Analytica Chimica Acta 1088 (2019) 79e88

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca

Validated quantitative cannabis profiling for Canadian regulatory

compliance - Cannabinoids, aflatoxins, and terpenes
Alistair K. Brown a, Zhe Xia a, Patrique Bulloch a, Ifeoluwa Idowu a, Olga Francisco a,
Jorg Stetefeld a, Jake Stout b, Jeff Zimmer c, Chris Marvin d, Robert J. Letcher e,
Gregg Tomy a, *
a
University of Manitoba, Department of Chemistry, Winnipeg, MB, R3T 2N2, Canada
b
University of Manitoba, Department of Biological Sciences, Winnipeg, MB, R3T 2N2, Canada
c
Saskatchewan Research Council, 143-111 Research Drive, Saskatoon, SK, S7N 3R2, Canada
d
Environment and Climate Change Canada, National Water Research Institute, Burlington, ON L7S 1A1, Canada
e
Environment and Climate Change Canada, Ecotoxicology and Wildlife Health Division, National Wildlife Research Centre, Carleton University, Ottawa, ON
K1A 0H3, Canada

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

Validated method developed for an-

alyses of organic chemicals in
cannabis.

Cannabinoids and aflatoxins
measured by LC/MS/MS; terpenes by
GC/MS/MS.

Matrix effects evident for cannabi-
noids; isotope dilution needed to
ensure accuracy.

Method achieves the reporting limits
(2 mg kg1) for aflatoxins.

Extractions performed on <1 g ma-
terial with minimal solvent
consumption.

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.
1. Introduction
The United States federal government currently categorises
cannabis as a Schedule 1 illegal drug under the Controlled Sub-
stances Act (1970) and thus there are no federal guidelines for
contaminant restriction and analysis of the medicinal phyto-
chemicals in cannabis products [1]. However, since cannabis has
become recognised as a valuable commercial crop, legislation
should be forthcoming and standardised analytical testing will
become mandatory. In October 2018, the federal government of
Canada legalised the recreational use of cannabis, with regulations
stipulating strict production safety and consumer protection
guidelines under the Cannabis Act and Regulations (2018), which in
turn are facilitated through the Cannabis Tracking and Licensing
System [2]. For federally-regulated licenced producers of cannabis
to remain compliant with these regulations, it is essential to
develop quantitative analytical methods to easily and accurately
quantify both the medicinal phytochemicals produced by this plant
(collectively termed the cannabinoids), and potential fungal toxin
contaminants in cannabis products [2]. Previous analytical
methods were used to characterise medical cannabis for years prior
to legalisation [3]. Cumulatively, the total suite of analytes neces-
sary for Canadian regulatory compliance at the federal level is
cannabinoids, fungal toxins, pesticide residues, and metals content.
However, there presently exists fundamental analytical challenges,
as the methods used were unable to unequivocally address
cannabis potency (due to identical molecular masses of many of the
cannabinoids), insufficient reporting limits for fungal contaminants
[4,5] and poor chromatography for many of the terpenes [6].
Cannabis produces the cannabinoids as acidic molecules, and
quantitatively the major cannabinoids include D-9 tetrahy-
drocannabinolic acid (THCA), cannabidiolic acid (CBDA), cannabi-
gerolic acid (CBGA), and cannabichromenic acid (CBCA) [7]. Upon
heating, these molecules spontaneously decarboxylate to the
‘neutral’ cannabinoids D-9 tetrahydrocannabinol (THC), cannabi-
diol (CBD), cannabigerol (CBG), and cannabichromene (CBC). The
ratio of the acidic cannabinoids varies amongst different cannabis
varieties, with THCA being more abundant in marijuana cannabis
varieties and CBDA being more abundant in hemp cannabis vari-
eties [8]. Dried flowers from all cannabis plants typically contain
these cannabinoids in the mass percent range [9].
Cannabinoids were typically quantified using liquid chroma-
tography (LC) in conjunction with either UV-DAD [7,10e13] and MS
[14e17], and/or by gas chromatography (GC-MS) [6,8,10]. However,
it became evident that potency levels could be problematic as a
consequence of inaccuracies due to the isomeric nature of many
cannabinoids. The acidic cannabinoids THCA, CBDA, and CBCA all
have an identical molecular mass of 358.5 g mol1, whereas the
neutral cannabinoids THC, CBD, and CBC have an identical molec-
ular mass of 315.3 g mol1. Thus, these molecules possess similar
LC-electrospray positive ion mass spectra. Furthermore, these
cannabinoids are only a few of >90 that have been identified in
cannabis [9]. The heat-labile nature of the acidic cannabinoids
highlights the importance of using a non-destructive, low-tem-
perature method in order to achieve the necessary quantitative
accuracy of these molecules. Moreover, careful thought regarding
internal standards with which to quantify also becomes important
© 2019 Elsevier B.V. All rights reserved.
given the potential for variances in instrument response [18]. Thus,
a scheduled (or dynamic) multiple reaction monitoring (MRM)
approach that uses a specific time window spanning calculated
retention times is essential to isolate ion transitions that are diag-
nostic of the target cannabinoid.
Cannabis is susceptible to a number of fungal endophytes and
pathogens, including Aspergillus species that produce the potent
class of carcinogens known as the mycotoxins. Health Canada and
the federal Cannabis Act (2018) do not specifically outline regula-
tory limits for the types or levels of mycotoxins in cannabis prod-
ucts; however it is recommended that in the pursuit of complete
regulatory compliance that a standardised benchmark (e.g. the
European Pharmacopoeia) be followed comprehensively. Specif-
ically, the Canadian Food and Drug Act states that it is up to the
discretion of the licensed producer to decide which publication
listed within Schedule B it should adhere to for analytical testing of
mycotoxins in cannabis products. However, any document chosen
must be fully adhered to in its entirety in the Food and Drug Act
[19]. Contaminant challenges are non-trivial with a plant such as
cannabis, which has the potential to present microbial contami-
nation in addition to organics [5]. Analyses for fungal mycotoxins,
pesticides, and metals are necessary for Canadian federal compli-
ance [2]. Mycotoxins such as aflatoxins are typically found in the
high pg mL1 to low ng mL1 range; with EU (EU 2006) regulatory
levels of aflatoxins B1 of 2e12 mg kg1, and combined total B1, B2,
G1, and G2 of 4e15 mg kg1 [20]. Thus, effective sample extraction
methods are necessary to isolate the target analytes, to minimise
background contamination, to maximise sensitivity of mass anal-
ysis, and to minimise LC ionisation matrix effects from this complex
sample material. Pesticides (beyond the scope of this study) pose a
difficult challenge as there are currently 95 different pesticides that
require analysis as part of the regulatory control of cannabis [4],
some of which are not analysable by LC-MS/MS due to inefficient
ionisation [21].
Terpenes are volatile aromatic compounds that are responsible
for the flavour profile of cannabis, and have also been implicated in
the medicinal effects of cannabis either explicitly (e.g. anti-
convulsant, anti-inflammatory, anti-depressant, anticancer) [22],
or in concert with cannabinoids (e.g. increasing the biological/
physiological effects) which is commonly referred to as the
‘entourage effect’ [23]. A major analytical challenge is the co-
elution of various unknown terpenes along with targeted com-
pounds given the presence of hundreds of ancillary components
found in the cannabis plant (e.g. terpenes, flavonoids, fatty acids,
phenols, pigments) [9]. Thus, it is important from a quantitative
standpoint to know what proportion of the identified MS signal is
due to the targeted terpene of interest, and the possible qualitative
elucidation of new/unidentified terpenes within cannabis strains
[6]. Additionally, the lack of available commercial labelled stan-
dards for terpenes makes unambiguous quantitation problematic,
in that isotopically labelled surrogates become necessary, thereby
possibly affecting values due to unequal responses during mass
analysis (i.e. LC-ionisation matrix effects).
With all of these analytical challenges taken into consideration
and using the methods contained herein, the objective of the pre-
sent study is to provide a validated quantitative analytical method
for cannabinoids, aflatoxins, and terpenes from an ISO 17 025
 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88
accredited laboratory perspective to address major regulatory
[1,2,4], biological [3,22,23], and environmental concerns [24].
Regulatory compliance in a public safety context for licensed pro-
ducers notwithstanding, biological questions such as monitoring
for batch and plant strain consistency, novel medicinal properties
[22,23] and information relating to growth stage development and
plant components (e.g. inflorescence compared to stem, or roots in
comparison to higher stem) become important [14]. Additionally,
with an estimated increase in the cannabis consumption by the
Canadian population, questions surrounding the environmental
risk assessment of various cannabis and/or contaminants on non-
target organisms in the aquatic environment are increasingly
becoming a concern [24].
2. Experimental section
2.1. Materials and reagents
Dried cannabis material was purchased locally from a licensed
producer in the city of Winnipeg. Unlabelled analytical standards of
cannabinol (CBN), cannabidiol (CBD), cannabigerol (CBG), canna-
bichromene (CBC), cannabidiolic acid (CBDA), cannabigerolic acid
(CBGA), delta-9-tetrahydrocannabinol (THC), and delta-9-
tetrahydrocannabinolic acid A (THCA), and labelled standards
delta-9-tetrahydrocannabinol-d3 (THC-d3), cannabinol-d3 (CBN-
d3), cannabidiol-d3 (CBD-d3), b-myrcene-d6, aflatoxin B1, aflatoxin
B2, aflatoxin G1, and aflatoxin G2 were purchased from Toronto
Research Chemicals (Toronto, Canada). All native terpene com-
pounds were purchased from Sigma Aldrich (St. Louis, MO, USA).
Acetonitrile, methanol, hexane, were all purchased from Fisher-
Scientific and of LC-grade. Ultrapure Milli-Q (18 MU-cm) was ob-
tained from a Synergy™ Milli-Q purification system from Millipore
(Billerica, MA).
2.2. Preparation of dry cannabis
Dry cannabis from the same batch was ground via mortar and
pestle with liquid nitrogen into a fine powder. Homogenized ma-
terial was transferred into several clear scintillation vials and cap-
ped and stored at 20  C for the duration of the method
development. Subsamples of this larger batch were used in the
cannabinoid and terpene analyses. Aflatoxin analyses were done
using whole dried cannabis “buds” with as minimal handling as
possible. Aflatoxins are surficial chemicals that are typically found
at the inflorescence and stem junction, and not inside the cells of
the flower itself, in comparison to food stuffs where aflatoxins are
found within the manufactured product itself [20]. The challenge of
minimising matrix effects during mass analysis was completely
dependent on this minimalist method.
2.3. Extraction
Cannabinoids. 0.100 g of homogenized material was added to
10 mL centrifuge tubes along with 1.00 mL of acetonitrile. Tubes
were capped and vortexed for 1 min, sonicated for 5 min, centri-
fuged for 5 min at 4500 g. A 10 mL aliquot of supernatant was
diluted in 1 mL of 50% acetonitrile: 50% water, and vortexed for 30 s
to yield a 100-fold dilution. Another 10 mL fraction of that mixture
was diluted again in another 1 mL of 50% acetonitrile: 50% water,
and vortexed for 30 s for a total 10 000 fold dilution. This mixture
was transferred to an ashed LC vial using an ashed pipette, and
subjected to LC-MS/MS protocol described below.
Aflatoxins. 0.100 g of whole dried cannabis bud was added to
30 mL glass test tubes along with 1.00 mL of acetonitrile. Tubes
were capped and rolled vertically by hand for 30 s to completely
81
coat and submerge the bud with solvent, then sonicated for 5 min.
The tubes were rolled vertically by hand once again for 30 s, and the
solvent pipetted directly into an LC vial and subjected to LC-MS/MS
protocol described below.
Terpenes. 0.020 g of homogenized material was added to a
1.7 mL microcentrifuge tube along with 1.00 mL of hexane. Tubes
were capped and vortexed for 2 min, sonicated for 10 min, centri-
fuged for 5 min at 4500 g. The supernatant was transferred to a GC
vial using an ashed pipette. Another 1 mL hexane was added to the
microcentrifuge tube; and extraction was repeated once again. The
two aliquots of extracts were combined and vortexed. A 50 mL
aliquot of the combined extracts was diluted to final volume of 1 mL
with toluene (40 fold dilution).
2.4. Chromatography and MS analysis
Cannabinoids and aflatoxins. A volume of 10 mL of sample
extract was injected using a CTC Pal autosampler; and liquid
chromatography was conducted using an Agilent 1100 binary pump
(G1312A) and a binary gradient of MilliQ water (solvent A) and
methanol (solvent B). Separation was achieved using an XSelect®
HSS T3 C18 column (2.5 mm dp x 2.1 mm  50 mm) in conjunction
with a VanGuard® HSS T3 C18 cartridge (3 mm dp x
2.1 mm  5 mm). Mass analysis was conducted in positive mode
electrospray ionisation via a triple quadrupole mass spectrometer
(Sciex API 365) retrofitted with an HSID Ionics EP 10 þ orthogonal
ionisation source. A scheduled MRM (20 s window) was explicitly
used to confidently identify and quantify the various cannabinoid
isomers by retention time and baseline separation. A slightly
different elution gradient was the most effective at maximising
peak height (to lower LODs) of the aflatoxins in addition to the
necessity of using a different extraction method as outlined above
in 2.3. Salient to effective retention and separation of cannabinoids
and aflatoxins was the necessity to start the gradient with greater
water content to promote sorption to the stationary phase, and
hold at greater organic solvent content to promote desorption to
the mobile phase while minimising peak tailing and maximising
peak symmetry. All flow rates were 0.3 mL min1, and column
temperature was 42  C. The cannabinoid gradient was as follows:
0e3.00 min hold at 10%B, 3.01e6.00 min linear ramp to 90%B,
6.01e12.00 min hold at 90%B, and re-equilibrate
12.01e15.00 min at 10%B. The aflatoxin gradient was as follows:
0e3 min linear ramp from 10% to 90%B, 3.01e7.00 min hold at 90%
B, and re-equilibrate 7.01e10.00 min at 10%B. A complete list of the
quantifier and qualifier MRM transitions and collision potentials
can be found in Table S1.
Terpenes. An Agilent 7890 GC coupled with a triple quadrupole
mass spectrometer fitted with electron ionisation (EI) source was
used for the MS/MS acquisition. The ionisation energy was set to
50 eV to preserve the molecular ion for monoterpenes (m/z 135.9)
as terpenes are susceptible to fragmentation; the ion abundance of
the molecular ion for most terpenes is too low to set MRM transi-
tions under ionisation energy of 70eV. Separation was achieved
using an Agilent DB-5 MS ultra-inert column
(30 m  0.25 mm  0.25 mm) with helium as the carrier gas at a
constant flow rate of 1.2 mL min1 1 mL of sample was injected with
a PAL RSI 85 autosampler at 250  C in splitless mode. The oven
temperature was held at 50  C from 0 to 1.00 min, 1.01e6.00 min
linear ramp to 100  C at 10  C/min, 6.01 mine20.00 min linear ramp
to 170  C at 5  C/min, 20.01e24.00 min hold at 170  C,
24.01e26.5 min linear ramp to 300  C at 50  C/min, and finally held
26.51e30.6 min at 300  C. Both transfer line and source tempera-
ture were set at 320  C and UHP nitrogen was used as the collision
gas at 60 psi. Before MRM conditions could be established it was
first necessary to record full-scan EI mass spectra (m/z 50e250) for
82 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88
each analyte (1 ng mL1, 1 mL injection). Once the most diagnostic
and abundant precursor ion was determined it was then purposely
selected and fragmented in our collision cell (Q2) to generate
product ions. To improve detection limits, pseudo-MRM transitions
were used for some compounds.
2.5. Method performance characteristics
The Eurachem Guide-The Fitness for the Purpose of Analytical
Methods (2014) 2nd edition outlines the various performance
characteristics necessary for full validation [25], namely: selectivity,
accuracy, LOD/LOQ, working range, precision, and ruggedness.
Selectivity. Selectivity for cannabinoids and aflatoxins was
determined through a scheduled (dynamic) MRM via LC-MS/MS
with quantitation achieved by selecting the m/z ion transition of
greatest abundance for quantification and the second greatest
abundant ion transition used for qualification. Of utmost impor-
tance to this method was the necessity to create a binary solvent
gradient that baseline separated the isomers comprising the
various individual cannabinoids. Individual unlabelled standards of
cannabinoids were subjected to the same gradient and retention
times isolated by selective reaction monitoring (SRM). The MRM
was then constructed in light of these specific SRMs. For terpenes,
selectivity was determined by comparing the selected ion moni-
toring (SIM) transitions of each individual terpene and constructing
a dynamic MRM on GC-MS/MS. To isolate the various co-eluting
isomers the temperature gradient was effectively doubled in time
(i.e. 15 mine30.6 min) to allow for best separation for
quantification.
Limits of Detection and quantitation. LOD/LOQs for cannabi-
noids were calculated by spiking very small amounts of deuterated
labelled standards into cannabis plant material (n¼10) and stan-
dard deviations of that signal calculated. The s0’ was calculated by
the ratio s0’ ¼ s0/√n. LOD was determined as 3* s0’, and LOQ as 10*
s0’. Aflatoxins were calculated in a similar way with the exception of
using unlabelled standards instead of labelled. Working range
(linearity) was determined by linear regression (R2 values) of the
calibration curves associated with each method. Cannabinoids and
terpenes were determined using external calibration curves, and
aflatoxins were determined using unlabelled B1, B2, G1, G2 stan-
dards and matrix-matched calibration curves. LOD/LOQs for ter-
penes were calculated by determining the s0 of native terpenes in
cannabis plant material (n¼10). For the terpenes not detected in
plant material, analytical detection limits (ADLs) was used for
calculating LOD, which is calculated by determining the standard
deviation (s0) of 50 ng mL1 native terpene solution (n¼10). The s0’
was calculated as above for cannabinoids and aflatoxins. Labelled b-
myrcene-d6 was used as instrumental performance internal stan-
dard (IPIS) to account for the variability among different injections.
Calibration range. For cannabinoids an eleven point external
calibration curve was constructed from 0.1 to 9000 ng mL1 with
50 ng mL1 of internal standard mixture added to each point. Thus,
concentration of each point was calculated by signal proportion-
ality via relative response factors. For aflatoxins a six point matrix-
matched calibration curve was constructed from 0.1 to 20 ng mL1.
For terpenes a six point external calibration curve was constructed
from 50 to 1000 ng mL1. The concentration of the IPIS, b-myrcene-
d6, was added at 125 ng mL1 for each calibration point. Calibration
standards were run randomly and in triplicate along with a solvent
blank injection. The signal response of each analyte was normalised
to that of b-myrcene-d6 and plotted as a function of concentration.
Linearity was evaluated by the magnitude of R2 (correlation coef-
ficient) and the level of significance (i.e., p-value) for each of the
analytes. Residual plots were also examined to ensure that data was
randomly distributed about zero.
Accuracy and matrix effects. Aflatoxins are most likely found in
the ng g1 range. Thus, in order to assess accuracy of aflatoxins,
10 mL of a 1000 pg mL1 solution containing all 4 native aflatoxins
was spiked onto 0.10 g of whole bud cannabis, and extracted
through the complete method. Cannabinoid and terpenes assessed
in this study were pre-determined to be found in the mg g1 range.
Therefore, only isotopically labelled cannabinoid standards were
quantified throughout method development for 4 main reasons: 1)
Available unlabelled standards were in the 1 mg mL1 (which is still
in the 1% (w/v) range of cannabis). 2) The cost associated with
spiking numerous aliquots of entire standard ampules per extrac-
tion was prohibitive. 3) Each standard was only available in
methanol, which was not the extraction solvent of choice (aceto-
nitrile). Thus, if the entire standard solution was spiked onto the
cannabis material, the extraction process would have commenced,
and undesirable matrix components would have leached from the
material into solution. 4) Cannabis contains hundreds of matrix
components including fatty acids, pigments, sterols, and sugars.
Using another plant devoid of any cannabinoid or terpene in order
to conduct recovery studies would not accurately reflect the impact
of unique matrix effects on extraction capabilities or the ionisation
process using ESI. To elucidate the extraction efficiency and
response of labelled cannabinoids, smaller aliquots of deuterated
cannabinoid internal standard mixture were spiked onto the plant
and the signal areas assessed. External and matrix-matched cali-
bration curves of the labelled cannabinoids were made at 34, 67,
125, 250, and 500 ng mL1, and ion signal suppression assessed at
each point. First, external calibration curves in 50% acetonitrile: 50%
deionised water. Second, ground cannabis was spiked with labelled
cannabinoids, extracted with 1 mL acetonitrile, and diluted 100 fold
in 50% acetonitrile: 50% deionised water (called 100 fold dilution).
Third, ground cannabis was extracted with 1 mL acetonitrile, and
diluted 100 fold in 50% acetonitrile: 50% deionised water, then
spiked with labelled cannabinoids (called post 100 fold dilution).
Then 7 separate spikes of 100 ng mL1 were added to ground
cannabis and extracted; then interpolated on the corresponding
“post 100 fold dilution” matrix-matched calibration curve under
identical conditions. This was used as the assessment for accuracy/
recovery of CBD-d3, CBN-d3, and THC-d3. Matrix effects were
assessed by comparing the signal area ratio between the external
solvent calibration curve points and the post 100 fold dilution
points, and calculated proportionality ratio. Additionally, signal
area ratios were also calculated between the 100 fold dilution and
the post 100 fold dilution spikes to see if the signal area increased
100 fold as well, and dilution matrix effects calculated as well.
Similar to cannabinoids, for terpenes, small aliquots of b-myrcene-
d6 internal standard were spiked onto plant material prior to
extraction and the signal areas assessed. External calibration curves
of b-myrcene-d6 were made at 34, 67, 125, and 250 ng mL1, and
signal suppression also assessed at each point. Replicate measure-
ments (n ¼ 10) of plant extracts were analysed and the b-myrcene-
d6 was quantified by the above external calibration method.
Precision. Precision was calculated by extracting and quanti-
fying each analyte from real matrix using the whole methods in
triplicate over a 24 h period (intraday) and over 3 consecutive days
(interday). For cannabinoids and aflatoxins the same 100 ng mL1
spikes of internal standard mixture used to assess accuracy and
precision. For terpenes, 125 ng mL1 spikes of the internal standard
b-myrcene-d6 used in accuracy were used for precision assessment.
Ruggedness. The ruggedness of each analytical method was
determined by purposefully changing either the mass of cannabis
extracted or the volume of solvent used in extraction by a specific
nominal amount, in triplicate. Results were then normalised to the
unit mass of cannabis (i.e. w/w). The variations in the methods
were conducted both above and below the optimised values to
 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88 83

provide a comprehensive look at ruggedness. For cannabinoids

0.10 g of ground cannabis was extracted with either 0.9 or 1.1 mL of
acetonitrile. For aflatoxins 0.080 or 0.12 g of whole bud was
extracted using either 0.9 or 1.1 mL of acetonitrile. For terpenes
0.010 or 0.050 g of ground cannabis was extracted using 1.0 mL
hexane; 0.020 g of ground cannabis either 0.8 or 1.2 mL of hexane.

2.6. Statistical analysis

Statistical analysis was conducted by GraphPad Prism® 5 to run

paired t-tests with Dunnett's post-hoc tests to compare means of
treatments to values to control values for accuracy/recovery, and
ruggedness, and one-way ANOVA to compare differences in slope
between solvent external calibration curves and matrix-matched
calibration curves (matrix effects). All other statistical analyses for
determination of LOD/LOQ and precisions were conducted using
Excel functions within Microsoft Office 2010®.

3. Results and discussion

As stated earlier, the method validation was conducted in

accordance with Eurachem guidelines In addition the European
Pharmacopoeia was adhered to in order to assess the analytical
method as being able to meet the reporting limits needed for
Health Canada approval.
Assessing method performance characteristics. Paramount to
the method validation was the assessment of selectivity and
confirmation of analytes across all three classes of compounds due
to mutual interferences and potential misinterpretation of the data.
This confirmatory aspect of this validation alone makes this method
valuable for quantitative purposes. All cannabinoids (Fig. 1) and
aflatoxins (Fig. 2) were confirmed using a qualifier transition ion.
Given the identical molecular masses (315.3 m/z) and ion transi-
tions (259.4 and 193.5 m/z) of CBD, THC, and CBC, single reaction
monitoring was independently conducted in solvent standards to
elucidate their respective retention times. A 20 s scheduled acqui-
sition window was established for each analyte in order to generate
an accurate MRM. Similarly, isotopically-labelled internal standards
CBD-d3, CBN-d3, and THC-d3 individual transitions were also
extracted from cannabis to confirm analyte elution patterns. Eight
various cannabinoids, in order of elution, CBGA, CBDA, CBG, CBD,
CBN, THCA, THC, and CBC, and four aflatoxins, G2, G1, B2, and B1
were quantified in this study using this methodology.
Deuterated isotopically-labelled standards of CBD-d3, CBN-d3,
and THC-d3 were used to quantify the eight cannabinoids. It was
Fig. 1. Liquid chromatography-electrospray(þ)-tandem MS analysis and the selective
determined that THC and THC-d3 were both equi-responsive to one reaction monitoring (SRM) of all eight cannabinoids quantified in this study: canna-
another but had disproportionately greater signal response to all bigerolic acid (CBGA), cannabidioloic acid (CBDA), cannabigerol (CBG), cannabidiol
other cannabinoids. Thus, THC-d3 was solely used in the quantifi- (CBD), cannabinol (CBN), tetrahydrocannabinolic acid (THCA), tetrahydrocannabinol
cation of endogenous THC. Similarly, CBN-d3 was used solely to (THC), and cannabichromene (CBC). Each analyte is listed at their respective retention
time, with its signal area equivalent to a concentration of 500 ng mL1 standard in
quantify endogenous CBN, and CBD-d3 was used to quantify all acetonitrile, and the quantifier ions signal area highlighted in bold. These SRMs were
others based on signal abundance similarity. used to construct the scheduled multiple reaction monitoring acquisition method.
For GC-EI-MS/MS analysis of terpenes, the initial temperature
gradient was set to 50  C 0e1.00 min, 1.01e6.00 min linear ramp to
200  C at 15  C/min, 9.00e11.00 min linear ramp to 300  C at 50  C/ hold at 170  C, 24.01e26.6 min linear ramp to 300  C at 50  C/min,
min, and finally held 11.00e15.00 min at 300  C. However, co- and finally held 26.61e30.6 min at 300  C. As can be seen in the
elution challenges were preliminarily apparent as shown in the bottom panel of Fig. 3, b-pinene (peak 4) and b-myrcene (peak 5)
top panel of Fig. 3. The two isomers of monoterpenes, b-pinene have been teased out apart; peaks 9 (ocimene and eucalyptol) has
(peak 4) and b-myrcene (peak 5), co-eluted and were not distin- baseline-separation.
guishable. Ocimene and eucalyptol co-eluted and comprised the Additionally, ten plant extracts were injected on GC-MS/MS and
single peak 9; additionally, peaks 9 (ocimene and eucalyptol) and terpenes were quantified separately using both MRM method and
10 (ocimene-2) also do not demonstrate baseline separation. SIM method monitoring ions using the most abundant ion at m/z
In order to achieve adequate separation between b-pinene and 93.0. After quantifying all the native terpenes, and using t-tests
b-myrcene, the temperature gradient was changed to 50  C (t ¼ 0.05, two-tail), we found that there were significant differences
0e1.00 min, 1.01e6.00 min linear ramp to 100  C at 10  C/min, 6.01 (p > 0.05) in concentrations of 10 terpenes (camphene, sabinene, b-
mine20.00 min linear ramp to 170  C at 5  C/min, 20.01e24.00 min pinene, ocimene, L-()-fenchone, linalool, a-terpineol, ()-trans-
84 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88

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
two approaches. Not surprisingly, the chromatogram obtained us-
ing the SIM method resulted in an extracted ion chromatogram that
had more interferences (Fig. 4, bottom panel, and Fig. S1) and
generally contained more background signal especially between 17
and 18 min making the quantification of cis-nerolidol (peak 27)
very challenging. Results of our terpene measurements comparing
SIM and MRM supports this and implies that using a SIM method to
quantify terpenes will introduce a bias relative to MRM; specif-
ically, for monoterpenes the bias is greater for SIM than MRM while
for all other terpenes the bias is smaller.
The identities of the species that appear in the ion chromato-
gram (peaks labelled A-Y) remain unknown. While beyond the
scope of this study, high resolution mass spectrometry will un-
doubtedly help to elucidate the nature of these species.
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
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
presence of aflatoxins in herbal drugs, 2 mg kg1 for B1 and
4 mg kg1 for total aflatoxins. The same limit is set for the presence
of aflatoxins by the British Pharmacopoeia; and there is no mention
of ochratoxins for mandatory regulation. Health Canada has no
standards itself in the regulation of cannabis but strongly “sug-
gests” that regardless of which document is followed, that the
document be adhered to in its entirety [4]. As can be seen in Table 1,
LODs/LOQs obtained in this method were all 2 mg kg1; they 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
respectively, and thus would be equal to or lesser than the
reporting limits outlined by the European Pharmacopoeia above.
Cannabinoids were assessed using an external calibration curve
composed solely by a mixture CBD-d3, CBN-d3, and THC-d3 spiked
onto ground cannabis as outlined in 2.9.4 above, from 34 to
500 ng mL1. Then 7 replicates of 100 ng mL1 were spiked onto
ground cannabis, extracted, and diluted 100 fold. Using this
100 ng mL1 spike, the standard error (s0’) was calculated as in 2.9.2
above. Resultant LODs/LOQs were 610/2100 ng g1 (CBD-d3), 3100/
10 300 ng g1 (CBN-d3), 7200/24 000 ng g1 (THC-d3).
For terpenes with exception of caryophyllene, all terpenes had
LODs/LOQs 0.25/0.084 mg g1. The LODs/LOQs ranged from
0.00089/0.0030 mg g1 to 0.82/2.7 mg g1 for g-terpinene and
caryophyllene respectively.
Calibration standards for cannabinoids (solvent), aflatoxins
(matrix-matched), and terpenes (solvent) were injected in tripli-
cate starting from lowest concentration to greatest with lab blanks
injected between each level to prevent the potential for carry-over
in each subsequent set of injections. The correlation coefficients
(R2) were all >0.99 for cannabinoids and aflatoxins, and >0.90 for
terpenes, with p-values <0.0001. Residual plots demonstrate data
normality scattered around zero. Thus, the linear regression models
for each analyte are acceptable.
Accuracy was conducted using three different methods as noted
above: aflatoxins were calculated using the nominal spike and re-
covery method of unlabelled analytes using a matrix-matched
calibration curve. Given the great degree of ion suppression in ESI
85
due to cannabis matrix, it was critical to use a matrix-matched
calibration curve [18]. Fundamental to the compensation of ma-
trix effects was the understanding that fungal mycotoxins are sur-
ficial contaminants, in that ground cannabis is not necessary for
extraction. This was the key to obtaining satisfactory recoveries,
and was easily the most difficult and time consuming part of
method development for all analytes reported in this document.
Numerous different methods were tried before settling on the
simplest approach, and the following were not successful: extrac-
tion as usual and passage through 3 cc, 60 mg C18 solid phase
extraction (SPE) cartridges, approximately 50% of aflatoxins were
scavenged; extraction as usual and Quick, Easy, Cheap, Effective,
Rugged and Safe (QuEChERS) using AOAC 2007.01 rated 2 mL Agi-
lent Bond-Elut™ tubes comprised of 50 mg primary secondary
amine (PSA), 50 mg graphite carbon black (GCB), 150 mg MgSO4,
scavenged 100% of aflatoxins; 1 g cannabis extracted as above but
using 10 mL of acetonitrile and 6 g of the same QuEChERS sorbent
as above but in dSPE loose salts package, scavenged 100% of afla-
toxins; extraction as usual and aflatoxins spiked afterwards, then
QuEChERS using Bond-Elut™ tubes, scavenged 100% of aflatoxins.
In addition various solvents were tested: water, 50% acetonitrile:
50% water, toluene, ethyl acetate, and methanol alone, all yielded
no positive results. In support of these findings, the comprehensive
mycotoxin analysis review by Zhang et al. (2018) [18] noted that
GCB is prone to adsorption of planar structures such as aflatoxins.
The effect of plastic centrifuge tubes or glass test tubes used in
the extraction process on recovery of aflatoxins was assessed and
found no statistical difference. Therefore, the main consideration in
extraction of aflatoxins was the shape of the bottom of the test tube
used. Larger round bottom test tubes volume (e.g. 30 mL vs. 10 mL)
were more effective at completely submerging the cannabis bud in
acetonitrile than smaller ones (and also conical bottoms), due to the
increased diameter, thus maximising submersion during extrac-
tion. Seven replicates over 3 days of 10 mL of 1000 ng mL1 unla-
belled mixture was spiked onto whole bud cannabis, and allowed to
acclimate for 15 min for a final target concentration of 10 ng mL1
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
respectively, and there was no significant difference (P ¼ 0.0175) to
the control values. Therefore, satisfactory recoveries using the
matrix-matched calibration curves were obtained.
Given that terpenes and cannabinoids are found at the mass
percent range, spiking with lesser nominal amounts (e.g. ng g1
spike in comparison to mg g1 endogenous) and attempting to
quantify proportional recovery was very problematic. Thus, ter-
penes were calculated using internal standard calibration curves
using b-myrcene-d6 as a surrogate standard for all analytes.
Cannabinoid concentrations were also assessed using internal
standard calibration curves. However, it was salient for quantita-
tion, and to the accuracy of potency values provided to licensed
producers, to validate the mass of deuterated analyte to the cor-
responding signal peak area, to ensure complete recovery using the
method.
As noted earlier, external calibration curves (solvent alone) were
constructed, in triplicate, using the deuterated analyte mixture
from 34 to 500 ng mL1. Deviations from zero, linearity (R2), and
significant difference from one slope to another was assessed.
Subsequently, at each calibration point (i.e. 34, 67, 125, 250, and
500 ng mL1) the mean signal ratios (and SD) between solvent
standards and post 100 fold dilution were calculated; this became
the mean matrix effects. Then seven separate 100 ng mL1 post
dilution spikes of deuterated analytes were extracted and the mean
signals interpolated on the post dilution curves. This was essential
in calculating accuracy, and extraction efficiency, of cannabinoid
content of CBD-d3, CBN-d3, and THC-d3. If these recoveries were
acceptable then the assumption of extraction efficiency and accu-
racy of endogenous cannabinoids would be valid as well.
In total, upon linear regression (Fig. 5 and Fig. S2), all deuterated
external calibration curves were significantly different from 0 (all P-
values <0.001), linear (all R2 values > 0.99), and all slopes were
significantly different from one another from solvent standards to
post 100 fold dilution extracts to pre 100 fold dilutions as well (all
P-values <0.0001). Also no deviation from linearity was calculated
 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88
Fig. 5. Liquid chromatography-electrospray þ -tandem MS analysis and matrix effects
on the absolute quantitation of deuterated standards of cannabidiol-d3 (CBD-d3). The
absolute signal area of each point (34, 67, 125, 250 and 500 ng mL1) of deuterated
standard mixture was plotted vs treatment: pure solvent (red), post 100 fold dilution
(blue), and pre 100 fold dilution (green). Linear regression was conducted on each
treatment and slopes compared. (For interpretation of the references to colour in this
figure legend, the reader is referred to the Web version of this article.)
for any calibration curve. As can be seen in Fig. 5 and Fig. S2, a mean
matrix suppression of 77% ± 9% (CBD-d3), 48 ± 7% (CBN-d3), and
37% ± 6% (THC-d3) was seen between all solvent standards and post
100 fold dilution extracts. The close precision (i.e. <10%) of all signal
area ratios indicated that suppression was consistent across the
working range of 34e500 ng mL1. Therefore identical matrix ef-
fects were seen with the 7 replicates of 100 ng mL1 labelled
cannabinoid mixtures spiked into diluted cannabis extracts, and
accuracy assessed. 145% (CBD-d3), 92% (CBN-d3), and 92% (THC-d3)
recovery was interpolated using the post 100 fold dilution curves.
Thus, any nominal spike of internal standards for subsequent
quantification of cannabinoids should have maintained a satisfac-
tory recovery throughout the method. This was the first time to the
best of our knowledge that a thorough assessment of matrix effects
based on dilution has been conducted.
Now that the relative signal intensity to concentration was
corroborated, 11-point internal calibration curves from 0.1 to
9000 ng mL1 were used to quantify endogenous cannabinoids by
back-calculating with the 100 fold dilution factor and dividing by
the sample mass to obtain the mass in dry weight per gram. The
same seven replicates of 100 ng mL1 internal standard spiked
cannabis samples were quantified and a mean concentration of
1.5  105 ng mL1 ± 5% equivalent to 0.0015 g g1 dry weight, or
0.15% THC was calculated. Similarly, CBD, CBN, CBG, CBC, concen-
trations of 0.006%, 0.022%, 0.006%, and 0.005% respectively were
calculated.
In the similar fashion as for the cannabinoids above, for ter-
penes, the recovery of b-myrcene-d6 was calculated to be
121 ± 24%, which was above the target value, and indicated
exhaustive recovery. This also directly implied that the matrix ef-
fects were negligible for terpene analysis at our 40 fold dilution
factor.
Precision (RSD) for aflatoxins was calculated using 3 separate
extractions of 10 ng mL1 over 24 h (intraday), and over 3 consec-
utive days (interday). They were 10.7/3.7% for B1, 3.6/1.8% for B2,
8.0/4.1% for G1, 5.4/6.7% for G2. For labelled cannabinoids, the
signal area responses were assessed the same way; they were 6.3/
2.8% for CBD-d3, 4.9/1.8% CBN-d3, 3.0/1.6% for THC-d3. For terpenes
with the exception of eucalyptol, g-terpinene, and a-cedrene, the
precision for all target analytes was <30% for both intra- and inter-
day repeatability studies. Intraday repeatability ranged from 15.1%
(b-myrcene) to 25.5% (g-terpinene), and interday repeatability
ranged from 18.4% (sabinene and b-pinene) to 34.2% (eucalyptol).
87
The two variables of mass of cannabis extracted and volume of
solvent used in the development of these methods were assessed.
The Eurachem guideline for ruggedness states explicitly that
ruggedness test for in-house developed methods must be per-
formed [25]. There were no significant differences in the perfor-
mance of our methods as a result of any of the deliberate changes
applied, and thus confirmed robustness.
4. Conclusions
To our knowledge, the work presented here represents the first
attempt to fully validate analytical methods for the determination
of cannabinoids, aflatoxins and terpenes in cannabis plant. The
method performance characteristics were rigorously tested and the
validation parameters were established in accordance with Eur-
achem Guide to Method Validation and Quality in Analytical
Chemistry. A minimalist approach to sample extraction and clean-
up was developed so as to streamline the analysis of these com-
pounds. Because of their high content, dilution and use of mass-
labelled internal standards for analysis of cannabinoids and ter-
penes yielded results that were fit for purpose. Caution must be
adopted in the analysis of cannabinoids as matrix effects can be
significant and vary for the individual cannabinoids. For aflatoxins,
which are surficial contaminants, a less aggressive extraction
approach was validated which mitigated deleterious effects from
co-extracted matrix material.
Declaration of competing interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
CRediT authorship contribution statement
Alistair K. Brown: Conceptualization, Methodology, Validation,
Formal analysis, Investigation, Writing - original draft, Writing -
review & editing, Visualization. Zhe Xia: Conceptualization,
Methodology, Validation, Formal analysis, Investigation, Writing -
original draft, Writing - review & editing, Visualization. Patrique
Bulloch: Conceptualization, Methodology, Validation, Formal
analysis, Investigation, Visualization. Ifeoluwa Idowu: Validation,
Investigation. Olga Francisco: Validation, Investigation. Jorg Ste-
tefeld: Validation, Investigation. Jake Stout: Validation, Investiga-
tion, Writing - review & editing, Visualization. Jeff Zimmer:
Validation, Investigation, Writing - review & editing, Visualization.
Chris Marvin: Validation, Investigation, Writing - review & editing,
Visualization. Robert J. Letcher: Validation, Investigation, Writing -
review & editing, Visualization. Gregg Tomy: Conceptualization,
Methodology, Validation, Formal analysis, Investigation, Writing -
original draft, Writing - review & editing, Visualization, Supervi-
sion, Project administration, Funding acquisition.
Acknowledgements
This work was supported by the Natural Sciences and Engi-
neering Research Council of Canada Engage Grant [EGP-522299-
2017] to GTT. We thank Patrick Romeo at North Star Cannabis
Laboratories for allowing us to process plant material at his licensed
facility (Health Canada license #: LIC-LG48X1TKZ6-2019).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.aca.2019.08.042.
88 A.K. Brown et al. / Analytica Chimica Acta 1088 (2019) 79e88

References [13] S. Zivovinovic, R. Alder, M.D. Allenspach, C. Steuer, Determination of canna-

binoids in Cannabis sativa L. samples for recreational, medical, and forensic
purposes by reversed-phase liquid chromatography-ultraviolet detection,
[1] L.N. Sandler, J.L. Beckerman, F. Whitford, K.A. Gibson, Cannabis as conundrum,
J. Anal. Sci. Tech. 9 (2018) 1e10.
Crop Protect. 117 (2019) 37e44.
[14] W. Gul, S.W. Gul, S. Chandra, H. Lata, E.A. Ibrahim, M.A. ElSohly, Detection and
[2] H. Canada, in: H. Canada (Ed.), Cannabis Tracking and Licensing System-
quantification of cannabinoids in extracts of cannabis sativa roots using LC-
Monthly Reporting Guide for Federal License Holders, Government of Canada,
MS/MS, Planta Med. 84 (2018) 267e271.
2018. Online Resource.
[15] J. Kim, A. de Castro, E. Lendoiro, A. Cruz-Landeira, M. Lopez-Rivadulla,
[3] E.P. Baron, P. Lucas, J. Eades, O. Hogue, Patterns of medicinal cannabis use,
M. Concheiro, Detection of in utero cannabis exposure by umbilical cord
strain analysis, and substitution effect among patients with migraine, head-
analysis, Drug Test. Anal. 10 (2018) 636e643.
ache, arthritis, and chronic pain in a medicinal cannabis cohort, J. Headache
[16] Q.F. Meng, B. Buchanan, J. Zuccolo, M.M. Poulin, J. Gabriele, D.C. Baranowski,
Pain 19 (2018) 1e28.
A reliable and validated LC-MS/MS method for the simultaneous quantifica-
[4] H. Canada, in: H. Canada (Ed.), Mandatory Cannabis Testing for Pesticide
tion of 4 cannabinoids in 40 consumer products, PLoS One 13 (2018).
Active Ingredients- List and Limits, Government of Canada, 2018, pp. 1e6.
[17] I. Di Marco Pisciottano, G. Guadagnuolo, V. Soprano, M. De Crescenzo, P. Gallo,
Online Resource.
A rapid method to determine nine natural cannabinoids in beverages and food
[5] L.M. Dryburgh, N.S. Bolan, C.P.L. Grof, P. Galettis, J. Schneider, C.J. Lucas,
derived from Cannabis sativa by liquid chromatography coupled to tandem
J.H. Martin, Cannabis contaminants: sources, distribution, human toxicity and
mass spectrometry on a QTRAP 4000, Rapid Commun. Mass Spectrom. 32
pharmacologic effects, Br. J. Clin. Pharmacol. 84 (2018) 2468e2476.
(2018) 1728e1736.
[6] N.A. dos Santos, L.V. Tose, S.R.C. da Silva, M. Murgu, R.M. Kuster, R.S. Ortiz,
[18] L. Zhang, X.W. Dou, C. Zhang, A.F. Logrieco, M.H. Yang, A review of current
F.A.O. Camargo, B.G. Vaz, V. Lacerda, W. Romao, Analysis of isomeric canna-
methods for analysis of mycotoxins in herbal medicines, Toxins 10 (2018).
binoid standards and cannabis products by UPLC-ESI-TWIM-MS: a compari-
[19] G.o. Canada, in: D.o. Justice (Ed.), Food and Drugs Act- Schedule B, Govern-
son with GC-MS and GC x GC-QMS, J. Brazilian Chem. Soc. 30 (2019) 60e70.
ment of Canada, 2018. Online Resource.
[7] Y.H. Wang, B. Avula, M.A. ElSohly, M.M. Radwan, M. Wang, A.S. Wanas,
[20] A. Alshannaq, J.H. Yu, Occurrence, toxicity, and analysis of major mycotoxins
Z. Mehmedic, I.A. Khan, Quantitative determination of (9) -THC, CBG, CBD,
in food, Int. J. Environ. Res. Public Health 14 (2017) 1e21.
their acid precursors and five other neutral cannabinoids by UHPLC-UV-MS,
[21] J.R. Moulins, M. Blais, K. Montsion, J. Tully, W. Mohan, M. Gagnon,
Planta Med. 84 (2018) 260e266.
T. McRitchie, K. Kwong, N. Snider, D.R. Blais, Multiresidue method of analysis
[8] F. Pellati, V. Brighenti, J. Sperlea, L. Marchetti, D. Bertelli, S. Benvenuti, New
of pesticides in medical cannabis, J. AOAC Int. 101 (2018) 1948e1960.
methods for the comprehensive analysis of bioactive compounds in cannabis
[22] T. Nuutinen, Medicinal properties of terpenes found in Cannabis sativa and
sativa L. (hemp), Molecules 23 (2018) 2639e2660.
Humulus lupulus, Eur. J. Med. Chem. 157 (2018) 198e228.
[9] C. Citti, D. Braghiroli, M.A. Vandelli, G. Cannazza, Pharmaceutical and
[23] S. Blasco-Benito, M. Seijo-Vila, M. Caro-Villalobos, I. Tundidor, C. Andradas,
biomedical analysis of cannabinoids: a critical review, J. Pharm. Biomed. Anal.
E. Garcia-Taboada, J. Wade, S. Smith, M. Guzman, E. Perez-Gomez, M. Gordon,
147 (2018) 565e579.
C. Sanchez, Appraising the "entourage effect": antitumor action of a pure
[10] M.W. Giese, M.A. Lewis, L. Giese, K.M. Smith, Development and validation of a
cannabinoid versus a botanical drug preparation in preclinical models of
reliable and robust method for the analysis of cannabinoids and terpenes in
breast cancer, Biochem. Pharmacol. 157 (2018) 285e293.
cannabis, J. AOAC Int. 98 (2015) 1503e1522.
[24] M. Parolini, S. Castiglioni, S. Magni, C. Della Torre, A. Binelli, Increase in
[11] L.A. Ciolino, T.L. Ranieri, A.M. Taylor, Commercial cannabis consumer products
cannabis use may indirectly affect the health status of a freshwater species,
part 2: HPLC-DAD quantitative analysis of cannabis cannabinoids, Forensic Sci.
Environ. Toxicol. Chem. 36 (2017) 472e479.
Int. 289 (2018) 438e447.
[25] E.M.V.W. Group, in: B. Magnusson, U. Ornemark (Eds.), Eurachem Guide: the
[12] B. Patel, D. Wene, Z.H. Fan, Qualitative and quantitative measurement of
Fitness for Purpose of Analytical Methods e A Laboratory Guide to Method
cannabinoids in cannabis using modified HPLC/DAD method, J. Pharm. Bio-
Validation and Related Topics, 2014. www.eurachem.org.
med. Anal. 146 (2017) 15e23.