Phytochemistry 71 (2010) 20582073

Contents lists available at ScienceDirect

Phytochemistry
journal homepage: www.elsevier.com/locate/phytochem

Metabolic ngerprinting of Cannabis sativa L., cannabinoids and terpenoids

for chemotaxonomic and drug standardization purposes
Justin Thomas Fischedick a,, Arno Hazekamp a, Tjalling Erkelens b, Young Hae Choi a, Rob Verpoorte a
a
Division of Pharmacognosy, Section Metabolomics, Institute of Biology, Leiden University, Leiden, The Netherlands
b
Bedrocan BV, P.O. Box 2009, 9640CA Veendam, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Cannabis sativa L. is an important medicinal plant. In order to develop cannabis plant material as a medic-
Received 26 July 2010 inal product quality control and clear chemotaxonomic discrimination between varieties is a necessity.
Received in revised form 20 September Therefore in this study 11 cannabis varieties were grown under the same environmental conditions.
2010
Chemical analysis of cannabis plant material used a gas chromatography ame ionization detection
Available online 30 October 2010
method that was validated for quantitative analysis of cannabis monoterpenoids, sesquiterpenoids,
and cannabinoids. Quantitative data was analyzed using principal component analysis to determine
Keywords:
which compounds are most important in discriminating cannabis varieties. In total 36 compounds were
Cannabis sativa
Cannabis
identied and quantied in the 11 varieties. Using principal component analysis each cannabis variety
Cannabinoids could be chemically discriminated. This methodology is useful for both chemotaxonomic discrimination
Terpenoids of cannabis varieties and quality control of plant material.
Metabolic ngerprinting 2010 Elsevier Ltd. All rights reserved.
Chemotaxonomy
Gas chromatography

1. Introduction AIDS patients, and multiple sclerosis. The increased medical inter-
est in these substances has prompted the development of various
Cannabis sativa L. (cannabis) is an annual dioecious plant cannabis based medicines such as the oral D9-THC preparation
belonging to the family Cannabaceae. Cannabis has a long history Marinol (Solvay Pharmaceuticals, Belgium), a synthetic analogue
of human use as a medicinal plant, intoxicant, and ritual drug of D9-THC Nabilone (Valeant Pharmaceuticals International,
(Russo, 2007). Today most nations worldwide regard cannabis as USA), and Sativex (GW Pharmaceuticals, UK) an oral mucousal
an illegal drug of abuse. Despite the abuse potential of cannabis re- spray containing 1:1 ratio of D9-THC and cannabidiol (CBD) (Ben
search into its chemistry and pharmacology has demonstrated that Amar, 2006; Hazekamp and Grotenhermen, 2010). Since 2003 The
it also has medical properties. Chemical analysis of cannabis in the Netherlands has allowed the distribution of standardized herbal
1940s and 1960s led to the discovery of a unique group of terpeno- cannabis in pharmacies to patients with a prescription (Hazekamp,
phenolic secondary metabolites, known as cannabinoids, of which 2006). In the USA 14 states have legalized under state law the use
trans-()-D9-tetrahydrocannabinol (D9-THC) was shown to be the of medical cannabis. In order to facilitate research into clinical
primary psychoactive ingredient (Pertwee, 2006). At least 90 plant safety and effectiveness the American Medical Association (AMA)
cannabinoids, also known as phytocannabinoids, have been iso- has recently called for the rescheduling of cannabiss legal status
lated from cannabis (Ahmed et al., 2008; ElSohly and Slade, from Schedule I to Schedule II (Hoffmann and Weber, 2010). These
2005; Radwan et al., 2009). In the early 1990s the G-protein cou- developments highlight the urgency to dene the criteria neces-
pled cannabinoid receptors (CB) were discovered. Two types of sary for the chemotaxonomic classication of medicinal cannabis
cannabinoid receptors CB1 and CB2 revealed a receptor based for drug standardization and clinical research purposes.
mechanism for the action of D9-THC (Pertwee, 2009). There has been considerable debate over whether or not whole
Clinical trials into cannabis, pure cannabinoids, and synthetic herbal cannabis has any additional therapeutic benets when
analogues have demonstrated some effectiveness as analgesics compared to pure cannabinoids (ElSohly et al., 2003; Llan et al.,
for chronic neuropathic pain, appetite stimulants for cancer or 2005; McPartland and Russo, 2001; Russo and McPartland, 2003;
Wachtel et al., 2002). However, there is some evidence that certain
cannabis preparations exhibit different effects when compared to
Corresponding author. Tel.: +31 71 527 4500; fax: +31 715274511. pure cannabinoids (Fairbairn and Pickens, 1981; Johnson et al., 1984;
E-mail address: [email protected] (J.T. Fischedick). Pickens, 1981; Ryan et al., 2006; Segelman et al., 1974; Whalley

0031-9422/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.phytochem.2010.10.001
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2059

et al., 2004; Wilkinson et al., 2003). Both the terpenes and minor cannabis varieties grown under standardized environmental con-
cannabinoids present in cannabis are known to have various bio- ditions. Principal component analysis (PCA) was used to identify
logical activities (McPartland and Russo, 2001). A lack of detailed the compounds most important in distinguishing cannabis varie-
chemical characterization beyond D9-THC, CBD or cannabinol ties. We also studied the variation on cannabis chemical proles
(CBN) quantication is shown in the above mentioned preclinical as a result of growing plants in different batches and with devia-
as well as clinical research making it difcult to compare results tions in growth time. This study establishes useful criteria for qual-
across studies (Ben Amar, 2006; Hazekamp and Grotenhermen, ity control and standardization of cannabis varieties for clinical
2010). It is not possible to draw any strong conclusions about what studies as well as chemotaxonomy.
components other than D9-THC and occasionally, depending on
the study design CBD, present in cannabis preparations may have
2. Results and discussion
an inuence on the drugs effects.
Cannabinoids are produced biosynthetically in cannabis as their
2.1. Plant material
carboxylic acid derivatives and are known as cannabinoid acids.
Cannabinoid acids degrade into their neutral counterparts through
Bedrocan BV (Groningen, The Netherlands) is a company li-
the action of heat, sunlight, and storage (Taura et al., 2007). Canna-
censed and contracted by the Dutch government to produce stan-
bis is most commonly administered by smoking the dried ower
dardized cannabis plant material under Good Agricultural Practice
buds due to the avoidance of rst pass metabolism of orally admin-
(GAP) conditions to be supplied to patients on prescription,
istered D9-THC as well as ease of self-titration by the user or pa-
through pharmacies (OMC, 2010). All plant material in these
tient (Williamson and Evans, 2000). In a recent study we
experiments was grown by Bedrocan BV. The varieties Bedrocan
demonstrated that cannabis ethanol extracts, smoke, and vapor
(Bedrocan), Bedropuur (Bedropuur), and Bediol (Bediol), have
produced by a vaporizing device are composed of a complex mix-
been breed by Bedrocan BV for use in medicine or research. All
ture of terpenoids and cannabinoids (Fischedick et al., 2010).
other varieties grown in this study are currently used for research
Therefore quality control methods for the major volatile com-
purposes only. In total 11 cannabis varieties were grown (Table 1).
pounds in cannabis should be utilized prior to and during clinical
Standard growth conditions are dened as the optimum vegetative
studies of cannabis administered with a vaporizing device or by
and owering growth times for each variety. The morphological
smoking.
type classication for each variety is based on morphological traits
Two morphological types of cannabis are commonly recognized,
as well as knowledge Bedrocan BV has of the varieties origin and
C. sativa being taller and more highly branched typically represent-
breeding history. Hybrids are described as having either equal
ing ber type varieties and Cannabis indica being shorter with
morphological traits from C. indica or C. sativa (i.e. hybrid indica/
broader leaves typically representing strains used for recreational
sativa) or having traits of both but mostly having traits representa-
or medicinal purposes. Whether or not these two morphotypes
tive of one of the morphotypes (i.e. hybrid mostly sativa). The let-
are different species is still a matter of debate (Russo, 2007). A
ter codes have no meaning other than to distinguish between
third subtype, Cannabis ruderalis has also been recognized, and is
varieties. All plants were grown from clones of a motherplant. A
described as having low levels of cannabinoids with a bushy
motherplant is dened as a female cannabis plant from one dis-
appearance (Hillig and Mahlberg, 2004). Today many cannabis
tinct variety used for cloning (vegetative propagation) only.
varieties used recreationally and for medical purposes are hybrids
Two female cannabis plants were grown for each batch and
of the various cannabis morphotypes mostly C. sativa and C. indica.
each growth treatment. Five random samples of dried ower mate-
Chemotaxonomic evaluation of cannabis has led to the recognition
rial were selected for the analysis of each batch and each growth
of three chemotypes, a drug type with higher levels of D9-THC, a
treatment. The purpose of growing plants in different batches
ber type with higher CBD, and an intermediate type with similar
and with deviations from standard growth conditions was to test
levels of each (Fetterman et al., 1971; Small and Beckstead,
the robustness of our chemical classication as well as determine
1973a,b). More recent studies using gas chromatography (GC)
the reproducibility of a cannabis varieties chemical prole. The
analyzing cannabinoids (Hillig and Mahlberg, 2004) or terpenoids
AO variety was grown in ve batches at the same time. Each batch
(Hillig, 2004) have been performed for chemotaxonomic purposes.
1 originated from a different seed from the same cannabis variety.
H NMR has been used to ngerprint cannabis aqueous extracts
Seeds were grown and female plants were selected for cloning.
and tinctures (Politi et al., 2008) as well as to chemically differen-
Each number for the AO variety thus denotes a different original
tiate cannabis cultivars (Choi et al., 2004). However, none of these
seed and its subsequent female clones. Therefore each AO batch
methods offer validated quantitative methods for the analysis of
was not genetically identical. For all other varieties the plants
cannabis terpenoids and cannabinoids simultaneously. Further-
grown were genetically identical. The AO7 batch was grown for
more the sample preparation used by Hillig (2004) for terpenoid
an extra week in the owering state. The varieties AG, AE, Ai94
analysis utilized extensive sample drying (2 months at room tem-
were each grown in three separate batches (1, 2, and 3). Batches
perature) and heating at 30 C prior to analysis. This would have
1 and 2 were grown about a month apart while batch 3 was grown
resulted in a higher rate of volatilization for the monoterpenoids
at the same time as 2 except with an extra week of vegetative
thus biasing the chemotaxonomic evaluation towards the less vol-
growth and an extra week of owering (Table 1). Bedrocan was
atile sesquiterpenoids.
grown in two batches at the same time. One batch had its lower
Metabolic ngerprinting, also known as metabolic proling, is a
branches clipped (c) while the other batch was grown under stan-
targeted analytical approach which aims to quantify a group or
dard conditions (Table 1). Bedropuur was grown in four batches at
groups of compounds found in an organism or group of organisms.
the same time with one batch grown under standard conditions
Metabolic ngerprinting with GC, HPLC, coupled with mass spec-
and the other three batches grown with deviations from standard
trometry, or 1H NMR is useful for studying plant biochemistry, che-
conditions (Table 1).
motaxonomy, ecology, pharmacology, and quality control of
medicinal plants (Van der Kooy et al., 2009). To metabolically n-
gerprint cannabis we validated a GCame ionization detection 2.2. Method validation
(GCFID) method for monoterpenoids, sesquiterpenes, and can-
nabinoids. The analytical method was used to study the chemical Results of GC method validation are summarized in Table 2. For
composition and variability of terpenoids and cannabinoids in 11 precision the percent relative standard deviation (RSD) of the peak
2060 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

Table 1
Cannabis plant information and growth conditions. Vegetative and owering columns show the number of days each sample was grown in each stage under standard conditions.

Variety Vegetative Flowering Morphological type Growth conditions

AG (1, 2, 3) 37 54 Hybrid indica/sativa 1 and 2 standard
3 vega +1 wkb, c +1 wk
AE (1, 2, 3) 37 54 Hybrid mostly sativa 1 and 2 standard
3 veg +1 wk, +1 wk
Ai94 (1, 2, 3) 37 54 Hybrid mostly sativa 1 and 2 standard
3 veg +1 wk, +1 wk
AO (1, 3, 5, 6, 7) 37 47 Indica 1,3,5,6 standard (four seeds)
7 +1 wk
AN 37 54 Indica Standard
AD 37 47 Indica Standard
AM 37 40 Indica Standard
AF 37 54 Indica Standard
Bedropuur (A, B, C, D) 37 40 Indica A veg 1 wk, +1 wk
B standard
C veg +1 wk
D veg +1 wk, +1 wk
Bedrocan (C) 37 54 Hybrid indica/sativa Bedrocan standard
C lower branches clipped
Bediol 29 54 Indica/sativa/ruderalis Standard
a
veg = vegetative.
b
wk = week.
c
= owering.

Table 2
Validation results.

Extract Intraday (n = 5) Interday (n = 15) TPa Mr = 136 RF

RSD RSD
Reproducibility Comparison RF
b-Pinene 0.4 0.5 c-Terpinene 4604.1
Myrcene 0.5 0.4 (R)-Limonene 4598.1
Limonene 0.7 0.5 a-Pinene 4625.3
1-Octanol 1.3 0.6 b-Pinene 4753.1
Linalool 1.2 0.6 (S)-Limonene 4648.8
b-Caryophyllene 1.3 0.5 Average 4645.9
Humulene 1.3 0.5 RSD 1.4
d-Guaiene 4.2 1.6 TP with oxygen
SQ(1) 1.9 0.6 Camphor 3894
SQ(2) 4.9 1.1 Linalool 3936.1
Elemene 1.7 0.6 1,8-Cineol 3827.2
Guaiol 1.5 0.7 Average 3885.8
c-Eudesmol 1.4 1.3 RSD 1.4
THCV 1.1 1.6 SQb Mr = 204
CBC 2.6 1.3 b-Caryophyllene 4753.9
CBGM 1.9 0.4 Humulene 4661.2
D9-THC 1.6 0.9 Average 4707.6
CBG 2 0.8 RSD 1.4
Pure compound SQ with oxygen
c-Terpinene 0.5 1.9 Caryophyllene-oxide 4285.3
Precision RSD (n = 120) Cannabinoids
1-Octanol 2.8 D9-THC 3490.1
Extraction efciency % Remaining (n = 3) D8-THC 3673.7
9
D -THC 2 CBD 3502.1
Accuracy % Recovery (n = 5) RSD CBN 3520.9
b-Pinene 102 3.2 CBG 3614.3
Linalool 102 2.2 Average 3560.2
b-Caryophyllene 100 3.5 RSD 2.2
a
TP = terpenoid.
b
Sesquiterpenoid.

area of 1-octanol from the 120 cannabis samples analyzed was cal- analyses. All compounds had a RSD of <5%. The reproducibility of
culated. The low RSD of 1-octanol (2.8%) indicates that the method a pure compound, c-terpinene had a RSD of <2%. These low RSD
was precise in terms of needle injections and FID response over the values indicate that the method is reproducible for the analysis
duration of the analytical period. This period consisted of 130 h of of cannabis terpenoids and cannabinoids.
GC time excluding calibration curves and other validation analyses. The extraction method chosen for this study had previously
Reproducibility was determined by comparing peak areas of each been demonstrated to be exhaustive and exhibit a high recovery
compound in a Bedropuur extract for both intraday and interday for the quantitative analysis of cannabinoids by HPLC (Hazekamp,
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2061

2007). Therefore we sought to determine whether or not the quiterpenoids with oxygen was 0.5 mg/g, and for cannabinoids was
extraction procedure utilized for cannabinoids was also exhaustive 0.6 mg/g. The signal to noise ratio was greater then 1:10 for all
for the terpenoids present in cannabis. Bedrocan was selected be- compounds present at a concentration above their LOQ. A signal
cause it has been shown to contain high levels of D9-THC and ter- to noise ratio of 1:5 was selected as the limit of detection (LOD)
penoids (Fischedick et al., 2010). By the fourth ethanol extract only and is therefore half that of the LOQ for each compound group.
2% D9-THC compared to the total peak area of D9-THC in the rst The response factor (RF) for each compound is shown in Table 2.
three extracts remained. This was consistent with previous results The low variability in RF among compounds with similar mass
concerning the recovery of cannabinoids with this extraction and chemical structure using the FID is consistent with other re-
method (Hazekamp, 2007). No other residual compounds were search done on the quantication of essential oils with GCFID
detected in the forth extract indicating that the method is also (Bicchi et al., 2008). Therefore terpenoids or cannabinoids for
exhaustive for the extraction of terpenoids in cannabis. Accuracy which no reference is available could be accurately quantied with
of the extraction method was demonstrated by determining the components of similar or identical molecular mass/formula.
recovery of spiked terpenoids. We selected Bedrobinol plant
material for this experiment because in previous studies 2.3. Qualitative and quantitative analysis of cannabis varieties
(Fischedick et al., 2010) this plant material was shown to have
low levels of b-pinene and b-caryophyllene with no detectable lev- Compounds were identied by comparing their mass spectra,
els of linalool. All terpenoids were completely recovered indicating and retention times with authentic references as well as literature
that the method is accurate for the analysis of cannabis terpenoids reports (Adams, 1989; Hillig, 2004; Komori et al., 1968; Ross and
(Table 2). ElSohly, 1996; Rothschild et al., 2005). The NIST library was also
Linear standard curves (r2 > 0.99) for all compounds tested used to assist in compound identication. A summary of quantita-
could be generated in the range of 0.01 mg/ml to 1 mg/ml. For tive data for all compounds in all 11 cannabis varieties is shown in
D9-THC linear standard curves up to 2 mg/ml could be generated. Table 3. The compounds c-terpinene, limonene, a-pinene,
The limit of quantication (LOQ) for monoterpenoids with a molec- b-pinene, linalool, b-caryophyllene, humulene, D9-THC, trans-()-
ular weight (Mr) of 136 was 0.4 mg/g, monoterpenes with oxygen D8-tetrahydrocannabinold (D8-THC), CBD, and cannabigerol
was 0.5 mg/g, sesquiterpenoids with a Mr of 204 was 0.4 mg/g, ses- (CBG) were quantied using their standard curve RF values. All

Table 3
Quantitative data for cannabis strains. Compounds with no had standard deviations of <100 lg.

Compound RRta AO Bedropuur Bedrocan Bediol AG AE Ai94 AN AF AM AD

Number of samples 25 20 10 5 15 15 15 5 5 5 5
a-Pinene 0.26 6.7 2.4 0.5 0.8 0.1 1.7 0.1 2.8 0.4 Trc 0.8 0.5 Tr 0.5 3.2 0.1 NDb
b-Pinene 0.31 1.9 0.6 0.7 0.1 1.2 0.2 0.8 0.1 1.4 0.2 Tr 0.5 0.1 Tr 0.7 1.2 Tr
Myrcene 0.33 14.8 7.3 1.6 0.7 6.1 1 19 1.9 13 1.6 0.4 0.1 7.1 0.8 0.8 0.2 3.4 0.1 6.7 0.6 13.8 0.9
a-Phellandrene 0.35 ND ND 0.6 0.1 Tr ND Tr Tr ND Tr ND ND
D3-Carene 0.36 ND ND 0.5 0.2 Tr ND Tr Tr ND Tr ND ND
a-terpinene 0.36 ND ND 0.4 Tr ND Tr Tr ND Tr ND ND
b-Phellandrene 0.38 ND ND 2.1 0.3 0.7 0.1 ND Tr 0.5 ND 1.0 ND Tr
Limonene 0.38 2.5 0.7 4.9 0.8 ND ND 2.4 0.4 ND ND 2.3 0.3 ND 0.7 0.1 ND
cis-Ocimene 0.40 1.4 0.4 ND 3.9 0.6 1 0.1 0.6 0.1 0.7 0.1 1 0.1 ND 1.0 0.1 ND ND
Terpinolene 0.47 ND ND 11.3 1.7 3.7 0.4 ND 1.9 0.9 2.9 0.3 ND 5.4 0.2 ND ND
Linalool 0.48 0.9 0.1 1.2 0.2 Tr ND Tr ND Tr Tr Tr Tr Tr
Terpineol 0.63 Tr Tr 0.7 0.1 0.6 ND Tr Tr Tr Tr Tr ND
TP(1) 0.86 ND ND 0.7 0.2 Tr ND Tr ND ND Tr ND ND
b-Caryophyllene 1.00 3.5 2.4 2.6 0.2 1.6 0.2 0.8 0.1 1.4 0.4 0.5 0.3 0.8 0.4 1.4 0.2 0.5 0.5 2.0 0.1
a-Guaiene 1.03 ND Tr Tr Tr 0.5 ND Tr ND Tr Tr Tr
Humulene 1.05 1.2 0.9 0.8 0.1 0.9 0.1 0.6 0.1 0.7 0.1 Tr 0.7 0.1 Tr Tr ND 0.6
d-Guaiened 1.13 ND 0.7 0.1 0.8 0.1 0.8 0.7 0.2 ND ND ND ND ND 0.6
SQ(1) 1.18 0.5 0.1 0.6 0.1 0.6 0.1 Tr Tr ND ND 1.3 0.1 Tr ND ND
SQ(2) 1.19 0.6 0.2 ND ND ND ND ND ND ND ND ND ND
SQ(3) 1.19 ND 1.0 0.2 0.6 0.1 Tr 0.5 0.1 Tr ND 1.8 0.1 ND ND ND
Elemene 1.21 1.1 0.3 2.3 0.4 1.3 0.3 Tr 0.6 0.1 0.8 0.2 ND 2.7 0.6 Tr Tr ND
Guaiol 1.27 0.6 0.2 0.6 ND ND ND 0.4 ND ND Tr ND ND
c-Eudesmol 1.30 0.7 0.2 0.6 ND ND ND 0.4 ND ND Tr ND ND
b-Eudesmol 1.34 0.4 Tr ND ND ND Tr ND ND Tr ND ND
Agarospirol 1.35 0.5 0.1 Tr ND ND ND Tr ND ND Tr ND ND
Bulnesold 1.37 0.5 ND ND ND ND ND ND ND ND ND ND
a-Bisabolol 1.39 0.5 0.1 Tr ND ND ND ND ND 0.5 0.1 ND ND Tr
CBDV 2.04 ND ND ND Tr ND ND 1.5 ND ND ND ND
THCV 2.15 1.3 0.4 1.3 0.3 1.3 0.2 Tr 0.7 0.1 Tr ND 0.6 0.7 0.1 1 0.1 0.6
CBD 2.27 Tr Tr Tr 79.8 1.8 ND ND 73.6 2.1 ND1 ND Tr ND
CBC 2.27 2.1 0.4 2.1 0.3 2.3 0.1 5.4 0.1 1.4 0.1 1.6 0.8 4.6 0.3 0.9 0.1 0.9 1.4 0.1 2.2 0.2
CB(1) 2.32 0.8 0.5 1.1 0.2 1.6 0.3 1.2 0.7 ND ND Tr Tr Tr Tr
CBGM 2.32 ND 1.1 0.1 ND ND ND ND ND 1.8 0.3 2.6 0.1 Tr Tr
D8-THC 2.33 Tr 0.7 0.1 Tr Tr ND ND 1.2 0.1 ND 0.6 Tr Tr
199.2 181 207.5 61.5 144.1 26 3.4 95.2 87.2 120.1 134.9
D9-THC 2.37 33 21 19 1.9 16 7 0.6 6.7 3.4 5.5 6.6
CBG 2.42 10 1.6 4.1 1.3 11.2 2 1.7 0.2 2.8 0.6 1 0.4 ND 1.9 0.2 3.0 0.3 2.1 0.1 1.2 0.1
a
RRt relative retention time to b-caryophyllene.
b
ND = not detected (<LOD).
c
Tr = trace levels detected (<LOQ).
d
Putative identication.
2062 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

other compounds were quantied using the average RF for com- involved in enhancing the levels of these compounds originate
pounds with the closest molecular mass/formula. In total 36 com- from C. indica and is not commonly present in C. sativa subtype
pounds were quantied. (Hillig and Mahlberg, 2004). CBD was only detected in trace
The sesquiterpenoids d-guaiene and bulnesol are reported as amounts (<0.6 mg/g) or not at all among the high, medium, and
putatively identied because they have not been reported in can- low D9-THC varieties making them representatives of the drug/
nabis previously and a reference compound was not available for D9-THC chemotype. Bediol is representative of the intermediate
structural conrmation. Five compounds could not be identied chemotype and Ai94 is representative of the ber/CBD chemotype.
so we report their characteristic mass ions. Unknown monoterpe- The levels of D9-THC and CBD alone do not chemically distinguish
noid TP(1) m/z: 152 [M+], 91, 84, 69 (base). The unknown sesquit- the high or medium D9-THC containing varieties well from one
erpenoids (SQ) SQ(1) m/z: 204 [M+], 161 (base), 133, 105; SQ(2) m/ another.
z: 236 [M+], 204, 161, 119, 93 (base); SQ(3) m/z: 204 [M+], 161 Therefore to further chemically classify cannabis principle com-
(base), 122, 93. The unknown cannabinoid CB(1) m/z: 356 [M+], ponent analysis (PCA) was used. PCA is a multivariate projection
313, 297 (base), 243, 231. The monoterpenoid TP(1), we suspect method which extracts and displays systemic variation from a
is oxygen substituted due to its Mr of 152. The unknown sesquit- set of matrix data consisting of observations and variables (Eriks-
erpenoids SQ(1) and SQ(3) appear to have been reported as un- son et al., 2006). The 36 compounds were the variables and their
knowns in previous studies (Hillig, 2004; Ross and ElSohly, mg/g levels the observations. Initially all the cannabis samples
1996). SQ(2) is a sesquiterpenoid with unknown substitution. No were analyzed by PCA (Fig. 2). Principal component 1 (PC1) and
detectable levels of the D9-THC breakdown product CBN were de- PC2 explains 44% of the variance. The highest D9-THC containing
tected in any samples. This indicates that the drying and storage varieties Bedropuur, Bedrocan, and AO are separated along the po-
process used in this study resulted in no signicant amount of sitive PC1. The Bedrocan variety was also well separated along neg-
D9-THC degradation except perhaps that of trans-()-D9-tetrahy- ative PC2. The compounds responsible for making Bedrocan
drocannabinbolic acid A (THCA) into D9-THC. different according to the loading plot are terpinolene, b-phelland-
Higher levels of cannabinoids were positively correlated to rene, a-phrellandrene, terpineol, cis-ocimene, and D3-carene.
higher levels of terpenoids (Fig. 1). Both the cannabinoids and ter- Bedrocan contained higher levels of these compounds compared
penoids of cannabis are localized primarily in glandular trichomes with other varieties. Terpinolene was a very dominant monoterpe-
(Malingre et al., 1975; Taura et al., 2007). This may explain why noid (11.3 mg/g) in the Bedrocan variety (Table 3). Bediol partially
in some varieties their levels are correlated however it does not separated along the negative PC2 also contained terpinolene,
prove that their biosynthesis is necessarily correlated. This is b-phellandrene, a-phellandrene, terpineol, and D3-carene but in
demonstrated by the Bedropuur variety whose D9-THC levels lower levels than Bedrocan. This observation is interesting because
are high but its terpenoid levels are similar to the varieties AD the Bediol variety was bred by hybridizing the Bedrocan variety
and AG. This suggests that it is possible to breed cannabis that with higher CBD containing varieties.
contains high levels of D9-THC but not necessarily higher levels The loading plot along the positive PC1 and positive PC2 shows
of terpenoids. that Bedropuur and AO contained more of the sesquiterpene alco-
hols guaiol, c-eudesmol, b-eudesmol (Fig. 2). The study by Hillig
2.4. Metabolic ngerprinting of cannabis (2004) reported that guaiol, c-eudesmol, and b-eudesmol were
characteristic terpenoid compounds of the C. indica varieties origi-
It is clear from the data that each cannabis variety is both qual- nating from Afghanistan. These sesquiterpenoid alcohols appear to
itatively and quantitatively different (Table 3). The levels of D9- be important in distinguishing C. indica varieties from one another
THC ranged from 20.8% (Bedrocan) to 0.3% (Ai94). Bedrocan, Bed- because the AG, AN, AM, and AD varieties which are also C. indica
ropuur, and AO all contained high levels of D9-THC (>15%). AG, morphotypes did not contain detectable levels of these com-
AM, AD, AN, and AF all contained a medium level of D9-THC pounds. AF another C. indica only contained trace amounts of these
(<15%, >5%). Bediol also contained a medium level of D9-THC compounds (Table 3). In order to distinguish Bedropuur and AO
(6%) however its relatively high level of CBD (8%) makes it unique further PC1 and PC3 were compared (Fig. 3). PC3 was able to ex-
compared to the other varieties. AE and Ai94 contained low plain an additional 15% of the variance. The Bedropuur variety con-
amounts of D9-THC (<5%). Ai94 contained a relatively high level tained higher levels of limonene as well as the sesquiterpenoid
of CBD (7.4%) compared to the other varieties. Ai94 also contained elemene while the AO variety contained higher levels of myrcene
the C3 side chain variant of CBD, ()-cannabidivarin (CBDV). Bediol and a-pinene. Also along PC3 information was obtained about
only contained trace levels of CBDV. The levels of propyl side chain the AN variety which contains a medium level of D9-THC
analogues of cannabinoids have been reported to be of chemotax- (95.2 mg/g) but higher levels of the sesquiterpenoids SQ(1),
onomic signicance. It has been hypothesized that the enzymes SQ(3), and elemene when compared to all other varieties.

Fig. 1. Correlation of cannabinoid versus terpenoid levels.

 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2063

Fig. 2. PCA of all cannabis varieties. PC1 versus PC2, scatter plot (top) and loading plot (bottom). (1) AO, (2) Bedropuur, (3) Bedrocan, (4) Bediol, (5) AG, (6) AE, (7) Ai94, (8) AN,
(9) AF, (10) AM, and (11) AD.

The medium D9-THC varieties AG, AF, AM, and AD were not well most likely because it has the lowest amount of D9-THC compared
separated along PC1, PC2, or PC3. Therefore these varieties were to the other medium varieties. Each medium D9-THC variety is
reanalyzed by PCA with all other varieties excluded (Fig. 4). AG clustered with itself except AG because it was grown in different
and AD had higher and very similar levels of D9-THC, myrcene, batches which caused small differences in chemical prole. The
and b-caryophyllene compared with AM and AF. AG and AD were morphotype of each medium D9-THC variety does not seem impor-
distinguished along PC2. AG contained more a-pinene, b-pinene, tant in relating these varieties.
limonene, a-guaiene, elemene, and SQ(3). AD however contained These observations represent a signicant improvement com-
low levels of monoterpenoids and sesquiterpenoids in general pared with other methodologies, discussed in the introduction,
and only slightly higher levels (<1.0 mg/g) of myrcene, b-caryo- using chemotaxonomy to discriminate cannabis varieties. By using
phyllene, and CBC compared to AG. AF contained the highest levels quantitative data on cannabinoid and terpenoid levels it was pos-
of cannabigerol monomethyl ether (CBGM) while AM contained sible to chemically distinguish each variety from one another with
higher levels of myrcene and a-pinene. the aid of PCA. Both Hillig (2004) and Hillig and Mahlberg (2004)
Hierarchical cluster analysis (HCA) was used to conrm the PCA had difculty discriminating drug type cannabis accessions from
analysis (Fig. 5). The AO batches are all clustered together and each one another. Furthermore the conclusion in the study of Hillig
genotype (different seed) is also grouped together. The Bedropuur (2004) that sesquiterpenoids were more important then monot-
batches are clustered together and AN was the next similar variety. erpenoids in chemically differentiating cannabis varieties is not
Both Bedropuur and AN were separated along the negative PC3 accurate. In this study monoterpenoids were able to distinguish
(Fig. 3) because of the presence of the cannabinoid CBGM as well varieties which had similar sesquiterpenoid levels and similar can-
as numerous similarities in monoterpenoids and sesquiterpenoids nabinoid levels such as AO and Bedropuur as well as a number of
(Table 3). Bedrocan was in its own group which is consistent with the medium D9-THC varieties.
PCA analysis. The clipped Bedrocan batch exhibits some differences
according to HCA when compared to unclipped. Bediol and Ai94 2.5. Effect of growing cannabis in different batches and growth cycle
are related due to higher levels of CBD but each clustering on their deviations
own. Ai94(2) however was clustered closer too AE and AF most
likely because of small differences caused by growing this variety The effect on chemical prole from growing cannabis varieties
in different batches. The medium D9-THC varieties are all clustered in separate batches about a month apart as well 1 week extra veg-
close to one another except AF. AF was closer to AE and Ai94(2) etative and owering periods was studied in the AG, AE, and Ai94
2064 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

Fig. 3. PCA of all cannabis varieties. PC1 versus PC3, scatter plot (top) and loading plot (bottom). (1) AO, (2) Bedropuur, (3) Bedrocan, (4) Bediol, (5) AG, (6) AE, (7) Ai94, (8) AN,
(9) AF, (10) AM, and (11) AD.

varieties. A comparison of the compounds within the AG varieties the other three batches. Bedropuur D had the lowest amount of
batches is shown in Fig. 6. The differences between each batch D9-THC. Bedropuur A had lower concentrations of the b-caryo-
were minor with no clear distinction between them. The largest phyllene, elemene, and CBG when compared to batches B and C.
differences are between AG(1) and AG(2) with the level of myrcene These results demonstrate that alterations in growth cycle time
being 1.8 mg/g higher on average in AG(1) compared with AG(2), can cause changes in the chemical prole of cannabis plants
b-caryophyllene being 0.5 mg/g higher on average in AG(2) com- grown under environmental conditions that were otherwise the
pared to AG(1), CBG being 0.8 mg/g higher in AG(2) compared to same. Alterations in growth cycle time appear to cause more dif-
AG(1), and D9-THC being 17.3 mg/g higher in AG(2) compared to ferences in a cannabis varieties chemical prole then growing
AG(1). Most compounds in the AE batches did not differ much in the plant material in different batches. However more experi-
concentration (<1.0 mg/g), except terpinolene and D9-THC ments with more varieties, grown with more deviations in growth
(Fig. 7). The Ai94 batches also only had minor differences (Fig. 8). cycle time, and more replicates would be needed to conrm these
These results importantly demonstrate that genetically identical observations.
cannabis plants grown in batches at separate times under stan- Clipping the lower branches on the Bedrocan variety caused
dardized environmental conditions are reproducible in terms of some compounds to be present at lower concentrations (Fig. 10).
terpenoid and cannabinoid concentrations. These compounds include myrcene, cis-ocimene, b-caryophyllene,
A detailed look into the chemical variation among the Bed- elemene, CBG, and D9-THC. This suggests that by clipping the low-
ropuur batches is shown in Fig. 9. Batches A, C, and D differed in er branches, which would allow more water and nutrients to ow
the concentrations of certain compounds compared with the stan- to the upper parts of the plant closest to the light, does cause some
dard batch B. The levels of limonene were lower in A, C, and D. changes in the chemical prole. Further experiments would be
Myrcene was lower in A and D compared with B and C. The levels needed to determine if this represents a consistent pattern and ex-
of D9-THC were about 30 mg/g higher in Bedropuur C compared to plain why it occurs.
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2065

Fig. 4. PCA of the varieties AG, AD, AM, and AF. PC1 versus PC2 scatter plot (top) and loading plot (bottom).

Fig. 5. Hierarchical clustering analysis of all cannabis samples.

2066 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

Fig. 6. Comparison of AG batches. Compounds that are missing in certain batches were present at levels <LOQ.

The different AO batches exhibited the greatest quantitative dif- observed in the HCA. Cannabis plants from seeds representing dif-
ferences in chemical prole compared with all other varieties ferent genotypes but the same variety can differ considerably in
(Fig. 11). AO batches exhibited a wide range of concentrations for quantitative chemical prole. Future research should aim to deter-
a-pinene, myrcene, b-caryophyllene, and D9-THC. The different mine if cannabis could be grown in such a reproducible manner for
AO batches could even be clearly distinguished by PCA (Fig. 12). many years. As a preliminary indication of chemical prole repro-
This observation shows that by metabolically proling cannabis ducibility a previous study in our laboratory using similar method-
strains based on cannabinoid and terpenoid levels it is also possi- ology analyzed the Bedrocan variety. This plant material was
ble to distinguish separate genotypes of the same variety. grown about 1 year previously to the batches analyzed in this
Overall these experiments demonstrate that the best way to study. This batch had similar levels of the main compounds ob-
grow reproducible batches of cannabis is by using genetically iden- served in the present study (Fischedick et al., 2010).
tical plant material grown from clones, under standardized envi-
ronmental conditions, with the same growth cycle. Deviations in 3. Concluding remarks
growth cycle and clipping of lower branches can cause quantitative
differences, although minor in absolute terms, in chemical prole. In this study a simple quantitative GCFID method was vali-
These deviations can obscure their chemical classication as was dated for the quantitative analysis of cannabis monoterpenoids,
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2067

Fig. 7. Comparison of AE batches. Compounds that are missing in certain batches were present at levels <LOQ.

Fig. 8. Comparison of Ai94 batches. Compounds that are missing in certain batches were present at levels <LOQ.

sesquiterpenoids, and cannabinoids. Quantitative GC data was of medicinal cannabis. Our methodology also appears to be able to
used to chemically discriminate cannabis varieties with the aid overcome the difculties in chemotaxonomic analysis of cannabis
of principal component analysis. Our results show for the rst time observed by other researchers in distinguishing drug type cannabis
using validated methodology the absolute (mg/g) levels of cannab- varieties from one another. These techniques should be applied on
inoids and terpenoids in cannabis simultaneously. This data can be a wider range of cannabis samples representing both geographi-
useful for guiding pharmacological or clinical studies that want to cally and morphologically distinct varieties. By combining genomic
examine the potential interactions of the volatile constituents of approaches with metabolic ngerprinting it may be possible to
cannabis. The chemical prole of cannabis varieties could poten- elucidate exactly which biochemical pathways differ in various
tially be more closely correlated to therapeutic effectiveness. The cannabis varieties and how these differences lead to the observed
reported methodology could be implemented in the quality control chemical prole.
2068 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

Fig. 9. Comparison of Bedropuur batches. Compounds that are missing in certain batches were present at levels <LOQ.

4. Experimental lene, citronellal, geranyl acetate, pulegone, citral, a-terpinene,

a-fenchyl alcohol, calamanene, c-cadinene, bornyl acetate, a
4.1. Chemicals mixture of cis/trans-ocimene, a-cedrene, a-phellandrene, nerol,
b-phellandrene, nerolidol, and piperitone-oxide were from a chem-
Reference terpenoids of caryophyllene-oxide, camphor, ical bank of the authors. The cannabinoid references for D9-THC,
a-bisabolol, b-pinene, myrcene, a-pinene, c-terpineol, (R)-limonene D8-THC, CBD, cannabigerol (CBG), cannabichromene (CBC), trans-
(limonene), (S)-limonene, 1,8-cineol, carvacrol, and b-caryophyl- ()-D9-tetrahydrocannabivarin (THCV), and cannabinol (CBN)
lene were purchased from SigmaAldrich (Steinheim, Germany). were puried and quantied by PRISNA BV as previously described
Terpineol mixture of isomers, a-humulene, and linalool were pur- (Leiden, The Netherlands) (Hazekamp et al., 2004a,b). All cannabi-
chased from Fluka (Steinheim, Germany). Geraniol was purchased noids references were >98% pure, except THCV. Absolute ethanol
from Chromadex (Irvine, California, USA). Camphene, a-thujene, (EtOH) used for extraction and sample preparation was of
sabinene, terpinene-4-ol, 1,4-cineol, D3-carene, p-cymene, terpino- analytical reagent (AR) grade (Biosolve BV, Valkenswaard, The
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2069

Fig. 10. Comparison of Bedrocan batches. Compounds that are missing in certain batches were present at levels <LOQ.

Netherlands). 1-Octanol (HPLC grade) was purchased from Sigma ing only roots, stems, and leaves. After an optimized vegetative
Aldrich (Steinheim, Germany). period plants are switched to 12 h of uninterrupted light per day
which induces owering. The period for which each variety exists
4.2. Plant material in each phase can differ and has been optimized by Bedrocan BV
for efcient growth.
Cannabis plant material was grown indoors. The plant material Environmental conditions for all varieties were the same. Plants
was produced by taking cuttings from standardized plants (mother were harvested after a standardized amount of days when the pis-
plants) kept under vegetative conditions. Cannabis plants were tils faded from white to brown and the branches started to hang.
grown in two growth cycles. First a vegetative period in which The plants were then dried under the same environmental condi-
plants are grown under 18 h of uninterrupted light per day produc- tions. After 1 week drying the plant material lost 73% of its weight.
2070 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

Fig. 11. Comparison of AO seed batches. Compounds that are missing in certain batches were present at levels <LOQ.

The plants were then processed by removing leaves from the buds with a few milliliters of EtOH into the falcon tube. The volume was
and clipping buds from the main stems. Remaining plant material then brought up to 45 ml with ethanol. Falcon tubes were placed
(buds) was packaged into 50 ml falcon tubes and stored at 20 C on a Yellow Line Orbital Shaker OS 2 Basic (IKA GmbH, Staufen,
until extraction. Germany) at 400 rpm (rpm) for 15 min. Samples were centrifuged
briey for 30 s at 2000 rpm. Supernatant was collected in a 100 ml
4.3. Sample preparation glass volumetric ask. Samples were extracted two more times
with 25 ml ethanol. As an internal standard 1 ml of an EtOH soln.
Cannabis plant material was weighed to the nearest mg with a containing 1-octanol (1%) was added to the volumetric asks.
typical weight range of 0.91.1 g. The plant material was crushed Samples were nally brought to a volume of 100 ml with ethanol.
with a metal spoon within a falcon tube and the spoon was rinsed Samples were ltered into 20 ml glass vials with a PTFE syringe
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2071

Fig. 12. PCA loading plot PC1 versus PC2 of AO seed batches.

lter (0.45 lM, 25 mm diameter). Samples were stored air tight in gram of the National Institute of Standards and Technology, Dis-
the dark at 20 C until analysis. tributed by Agilent Technologies) was used to assist compound
identication.
4.4. GCFID
4.6. Standards preparation
An Agilent GC 6890 series equipped with a 7683 autosampler, a
DB5 (30 m length, 0.25 mm internal diameter, lm thickness Mono and sesquiterpenoid references were weighed to 50 mg in
0.25 lm, J&W Scientic Inc., Folsom, California, USA) column and a tarred volumetric ask using a Satorius analytical balance A200S
a ame ionization detector (FID) was used for quantitative analy- 0.0001 mg (Satorius Mechatronics, Utrecht, The Netherlands). Vol-
sis. The injector temperature was set to 230 C, an injection volume ume was brought to 25 ml with EtOH to make 2 mg/ml stock solu-
of 4 ll, a split ratio of 1:20 and a carrier gas (N2) ow rate of 1.2 ml/ tions. Stock solutions were used to make dilutions for standard
min. The oven temperature program began at 60 C with a ramp curves. Stocks were stored at 20 C in sealed glass vials in the
rate of 3 C/min. The nal temperature was set to 240 C which dark until needed. Cannabinoid references were supplied already
was held for 5 min making a total run time of 65 min/sample. quantied in EtOH. References were diluted in EtOH to make stan-
The FID detector temperature was set to 250 C. The GCFID was dard curves. Cannabinoid references were stored at 20 C in am-
controlled by GC Chemstation software version B.04.01[481] (Agi- ber sealed glass vials in the dark until needed.
lent Technologies Inc., Santa Clara, California, USA).

4.7. Method validation

4.5. GCMS

4.7.1. Reproducibility
GCMS analysis was performed on an Agilent 7890A series gas
Intraday reproducibility was determined by injecting an aliquot
chromatograph equipped with a 7693 autosampler, an HP5-ms
of a cannabis extract ve times from the same vial in a single day
column (30 m length, 0.25 mm internal diameter, lm thickness
and a reference sample of c-terpinene (1 mg/ml) ve times from
0.25 lm, Agilent Technologies Inc., Santa Clara, California, USA)
the same vial in a single day (n = 5). Interday reproducibility was
and a single quadropole mass spectrometer 5975C. The MS source
determined by taking a fresh aliquot of the same cannabis extract
was set to 230 C, the single quad temperature was 150 C, and the
and c-terpinene reference and injecting ve times for an additional
transfer line temperature was set to 280 C. The GC-column was
two days (n = 15) using fresh aliquots on each day. All injections
linked to the MS via a quickswap (Agilent Technologies Inc., Santa
performed on the GCFID.
Clara, California, USA) and restrictor (0.11 mm internal diameter,
Agilent Technologies, Santa Clara, USA). The injector temperature
was 230 C with an injection volume of 2 ll, a split ratio of 1:20 4.7.2. Extraction efciency
and a carrier gas (He) ow rate of 1.2 ml/min. The oven tempera- Three 1 g samples of a batch of Bedrocan that had been used in
ture program was the same as the GCFID. The mass range ana- previous studies (Fischedick et al., 2010) and stored for 7 months
lyzed by the mass spectrometer was 50500 amu. The GCMS at 4 C in the dark was extracted with the procedure outlined
was controlled by Enhanced Chemstation software version above. After three extractions a fourth extraction was performed
E.02.00.493 (Agilent Technologies Inc., Santa Clara, California, on each Bedrocan sample with an additional 25 ml of ethanol
USA). The NIST library version 2.0f (Standard Reference Data Pro- and analyzed for residual compounds by GCFID.
2072 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073

4.7.3. Accuracy Eriksson, L., Johansson, E., Kettaneh-Wold, N., Tyrgg, J., Wikstrm, C., Wold, S., 2006.
Multi- and Megavariate Data Analysis Part 1: Basic Principles and Applications,
Accuracy was determined by checking the recovery of the extrac-
second ed. Umetrics, Ume, Sweden.
tion method with spiked monoterpenoids and sesquiterpenoids. Fairbairn, J.W., Pickens, J.T., 1981. Activity of cannabis in relation to its D0 -trans-
Five 1 g samples of a batch of Bedrobinol used in previous studies tetrahydrocannabinol content. Br. J. Pharmacol. 72, 401409.
(Fischedick et al., 2010) and stored for 7 months at 4 C in the dark Fetterman, P.S., Keith, E.S., Waller, C.W., Guerrero, O., Doorenbos, N.J., Quimby,
M.W., 1971. Mississippi-grown Cannabis sativa L. preliminary observation on
was extracted as outlined above. Five 1 g samples of the same batch chemical denition of phenotype and variations in tetrahydrocannabinol
of Bedrobinol were spiked with 50 ll of the pure references of content versus age, sex, and plant part. J. Pharm. Sci. 60, 12461477.
b-pinene, linalool and b-caryophyllene while in falcon tubes then Fischedick, J., Van der Kooy, F., Verpoorte, R., 2010. Cannabinoid receptor 1 binding
activity and quantitative analysis of Cannabis sativa L. smoke and vapor. Chem.
extracted as described above. Five volumetric asks were spiked Pharm. Bull. 58, 201207.
with 50 ll of the pure references of b-pinene, linalool and b-caryo- Hazekamp, A., 2006. An evaluation of the quality of medicinal grade cannabis in the
phyllene and brought to 100 ml with EtOH. All samples analyzed Netherlands. Cannabinoids 1, 19.
Hazekamp, A., 2007. Cannabis Extracting the Medicine. Ph.D. Thesis, Universiteit
by GCFID. Percent recovery was calculated by subtracting the peak Leiden, Leiden, The Netherlands.
area of each terpenoid from the spiked samples minus the un-spiked Hazekamp, A., Grotenhermen, F., 2010. Review on clinical studies with cannabis and
controls. This number was then divided by the peak area of pure cannabinoids 20052009. Cannabinoids 5, 121.
Hazekamp, A., Choi, Y.H., Verpoorte, R., 2004a. Quantitative analysis of
references diluted in 100 ml ethanol and multiplied by 100. cannabinoids from Cannabis sativa using 1H-NMR. Chem. Pharm. Bull. 52,
718721.
4.7.4. Linear range, LOD, LOQ, and RF Hazekamp, A., Simons, R., Peltenburg-Looman, A., Sengers, M., Van Zweden, R.,
Verpoorte, R., 2004b. Preparative isolation of cannabinoids from Cannabis sativa
The linear range was determined empirically by injecting stan- by centrifugal partition chromatography. J. Liq. Chromatogr. Relat. Technol. 27,
dard compounds in a range of 0.012 mg/ml. The LOD and LOQ 24212439.
were determined empirically and using signal to noise calculations Hillig, K.W., 2004. A chemotaxonomic analysis of terpenoid variation in cannabis.
Biochem. Syst. Ecol. 32, 875891.
with Chemstation software. The detector response for monoterpe- Hillig, K.W., Mahlberg, P.G., 2004. A chemotaxonomic analysis of cannabinoid
noids, sesquiterpenoids, and cannabinoids was determined by run- variation in cannabis (cannabaceae). Am. J. Bot. 91, 966975.
ning standard curves (0.021.0 mg/ml) of c-terpinene, limonene, Hoffmann, D.E., Weber, E., 2010. Medical marijuana and the law. N. Engl. J. Med.
362, 14531457.
a-pinene, b-pinene, (S)-limonene, camphor, linalool, 1,8-cineol, b- Johnson, J.M., Lemberger, L., Novotny, M., Forney, R.B., Dalton, W.S., Maskarinec,
caryophyllene, humulene, caryophyllene-oxide, D9-THC, D8-THC, M.P., 1984. Pharmacological activity of the basic fraction of marihuana whole
CBD, CBN, and CBG in duplicate. smoke condensate alone and in combination with delta-9-
tetrahydrocannabinol in mice. Toxicol. Appl. Pharmacol. 72, 440448.
Komori, T., Nohara, T., Hosokawa, I., Kawasaki, T., 1968. Cannabigerol monomethyl
4.7.5. Instrumental precision ether, a new component of hemp. Chem. Pharm. Bull. 16, 11641165.
The variation in peak area of the internal standard 1-octanol for Llan, A.B., Gevins, A., Coleman, M., ElSohly, M.A., de Wit, H., 2005.
Neurophysiological and subjective prole of marijuana with varying
all cannabis samples was used to determine precision of the GC
concentrations of cannabinoids. Behav. Pharmacol. 16, 487496.
FID. Malingre, T., Hendricks, H., Batterman, S., Bos, R., Visser, J., 1975. The essential oil of
Cannabis sativa. Planta Med. 28, 5661.
McPartland, J.M., Russo, E.B., 2001. Cannabis and cannabis extracts: greater then the
4.7.6. Data analysis sum of their parts? J. Cannabis Ther. 1, 103132.
Principal component analysis (PCA) was performed on SIMCA- OMC: Ofce of Medicinal Cannabis, 2010. Ofcial Website: <www.cannabis
P+ version 12.0.0.0 (Umetrics, Ume Sweden). Unit variance scaling bureau.nl>.
Pertwee, R.G., 2006. Cannabinoid pharmacology: the rst 66 years. Br. J. Pharmacol.
was used. Hierarchical clustering analysis was also done on SIMCA-
147, 163171.
P+ software and used PCs 16 with the Ward method sorted by Pertwee, R.G., 2009. Emerging strategies for exploiting cannabinoid receptor
size. agonists as medicines. Br. J. Pharmacol. 156, 397411.
Pickens, J.T., 1981. Sedative activity of cannabis in relation to its D0 -trans-
tetrahydrocannabinol and cannabidiol content. Br. J. Pharmacol. 72, 649
Acknowledgements 656.
Politi, M., Peschel, W., Wilson, N., Zloh, M., Prieto, J.M., Heinrich, M., 2008.
Direct NMR analysis of cannabis water extracts and tinctures and semi-
We thank Bedrocan BV for supplying the plant material used in
quantitative data on D9-THC and D9-THC-acid. Phytochemistry 69, 562
this study. We also thank the STW foundation for nancial support 570.
of this study. Radwan, M.M., ElSohly, M.A., Slade, D., Admed, S.A., Khan, I.A., Ross, S.A., 2009.
Biologically active cannabinoids from high-potency Cannabis sativa. J. Nat. Prod.
72, 906911.
Appendix A. Supplementary data Ross, S.A., ElSohly, M.A., 1996. The volatile oil composition of fresh and air-dried
buds of Cannabis sativa. J. Nat. Prod. 59, 4951.
Rothschild, M., Bergstrm, G., Wngberg, S., 2005. Cannabis sativa:
Supplementary data associated with this article can be found, in volatile compounds from pollen and entire male and female plants of
the online version, at doi:10.1016/j.phytochem.2010.10.001. two variants, Northern lights and Hawaian indica. Bot. J. Linnean Soc. 147,
387397.
Russo, E.B., 2007. History of cannabis and its preparations in saga, science, and
References sobriquet. Chem. Biodiversity 4, 16141648.
Russo, E.B., McPartland, J.M., 2003. Cannabis is more than simply D9-
Adams, R.P., 1989. Identication of Essential Oils by Ion Trap Mass Spectrometry. tetrahydrocannabinol. Psychopharmacology 165, 431432.
Academic Press Inc., New York. Ryan, D., Drysdale, A.J., Pertwee, R.G., Platt, B., 2006. Differential effects of cannabis
Ahmed, S.A., Ross, S.A., Slade, D., Radwan, M.M., Zulqar, F., ElSohly, M.A., 2008. extracts and pure plant cannabinoids on hippocampal neurones and glia.
Cannabinoid ester constituents from high-potency Cannabis sativa. J. Nat. Prod. Neurosci. Lett. 408, 236241.
71, 536542. Segelman, A.B., Soa, R.D., Segelman, F.P., Harakal, J.J., Knobloch, L.C., 1974. Cannabis
Ben Amar, M., 2006. Cannabinoids in medicine: a review of their therapeutic sativa L. (marijuana) V: pharmacological evaluation of marijuana aqueous
potential. J. Ethnopharmacol. 105, 125. extract and volatile oil. J. Pharm. Sci. 26, 962964.
Bicchi, C., Liberto, E., Matteodo, M., Sgorbini, B., Mondello, L., Zellner, B.A., Costa, R., Small, E., Beckstead, H.D., 1973a. Cannabinoid phenotypes in Cannabis sativa. Nature
Rubiolo, P., 2008. Quantitative analysis of essential oils: a complex task. Flavour 245, 147148.
Fragr. J. 23, 382391. Small, E., Beckstead, H.D., 1973b. Common cannabinoid phenotypes in 350 stocks of
Choi, Y.H., Kim, H.K., Hazekamp, A., Erkelens, C., Lefeber, A.W.M., Verpoorte, R., cannabis. Lloydia 36, 144165.
2004. Metabolic differentiation of Cannabis sativa cultivars using 1H-NMR Taura, F., Sirikantaramas, S., Shoyama, Y., Shoyama, Y., Morimoto, S., 2007.
spectroscopy and principal components analysis. J. Nat. Prod. 67, 953957. Phytocannabinoids in Cannabis sativa: recent studies on biosynthetic
ElSohly, M.A., Slade, D., 2005. Chemical constituents of marijuana: the complex enzymes. Chem. Biodiversity 4, 16491663.
mixture of natural cannabinoids. Life Sci. 78, 539548. Van der Kooy, F., Maltese, F., Choi, Y.H., Kim, H.K., Verpoorte, R., 2009. Quality
ElSohly, M.A., Wachtel, S.R., de Wit, H., 2003. Cannabis versus THC: response to control of herbal material and phytopharmaceuticals with MS and NMR based
Russo and McPartland. Psychopharmacology 165, 433434. metabolic ngerprinting. Planta Med. 75, 763775.
 J.T. Fischedick et al. / Phytochemistry 71 (2010) 20582073 2073

Wachtel, S.R., ElSohly, M.A., Ross, S.A., Ambre, J., de Wit, H., 2002. Comparison of the Wilkinson, J.D., Whalley, B.J., Baker, D., Pryce, G., Constanti, A., Gibbons, S.,
subjective effects of D9-tetrahydrocannabinol and marijuana in humans. Williamson, E.M., 2003. Medicinal cannabis: is D9-tetrahydrocannabinol
Psychopharmacology 161, 331339. necessary for all its effects? J. Pharm. Pharmacol. 55, 16871694.
Whalley, B.J., Wilkinson, J.D., Williamson, E.M., Constanti, A., 2004. A novel Williamson, E.M., Evans, F.J., 2000. Cannabinoids in clinical practice. Drugs 60,
component of cannabis extracts potentiates excitatory synaptic transmission 13031314.
in rat olfactory cortex in vitro. Neurosci. Lett. 365, 5863.