Novel Δ⁸-tetrahydrocannabinol vaporizers contain unlabeled adulterants, unintended byproducts of chemical synthesis, and heavy metals

Abstract

Cannabis e-cigarettes containing Δ⁸-tetrahydrocannabinol (Δ⁸-THC) produced synthetically from hemp-derived cannabidiol (CBD) have recently risen in popularity as legal means of cannabis consumption, but questions surrounding purity and unlabeled additives have created doubts of their safety. Herein, NMR, GC-MS, and ICP-MS were used to analyze major components of 27 products from 10 brands, and it was determined none of these had accurate Δ⁸-THC labelling, 11 had unlabeled cutting agents, and all contained reaction side-products including olivetol, Δ⁴⁽⁸⁾-iso-tetrahydrocannabinol, 9-ethoxyhexahydrocannabinol, Δ⁹-tetrahydrocannabinol (Δ⁹-THC), heavy metals, and a previously unidentified cannabinoid iso-tetrahydrocannabifuran.

Graphical Abstract

Cannabis e-cigarettes (CECs) are a non-combustion inhalation method which uses technology adapted from electronic nicotine delivery systems. CECs vaporize an oil rich in Δ⁹-THC, the psychoactive principle of Cannabis sativa, and release hundreds of chemical breakdown products including carcinogenic and irritating gases such as isoprene, benzene, methacrolein, and methyl vinyl ketone.^1,2 CECs are popular with teens and young adults in the United States, with 23.7% of 12^th graders having reported lifetime cannabis vaping in 2019.³ CEC use was shown to be independently associated with higher odds of respiratory symptoms such as wheezing.⁴ The 2019 outbreak of e-cigarette or vaping product use-associated lung injury (EVALI) was centered around CECs, and though vitamin E acetate was identified as a potential causative agent, other ingredients or aerosol components were not ruled out.⁵ Some EVALI CECs were later found to contain unnatural cannabinoid distributions suggesting these were of synthetic origin.^6,7 Niche online communities that recount intoxication experiences by minor and/or synthetic cannabinoids have existed likely for decades,⁸ but it is only recently that cannabinoids other than those naturally-occurring have reached broad commercial availability.⁹ At the forefront of this trend is Δ⁸-THC (Chart 1).

Chart 1.

Relevant structures

Δ⁸-THC is an isomer of Δ⁹-THC not produced biosynthetically¹⁰ but present at low levels in most cannabis products as a result of spontaneous isomerization given its higher thermodynamic stability and resistance to oxidative degradation than Δ⁹-THC.¹⁰ Recent federal regulations that are permissive of hemp-derived products¹¹ have resulted in a rapid growth in usage of Δ⁸-THC CECs that are abundantly available to consumers through brick-and-mortar and online sources. Extensive hospitalizations involving suspected Δ⁸-THC consumption have been recently documented.¹² Δ⁸-THC is synthesized via acid-catalyzed cyclization of CBD.^13,14 Though Δ⁹-THC is the direct product of CBD cyclization (Scheme 1), Δ⁸-THC is favored as a major product at longer reaction times.^15,16

Scheme 1.

Routes of formation of Δ⁸-THC (upper), Δ⁴⁽⁸⁾-iso-THC (middle), and iso-THCBF (lower) from CBD via acid catalysis.

Hydrochloric acid (HCl), sulfuric acid, p-toluenesulfonic acid, boron trifluoride,and camphorsulfonic acid, among others, are viable catalysts, but the acids, solvents, and purification steps used by manufacturers are not known.^15,16 In order to address this emerging class of products, available flavor formulations from different Δ⁸-THC brands were obtained. Proton nuclear magnetic resonance spectroscopy (¹H NMR) was chosen as the primary analytical tool given this instrument’s capability for characterizing analytically-challenging vaporizer adulterants¹⁷ without the need for derivatization or developing dedicated chromatographic methods which may be necessary complex samples. Quantitative ¹H NMR (QNMR) was used to report component levels in these products, a facile but direct quantitative method whose limitations include the fact that some components cannot be identified or quantified due to spectral overlap of their resonances, and its inherently low sensitivity precludes identification of ultra-trace impurities.

Medium chain triglyceride oil was identified in B5 (3.71 ± 0.06%, $\overline{x}$ ± SEM), B6 (3.48 ± 0.06%), B8 (2.94 ± 0.05%), and B9 (5.6±0.1%). Triethyl citrate (TEC) was identified in F20 (6.3 ± 0.06%), F21 (6.27 ± 0.03%), F22 (6.5 ± 0.1%), G23 (7.28 ± 0.05%), G24 (6.2 ± 0.1%), I26 (11.1 ± 0.1%), and J27 (5.34 ± 0.06%). Δ⁴⁽⁸⁾-iso-Tetrahydrocannabinol (Δ⁴⁽⁸⁾-iso-THC) is a previously described byproduct of acid-catalyzed CBD cyclization (Scheme 1)¹⁸ and was detected in all 27 samples ranging from 2.36 ± 0.05% to 12.79 ± 0.06% with $\overline{x}$ ± SD of 5.4 ± 3.5 % in n=16 where quantification was possible (see SI). Olivetol (5-pentyl-1,3-benzenediol, Chart 1) was identified in 22/27 products, but its quantification was not possible (see SI). Olivetol has been previously shown in EVALI-associated CECs that also contain unnatural cannabinoid distributions⁷ and is likely a byproduct of chemical synthesis. Olivetol is a synthetic precursor to tetrahydrocannabinols,^19,20 and its presence could indicate the use of these pathways for production. 9-Ethoxyhexahydrocannabinol (9-EtO-HHC) is a known byproduct of CBD cyclization in ethanol²¹ and was detected in D13 and D14. 9-EtO-HHC presence is correlated with lower levels of Δ⁸-THC (p<0.01) and higher levels of Δ⁴⁽⁸⁾-iso-THC (p<0.01) than in D15 and D16, suggesting ethanol may favor Δ⁴⁽⁸⁾-iso-THC formation as opposed to Δ⁸-THC. Bornyl chloride (Chart 1), a known reaction product of HCl and β-pinene,²² was tentatively identified by GC-MS (Figure S23) in A2 and A3, and is indicative of HCl as a cyclization catalyst. However, its absence in other products does not rule out the use of HCl, as its presence in A2 and A3 may simply be evidence of starting material impure with β-pinene. The potential for bornyl chloride to generate HCl gas when pyrolyzed²³ could present a significant inhalation hazard.

In addition to the above, a molecule which, to the best of the authors’ knowledge, has never been previously described was also identified. The cannabinoid (5aR,9aS)-5a-isopropyl-8-methyl-3-pentyl-5a,6,7,9a-tetrahydrodibenzo[b,d]furan-1-ol or iso-tetrahydrocannabifuran (iso-THCBF, Chart 1) is likely the result of a hydride shift in the carbocation intermediate (Scheme 1). iso-THCBF was isolated from Δ⁸-THC CEC products and characterized by mass spectrometry and 1D and 2D NMR (see SI). iso-THCBF was present in nearly all products tested but was not quantifiable in products containing TEC due to spectral overlap.

CEC screening by inductively coupled plasma-mass spectrometry (ICP-MS) shows the existence of metals such as magnesium (599 ± 391 ppb $\overline{x}$±SD n=10), chromium (446 ± 758 ppb), nickel (380 ± 364 ppb), copper (509 ± 1143 ppb), zinc (1.8 ± 2.1 ppm), mercury (160 ± 162 ppb), lead (42 ± 28 ppb), and others (see SI). These metals are likely leachates from vaporizer components or production materials, and their inhalation could cause deleterious effects on the respiratory tract that stem from the generation of reactive oxygen species.^24,25 ICP-MS identified elevated levels of silicon (205 ± 108 ppm), a finding that has been previously shown for EVALI-associated CECs.²⁶ Silica gel may be used as a purification medium or decolorizing agent, and its potential delivery to the respiratory tract from these products are subject of further investigation.

QNMR indicates that Δ⁸-THC levels can vary as much 40% from the labelled value (Table 1), suggestive of poor testing capabilities and falsified results. For brand A, the average of the sums of Δ⁸-THC and Δ⁴⁽⁸⁾-iso-THC for each product is not significantly different from the average reported Δ⁸-THC content (p<0.01), suggesting the analysis method (HPLC-UV as stated in the certificate of analysis) cannot discriminate the two. Brands B-E appear to use one lab result for all their products when these not only have variable levels of Δ⁸-THC, but also contain distinct levels of byproducts indicating different manufacturing methods in products that otherwise appear identical except for flavor formulation. Significant levels of understudied (Δ⁴⁽⁸⁾-iso-THC, 9-EtO-HHC) and novel (iso-THCBF) cannabinoids present a danger to users as these compounds are not well characterized pharmacologically and could cause unexpected levels of intoxication. High levels of unlabeled cutting agents present a further complication given the little safety information available. Further chemical, pharmacological, and toxicological testing of these and similar products is necessary.

Table 1.

Major components of 27 products (P) from 10 brands (B). Levels are mass% ± standard error of the mean (SEM).

B	P	Δ8-THC Reported	Δ8-THC Measured	Δ4(8)-iso-THC	iso-THCBF
A	1	83.2	76±1	4.24±0.07	1.22±0.02
	2	83.2	79.5±0.1	3.45±0.03	1.25±0.05
	3	86.15	81.2±0.6	3.48±0.03	1.31±0.04
	4	84.66	79.5±0.8	3.9±0.1	<LOD
B	5	93.0821	75±2	NQ	0.88±0.05
	6	93.0821	77±1	NQ	<LOQ
	7	93.0821	62.7±0.7	7.96±0.09	<LOD
	8	93.0821	77±1	NQ	<LOQ
	9	93.0821	78±2	NQ	0.63±0.03
C	10	90	74.8±0.2	4.79±0.03	0.957±0.004
	11	90	77.4±0.4	4.23±0.04	0.9±0.02
	12	90	79.4±0.4	3.74±0.05	0.9±0.02
D	13	90	54.8±0.2	12.79±0.06	1.67±0.02
	14	90	53.6±0.5	12.51±0.05	1.65±0.04
	15	90	78.0±0.7	4.13±0.04	1.6±0.02
	16	90	79.9±0.7	2.36±0.05	0.73±0.02
E	17	77.71	78.8±0.4	3.01±0.03	0.61±0.02
	18	77.71	80.3±0.7	2.851±0.005	0.415±0.008
	19	77.71	79.7±0.2	2.75±0.05	0.472±0.005
F	20	80.85	76.8±0.3	NQ	NQ
	21	84.02	77.1±0.7	NQ	NQ
	22	81.87	77±1	NQ	NQ
G	23	85.000	70.9±0.7	NQ	NQ
	24	81.240	78.9±0.9	NQ	NQ
H	25	78.43	61.6±0.3	10.79±0.05	1.54±0.01
I	26	NA	72.2±0.3	NQ	NQ
J	27	NA	73.4±0.2	NQ	NQ

NQ: identified but not quantifiable. ND: not detected; NA: non-available data; LOD: limit of detection (signal-to-noise ≤ 3); LOQ: limit of quantification (signal-to-noise ≤ 12).

ABBREVIATIONS

CEC

cannabis e-cigarette

EVALI

e-cigarette or vaping product-use associated lung injury