With more and more states legalizing medical marijuana every year, many analytical chemists are now bringing their expertise to the field to help ensure that marijuana products are tested properly for potency and for contaminants such as pesticides. To get a sense of some of the current issues in the analysis of medical cannabis, we spoke to five scientists working in this field: Jeff Dahl, an applications scientist at Shimadzu Scientific Instruments; Joe Konschnik, a business development manager for the food and beverage market at Restek Corporation; Andre Santos, the Americas market development manager at Agilent Technologies; Alicia D. Stell, the CEM product development lab manager; and Xiaoyan Wang, a research scientist at UCT.
What are the biggest challenges in sample cleanup or preparation for cannabis analysis?
Dahl: Samples come in three major forms: dried flower, cannabis extracts and concentrates, and edibles. This diversity makes sample preparation more difficult because extraction and cleanup methods may have to be adapted to each type of product. Further, edibles may be quite different from each other and from the original plant, so that extraction efficiencies, potential matrix effects, and interferences can be unpredictable. Laboratories have limited time and budgets, so analysis has to be easy, fast, affordable, and effective.
Konschnik: I’ve worked closely with cannabis testing labs for the past five years, and in my experience the most significant challenges are matrix related. First, isolating the compounds of interest from a very complicated sample matrix can be difficult. Secondly, sample cleanup of varying matrices, from flowers to edibles, can make analyses extremely complex and increase variability. Lastly, detection of analytes in the presence of high concentrations of native active chemicals of interest, such as tetrahydrocannabinols (THCs), can often interfere with quantitation.
Santos: As with other botanical matrices, cannabis presents challenges because of its complex chemical composition that includes sugars, fatty acids, terpenes, and so on, and its variability. The sample cleanup and preparation will also depend on the analysis to be performed (such as potency, residual solvents, terpenes, or pesticides) and on the type of matrix (such as edibles, beverages, or plant material).
Stell: I believe the biggest challenges in sample cleanup or preparation for cannabis analysis will be the difficult sample matrix and the need for small sample size.
Wang: Sample homogeneity is one of the biggest challenges in cannabis testing. An adequately sized, representative sample should be submitted, because different parts of the cannabis plant contain different amounts of cannabinoids; for example, the flowers and buds have the highest potency. Other major issues at play include the complexity of the various matrices being analyzed along with the wide polarity range of compounds often being monitored.
What sample cleanup or preparation approaches do you recommend for cannabis analysis?
Dahl: For potency analysis using liquid chromatography with UV detection (LC–UV), we found solvent extraction with methanol is effective with a wide variety of samples. For pesticides in dried flower, we extract using a QuEChERS (quick, easy, cheap, effective, rugged, and safe) method and clean up the samples with dispersive solid-phase extraction (dSPE). For pesticides in cannabis concentrates we dissolve in methanol, filter, and directly inject. We found this approach was not only effective but robust as well. If lower detection limits were required, we might have to use additional cleanup and concentration. Grinding cannabis is sometimes difficult, with limited sample amounts usually available.
Konschnik: We’ve found that QuEChERS and dSPE have been effective at helping chemists effectively and quickly extract the analytes of interest from cannabis plant material and marijuana-infused products (MIPs), then subsequently clean up the extracts for analysis. We expect that future applications with solid-phase microextraction (SPME) may also be advantageous as a sample preparation approach.
Santos: That depends on the analyte being tested. For pesticides, QuEChERS followed by dSPE or our lipid matrix removal product for cleanup are possible approaches. However, depending on the matrix, different extraction or cleanup approaches may be required. For residual solvents, for example, headspace and SPME coupled to gas chromatography with flame ionization detection (GC–FID) or mass spectrometry detection (GC–MS) are the most common methods.
Stell: We will be starting our cleanup or preparation approach with QuEChERS; however, I believe no matter the approach, a modified method is going to be needed.
Wang: I would suggest QuEChERS, a multiresidue method that can extract cannabinoids, pesticides, mycotoxins, and other organic contaminants all in one procedure. Following extraction, dilutions can be made for potency analysis. Dilution is often required because of the high cannabinoid concentrations commonly encountered, which can ultimately lead to signal saturation in the analytical detector. For chemical residue analysis, dSPE using different sorbents, can be performed to remove matrix coextractives.
In developing methods for cannabis analysis, have you had to work with a surrogate sample, because of difficulty in obtaining cannabis samples? If so, have you identified a surrogate that you believe is representative?
Dahl: We use authentic cannabis for most of our work, which is done on location in state-certified labs. We have tried dried hops cones as a surrogate; however, there is a very different wax and resin profile, and hops has a high percentage by weight of bitter acids. Hemp flowers may be another option. We have not found any candidate surrogate for cannabis concentrates.
Konschnik: For some applications we have found suitable surrogates, but for others we have not. For example, we’ve used hops as a simulated plant matrix for the analysis of terpenes by headspace GC–FID. There really isn’t a good substitute for cannabis plant material for other analyses routinely performed on marijuana matrices because it’s not just the plant material that presents analytical challenges, but also the large number of native compounds present, which can interfere with analysis of targeted compounds.
Stell: Yes, obtaining cannabis samples is difficult, so we will be starting our approach with a surrogate sample and will be using hops.
Wang: UCT has been fortunate enough to have steady access to edibles, topical oils, and plant material to conduct our pesticide, mycotoxin, and cannabinoid testing. However, both hops and hemp seed are often used by laboratories that are having trouble obtaining cannabis samples for method validation. The actual texture and plant structure of hops is similar to that of cannabis, allowing scientists to mimic as best they can some of the sample preparation issues that often accompany this category of analysis.
What are the biggest challenges in the analysis of pesticides in cannabis?
Dahl: Cannabis is a high-pigment and high-wax matrix that can sometimes suppress signal and cause interference. Reaching low detection limits can be difficult if the background or interferences are high. Since pesticides may be ionized in positive or negative mode, the mass spectrometer should be capable of high-speed polarity switching and fast multiple reaction monitoring (MRM) event times to accommodate a large number of compounds. Without ultrafast mass spectrometry (MS), each sample might require multiple runs. Finally, data processing requires a little extra care to ensure confidence in pesticide detections.
Konschnik: First, deciding which pesticides require analysis, then finding certified reference materials (CRMs) for each can be challenging. Most states where medical cannabis is legal have defined their lists, but some have not. Second, along with the typical challenges associated with pesticides extraction and analysis in plant material, the presence of native compounds at percentage levels can interfere with quantitation, especially when determining trace detection limits for certain pesticides. Meeting acceptable quality control criteria can be a challenge for laboratories with limited pesticides analysis experience.
Santos: Again, matrix complexity is one of the biggest challenges in testing cannabis for pesticides. So, depending on the matrix being tested, different sample preparation methods will be required. Also, depending on whether the pesticides to be tested are GC- or LC-amenable, both GC–MS/MS and LC–MS/MS may be required. Since the regulations around cannabis are still evolving and there are still very few pesticides, if any, approved for use in cannabis, a laboratory may be faced with the challenge of testing for hundreds of pesticides.
Stell: The biggest challenge is the unknown, as there does not seem to be a consensus on which pesticides need to be analyzed.
Wang: One of the biggest challenges in the analysis of pesticides in cannabis is to reduce the matrix coextractives so that fewer interferences will be introduced into the analytical instruments. Marijuana samples, including cannabis edibles, are complex matrices, containing lipids, organic acids, cannabinoids, sugars, food dyes, and natural pigments (such as chlorophylls and anthocynins) that can be coextracted with pesticides that interfere with the pesticide analysis. Another major issue commonly encountered is the extremely high concentration of the cannabinoids suppressing or masking the pesticide peaks, or both.
What analytical methods (such as LC–MS, GC–MS, supercritical fluid chromatography) are you applying to the analysis of pesticides in cannabis, and why?
Dahl: We use LC–MS/MS for most pesticide analysis in cannabis. A triple-quadrupole MS system in MRM mode is best because it gives the optimum combination of cost, sensitivity, and selectivity. Modern mass spectrometers are easy to use, robust, and affordable. Most pesticides of interest required by state lists can be done by LC–MS, but there are a few that may require GC–MS.
Konschnik: We’re seeing most labs using a combination of all techniques you’ve mentioned. We are seeing mostly LC–MS/MS being used for pesticide analysis. There is some interest in, and recent work done, using GC–MS, as well. The difference between the two is often cost- and analyte-specific. Until there are firm guidelines and regulation on the testing method requirements, it is difficult to confidently say one is better than the other.
Santos: Both LC–tandem quadrupole MS and GC–tandem quadrupole MS can be used, combined with extensive MRM databases. Actually, both GC–MS and LC–MS may be required because depending on the molecule to be analyzed one may not be able to generate ions and obtain a reasonable signal by LC–MS, either with electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI), thus requiring electron ionization (EI).
Stell: We currently have access to a GC–MS system for this analysis, so we are initially using that; however, we believe with time we will need an LC–MS system to be able to analyze everything necessary.
Wang: We have executed LC–MS/MS for the analysis of pesticides in cannabis and edibles. While LC–MS/MS can detect hundreds of pesticides at the needed sensitivity levels, GC–MS/MS is also favorable because there are some pesticides that do not fragment efficiently via LC–MS/MS or are simply just more GC-amenable. Both instrument techniques are recommended for cannabis pesticide analysis, such as for the medical marijuana pesticide list recently issued by the Nevada Department of Agriculture, which contains both LC- and GC-amenable pesticides.
What do you think are the areas with the most opportunity for further development in terms of analytical methods for cannabis analysis?
Dahl: Sample preparation could be improved by simple methods to selectively strip cannabinoids, resin, wax, and so on from the sample. Some concentrates are already dewaxed during their production, so wax removal is not the only concern. Methods to extract mycotoxins are also needed. Like any lab, cannabis testing labs can benefit from automation, including on-line sample preparation. Automation decreases workload and increases accuracy and precision.
Konschnik: The vast chemical diversity of this plant species has just begun to be characterized. With some labs analyzing for only THCs, fewer than 10 cannabinoids, and 20 terpenes, I can’t help but think further analytical developments will reveal new structural isomers among hundreds of compounds present. Along with the extraction, purification, and subsequent concentration of cannabis oils, better analytical techniques will also be needed to further characterize them, since contaminants are concentrated along with the native active compounds.
Santos: I think there is a great opportunity around the harmonization and standardization of methods and quality assurance at the laboratories—for example, with accreditation by ISO 17025—as the regulations evolve. Improvements in the sample preparation workflows will continue to be important as well. I also believe that there are opportunities around testing of trace metals by both inductively coupled plasma–optical emission spectroscopy (ICP-OES) and ICP–mass spectrometry (ICP-MS), and around adulteration and authenticity.
Stell: Given how difficult the sample matrix is for cannabis analysis, I believe the area of sample cleanup or preparation offers the most opportunity for further development.
Wang: The first is cleanup options, such as novel sorbents, push-through filters and cartridges, spin filters, and well plates for cleaner extracts or higher sample throughput. The second is methods that can isolate cannabis content from pesticide content to prevent interference. The third is uniform standards and analytical regulations for the testing for potency, pesticides, and other residual content in both plant and edible material. The fourth is robotic or liquid-handling platforms that can fully automate the QuEChERS extraction technique.
This interview was originally published in the "2016-2017 Annual Industry Trends and Directory Issue," a supplement to LCGC North America in August 2016.