Development and Optimization of an HPLC-PDA Method for the Determination of Major Cannabinoids in Hemp (Cannabis sativa L.) Essential Oil Obtained by Hydrodistillation

The use of hemp (Cannabis sativa L.) essential oil (EO) has shown a significant increase in interest and use during recent years. In this work, a new and simple reversed-phase HPLC with photodiode-array (PDA) detector method has been developed and optimized for the detection and quantification of cannabidiolic acid (CBDA), cannabidiol (CBD), cannabinol (CBN), Δ9-tetrahydrocannabinol (Δ9-THC), and Δ9-tetrahydrocannabinolic acid (THCA). The cannabinoids were extracted from the EO by partition with n-hexane and water, followed by sonication, evaporation to dryness under nitrogen, and reconstitution with methanol:chloroform (9:1, v/v) before HPLC-PDA analysis. The method shows good selectivity and robustness, linearity in the range 0.5–100 mg L−1 with R2 higher than 0.999 for all cannabinoids analyzed, LOD of 0.11–0.16 mg L−1, and LOQ of 0.35–0.48 mg L−1. The recovery was between 78 and 100% and the intra-day and intermediate precision, expressed as relative standard deviation (RSD), was < 4% and 4–10%, respectively.


Introduction
The Cannabis genus belongs to the Cannabaceae family, order of the Rosales, and it is considered a monotypic genus that includes only the Cannabis sativa L. species with a wide genetic variability (Hazekamp 2007;Clarke and Merlin 2016;Koren et al. 2020). This species includes some subspecies, such as industrial or fiber hemp (Glivar et al. 2020) and drug or marijuana strains; the latter are cultivated and used only for medical-therapeutic purposes.
The cultivation of industrial hemp is regulated by national laws and allowed in several European countries only for monoecious and dioecious varieties registered in the EU plant variety database, containing less than 0.2% of the psychoactive Δ9-THC (Commission of the European Communities 2004). This crop, native to Asia and Indian subcontinent, today is grown in many countries (Sandler and Gibson 2019), with a strong increase in the use of its processed products in the manufacturing of paper, varnishes, inks, and biofuel, and in the phytoremediation, food and animal feed, medicine, nutraceuticals, cosmetics, and construction industries (Callaway 2004;Rehman et al. 2013).
Hemp EO is considered a niche product with high benefits and broad potential uses (Mediavilla and Steinemann 1997;Bertoli et al. 2010) from the cosmetic to the 1 3 medical-pharmaceutical industry due to its natural antioxidant and antimicrobials (against Candida spp.) properties (Novak et al. 2001;Nissen et al. 2010;Nafis et al. 2019). Hemp EO has also been shown to have anti-leishmaniasis activities (Wanas et al. 2016), insecticidal and larvicidal activity against Aedes aegypti larvae (Górski et al. 2016;Bedini et al. 2016;Wanas et al. 2020), nematocide (Mukhtar et al. 2013), fungicide properties (Wielgusz et al. 2012), and allelopathic effect (Synowiec et al. 2016) finding utilization also in crop protection (Górski et al. 2016). Its use in the food sector is also very interesting and promising as flavoring for beverages and additives in pastry and catering (Meier and Mediavilla 1998).
Among the several methods employed to obtain EO from plant tissue, hydrodistillation is widely used as the official technique and adopted by the Official Pharmacopoeia of the Italian Republic (Commissione Permanente per la Revisione e la Pubblicazione della Farmacopea Ufficiale 2008). This technique extracts most of the volatile compounds present in the aromatic resins located mainly on the bracts that surround female flowers and fruits of the hemp plant (Kim and Mahlberg 1991;Meier and Mediavilla 1998;Andre et al. 2016). Hemp EO is a complex mixture of many volatile compounds, mainly terpenes and other minor substances like terpenoids that play different biological roles in the Cannabis sativa plants (Hazekamp 2007). Terpenes are a complex mixture of volatile compounds as monoterpenes, which are the main responsible for the fragrance (e.g., α-pinene, myrcene, and terpinolene) and sesquiterpenoids (α-humulene, (E)-caryophyllene, and caryophyllene oxide) (Hillig 2004;Happyana and Kayser 2016). Cannabinoid compounds can be hydrodistilled together with terpene constituents; however, their concentration in the hemp EO is lower if compared to the inflorescence, being only partially water-distillable due to their low volatility (Kim and Mahlberg 1991;Mediavilla and Steinemann 1997).
Recently, Fiorini et al. (2020) applying the microwaveassisted extraction (MAE) method associated with hydrodistillation to hemp dried inflorescences obtained a higher yield of EO with an enrichment in bioactive compounds, especially cannabidiol (CBD), with respect to the conventional hydrodistillation.
Cannabinoids, exclusively present in Cannabis (Raharjo and Verpoorte 2004), reach their maximum content in plant resin close to flowering time (Pacifico et al. 2007). They are a group of terpenophenolic compounds characterized by a structure with 21 carbon atoms (Mukhtar et al. 2013) and currently about 113 compounds have been identified (Aizpurua-Olaizola et al. 2016). The most common types of cannabinoids present in the plant shared a carboxyl group: Δ9-tetrahydrocannabinolic acid (THCA), cannabidiolic acid (CBDA), and cannabicromenic acid (CBCA). The carboxyl group however is not very stable and is easily lost as carbon dioxide, upon exposure to heat and light, particularly during collection, processing, and storage. The loss of the carboxyl group results in the conversion of the acids into their corresponding neutral and active form Δ9-THC, CBD, and CBC. Moreover, some cannabinoids are the result of the degradation of other cannabinoids, such as cannabinol (CBN) produced by the non-enzymatic oxidation of Δ9-THC under the effect of heat and light (De Backer et al. 2009). Δ9-THC is the main active compound responsible for psychoactive and physiological effects in humans, and many countries, including Italy, have recently established the maximum limits of total Δ9-THC admissible in food at 2 mg kg −1 for seed, flour, and supplements containing foods derived from hemp (Ministero della Salute 2020). Among non-psychotropic cannabinoids, CBD is undoubtedly the most interesting for its therapeutic and medicinal proprieties (Vági et al. 2019) and several new CBD products are utilized as alternative treatments, especially as beneficial for the symptomatology of health-related problems such as pain, anxiety, and inflammation (Corroon and Phillips 2018;Poleg et al. 2019).
Nowadays, high-performance liquid chromatography (HPLC) and gas chromatography mass spectrometry (GC-MS) are the two most utilized and validated techniques for the analysis of cannabinoids in Cannabis plant materials and derived food products (inflorescence, seeds, seed oil, flour, etc.; De Backer et al. 2009;Citti et al. 2018), with the exception of EO. In particular, HPLC-based methods allow a better speciation between the acidic and decarboxylated forms of cannabinoids with respect to those employing GC-MS, in which high temperatures, necessary for the analysis, have as a consequence the decarboxylation of the acid compounds, leading to possible overestimation of their corresponding neutral forms (De Backer et al. 2009;Citti et al. 2018).
The aim of this work was to develop a simple and reliable HPLC-PDA method for the detection and quantification of five cannabinoids, namely CBD, CBDA, Δ9-THC, THCA, and CBN, focusing on hemp EO obtained by hydrodistillation.

Preparation of Standard Solutions
Working standard solutions of analytes were obtained after dilution of the reference standards with a mixture of MeOH:chloroform (9:1, v/v). All stock and working standard solutions were stored at −20 °C, without showing any degradation up to 90 days.
Quality control (QC) samples were prepared by spiking hemp EO samples (20 mg) at three concentration levels for all cannabinoids. The concentration levels were denoted as low-quality control (LQC) with a concentration of 0.2 mg g −1 for CBDA, CBN, Δ9-THC, and THCA; mediumquality control (MQC) with a concentration of 1.0 mg g −1 for CBDA, CBN, Δ9-THC, and THCA; high-quality control (HQC) with a concentration of 2.5 mg g −1 for CBDA and THCA and 5.0 mg g −1 for CBN and Δ9-THC. In the case of CBD, due to the high amount of this compound in the EO, only two concentration levels were prepared (LQC and HQC) with a concentration of 1.0 and 5.0 mg g −1 , respectively.

Cannabis Samples
The female inflorescences of Antal, a hemp dioecious variety, were utilized. Plants were grown at an experimental farm of Udine University, Udine, Italy (N 46°04′00″, E 13°23′00″, 109 m a.s.l.) in 2016. The European Union only allows the cultivation of hemp varieties registered in the EU plant variety database (European Parliament and of the Council 2013), with a Δ9-THC content not exceeding 0.2%, such as Antal. Five samples of female hemp inflorescences were collected manually at full flowering and immediately stored at −20 °C.

Essential Oil Extraction and Sample Preparation
The frozen fresh inflorescences were thawed and submitted to EO extraction by hydrodistillation, using a Clevenger-type apparatus, fitted to essential oils with low density. Briefly, 100 g of inflorescence were cut with scissors and placed in a round bottomed flask and 600 mL of distilled water were added. The flask was connected to the apparatus and the condensate flow was adjusted to 2 mL min −1 . Distillation continued for 3 h (Vuerich et al. 2019 Twenty milligrams of EO were weighed in 15-mL polypropylene tubes; then, 50 μL of a benzo[ghi]perylene acetone solution (1120 mg L −1 ), as internal standard (IS), 1 mL of n-hexane, and 1 mL of water were added and the sample was briefly vortexed and sonicated for 15 min in an ultrasound bath, LBS 2 (Falc Instruments Srl, Treviglio (BG), Italy). After sonication, the sample was centrifuged at 1157g for 15 min on a Centrifuge 5804R (Eppendorf AG, Germany). Then, 700 μL of supernatant was transferred in a glass tube, evaporated to dryness under nitrogen, reconstituted with 500 μL of MeOH:chloroform (9:1, v/v), vortexed, and filtered through syringe filters (PTFE 0.22 μm, 25 mm, DTO Servizi Srl, Venice, Italy). The filtrate was transferred into an autosampler vial and 10 μL was injected into HPLC.
The same preparation procedure, including all solvents but without EO, was used to obtain reagent blank in order to exclude the presence of interfering compounds eluted with the same retention times of the analytes in the HPLC analysis.

HPLC-PDA Chromatographic Conditions
The HPLC system included a Shimadzu LC-20AT pump, a vacuum degasser, a Prominence SPD-M20A photodiodearray detector, a Prominence SIL-20AC HT autosampler (10 μL loop), and a Prominence CTO-20AC column oven set at 25 °C (Shimadzu Corporation, Kyoto, Japan). The HPLC separation was achieved using a Poroshell 120 EC-C18 column (4.6 × 150 mm, 2.7 μm particle size, Agilent Technologies, Santa Clara, CA). Equipment control, data acquisition, and integration were performed with Shimadzu LabSolutions (Ver. 5.54 SP2) Software. A binary gradient of 0.1% formic acid solution in water (A) and 0.1% formic acid solution in ACN (B) at a flow rate of 0.40 mL min −1 was optimized. The gradient started at 70% of B for 5 min, and then increased linearly to 77% of B in 7 min and at 90% of B in 0.10 min. This composition was kept constant up to 30 min before a linear decrease to the initial conditions (70% of B) in 0.10 min and held up to 60 min for the equilibration of the column. The optimum wavelength detection for all cannabinoids was found to be 220 nm and the detector slit width was 4 nm. Full spectra were recorded in the range 190-400 nm.

Analytical Performance of the HPLC Method
To assess the analytical performance of the method, the following analytical parameters were evaluated: selectivity, linearity, limit of detection, limit of quantitation, accuracy, precision, carry-over, stability, and robustness.

Selectivity
The selectivity of the method was assured by comparison of the UV spectra of the analytes with those of the corresponding standard compounds and by the peak purity check.

Linearity
Calibration curves were prepared at seven calibration levels: 0.5, 1.0, 5.0, 10, 20, 50, and 100 mg L −1 , for CBDA, CBD, CBN, Δ9-THC, and THCA, and 40 mg L −1 of IS. The standard solutions were freshly prepared before use. Each standard solution was analyzed six times and the average peak area ratios of each cannabinoid vs IS were plotted against the corresponding concentration ratios. Before each analysis session, a standard solution (10 mg L −1 ) was analyzed three times and the average relative response factor of cannabinoids with respect to the internal standard was calculated.

Limit of Detection (LOD) and Limit of Quantitation (LOQ)
LOD was determined by the analysis of reagent blank and EO diluted samples by establishing the minimum level at which the analytes could be detected, based on a signal-tonoise ratio of 3:1 (S/N = 3). LOQ was determined by following the same procedure and by establishing the minimum level at which the analytes could be quantified, based on a signal-to-noise ratio of 10:1 (S/N = 10). The parameters LOD and LOQ were calculated by using the instrument software.

Accuracy and Precision
The accuracy was assessed by recovery studies using the QC samples LQC, MQC, and HQC, prepared as reported in the "Preparation of Standard Solutions" section and expressed as percentage recovery (% R). Three separate replicates were carried out for each recovery level.
The precision was evaluated by analyzing three separate replicates of the LQC, MQC, and HQC samples on the same day (intra-day repeatability) and two separate replicates of the LQC and MQC solutions over a period of 3 days (intermediate precision). The results were expressed as relative standard deviation (RSD).

Carry-Over
To check the carry-over, the reagent blank prepared as reported in "Essential Oil Extraction and Sample Preparation" section was injected in the HPLC system in between the EO samples.

Stability
Stability experiments were performed in triplicate by keeping the MQC samples of CBDA, CBN, Δ9-THC, and THCA and the LQC sample of CBD at room temperature for 7 days and at −20 °C for 90 days.

Robustness
To evaluate the influence of small variations of the chromatographic parameters on the analytical results, a robustness study was carried out by varying, in a narrow range, the column oven temperature 25 ± 5 °C and the flow rate 0.40 ± 0.05 mL min −1 .

Statistical Analysis
Statistical analysis was performed using the software R v. 4.0.0 (R Core Team 2020). The R script for contour plot was adapted from https:// yuanb ofaith. github. io/ RK_ LCMS/ (Yuan et al. 2020).

Method Development
The HPLC-PDA chromatogram of a standard mixture of five cannabinoids at 60 mg L −1 and IS is reported in Fig. 1. The chromatogram shows a good separation of the cannabinoids with retention time in the range 15-26 min, whereas the IS retention time was 30 min. The cannabinoids elution order was CBDA, CBD, CBN, Δ9-THC, and THCA. The UV spectra (190-400 nm) of the cannabinoids standards, recorded by the PDA detector and used to prepare a spectral library, are reported in the same figure.
Attempts were carried out in order to analyze directly the EO, after a simple dilution, before the HPLC analysis. The resulting chromatograms were, however, characterized by the presence of many intense peaks due to monoterpene and sesquiterpene hydrocarbons, which interfered with the quantification of cannabinoids. Partition of EO components between water and n-hexane resulted in a cleaner extract, free of interfering compounds in the elution zones of the cannabinoids of interest. Moreover, for the optimization of the analytical method, preliminary analyses were carried out by using the QC samples in order to have an EO sample containing all the analytes of interest.
The chromatograms of an EO sample (A) and of the HQC sample (B) are showed in Fig. 2. The analytes were 1 3 The average hemp EO yield was 0.25 ± 0.03% (v/w, d.m.) (n = 5) similar to literature data reported for Cannabis sativa L. EO obtained by hydrodistillation (Hillig 2004;Nissen et al. 2010;Bertoli et al. 2010;Fiorini et al. 2019;Ascrizzi et al. 2019). The most abundant cannabinoid in hemp EO samples was CBD with an average concentration of 1074 ± 62 mg kg −1 (n = 5). This result is in agreement with the values reported by other authors (Bertoli et al. 2010;Ascrizzi et al. 2019) for EO obtained using the same extraction method (hydrodistillation) from different varieties and analyzed by GC-MS. On the contrary, the CBD concentration obtained in the present work was much lower than that obtained by Fiorini et al. (2019) (> 40,000 mg kg −1 ); however, these authors adopted different sample pre-treatments, including the exposure of dried inflorescences to microwave heating at 900 W power for 1 min, which proved to be the best method to increase the CBD content in EO.
Regarding the amount of the other cannabinoids in hemp EO, their concentration was lower than LOD.

Selectivity
By comparison of the EO and HQC sample chromatograms reported in Fig. 2 and the UV spectra of the compounds (Fig. 1) and by the purity check, no chromatographic interferences were observed between the cannabinoids and other compounds eluted in the region of interest, evidencing a good selectivity of the method.

Calibration Parameters-Linearity-LOD and LOQ
The calibration results of the method are reported in Table 1. The calibration curves were linear over the concentration range 0.5-100 mg L −1 with determination coefficients (R 2 ) higher than 0.999 for all cannabinoids; moreover, statistical analysis through ANOVA showed significant linear regression without deviation. The LOD ranged between 0.11 and 0.16 mg L −1 and the LOQ between 0.35 and 0.48 mg L −1 (Table 1).

Accuracy and Precision
Recovery tests were performed by using the quality control solutions, LQC, MQC, and HQC, and the results ranged from 78 to 100% (Table 2), indicating a satisfactory accuracy of the proposed method. Intra-day precision was determined at the same concentration levels with RSD < 4% evidencing a good repeatability (Table 2). Intermediate precision was measured at two concentration levels, LQC and HQC, over a period of 3 days with satisfactory RSD in the range 4-10% and 5-10% for LQC and HQC, respectively.

Stability
The EO samples were stable for 7 days at room temperature and 90 days at −20 °C.

Robustness
The robustness of the method was evaluated by analyzing the MQC sample before and after small variations of the chromatographic parameters. Changes in flow rate 0.40 ± 0.05 mL min −1 and column temperature 25 ± 5 °C did not influence the analytical results. The effect of perturbation of flow rate and column temperature on the concentrations of the cannabinoids is reported as contour plots in Fig. 3. The data have been normalized between 0 and 1. Darker regions indicate the highest variations in quantification. Δ9-THC showed the highest relative variation as response to temperature and flow rate modification, whereas CBD, CBDA, and  CBN were the least sensitive to variation in the chromatographic conditions.

Conclusions
The widespread use of hemp EO in the pharmaceutical, nutraceutical, food, and cosmetic sectors demands for reliable and simple analytical methods for its characterization, in particular, the quantification of CBD, the most abundant cannabinoid in hemp EO. Moreover, the acid forms and CBN concentrations are time-and temperature-dependent and CBN, which is considered a degradation product, could indicate the quality level of an industrial product. In this work, a new HPLC-PDA method has been developed for CBDA, CBD, CBN, Δ9-THC, and THCA quantification in hemp EO, a matrix less investigated with respect to other hemp-derived products. This analytical method evidenced good linearity, accuracy, and robustness and appears adequate for the quality control of hemp EO. The method was applied to the analysis of hemp EO obtained by hydrodistillation, but has much wider applications as it allows to quantify both the acidic and neutral forms of the cannabinoids in different hemp extracts, regardless of the preparation method adopted.