belled mixture was spiked onto whole bud cannabis, and allowed to
covery method of unlabelled analytes using a matrix-matched acclimate for 15 min for a final target concentration of 10 ng mL1
calibration curve. Given the great degree of ion suppression in ESI in 1 mL of acetonitrile. The cumulative mean results were
86 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88
Table 1
Using gas chromatography-electron impact-tandem MS (GC-EI-MS/MS) or liquid chromatography-electrospray þ -tandem mass spectrometry analysis (LC-ESI þ -MS/MS), the
method validation parameters for the analysis of aflatoxins, cannabinoids, and terpenes in order of chromatographic elution.
Compound Working Range (ng mL1) Accuracy (%) Linearity Precision (RSD %) LOD (mg kg1 dry weight) LOQ (mg kg1 dry weight) Matrix Effects (%)
Intraday Interday
Aflatoxins
G2 0.2e20 104 0.9926 5.4 6.7 0.53 1.8 NA
G1 89 0.9942 8.0 4.1 0.59 2.0 NA
B2 88 0.9953 3.6 1.8 2.0 6.7 NA
B1 104 0.9934 10.7 3.7 0.50 1.7 NA
Cannabinoids
CBGA 0.1e9000 NA 0.9940 NA NA NA NA NA
CBDA NA 0.9879 NA NA NA NA NA
CBG NA 0.9997 NA NA NA NA NA
CBD 145 0.9999 6.3 2.8 610 2100 41
CBN 92 0.9999 4.9 1.8 3100 10 300 47
THCA NA 0.9913 NA NA NA NA NA
THC 92 0.9999 3.0 1.6 7200 24 000 78
CBC NA 0.9989 NA NA NA NA NA
Terpenes LOD (mg kg1 dry weight) LOQ (mg kg1 dry weight)
a-Pinene 50e1000 NA 0.940 17.8 22.5 27 88 NA
Camphene NA 0.933 18.5 23.7 9.1 30 NA
Sabinene NA 0.937 15.9 18.4 3.5 12 NA
b-Pinene NA 0.934 16.9 18.4 39 130 NA
Myrcene 121 0.917 15.1 21.1 71 240 NA
3-Carene NA 0.938 NA NA 24a 81a NA
a-Terpinene NA 0.941 NA NA 27a 89a NA
Limonene NA 0.938 17.3 21.1 200 650 NA
Eucalyptol NA 0.931 25.1 34.2 7 23 NA
Ocimene total NA 0.937 16.9 21.9 17 55 NA
g-Terpinene NA 0.946 25.5 33.5 0.89 3 NA
Terpinolene NA 0.948 16.2 19.7 2.8 9.4 NA
Fenchone NA 0.948 18.1 22.1 5.9 20 NA
Linalool NA 0.927 15.2 20.6 160 520 NA
Fenchol NA 0.950 15.2 19.6 57 190 NA
Camphor NA 0.949 NA NA 25a 85a NA
Isoborneol NA 0.949 NA NA 25a 82a NA
Borneol NA 0.951 18.0 23.2 23 75 NA
Menthol NA 0.943 NA NA 27a 91a NA
a-Terpineol NA 0.948 16.0 21.0 73 240 NA
Citronellol NA 0.943 NA NA 29a 97a NA
Pulegone NA 0.949 NA NA 26a 87a NA
Geranyl Acetate NA 0.943 NA NA 30a 100a NA
Caryophyllene NA 0.946 16.2 21.0 820 2700 NA
a-Cedrene 0.939 19.4 30.1 12 39 NA
a-Humulene NA 0.945 16.4 21.2 220 750 NA
Valencene NA 0.954 NA NA 31a 100a NA
Nerolidol total NA 0.942 17.0 21.7 250 840 NA
Guaiol NA 0.949 22.8 22.9 13 44 NA
Cedrol NA 0.950 NA NA 25a 84a NA
a-Bisabolol NA 0.941 16.8 21.1 250 820 NA
a
Compounds are not detected in plant extract, use ADLs for LOD.
104 ± 12%, 88 ± 11%, 89 ± 14%, 104 ± 12% for B1, B2, G1, and G2 from 34 to 500 ng mL1. Deviations from zero, linearity (R2), and
respectively, and there was no significant difference (P ¼ 0.0175) to significant difference from one slope to another was assessed.
the control values. Therefore, satisfactory recoveries using the Subsequently, at each calibration point (i.e. 34, 67, 125, 250, and
matrix-matched calibration curves were obtained. 500 ng mL1) the mean signal ratios (and SD) between solvent
Given that terpenes and cannabinoids are found at the mass standards and post 100 fold dilution were calculated; this became
percent range, spiking with lesser nominal amounts (e.g. ng g1 the mean matrix effects. Then seven separate 100 ng mL1 post
spike in comparison to mg g1 endogenous) and attempting to dilution spikes of deuterated analytes were extracted and the mean
quantify proportional recovery was very problematic. Thus, ter- signals interpolated on the post dilution curves. This was essential
penes were calculated using internal standard calibration curves in calculating accuracy, and extraction efficiency, of cannabinoid
using b-myrcene-d6 as a surrogate standard for all analytes. content of CBD-d3, CBN-d3, and THC-d3. If these recoveries were
Cannabinoid concentrations were also assessed using internal acceptable then the assumption of extraction efficiency and accu-
standard calibration curves. However, it was salient for quantita- racy of endogenous cannabinoids would be valid as well.
tion, and to the accuracy of potency values provided to licensed In total, upon linear regression (Fig. 5 and Fig. S2), all deuterated
producers, to validate the mass of deuterated analyte to the cor- external calibration curves were significantly different from 0 (all P-
responding signal peak area, to ensure complete recovery using the values <0.001), linear (all R2 values > 0.99), and all slopes were
method. significantly different from one another from solvent standards to
As noted earlier, external calibration curves (solvent alone) were post 100 fold dilution extracts to pre 100 fold dilutions as well (all
constructed, in triplicate, using the deuterated analyte mixture P-values <0.0001). Also no deviation from linearity was calculated
A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88 87
4. Conclusions