Methods for quantification of cannabinoids: a narrative review

Background Around 144 cannabinoids have been identified in cannabis plant, among them tetrahydrocannabinol (THC) and cannabidiol (CBD) are the most prominent ones. Because of the legal restrictions on cannabis in many countries, it is difficult to obtain standards to use in research; nonetheless, it is important to develop a cannabinoid quantification technique with pharmaceutical applications for quality control of future therapeutic cannabinoids. Method To find relevant articles for this narrative review paper, a combination of keywords such as medicinal cannabis, analytical, quantification and cannabinoids were searched for in PubMed, EMBASE, MEDLINE, Google Scholar and Cochrane Library (Wiley) databases. Results The most common cannabinoid quantification techniques include gas chromatography (GC) and high-performance liquid chromatography (HPLC). GC is often used in conjunction with mass spectrometry (MS) or flame ionization detection (FID). The major advantage of GC is terpenes quantification however, for evaluating acidic cannabinoids it needs to be derivatised. The main advantage of HPLC is the ability to quantify both acidic and neutral forms of cannabinoids without derivatisation which is often with MS or ultraviolet (UV) detectors. Conclusion Based on the information presented in this review, the ideal cannabinoid quantification method is HPLC- MS/MS for the cannabinoids.


Introduction
Cannabis sativa L. is an annual herbaceous flowering plant indigenous to eastern Asia (De Backer et al. 2009). The phenotypes of cannabis are highly variable and the plant is accepted to have two subspecies: C. sativa subsp. sativa and C. sativa subsp. indica (Hillig and Mahlberg 2004;Knight et al. 2010). A third subspecies, C. sativa subsp. ruderalis, has been identified; however, it is not broadly recognized (Fischedick et al. 2010a;Hillig and Mahlberg 2004). Cannabis has been used for its therapeutic properties for thousands of years and it was introduced in western medicine in the nineteenth century until its prohibition in the US from mid-1930s (Aizpurua-Olaizola et al. 2014).
Because consumers have limited means to analyse the chemical composition of the cannabis products, consumers may be inadvertently purchasing products with undesired properties given that different cannabinoids produce different effects (Fischedick et al. 2010b). As a result, it is important to implement methods of quality control so that consumers can be certain that what they are consuming will have the desired effects (Dussy et al. 2005;Fischedick et al. 2010a;Fischedick et al. 2010b). As cannabis use becomes progressively accepted, it becomes increasingly important to quantify the cannabinoid profile and content of cannabis preparations to ensure the uniformity and quality of the preparations .
A variety of analytical techniques have been developed for quantification and qualification cannabinoids and other compounds in cannabis plant. Advances in analytical methods have also resulted in detection of various compounds from cannabis extracts in the last decade (eg terpenes). The purpose of this literature review is to explore cannabinoid quantification techniques and subsequently suggest an optimal method for pharmaceutical grade quantification.

Quantitative analysis of cannabinoids
Gas chromatography (GC) Gas chromatography (GC) is one of the most commonly used chromatographic methods in quantitative cannabinoid analysis (Hazekamp et al. 2009). Gas chromatography is typically completed in under 20 min at up to 300°C and makes use of stationary phases with low polarities, such as 5% diphenyl-and 95% dimethyl polysiloxane (Leghissa et al. 2018a). It is important to note that the total quantity of cannabinoids in a sample is the sum of the acidic and neutral components (Citti et al. 2018). Because gas chromatography requires high column temperatures, the acidic cannabinoids undergo decarboxylation during transit through the column (Citti et al. 2018;De Backer et al. 2009;Hazekamp et al. 2009). Thus, acidic cannabinoids cannot be determined unless they are derivatized prior to analysis (Hazekamp et al. 2009). Not only does derivatization preserve cannabinoid structure, but it also causes cannabinoids to become more volatile, thus improving peak shape (Leghissa et al. 2018a). Dussy and Hamberg (2005) suggested calculating the amount of neutral and acidic cannabinoids separately in order to accurately determine the total cannabinoid content. Gas chromatography resolves cannabinoids but detection on elution presents its own challenges and solutions.

GC-FID/MS
GC is normally coupled with mass spectrometry (MS) or flame ionization detection (FID) to detect and quantify cannabinoids (Citti et al. 2018;Hazekamp et al. 2009) (Tables 1 and 2). MS employs standardized electron ionization to fragment analytes, permitting the use of compound libraries to identify the parent analyte. FID provides more accurate cannabinoid quantification because it makes use of relatively cheap authentic standards while mass spectrometry usually requires equivalent deuterated standards, which are expensive and not available for all cannabinoids (Citti et al. 2018;Hazekamp et al. 2009).
As it is shown in Table 1, in the all mentioned references the carrier gas is helium in GC-MS because it provides higher efficacy than other gases such as hydrogen and nitrogen. In each study, according to the compound of interest, different column and different protocols were used. The high temperature in injection point is the con for preserving acidic form of cannabinoids. To validate all the quantification methods two parameters should be detected. Limit of detection (LOD) and limit of quantification (LOQ). These two parameters show the lowest concentration of the interest compound that can be reliably measured by an analytical method which are mentioned in Table 1 for GC-MS method. Leghissa et al. (2018b) used Multiple Reaction Monitoring (MRM) analysis of cannabis from a surrogate hops matrix by GC-MS with triple quadrupole mass spectrometry for the first time. They used silylated cannabinoids to avoid decarboxylation process due to high temperature in GC injection port. They found that this method is applicable to cannabinoids analysis from plant materials and cannabis products. The main achievement of their study is that, in this method, because the risk of interferences from the essential oils and waxes is reduced the extraction need less sample preparation in the laboratories compared to other techniques like SPE.
In another study, Gas Chromatography with Vacuum Ultraviolet spectroscopy (GC-VUV) was used which is gas chromatography with vacuum ultraviolet spectroscopy. The detection of cannabinoids and the cannabinoid metabolites was fast and simple, so that it can be used in rapid detection of them even without having a baseline for cannabinoids for comparison. This method has just one disadvantage which is high limit of detection (LODs). Due to this drawback, detecting analytes in biological matrices cannot be accomplished without pretreatments (Leghissa et al. 2018c).

Two-dimensional gas chromatography
Experience has shown that one-dimensional gas chromatography does not provide enough resolution to analyse complex cannabinoid mixtures (Aizpurua-Olaizola et al. 2016). Two-dimensional gas chromatography (GC × GC) has been found to be preferable over one dimensional GC for analyzing complex mixtures, such as cannabis extracts, in that it reveals more sample components (Dallüge et al. 2003;Groger et al. 2008;Omar et al. 2014). Additionally, GC × GC produces two sets of retention data for sample constituents and this can greatly aid analyte identification (Dallüge et al. 2003).
In the Table 2, nitrogen and helium are the carrier gases. In many studies, it is proved that nitrogen has the best efficacy, but it is time consuming. On the other hand, by using helium, the process is rapid and efficient, but the price is not affordable. The Initial and end temperatures, the type of columns and thedrawback are almost similar to GC-MS.

Liquid chromatography (LC)
High-performance liquid chromatography (HPLC) is a commonly used liquid chromatography (LC) technique in quantitative cannabinoid analysis (Hazekamp et al. 2009) ( Table 3). The most common columns used in HPLC consist of C18 stationary phases (Citti et al. 2018;Leghissa et al. 2018a) and methanol with 0.1% formic acid or water with 0.1% formic acid as mobile phases (Leghissa et al. 2018a). C18 columns have high resolution and can differentiate between cannabinoids (Citti et al. 2018;Citti et al. 2016). The use of formic acid in the mobile phase provides better peak shape and results than other mobile phases and improved resolution in the chromatographic analysis (Citti et al. 2016). Peschel and Politi (2015) ran two HPLC assays to identify major and minor cannabinoids. Extract profiling was based on the main cannabinoid (THC, CBD, CBG, and CBN) quantification and the presence of acids and flavones. In this research, they found good resolutions of THCA, CBGA, CBDA, THCVA, THC, CBG, CBD, and THCV by HPLC.
From Table 3, it is clearly obvious that C18 is the most popular column as it is mentioned earlier. The main difference between HPLC and GC is the operating temperature. That is why HPLC is used when preserving the acidic form of cannabinoids are matter. The only disadvantage of HPLC is, it is not able to analyse the volatile compounds like terpenes.     (Citti et al. 2018;Leghissa et al. 2018a), which is particularly useful in analyzing extracts from complex matrices such as cannabis. However, some cannabinoids, such as CBG and CBD are difficult to separate using UV detection especially in concentrations greater than 10% in the extract (Citti et al. 2018;De Backer et al. 2009). In the case of CBG and CBD, MS is preferred because it can differentiate between different cannabinoids based on the m/z value of their molecular ion (Citti et al. 2018). M/z value is not always unique, however; in an ongoing study, Citi, Braghiroli (Beal et al. 1995) found five cannabinoids with the same m/z of 315.2294; this value matches that of THC and CBD in Bediol® oil and ethanol extracts. Because some of these cannabinoids coelute, analysis of these compounds is difficult.

HPLC-ESI-qTOF/MS
HPLC-electrospray ionization-quadrupole time of flight (HPLC-ESI-qTOF) is very effective in identifying complex and common compounds and can identify the main component of the sample in addition to enhancing the signal to noise ratio in the peaks (Aminah Jatoi et al. 2002). Citti, Ciccarella (Aminah Jatoi et al. 2002) analyzed cannabinoid concentrations in olive oil, ethanol, and supercritical CO 2 and found that UV-DAD and qTOF detectors produced similar results, thus suggesting that these two detection systems are equally useful in cannabinoid analysis. Pellati and Brighenti (Brighenti et al. 2017;Pellati et al. 2018) used HPLC-ESI-MS both in positive and negative ion mode for the analysis of cannabinoids. By developing HPLC methods, they improved resolution, peak shape, and separation performance together with the improvement of the ionization in HPLC-ESI-MS (Brighenti et al. 2017;Pellati et al. 2018). Using the results from their study, they were able to distinguish cannabis plants grown indoors from those grown outdoors. These results suggest that HPLC can be used to successfully determine several cannabinoid profiles and that this method can be used to distinguish between cannabis varieties and growing conditions (Borgelt et al. 2013).

LC-MS/MS and APCI
Grauwiler, Scholer (2007) developed a method to simultaneously detect five cannabinoids in human plasma and urine using high-performance liquid chromatographytandem mass spectrometry (LC-MS/MS) and atmospheric pressure chemical ionization (APCI). Their method had a 25-min runtime with a 0.2 ng/mL lower limit of quantification on samples following human oral administration of 20 mg of THC. Although APCI methods are less sensitive than ESI methods, APCI methods were chosen instead of ESI methods because they produced fewer matrix effects. Limits of detection and limits of quantification were found to be acceptable even with APCI methods (Grauwiler et al. 2007).

UPLC-qTOF
Ultra-performance liquid chromatography allows researchers to use a thinner column compared to HPLC, and it can be used for particles less than 2 μm which leads to better separation with higher speed than conventional HPLC. Additionally, Aizpurua-Olaizola, Omar (Borgelt et al. 2013) identified seven unknown minor cannabinoids using UPLC-quadrupole time of flight mass spectrometry (UPLC-qTOF). Jung, Meyer (Brighenti et al. 2017) also implemented qTOF in their study to isolate and identify THCA and 12 of its metabolites in rat urine using LC-MS, LC-MS/MS and LC-qTOF MS. The use of qTOF results in increased accuracy of the detected ions, and when analysing extracts acquired from complex matrices using MS/ MS allows for increased cannabinoid specificity (Leghissa et al. 2018a).
Although MS offers many benefits, the use of qTOF mass spectrometers is ideal when trying to differentiate between two compounds with different compositions but the same nominal mass (Citti et al. 2018). qTOF mass spectrometers can provide accurate mass identification with a threshold less than 5 ppm for precursor and product ions; this allows for differentiation between isomers of cannabinoids (Aizpurua-Olaizola et al. 2014;Citti et al. 2018) such as Δ8-tetrahydrocannabinol and Δ9-tetrahydrocannabinol which have the same m/z because these cannot be differentiated by MS (Citti et al. 2018). Such isomers may have different therapeutic properties and may need to be separated for manufacture, so it is important to adopt an analytical technique that can differentiate between them.

Matrix-assisted laser desorption ionization mass spectrometry
Matrix-Assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS) is a new method which has been used in some studies for comparison with usual methods such as LCMS and GCMS in identification of cannabinoids metabolites (Beasley et al. 2016). Recently this method has attracted attention because compared to usual mentioned methods, the sample preparation is simpler, a narrower time frame of drug can be detected, and the sample amount is reduced. Beasley et al. (2016) have used MALDI-MS to detect the cannabinoids in a single hair sample. In this study, MALDI instrument was consist of MDS Sciex hybrid quadrupole time-of-flight mass spectrometer with an orthogonal MALDI ion source and a neodymium-doped yttrium aluminum garnet laser.

LC and GC methods comparison
Liquid chromatography (LC) often employs electrospray ionization (ESI) and atmospheric-pressure chemical ionization (APCI) as ionization sources (Grauwiler et al. 2007). These usually only generate a protonated molecule without diagnostic fragmentation; therefore MS/ MS is required to obtain diagnostic information when using LC. Additionally, because cannabinoids have phenolic and carboxylic functional groups that are not ionized effectively using ESI or APCI, GC-MS may offer greater sensitivity that LC-MS (Leghissa et al. 2018a). GC × GC provides greater separation power and analysis speed compared to coupled-column techniques such as liquid chromatography-mass spectrometry (LC-MS) (Dallüge et al. 2003).
However, unlike GC, HPLC allows for differentiation between acidic and neutral cannabinoids (Romano and Hazekamp 2013) because it does not require a derivatization step prior to analysis because high temperatures are not involved in the analysis (Aizpurua-Olaizola et al. 2014; Leghissa et al. 2018a). As a result, cannabinoids do not undergo decarboxylation in an HPLC column, so HPLC provides a more comprehensive chemical report of cannabis samples compared to GC. Table 4 shows LOD and LOQ values for some of the most well-known cannabinoids using HPLC and GC-MS.
Costs is another parameter for comparison these methods. Set up, maintenance and running costs are often important factors in selecting analytical techniques specially in industry settings. The running cost of LC, GC and HPLC are negligible but for equipment, LC is more expensive than GC and both are much more expensive than HPLC. Coupling of mass spectrometry with LC or GC can further increase the costs.

TLC (thin layer chromatography)
Hazekamp (2013) used the TLC method both for polar and non-polar systems. They used reversed-phase silica gel plates and normal phase silica gel plates for nonpolar and polar systems respectively. For more accurate results they used Fast-Blue B salt (4-benzoylamino-2, 5diethoxy benzene diazonium chloride hemi salt) which is a suitable coloring agent for visualization of cannabinoids at TLC plates. Fast-Blue B can determine acetylcholinesterase, αand β-glucosidase activity by changing to different colours which come from reacting of FBB with various compounds, however the colors depend on Table 4 limit of detection and limit of quantification for 5 common cannabinoids using both high performance liquid chromatography and gas chromatography for quantification. (Brighenti et al. 2017;Pellati et al. 2018 the concentration of constituents. As a result, they found that TLC is useful in rapid screening of many samples for the existence of cannabinoids, however, its performance is lower compared to other separation methods and the reproducibility of TLC depends on some parameters such as relative humidity (Romano and Hazekamp 2013).
FTIR Fourier transform infrared spectroscopy (FTIR) Hazekamp et al. (2005) measured cannabinoids with FTIR. They added KBr to the ethanolic solution of cannabinoids followed by vacuum ethanol evaporation because KBr does not show any absorption spectrum in IR region. Additionally, KBr has a 100% transmission window in the range of wave number at the FTIR spectroscopy. The IR spectra were measured in the range of 500 to 4000 cm − 1 . Compared to UV spectra, IR spectra presented more absorbance peaks (Hazekamp et al. 2005). Mutje et al. (2007) showed the existence of carbonyl and ester groups by the FTIR peak at 1775 and 1725 cm − 1 in composite samples of cannabis extract.

Nuclear magnetic resonance spectrometry (NMR)
Another alternative to GC and HPLC is NMR (Citti et al. 2018;Hazekamp et al. 2014). NMR is accurate and reproducible and unlike GS and HPLC, NMR is not sensitive to impurities, such a chlorophyll or lipids present in the sample (Hazekamp et al. 2014). Hazekamp, Choi (Casiraghi et al. 2018) developed a method for cannabinoid quantification using 1 H-NMR that does not require chromatographic purification and has a 5-min final analysis time. In that study, they analysed singlets in the δ 4.0-7.0 range in the 1 H-NMR spectrum and found that their technique was appropriate for the quantification of CBDA, THCA, CBG, CBGA, and possibly other cannabinoids as well. One of the major advantages of this technique is that reference standards are not required, meaning that this method can quantify cannabinoids that lack pre-existing reference standards and therefore cannot be analysed by other techniques. Although the results from NMR are promising, one major disadvantage to NMR is that high resolution instruments are very expensive (Citti et al. 2018).
Other parameters to consider when selecting a quantification method There are other aspects which must be considered when selecting a quantification method such as method performance for different types of cannabinoids and also common interfering substances and impurities. Figure 2 shows suggested methods for cannabis compounds and their impurities. There is no evidence to support which analytical method works best for a specific cannabinoid. However, generally, LC is the preferred method for cannabinoids and GC for terpenes. GC does not have the capability to quantify the acidic form of cannabinoids unless through derivatization while terpenes cannot be detected by LC because they are volatile compounds.
Another important factor to consider is analytical sample preparation which is the most time consuming and the most common cause for generating errors during analytical process. Sample preparation method, storage and handling are some of the parameters which can affect the results. For cannabis, the final sample should represent the original lot. So, not only the extraction method, but also the cultivation and processing steps play an important role in the analytical results.

Conclusion
Given that cannabis preparations from the same cannabis strain can vary by as much as 25% in cannabinoids composition, there is a clear need to develop an effective and efficient cannabinoid quantification technique so that clinicians can be certain about the chemical properties of the products they are administering (Hazekamp and Fischedick 2012). This literature review has explored a variety of cannabinoid quantification techniques. GC-MS is often employed for cannabinoid quantification (Ciolino et al. 2018;Groger et al. 2008;Hazekamp et al. 2009;Jung et al. 2009;Omar et al. 2013;Omar et al. 2014). However, quantification of cannabinoids via GC requires a derivatization step to avoid the decarboxylation of acidic cannabinoids (Citti et al. 2018;De Backer et al. 2009;Grauwiler et al. 2007;Hazekamp et al. 2009;Leghissa et al. 2018a). Performing GC without derivatization requires the calculation of total cannabinoid content from a combination of acidic and neutral cannabinoid content which can be an uncertain process (Dussy et al. 2005). HPLC-DAD and HPLC-UV provide alternatives to GC analysis but these detection techniques lack specificity and sensitivity (Galal et al. 2009;Grauwiler et al. 2007). The literature suggests that HPLC-MS/MS using ESI and APCI methods provide enough specificity and sensitivity to quantify cannabinoid content in all cannabis extracts (Aizpurua-Olaizola et al. 2014;Citti et al. 2018;Grauwiler et al. 2007).
There are multiple benefits to using HPLC-MS/MS over other analytical methods presented in this review. For example, HPLC can differentiate between acidic and neutral cannabinoids, unlike GC (Romano and Hazekamp 2013). MS offers several benefits over other detection methods as well. For example, MS can differentiate between different cannabinoids based on the m/z value of their molecular ion. It offers more specificity compared to UV detectors and can analyze extracts from complex matrices, such as cannabis (Citti et al. 2018;Leghissa et al. 2018a).