Use the “Subscribe” link at right. It just takes a few seconds!
After a long blog hiatus, I’d like to address one of the most frequently-asked questions we get from cannabis labs: “how do we quantitatively extract cannabinoids from edible matrices?”. With the huge number of matrices out there, limited peer-reviewed data, and lack of sufficiently concentrated reference standards for spiking, it’s no wonder that after several years, this is still a question in the cannabis analytical industry.
In February, my colleague, Julie Kowalski and I were lucky enough to spend two weeks at Trace Analytics developing methods for cannabis analysis. One of my focus areas during that time was to evaluate extraction efficiency for cannabinoids in edible matrices using a common sample preparation technique – QuEChERS. The QuEChERS approach is one of the simplest and most cost-effective sample preparation techniques available. It is used heavily in food safety labs for extraction of pesticides in food matrices, and has been applied across many other industries and matrices. In a nutshell, QuEChERS is simply a liquid-liquid extraction (LLE). The special thing about QuEChERS is that unlike most LLEs, the organic extraction solvent can be miscible with water because it is salted out during the QuEChERS extraction process. This improves extraction efficiency for water-soluble analytes and generally produces a cleaner extract than solvent extraction alone. Figure 1 shows the procedure we used for extraction of our edible matrices – in this case a caramel cookie:
Figure 1: Extraction Procedure for Cannabis Edibles
At the end of this process, we end up with four layers: an extract in acetonitrile that also contains some matrix components like fat, sugar, and pigment, insoluble matrix components like fiber and protein, water containing the bulk of the sugar and certain pigments, and excess salt from the extraction. Figure 2 points out the different parts of our QuEChERS extract:
Figure 2: QuEChERS Extract of Cannabis-Infused Caramel Cookie
After extraction, a cleanup step is commonly performed to further remove matrix interferences, but based on data that I don’t have room to show here, the common cleanup procedures result in loss of cannabinoid acids and CBN. You can find a detailed pdf version of the full procedure here (I included a plant extraction method in the document as well).
Now we have an extraction procedure, but how do we know if it works? Ideally, we’d like our extraction procedure to extract 100% of our analytes of interest. This is especially important for cannabis-infused edibles because if our extraction method only recovers a portion of cannabinoids, then we’ll have to determine what portion is recovered for each and every matrix that we test in order to ensure the validity of our results. Generally, when developing an extraction method, labs can measure recovery by spiking applicable amounts of a reference standard into a blank matrix and measuring how much they get back. Unfortunately, we don’t have that luxury for cannabis potency analysis. We can definitely get blank edible matrix, but due to the low concentration of the DEA exempt cannabinoid standards we must use, we are unable to spike the matrix at a representative level. For example, in order to spike 1g of edible matrix at 1mg THC/g, we would need to add 1mL of the most concentrated DEA-exempt THC standard available to that gram of brownie. By adding 1mL of our standard in organic solvent to 1g of brownie, we’ve effectively diluted our matrix by 50%. If we try to extract this, any recovery values we get won’t be representative of extraction from a ‘real’ sample. If we try to spike with additional cannabinoids, the situation only gets worse.
So how do we measure recovery for our extraction procedure without performing any spiking? Luckily, this is a common issue in the botanicals industry, and AOAC has guidance addressing this issue. The guidance suggests re-extraction of the same sample with fresh solvent. In order to do this, perform the following steps:
- After extraction, remove all of the solvent extract from the sample (i.e. all solvent for solvent extracts, and entire top layer for QuEChERS extracts).
- Add fresh solvent to the extracted sample as dictated by your method (if very few salts remain at the bottom of the tube for QuEChERS extractions, add ½ package of fresh salts – it will be messy).
- Extract sample according to your method.
If we removed all of the cannabinoids from our matrix during the first extraction step, we should see little to no cannabinoids in our re-extraction sample after performing the second extraction as detailed above. That’s exactly what we did for our work. We performed a re-extraction experiment on a variety of cannabis edible matrices covering sugar only, sugar + fat, and fat only samples. Figure 3 shows the results from our re-extraction experiment.
Figure 3: Re-extraction Experiment Results for Edible Matrices (mg/g)
Everything looked great, except for the butter matrix. It turned out that the true cannabinoid concentration in the butter was ~10mg/g, whereas in the rest of the matrices, the concentration was ~1mg/g. My hypothesis is that if we reduce the matrix amount for the butter to contain ~1mg total cannabinoids (i.e. use only 100mg of butter), the recovery will improve since the matrix:extraction solvent ratio will be much greater.
Stay tuned for my future blogs which will cover cleanups (or lack thereof) for edible matrices, derivatization for GC potency analysis, and multiple topics on residual solvents.
 AOAC International. AOAC Official Methods of Analysis: Appendix K. 2013. p.5
OK Jack, enough of the acronyms; this isn’t a word puzzle…
Recently I let ChromaBLOGraphy readers know of our work on using gas chromatography (GC) with atmospheric pressure ionization (API) – mass spectrometry (MS) for analysis of brominated diphenyl ethers (BDEs) employing nitrogen (N2) carrier gas, which was presented at BFR2016 (BFR standing for Brominated Flame Retardants, although the meeting has been expanded to include other flame retardants). The gist of it was that with a selective GC column like our Rtx-1614 you can use nitrogen carrier gas to get approximately the same separation in the same time as when using helium carrier, even though nitrogen is less efficient under helium flow rate conditions. The twist on this application is that the detector was a mass spectrometer. If you’re thinking that nitrogen carrier gas results in extremely low detectability when using MS, you’re right, but that’s under typical electron ionization (EI) conditions; we used API instead, where nitrogen is not only tolerated, it’s used in the ionization process. No sensitivity issues! In fact, we were able to get down to the sub pg level for BDEs. The “we” includes all my great co-authors, especially Karl Jobst at Ontario Ministry of Environment and Climate Change (MOECC) who did most of the work, Eric Reiner and Terry Kolic at MOECC (longtime friends and colleagues), and Jaap de Zeeuw of Restek who inspired our recent nitrogen carrier gas work. I also have to give a shout out to Cindy Ross, another colleague at Restek, who suggested years ago that nitrogen carrier gas work might be viable.
Here are some chromatograms for your inspection. More later…
OK, maybe it’s too much to hope that split ratio is exciting, but I think the information I’m laying down, all based on data I’ve recently collected in the lab, can be helpful to anyone working with GC. So please check out my recent article in The Column. Let me know if you have any questions, and many thanks for reading.
The recent addition of Radio Frequency Identification tags (RFID tags) to detector lamps allows automatic reference and usage data collection to your system software. This is convenient but comes at a cost: up to 40% more for an OEM-tagged lamp compared to a Restek replacement lamp. At the time of the writing of this blog post, Restek does not offer RFID-tagged lamps. But if you don’t need the bookkeeping feature, simply disable the tag requirement in your system software and save some money.
To allow the use of a non-tagged lamp in Lab Advisor, follow these instructions:
At your Lab Advisor home screen, choose “System Information > Instrument Control” from the left navigation column. A module list appears in the main window. Click on the detector module. Expand the “Controls” menu. Click on the “Special Commands” triangle at the bottom. Turn the “Lamp tag required” option off (circled option). A non-tagged lamp will not attempt to light if this option is ON. Be sure to click the “Save Session Results” button at the bottom of the screen. That’s it!
The software can still count your hours. Simply reset the counter in the “EMFs” menu option.
Except for the tag, Restek lamps are manufactured to meet OEM specifications and have stood the test of time in field use. We have been offering lamps for more than 15 years. In the rare case that you experience an issue with a Restek lamp, return it for a replacement. 100% satisfaction guarantee is Pure Satisfaction.
Pesticides in cannabis has been a hot topic lately and we have been getting many requests for help with this analysis. We did work on pesticide residue testing in cannabis about 6 years ago. At that time, we were limited to testing our methods with a small amount of seized material. Recently, we have been able to work with great collaborators, Shimadzu Scientific Instruments and Trace Analytics, to do more comprehensive method development and partial validation. Trace Analytics is a testing laboratory servicing the medical and recreation cannabis inustry in Spokane, Washington since 2015. With the help of our collaborators and availability of bulk samples, we were able to revisit method development and perform a partial validation. This would not have been possible without the hard work of all of the team members; Jeff Dahl, Caitlin Johnson, Derek Laine, Sara Minier, Amanda Rigdon, Gordon Fagras, and Jack Cochran.
We developed a modified quick, easy, cheap, effective, rugged and safe (QuEChERS) sample preparation method (stay tuned for a more discussion in a future blog) paired with LC-MS/MS analysis using a Shimadzu LCMS-8050 with Prominence HPLC. QuEChERS is designed to be generic and work for a wide variety of pesticides with diverse chemical properties. This is one of the reasons why QuEChERS is so popular for food safety testing. This approach takes advantage of the selectivity and sensitivity of LC-MS/MS allowing us to use a “good enough” sample preparation. QuEChERS is much less intensive than what would be needed if a non-MS/MS based method was used…another reason why it is popular.
We used a 1 µL injection to help with early eluting pesticides peak shapes since the extraction solvent is acidified acetonitrile and the initial mobile phase is mainly water. This also helps maintain column and instrument cleanliness allowing many injections to be made before maintenance is needed. At least two MS/MS transitions were monitored for each analyte.
Take a look at the method details of our sample prep protocol and analysis method.
We tested three different cannabis flower matrices. They included orange kush flower, permafrost flower and then a composite. The composite sample was simply a combination of flower and sugar leaves from several strains. We did this so we had enough “pesticide-free” sample to perform all of the spiking experiments. Each matrix was spiked at three levels (50, 200, 1000 µg/kg “dry” weight) with the Restek Oregon pesticide standard (more information in “sample prep protocol” page 2). These levels were chosen based on current and proposed regulatory limits from various states. Each spike level was performed in triplicate. The spiking scheme is shown in the table below. Triplicate data allows us to determine average recovery and relative standard deviation (RSD) for each level in each matrix as well as across the three matrices. Matrix-matched calibration was used for quantitation and both method and instrument internal standards were used.
The data shown here is for “dry” flower material and so we used a reduced sample amount (1.5 g) and a hydration step. Adding water is critical for the extraction chemistry to work properly. The AOAC QuEChERS extraction salts were used for the salting out step of the extraction. After the extraction step, we chose to use the “universal” formulation of the dispersive solid phase extraction (dSPE) to remove unwanted coextracted compounds. This formula contains 50 mg/mL extract each of primary secondary amine (PSA) and octadecyl (C18) sorbents and a moderate level of graphitized carbon black (GCB) (7.5 mg GCB per mL of extract). GCB removes chlorophyll which can cause instruments to become dirty if injected.
Take a look at recovery values and RSDs table.
Highlights of the data:
- For the large majority of pesticides, in all three matrices AND at all three spike levels, recovery was between 70-120% which is the desirable range for the food safety industry. It is expected that difficult pesticides or detection at very low levels may produce recovery below this range. In most of these cases, RSD values were less than 20% for a single matrix (n=9) and across all three matrices (n=27). This data indicates that this method is appropriate for multi-residue pesticide analysis in cannabis flower.
- There are some of the Oregon pesticides that are more amenable to GC analysis and we would suggest using GC-MS/MS testing of the extracts to cover bifenthrin, cyfluthrin, dichlorvos, MGK-264, and permethrin. We were able to determine recovery at the highest spike level (1000 ppb) with LC-MS/MS for bifenthrin, dichlorvos, MGK-264 and they fell within the 70-120% recovery and less than 20% RSD so this indicates that the sample preparation method is suitable.
- Abamectin, widely known by the trade name Avid, is a popular insecticide used to treat spider mites which are known to attack cannabis plants. Abamectin is heat sensitive and so the hot LC-MS/MS interface used for other pesticides caused the abamectin signal to be low. Abamectin can be tested with different LC-MS/MS interface parameters to obtain maximum signal and detect it a low levels. However, even using the multi-residue method, we were able to confirm recovery of 85% (7% RSD, n= 9) for the 1000 ppb level.
- Spiromesifen is an insensitive compound and detectability could be improved by increasing the injection volume for LC-MS/MS or by using GC-MS/MS analysis.
- Spinosad and spiroxamine show slightly lower recovery values ranging from 60% and higher but these recovery values are generally consistent with less than 20% RSD across all spike levels and matrices (only one exception). These compounds are slightly basic and recovery may be lower than other pesticides because acidic extraction conditions are used during sample preparation.
- Overall, this method is a great approach for multi-residue pesticide testing in cannabis as demonstrated by the acceptable recovery of nearly all of the almost 60 pesticides we tested.
Stay tuned for more information and details about pesticide analysis in cannabis.
Just wanted to let everyone know we will be hosting a #AskRestek Twitter Q&A session about QuEChERS next week at 3:00 p.m. EST (12:00 noon PST) on Tuesday, May 3.
QuEChERS is a great approach for sample preparation with lots of benefits. It is also very flexible which is nice but that can also make things a bit confusing. My colleagues Mike Chang and Jonathan “Munch” Keim… yes, the famous “Door Prize Guy” from NACRW will be the panelists. Mike Chang is our Sample Preparation Product Marketing Manager and Munch is the Education Program Manager. I will be on hand to help out with technical questions since i have used QuEChERS extensively.
Check out more details and instructions for attending the Q&A session here.
Well, it might be. In the early days of HPLC, THF was commonly used as a mobile phase solvent. It has eluting strength similar to acetonitrile, but just slightly stronger. Since it technically is an ether and is very miscible with water, it is sometimes useful with reverse phase HPLC. It also provides additional options for ternary mixes when methanol/water or acetonitrile/water mobile phases are not able to produce a fine tuned separation.
Like many organic solvents, there are some possible health hazards associated, which you can read about here. THF has a very low boiling point (66C) and emits fairly noxious fumes at room temperature, making it quite unpleasant to work with. You definitely need to use this in a hood.
A more concrete to reason to limit usage of THF for HPLC mobile phase is that it does have a tendency to swell PEEK (polyetheretherketone) material and may contribute to degradation over time. A side effect of the swelling could be increased system pressure, which may become an issue. The use of PEEK tubing and fittings has increased dramatically over the years, due to its ease of usage compared to stainless steel parts. However, THF should not be used with PEEK tubing, unless it is present only at low levels. I have read varying opinions on how much THF one should try using. Personally, I would prefer to stay less than 10% if using PEEK tubing. There is also the possibility that THF at higher levels can degrade other plastic-like materials, for example, pump seals. When using THF, it always best to check with the instrument manufacturer to ensure the proper seals are being used. Some pump seals are designed only for use with aqueous solutions and weaker organic solvents (these would be ones designated as “aqueous” or “reverse phase”).
As far as detection methods go, THF is OK to use for UV detection methods. Since its UV cutoff is around 212 nm, it usually does not produce interference. However, it is important to use HPLC-grade THF to avoid interference from stabilizers that often are used with other solvent grades. It is also important to make sure the THF is fresh, as the formation of peroxides over time will increase the UV background. Usage of THF for PDA and fluorescence detection is fairly similar and the same precautions exist. Some of the same concerns about purity, stabilizers and peroxides apply to most detection methods, including refractive index (RI) detectors.
Using THF with mass spec detectors presents some unique concerns. Agilent and Waters both suggest that its use for LC/MS should be very limited and special precautions should be taken:
ThermoScientific mentions similar precautions for Charged Aerosol Detectors below:
I hope you have found this information useful. Thank you for reading.
Nitrous oxide (N2O), is commonly known as “laughing” gas, but is also used as a component in fuels in rockets and as an aerosol propellant. N2O is itself a stable gas and can be analyzed relative easy via gas chromatography. Often it is confused with “nitric oxide”, (NO). NO is a very reactive gas. When oxygen is preset, it will immediate oxidize into NO2. NO2 can be easily recognized as it has a dark brown color. Also NO2 shows reactivity, meaning that the analysis of NO and NO2 via gas chromatography is not commonly done, see for details: http://blog.restek.com/?p=4583
Recently a summary of N2O analysis was published by Separation science. Separations are shown on different adsorbents like, Porous polymer, Alumina, Molsieve 5A and ShinCarbon materials.
Full Article can be found here: http://www.sepscience.com/Information/Archive/All-Articles/4365-/Analysis-of-Gases-via-Gas-Chromatography-Part-1-Nitrous-Oxide
Especially the alumina PLOT is interesting (Fig.1) as often CO2 is present and can interfere with the N2O measurement. CO2 is adsorbed completely by Alumina, resulting is a single N2O peak. CO2 can be removed periodically by conditioning at 200C.
I mentioned in a recent blog post that to maximize peak capacity in GCxGC the first dimension separation needs to be preserved by having a very short second dimension separation (short modulation time), often on the order of 2 sec or less, even. While maximizing peak capacity can be very important when trying to characterize a complex sample (e.g., for metabolomics or discovering emerging contaminants in the environment), maintaining the first dimension separation through short modulation times is even more critical when isomers are to be determined individually, including for mass spectrometry. That’s because isomers that coelute in the first dimension for GCxGC are very rarely separated in the second dimension. The second dimension column, no matter the alternate selectivity, is just too short.
The problem outlined above, first dimension coelution caused by a longer modulation time in GCxGC, and its solution, a faster modulation time to preserve the first dimension separation, are illustrated in the figure below. See how the tetrachlorobenzenes coelute when the modulation time is 2 sec? The peaks eluting from the first dimension column are “piling up” at the modulator, which might be OK if they were separated in the second dimension, but they are not. Modulate faster = preservation of the first dimension separation, in GCxGC. The general rule of thumb is modulate (slice) the first dimension peak at least 3 times.
Restek at BFR2016 in Toronto – APGC of Brominated Flame Retardants Using Helium and Nitrogen Carrier Gases
In only a few short weeks, I will be giving a presentation on the analysis of brominated flame retardants at BFR2016 in Toronto. My colleagues and I used an atmospheric pressure ionization mass spectrometer with gas chromatography on an Rtx-1614 (15m x 0.25mm x 0.10µm) column to look at polybrominated diphenyl ethers (PBDEs) in various samples. First though, we explored optimizing the chromatography for speed, while using efficient helium carrier gas. But we also looked at employing nitrogen carrier gas, since the APGC instrument we used can easily handle nitrogen carrier with no loss in sensitivity like would occur with the typical vacuum-pumped electron ionization MS system. By employing a selective GC column like the Rtx-1614, you can get the same analysis times (and PBDE retention times!) for helium and nitrogen carrier gases and still meet necessary separation criteria (e.g., separation of Br4 PBDE congeners 49 and 71 as per EPA Method 1614). Both helium and nitrogen approaches were facilitated by the EZGC Method Translator and Flow Calculator.