Tips and Challenges for Defensible Analysis of Pesticides in Cannabis Products
by
Marco Troiani
Savino Sguera
Digamma Consulting
As the cannabis industry becomes more regulated, analytical laboratories must generate data that results in their routinely passing regulatory audits. All analyses performed by the labs are challenging, but one of the biggest issues the industry faces is the analysis of pesticides.
Pesticide analysis can be more challenging than other analyses due to low action levels, in parts per billion (ppb) or nanograms (ng). Other contaminant evaluations, such as residual solvents, often have action levels in the parts per million (ppm) or microgram (µg) range, which is 1,000 times greater than ppb. Other organics, such as terpenes and cannabinoids, are present at higher levels such that percent (%) or milligrams-per-gram (mg/g) are used.
Heavy metal limits are also in the ppb range. The thorough digestion used in the analysis creates a more favorable signal-to-noise (S/N) environment in the inductively coupled plasma mass spectrometer (ICP-MS) than we see in LC-MSMS or GC-MSMS for pesticide analysis, where LC and GC are liquid and gas chromatography, respectively. Because there are only 92 naturally occurring elements, complete digestion of heavy metals allows for each to be measured accurately with little interference. In the analysis of pesticides, however, the number of interferences and similar compounds that may occur is extraordinarily high. This is why the technology used for pesticide analysis utilizes a series of mass filters to accurately detect analytes at trace levels.
ANALYTES
No singular ion source is sufficient to quantify all pesticides on most cannabis monitoring lists. Electrospray Ionization (ESI+/-) is typical for LC-MSMS, but is not effective for all analytes. A second source of ionization is currently necessary to detect remaining pesticides and varies between the following: Electron Impact (EI+) Ionization used for GC-MSMS and Atmospheric Pressure Chemical Ionization (APCI) for LC-MSMS or GC-MSMS.
Image01
As the cannabis industry becomes more regulated, analytical laboratories must generate data that results in their routinely passing regulatory audits. All analyses performed by the labs are challenging, but one of the biggest issues the industry faces is the analysis of pesticides.
Pesticide analysis can be more challenging than other analyses due to low action levels, in parts per billion (ppb) or nanograms (ng). Other contaminant evaluations, such as residual solvents, often have action levels in the parts per million (ppm) or microgram (µg) range, which is 1,000 times greater than ppb. Other organics, such as terpenes and cannabinoids, are present at higher levels such that percent (%) or milligrams-per-gram (mg/g) are used.
Heavy metal limits are also in the ppb range. The thorough digestion used in the analysis creates a more favorable signal-to-noise (S/N) environment in the inductively coupled plasma mass spectrometer (ICP-MS) than we see in LC-MSMS or GC-MSMS for pesticide analysis, where LC and GC are liquid and gas chromatography, respectively. Because there are only 92 naturally occurring elements, complete digestion of heavy metals allows for each to be measured accurately with little interference. In the analysis of pesticides, however, the number of interferences and similar compounds that may occur is extraordinarily high. This is why the technology used for pesticide analysis utilizes a series of mass filters to accurately detect analytes at trace levels.
ANALYTES
No singular ion source is sufficient to quantify all pesticides on most cannabis monitoring lists. Electrospray Ionization (ESI+/-) is typical for LC-MSMS, but is not effective for all analytes. A second source of ionization is currently necessary to detect remaining pesticides and varies between the following: Electron Impact (EI+) Ionization used for GC-MSMS and Atmospheric Pressure Chemical Ionization (APCI) for LC-MSMS or GC-MSMS.
Image02
EI+ is dependent on the following conditions: an inert, gaseous mobile phase, and analytes that can receive stable positive charge from loss of an electron. The mass spectra often reveal peaks of masses less than or equal to the molar mass. The nature of the EI+ source assures that fragments are common but adducts are not observed.
Nearly all pesticides on the California monitoring list for cannabis analysis are amenable to ESI+/- ionization and perform well using LC-MSMS. Some compounds require either EI+ ionization or APCI+/- ionization to be detected. These analytes include azoxystrobin, captan, chlordane, chlorfenapyr, chlorpyrifos, cyfluthrin, cypermethrin, methyl parathion, pentachloronitrobenzene, and permethrin.
Many analytes begin to degrade in solution under two common conditions: after removal from a freezer when temperatures subsequently rise and degradation reactions occur at higher rates; and when combined with other analytes or matrix components, wherein the half-life of the analyte in solution sharply decreases. Captan hydrolyzes in the presence of mildly basic water, and many other pesticides have basic chemical properties. A combined standard of all pesticides on a monitoring list most likely will not be accurate for more than a 24-hour period at room temperature. Therefore, multi-part standard mixes and matrix-matched calibration standards must be prepared daily to assure accuracy.
MATRIX
Cannabis testing labs face a challenge in finding matrix blanks. Although toxicology and pharmaceutical labs have well-controlled blank matrices, cannabis labs face variation similar to that seen in environmental labs. The major categories of matrices analyzed in cannabis labs are the following: flower, concentrates, edibles, and topicals.
Across all matrices, there are typically four major interferences observed: cannabinoids and terpenes, waxes and lipids, carbohydrates and amino acids, and polymers. Because each matrix class has varying ratios of the interfering compounds, it is necessary to matrix-match the calibration to ensure consistent recoveries.
Image03
Matrix-blank material must be selected from a pesticide-free material that matches the composition of interference compounds present in a true matrix. Organic cannabis is recommended for flower, but in lieu of that, buds of hops can serve as a proxy. Organic hemp oil is recommended for concentrates and is readily available at retail stores. Organic dry cereal is recommended for carbohydrate edibles and organic chocolate and coconut oil are recommended for lipids in edibles and topicals. Using the same matrix blank material for calibration and quality control allows for method variations to be controlled enough to comply with regulatory standards (+30% recovery) of California and within the EPA standards (+20%). It is recommended to test all matrix blank materials for the presence of analytes before adoption as a standard blank material.
Matrix-matched calibration is necessary because clean-up steps invariably cause analyte loss from extraction solvents. The best way to have the quality control samples match the client and calibration samples is to matrix calibrate with a matrix blank and run your quality control tests on that same blank.
Image04
Figure 4 provides data from a study presented at the American Chemical Society 2016 conference in Philadelphia, where the extraction efficiency of each analyte on the Nevada cannabis pesticide monitoring list is illustrated. Each analyte has a different percent recovery, and to correct for this variation, matrix-matched calibration is necessary. The percent recoveries of some pesticides are as low as 20%, well below the 80% recovery minimum in most analytical industries and the 70% minimum in California’s cannabis industry.
HOMOGENIZATION
Proper homogenization of each sample tested is required for reproducibility of reported data. Pesticide distribution is often not uniform, so samples should be homogenized to fine particle sizes and well mixed. Fine particle mesh also allows less acetonitrile to be sequestered in the plant matrix, and for a greater volume of acetonitrile to be collected after sample extraction. This enables a larger ratio of sample mass to extraction volume, which gives lower limits of detection for a method.
Image05
EXTRACTION
LC and GC systems have different vulnerabilities when it comes to matrix interferences, and require extraction clean-up approaches that protect each instrument. Interfering hydrophilic species (carbs, amino acids, proteins) are problematic for GC’s hot and dry environment, which causes Maillard reactions [1]. The Maillard reaction is a bond formation between carbohydrates and amino acids and is seen often in cooking when foods brown or caramelize. These reaction products are sticky and difficult to remove from the dry and hot environment in the GC system, making QuEChERS necessary for GC-MSMS maintenance. Because heat and exposure to oxygen is a necessary element of the Maillard reaction, the LC system’s cooler and wetter environment prevents formation of Maillard products. [1]
Interfering hydrophilic compounds are effectively removed from sample extract with QuEChERS salts. Interfering hydrophobic species (waxes, hydrocarbons) are problematic for LC because some have higher affinity for the C18 column than a mobile phase like methanol [2], and subsequently may irreversibly bind to the column, clog the ESI probe, or dirty the quadrupole, altering retention times or chromatographic resolution. Thankfully hydrophobic analytes can be removed with a lipophilic purge.
Image06
QuEChERS salts cause the loss of the ionic analyte daminozide because the pesticide’s high polarity causes it to bind preferentially to the dehydrating salts over the acetonitrile solvent. This step is incompatible with LC pesticide analysis, where daminozide is present on the pesticide monitoring list. Using 10-50 mg of graphitized carbon black (GCB) dispersed in 1.0-mL of extraction solvent used as a non-specific extract clean-up is preferred to QuEChERS when daminozide is on pesticide monitoring list. GCB clean-up in lieu of QuEChERS is not as effective but it’s compatible with LC systems. This method is not recommended on GC systems due to the inability to remove nearly all carbohydrates and proteins from extraction solvents, and the aforementioned Maillard reactions that can occur.
Image07
LC column packing material is typically functionalized with C18 moieties, which can cause issues with strongly hydrophobic components in a sample matrix. C18 is very aliphatic and binds strongly to certain compounds, such as aliphatic fatty acids, their triglycerides, and wax derivatives. Lipophilic clean-up steps, such as a hexane biphasic purge, may be necessary for proper LC-MSMS column maintenance. A GC system’s high heat allows for the off-gassing of aliphatic hydrophobic compounds due to their thermal lability and relatively low boiling points. Therefore, it’s important to purge and remove hydrophobic matrix interferences from extracted samples for LC analysis.
Image08
QuEChERS alone does not completely remove hydrophobic interferences, such as THC or waxes. In Figure 9, where relative THC concentration is compared across several extract preparations, sample A used QuEChERS only, while B-I involved various amounts of GCB as an additional step.
Image09
Both QuEChERS and lipophilic purges utilize liquid-liquid biphasic extraction to selectively remove compounds from an extract. When combined, as outlined in Figure 10, we can sequentially remove water and hydrophilic species, and then aliphatic hydrophobic compounds from an acetonitrile extract that is still enriched for pesticides.
Image10
A combined extraction approach involves a complex and coordinated clean-up. Sample are homogenized, dissolved in solvent, and the extract volumes are split for LC- and GC analyses. LC samples can be cleaned up with GCB and then purged of aliphatic compounds before LC injection. The GC samples are run through both stages of QuEChERS including dispersive solid-phase extraction with GCB added, and then treated with an aliphatic compound purge before injection.
QuEChERS cleanup is traditional in the regulated agricultural pesticide testing industry and an excess of validation documentation exists. The addition of the lipophilic purge was learned from the FDA-regulated olive oil industry. Olive oil has a massive hydrophobic component in the sample matrix, and analytical labs use GC and LC for quantifying pesticides contaminating the plants. [3] These hydrophobic interferences are similar to the cannabinoids, terpenes, and waxes seen in cannabis plants.
ANALYSIS
Other considerations for LC involve auto-sampler tray temperature, which can cause target or interference compounds to precipitate or clog vessels, injection volume, and needle washes. GC considerations involve the programmable temperature vaporizing (PTV) inlet and its necessity due to the range of optimal vaporization temperatures of certain analytes (captan, cypermethrin, cyfluthrin, etc.). These rapidly degrade when injected into a hot and constant temperature GC inlet, causing inconsistent results. With a PTV inlet, these analytes can be consistently delivered to the head of the GC column by slowly moving through the ideal vaporization temperature for each compound.
A thorough rinse of the GC sampling needle is also required to prevent jams which otherwise could frequently occur. An ideal rinse program incorporates hexane, isopropanol, and acetonitrile. This program works well because hexane and isopropanol are miscible, as are isopropanol and acetonitrile. Hexane and acetonitrile’s immiscibility is the basis for the biphasic lipophilic purge, but isopropanol provides an ideal middle-ground between the polarity of the two compounds to keep the injection needle free of clogs. The needle dwell time should be kept to a minimum to prevent burning of extract solvent on the needle exterior. This burnt material can re-circulate in extraction solvents and cause jams in the needle assembly.
When developing your chromatographic method, an initial rapid separation of analytes in solvent should be used as a baseline method. Further separation of analytes from matrix interferences can be performed for each matrix with unique chromatographic programs for each matrix class. Ion suppression and other variable “dark” influences on instrument response necessitate a matrix-matched calibration curve for each matrix type, even if no visible differences are identified on the chromatograms.
Some multiple reaction monitoring (MRM) channels show interferences in cannabis matrices, especially those with masses close to cannabinoids. Examples include myclobutanil, pyrethrins, acequinocyl, and spiromesifen. To preserve selectivity, additional chromatographic separation is necessary and which is why a separate chromatographic program may be necessary for each matrix class.
LC/MS-grade solvents are necessary for extraction and mobile phases because lower grades can be a source of organic contamination, which raises noise levels and causes some analytes to perform outside QC specifications. For LC-MSMS analytes revealing low response, such as acequinocyl and spiromesifen, organic mobile phase interference is a significant problem. Even among LC/MS-grade brands, some produce more noise than others and so experimentation is necessary to select optimal products.
LC/MS is prone to contamination from the glassware used in analysis. ESI+/- ionization, in particular, relies on cation adducts, often hydrogen adducts but also ammonium, sodium and potassium. Sodium adducts can be formed from the trace amounts of sodium present in the glass of vials and other vessels holding mobile phase or sample. These ions form adducts which are less stable than hydrogen or ammonium adducts, and hence produce a lower MRM response than the other adducts, while simultaneously deprecating their responses. The suppression of sodium from glassware will boost signal and increase performance of analyte quantitation.
Image11
MRMs are an excellent starting point for analyte detection in the mass spectrometer. For many analytes, Q1 and Q3 scans are necessary to discover ideal MRM for analyte detection. Q1 and Q3 are the names of the two quadrupoles that can be programmed to selectively filter ions by mass, and their operation in tandem is what allows for MRM scanning to occur. It is important to measure S/N and not just peak height in the MS. It is also important to determine the best S/N ratio in each matrix, as they are often not the same due to matrix and “dark” interferences. To compare potential fragments at different masses, a chromatogram is preferred to just spectral output to see correlation with analyte retention time and S/N measurement.
Image12
A component of mass spectrometry that is universal to LC and GC is the detection of certain compounds using isotopic abundances of specific elements. “A+2” elements, such as chlorine, bromine, and sulfur, are useful for identifying parent mass scans of compounds containing such atoms. “A+2” elements have appreciable quantities of heavier isotopes with two extra neutrons, increasing the observed m/z by 2. “A+2” elements are commonly found in synthetically produced pesticides, making them particularly useful in pesticide analysis. “A+2” elements have appreciable quantities of heavier isotopes with two extra neutrons, increasing the observed m/z by 2. Some typical ratios for “A+2” elements heavier isotopes are: 24.22% for 37Cl; 49.31% for 81Br; 4.29% for 34S; and 0.21% for 18O.
Reference
[1] Ellis, G. P. "The maillard reaction." Advances in carbohydrate chemistry, 1959, Volume 14: 63-134. [journal impact factor = 3.429; cited by 423]
[2] Krupczyńska, Katarzyna, and Bogusław Buszewski. "Peer Reviewed: Characterizing HPLC Stationary Phases." (2004): 226-A. [cited by 58]
[3] Ferrer, C., et al. "Determination of pesticide residues in olives and olive oil by matrix solid-phase dispersion followed by gas chromatography/mass spectrometry and liquid chromatography/tandem mass spectrometry." Journal of Chromatography A, 2005, Volume 1069(2): 183-194. [journal impact factor = 4.169; cited by 267]