This study, which evaluated the potential of a well-designed unidirectional aseptic processing approach to both hydrocarbon and supercritical carbon dioxide extract manufacturing, suggests that the extraction process is capable of deactivating or removing the tested microbial contaminants when highly contaminated Cannabis spp. Inflorescence is used as feed material.
The findings of this study are from the research initiative submitted to the Pennsylvania State Department of Health in accordance with House Bill 1024. The research study involved evaluating the effectiveness of solvent-based extraction methods for eliminating or substantially diminishing microbial contamination from feed materials (for example, Cannabis sativa) during the extract manufacturing process. This study investigated both supercritical carbon dioxide and hydrocarbon extraction technologies. Organic Remedies Inc.’s unidirectional aseptic manufacturing infrastructure and processes were used starting with contaminated plant material to prepare primary extracts followed by essential, decarboxylated, and distilled oils. Testing of extracts for microbial contamination according to Pennsylvania State regulations was carried out by a state certified laboratory. Results of testing associated with starting plant material revealed the presence of 28,000 cfu/g total aerobic bacteria, 17,975 cfu/g enterobacteriaceae, and “too numerous to count” cfu/g total yeast and mold. E. coli, salmonella spp., and mycotoxins were not detected. Results of testing all extracted materials showed that they were free of the tested microbial contaminants and fungal toxins. Data presented here show the results of testing for the hydrocarbon extraction pathway and the supercritical carbon dioxide pathway. In addition to testing for microbial contamination, the data show results for content of delta-9 tetrahydrocannabinol (THC), cannabidiol (CBD), total cannabinoids, and total terpenes, all of which were used to evaluate extraction efficiency.
Organic Remedies Inc., is a vertically integrated grower and processor located in Carlisle, Pennsylvania. As a licensed Clinical Registrant (CR), Organic Remedies Inc., participates with the Philadelphia College of Osteopathic Medicine, an Academic Clinical Research Center (ACRC), in research studies that explore the medicinal qualities of Cannabis sativa and its related formulated products. With the recent refinement of the Pennsylvania Medical Marijuana Regulations (1), ACRCs and CRs were allowed to undertake studies exploring whether the extraction process and associated refinement protocols are capable of producing materials that were within Department of Health quality standards when contaminated plants were utilized as feed material.
The current microbial contaminant limits for Cannabis spp. feed material intended for use in extracted product manufacturing are non-detect, non-detect, <10,000 cfu/g, <10,000 cfu/g, and <1,000 cfu/g for Salmonella spp., Escherichia coli, total aerobic bacteria, total yeast and mold, and bile tolerant gram-negative (Enterobacteriaceae) bacteria, respectively (1). Pennsylvania microbial limits (1) are relatively strict in comparison to other states’ standards; the rationale for such strict limits is a result of cannabis’ definition as a medicine in the current regulatory framework of the state. Moreover, no form of post-harvest sterilization is permitted; therefore, if a batch of flower or trim fails microbial testing it must be destroyed.In an efficiently operating indoor grow the cost of dried or cured plant material to the producer can be as much as $2.00/g; in a greenhouse setting, that cost can be as low as $0.75/g. As a result, plant material not conforming to PA microbial criteria has substantial negative impacts on business revenue, employment opportunities, and the cost of medical cannabis products.
Sterilization of cash crops is not uncommon in the food or nutraceutical industries; however, those products are very different from cannabis because they are usually fruits and herbs that are intended for oral consumption by the consumer. Furthermore, products such as fruits, vegetables, and herbs have very different morphological characteristics compared to Cannabis spp. inflorescence intended for vaporization or the manufacturing of products created via extraction. For the food and nutraceutical industries, numerous sterilization techniques are outlined in the United States Pharmacopeia (USP) (2) and permitted by the US Food and Drug Administration (FDA) and—when coupled with final product testing—can reliably ensure consumer safety. Some examples of terminal sterilization technologies that are permitted by the FDA include steam-sterilization, dry-heat sterilization, gas sterilization, ionizing radiation, and sterilizing filtration (2). Furthermore, the USP and FDA describe a process-based microbial risk mitigation technique for manufacturing called “Unidirectional Aseptic Processing” (2). That process, which does not diminish the importance of terminal sterilization and final product testing, incorporates numerous terminal sterilization techniques in the manufacturing process, appropriate air handling infrastructure, and intensive training of personnel (2).
As previously mentioned, Cannabis sativa differs from other cash crops regarding the intended use of its manufactured products. In fact, the entire manufacturing process—including the variety of final product formats—differs from other agricultural industries insofar as it is intended to produce botanical, extracted raw materials, and formulated products. Further complicating the manufacturing of Cannabis spp. products is that there is no single set of regulations governing the quality of raw material or final products; this lack of central governance is due to the classification of Cannabis sativa—particularly the cannabinoid Δ9 -tetrahydrocannabinol (Δ9 -THC)—as a Schedule I controlled substance (3).Therefore, regulatory frameworks are created on a state-by-state basis and recommendations set forth by the FDA or their associated advisory bodies may not be wholly adopted. For example, Pennsylvania does not allow for the use of federally accepted microbial remediation technology to sterilize contaminated plant material prior to market distribution. Additionally, the state does not allow Cannabis sativa growers or processors to utilize plant material contaminated with bacteria or yeast and mold in the manufacturing of raw materials or formulated products via solvent extraction. The logic behind these types of regulations is sound in that they are set up to be risk averse and to ensure that patients in the medical cannabis market of Pennsylvania have access to safe therapies; however, the current (restrictive) framework overlooks the capabilities of the extraction process including the associated infrastructure to deactivate or remove microbial contaminants.
In cooperation with the Philadelphia College of Osteopathic Medicine and with the approval of the Pennsylvania Department of Health, Organic Remedies Inc. designed and executed the proceeding study to investigate the potential of a well-designed solvent extraction and raw-material manufacturing process to deactivate or remove microbial contaminants from compromised feed material during the production of materials used to create consumer products.
Plant Material Preparation
Two batches of contaminated cannabis inflorescence (Triangle Kush x Fruity Pebbles, TKFP211007H/TKFP211007H1) were combined into a single batch of plant material large enough to support two experiments. Sub-batches were derived from the homogenized parent batch and were treated differently pre-extraction. For hydrocarbon extraction, the plant material's native bud structure was retained and batched in equal-mass replicates. Prior to carbon dioxide extraction, all parent plant material was ground using a Robocoupe Blixer 15 then batched into equal mass replicates.
The total amount of plant material allocated to hydrocarbon and supercritical fluid extraction was 9.67 kg and 12.98 kg, respectively. The mass of plant material in each hydrocarbon replicate (n=5) was 1.93 kg and the mass of each supercritical carbon dioxide replicate (n=5) was 2.60 kg.
Carbon Dioxide and Hydrocarbon Extraction
Hydrocarbon extraction was performed by cooling the extraction solvent to -20 oC before injecting the solvent into the closed-loop extraction system. The extraction process was conducted under dynamic flow (1.0 L/min.) conditions at 1 Bar. Each extraction lasted 15 min with temperature held constant at -20 oC by heat exchangers located before and after the extraction columns.
Supercritical carbon dioxide extraction was conducted under dynamic flow with the solvent pressurized to 285 Bar and heated to 55 oC; the flow rate of the carbon dioxide solvent was 275 g/min over 8 h resulting in a solvent to feed ratio of 51:1. The temperature and pressure conditions in the retrograde phases were 55 oC/115 Bar, 40 oC/60 Bar, and 24 oC/46 Bar, in the first, second, and third separators, respectively. The process conditions outlined above resulted in a carbon dioxide extraction output of two different fractions; those fractions were derived from separator 1 and separator 2. No material was obtained from separator 3. Concordantly, each extraction replicate yielded two fractions (denoted A and B) that were tested independently and later combined for downstream processing.
Winterization of the primary hydrocarbon extract was performed in the extraction solvent by maintaining a temperature of -80 oC for 12 h; the raw extract from each replicate was transferred to a vessel suitable to long term storage at -80 oC.Following the extended period of exposure to -80 oC, the raw extract was pumped at a rate of 3 L/min through the CUNO BHN2 bag (25 µm) filtration system at 1 Bar.
Combined extracts (such as the A and B fractions) from each carbon dioxide extraction replicate were dissolved in absolute ethanol at 40 oC with the assistance of a sonic bath. The volume of ethanol used during this process was 10x the mass of the combined extracts. To ensure that the raw extracts were fully dissolved, a hand blender was used. The ethanol and raw extract mixture was then put into a freezer at -40 oC for 12 h; following, the fats and waxes were removed using a 300 mm stainless steel Buchner funnel, Whatman 1 (11 µm) filter paper, and a vacuum pump and flask assembly. The residual fats on the filter were rinsed with clean, cold ethanol (-40 oC) until they were pearlescent.
To remove chlorophyll from the winterized hydrocarbon extracts, 650 g of pharmaceutical grade magnesium silicate was packed into a specialized column that was attached to the closed-loop extraction system. With that column in place, the winterized and filtered extract was pumped down through MgSiO3 at a rate of 1 L/min, at a pressure of 1.5 Bar, and 20 oC until all material was fed in through the system.
Clarification of the winterized and filtered supercritical carbon dioxide extract was conducted by direct addition of peat-derived activated carbon. Next, 35 g of activated carbon was added directly to the winterized extract that was dissolved in ethanol. The mixture was then heated to 50 oC and agitated with an immersion circulator for 30 min. Subsequently, the activated carbon was removed from the ethanolic mixture using a 300 mm Buchner funnel with Whatman 1 filter paper (11 µm) attached to a vacuum pump and flask system operating at 0.1 Bar.
In both extraction paradigms, sterilizing filtration was executed in a similar manner except for a single difference; that difference was the temperature that was maintained during the process. In the case of hydrocarbon extracts, the sterilizing filtration occurred at room temperature. For supercritical carbon dioxide extracts, the winterized and clarified mixture was cooled to -40 oC for 12 h before sterilizing filtration to ensure that all fats and waxes were captured. Both hydrocarbon and supercritical carbon dioxide extracts were pumped into the 3M cartridge filtration system (<0.2 µm) at a pressure of 1 bar until all extract moved through the system.
For hydrocarbon extracts, the solution was transferred to a 10 L rotary evaporator. The conditions for solvent recovery of the hydrocarbon solvent were 36 oC, 0.5 Bar, and 100 RPM. The solvent recovery system was continuously fed until all material was transferred and the essential oil of cannabis was the consistency of maple syrup.
For supercritical carbon dioxide extracts, a 10 L rotary evaporator was used and the conditions were 55 oC, 0.15 Bar, and 100 RPM. The solvent mixture was continuously fed into the system until all material was transferred and the essential oil was the consistency of maple syrup.
The purging process is unique to hydrocarbon manufacturing and is the terminal process for creating raw materials that are used in hydrocarbon-extract product manufacturing. The purging process was executed in a radiant heat vacuum oven coupled with a scroll-pump and cold trap assembly. The extract was purged at 30 oC and 0.07 Bar in a thin layer for 22 h.
Decarboxylation of the refined carbon dioxide cannabis extract was conducted in a boiling flask and stirring-mantle assembly coupled with a vacuum pump and cold trap. The decarboxylation was conducted over 2 h at a temperature of 165 oC and a vacuum level of 0.001 Bar with constant agitation.
Cannabinoid distillation was conducted in a wiped film system with a feed rate of 250 g/h at 0.08 mBar; the cannabinoids were condensed at 98 oC, and the terpenes were condensed at -15 oC or -60 oC.
Overview of Material Flow and Testing Points by Manufacturing Line
Supercritical Carbon Dioxide:
Analytical Testing of Extracted Materials
All testing of samples for cannabinoid potency (%), terpene content (%), microbial contaminants (cfu/g), and mycotoxins (µg/g) was undertaken by Steep Hill Pennsylvania, an International Organization for Standardization (ISO) and International Electrotechnical Commission (IEC) certified laboratory that is approved by the Commonwealth of Pennsylvania for testing of medical marijuana products.
The samples were homogenized and subdivided into the required weight for each analytical test. Each test was completed in replicates of five. Test portions were extracted with an organic solvent and prepared for analysis following Steep Hill PA’s standardized operating procedures. Cannabinoid analysis was conducted using high performance liquid chromatography (HPLC). Limits of quantitation for the cannabinoids CBD, cannabidiolic acid (CBDA), cannabinol (CBN), Δ9 THC, tetrahydrocannabinolic acid (Δ9 THCA), cannabidivarin (CBDV), tetrahydrocannabivarin (THCV), cannabigerol (CBG), cannabigerolic acid (CBGA), delta-8-tetrahydrocannabinol (Δ8 THC), and cannabichromene (CBC) were 0.52, 0.42, 0.28, 0.41, 0.41, 0.39, 0.39, 0.51, 0.50, 0.78, and 0.47 µg/g, respectively. Terpene analyses were conducted using gas chromatography–mass spectrometry (GC–MS) and the limit of quantitation varied between 0.055 and 0.072 µg/g depending on the identity of the analyte. Mycotoxin analyses were conducted using liquid chromatography–mass spectrometry (LC–MS); the limit of quantitation for Aflatoxin B1, Aflatoxin G1, Ochratoxin A, Aflatoxin B2, and Aflatoxin G2 were 1.44, 0.56, 1.44, 1.34, and 1.28 µg/g. Microbial testing was conducted using ready-to-use plates provided by 3M and BioRad. Third-party validations confirm that these platforms are reliable and derived data are repeatable. The manufacturers’ standard operating procedures (SOPs) for sample preparation and analysis were followed. The following ready-to-use agar products were used in this study:
Data Treatment and Statistical Design
Two data matrices were assembled from available data; the hydrocarbon and carbon dioxide matrices were 15x18 and 70x18, respectively. Independent variables in each matrix included the oil identification, replicate, and process-input material mass. Response variables included output mass and the cannabinoids CBD, CBDA, CBN, Δ9 THC, Δ9 THCA, CBDV, THCV, CBG, CBGA, Δ8 THC, and CBC; additionally, response variables included total yeast and molds, total aerobic bacteria, total enterobacteria, E. coli, and Salmonella spp. The variable “total mycotoxins” was calculated from individual values of the aflatoxins B1, B2, G1, G2 and Ochratoxin A.
Prior to conducting any calculations or analyses, all cannabinoid potency results were mathematically converted to account for the decarboxylation of acid form cannabinoids. This was done by multiplying the acid form percentage of a given cannabinoid by the decarboxylation coefficient (0.87) then adding that to the percentage of the same neutral form cannabinoid. Statistical analysis of each data set was undertaken using Repeated Measures Analysis and Tukey's honestly significant difference (HSD) post-hoc analysis to evaluate the influence of each manufacturing step in removing microbial contaminants. All analyses were executed in R (https://www.r-project.org/).
The parent material contained 17.27% total cannabinoids (that is, Δ9 THCA = 12.70% and Δ9 THC = 3.94%) and 1.28% total terpenes. Microbial tests indicated that the plant material was contaminated with 28,000 cfu/g total aerobic bacteria, 17,975 cfu/g enterobacteriaceae, and “too numerous to count” cfu/g total yeast and mold. Where “too numerous to count” were found for a specific group of microbial contaminants, a value of 1 x 106 cfu/g was tabulated in the data matrix for statistical purposes. No E. coli, Salmonella spp. or mycotoxins were detected in the parent material.
The most important finding of this study was that both hydrocarbon and supercritical carbon dioxide extraction of contaminated plant material resulted in manufactured raw materials that were free of viable microbial contaminants that were evaluated during this study. No statistical analyses were possible because there was no variance in the dataset following the extraction step; in short, all tested microbial contaminants were not detected after the parent plant material was processed via solvent extraction.
The essential oil manufacturing process, which included extraction, winterization, clarification, sterilizing filtration, and solvent recovery, extracted 1185.2 g of cannabinoids from the parent material that contained 1670.2 g of cannabinoids. The essential oil was free of all tested microbial contamination and mycotoxins following those manufacturing steps (see Figures 1–3 and Table I).
The purged oil was also found to be free of all tested microbial contaminants and mycotoxins (see Figures 1-3 and Table I); the transition of material from the essential oil phase to this phase was 93.9% efficient where 1112.3 g of cannabinoids were recovered during the purging process from the essential oil that contained 1185.2 g of cannabinoids.
Supercritical Carbon Dioxide
Five extractions, each with 2 fractions, were 84.6% efficient where 1730.8 g were extracted from the plant material that contained 2046.3 g of cannabinoids. Additionally, raw extracts from the supercritical carbon dioxide process were free of all tested microbial contaminants and fungal metabolites (see Figures 4-6 and Table II).
Following winterization, clarification, sterilizing filtration, and solvent recovery, the essential oil was found to be free of all tested microbial contaminants and mycotoxins (see Figures 4-6 and Table II).Overall, those processes were 78.8% efficient for cannabinoid retention where 1363 g of 1730.8 g of cannabinoids were retained through the four steps.
The decarboxylation step was found to be 97.2% efficient for cannabinoid retention and 1325.4 g of 1363 g of cannabinoids were recovered during the process. Additionally, the decarboxylated oil was found to be free of all tested microbial contaminants and fungal metabolites (see Figures 4-6 and Table II).
The cannabinoid distillation step was found to be 96.0% efficient; present in the decarboxylated oil was 1325.4 g of cannabinoids and the distillate oil contained 1272 g of cannabinoids. Finally, the distillate oil was free of all tested microbial contaminants and mycotoxins (see Figures 4-6 and Table II).
Overall, our results suggest that the tested microbial contaminants are not transferred from feed material to the raw materials produced during extraction and refinement and that those raw materials are suitable for use in product manufacturing.
Sterilizing of products during the manufacturing process or prior to distribution is not uncommon in the food, pharmaceutical, and medical device industries. The US FDA has reviewed a variety of sterilization methods for use in the biomedical and food industries such as irradiation, gas sterilization, steam or dry heat sterilization, sterilizing filtration, and liquid dips (4,5). Furthermore, the US FDA presented a challenge in 2019 to encourage companies to develop a new sterilization technique or set of techniques capable of replacing ethylene oxide in the food and medical device manufacturing industries because of its flammability and potential carcinogenicity (5-7). One promising technique for product sterilization is the application of supercritical carbon dioxide including additives such as water (8,9), hydrogen peroxide (10,11), peracetic acid (4,12), and ethanol (13). Additionally, supercritical carbon dioxide without additives has been shown to remove notable organisms such as Salmonella typhimurium (14), Escherichia coli (15,16), Listeria monocytogenes (17), Penicillium spp. (8), and Aspergillus spp. (18).
Hydrocarbon extraction utilizes a hydrocarbon solvent and submicron filtration (such as, sieve <0.2 µm) in the manufacturing process to produce products such as sugars, diamonds, liquid live resins, and diamond sauces. The results of our experiment utilizing contaminated plant material suggested that the techniques and infrastructure used here can create raw materials free of viable microbial contaminants and fungal metabolites that were tested here (no data shown for mycotoxins; 0 ppb at all stages for both technologies). Therefore, we assert that the hydrocarbon extraction, processing techniques, and infrastructure utilized in this study are suitable for creating raw material from contaminated feed material that meet or exceed current PA Department of Health regulations for extracted materials (1).
Supercritical carbon dioxide manufacturing also takes advantage of numerous aseptic processing technologies. At the extraction stage, supercritical carbon dioxide was utilized without the introduction of additives. It is unclear whether microbial deactivation or removal was due to supercritical carbon dioxide alone because there are a variety of materials interacting in the supercritical carbon dioxide during extraction (for example, water, cannabinoids, terpenes, and so on). However, it is apparent that the tested microbial contaminants were removed or deactivated during the first stage of our supercritical carbon dioxide manufacturing process. Furthermore, after the extraction process the extract was winterized, clarified, and filtered through a <0.2 µm sieve prior to solvent recovery; our results suggest that microbial deactivation or removal was maintained through those steps. Subsequent supercritical carbon dioxide manufacturing processes leverage heat to decarboxylate the acid-form cannabinoids and distill the neutral-form cannabinoids. Both of those processes were executed at temperatures greater than 100 oC for extended periods of time; our results suggest that, at both of those stages, microbial deactivation or removal is maintained and that those products are free of fungal metabolites. Therefore, we assert that the infrastructure and techniques leveraged during this study can produce raw materials that meet or exceed PA Department of Health’s current regulations (1).
This study, which evaluated the potential of a well-designed unidirectional aseptic processing approach to both hydrocarbon and supercritical carbon dioxide extract manufacturing, suggests that the extraction process is capable of deactivating or removing the tested microbial contaminants when highly contaminated Cannabis spp. inflorescence is used as feed material. Finally, the absence of viable microbial contamination tested herein was maintained throughout all raw material manufacturing steps in both the hydrocarbon and supercritical carbon dioxide schemes.
We would like to thank the Pennsylvania Department of Health for allowing us to undertake this study as well as the Pennsylvania Cannabis Coalition and the Organic Remedies Inc. board for financial support. Additionally, we would like to thank the Philadelphia College of Osteopathic Medicine and Duquesne University for their time and support of this research initiative. Finally, this work could not have been completed without the support of the Specialists and Technicians employed in the Organic Remedies Inc. laboratory.
M.R. June-Wells, W.P. Petroski, and E. Hauser are with Organic Remedies Inc., Department of Extraction in Carlisle, Pennsylvania. F.W. Fochtman is with Duquesne University Bayer School of Natural and Environmental Sciences in Pittsburgh, Pennsylvania. B. Balin is with the Philadelphia College of Osteopathic Medicine Center for Chronic Disorders of Aging in Philadelphia, Pennsylvania. D. Niesen is with Green Analytics North dba Steep Hill Pennsylvania in Harrisburg, Pennsylvania. Direct correspondence to email@example.com.
M.R. June-Wells, F.W. Fochtman, B. Balin, W.P. Petroski, D. Niesen, and E. Hauser, Cannabis Science and Technology® Vol. 5(9), 20-31 (2022).