When Things Go Wrong – Lab Testing: Could Your Results be Wrong?

We hope you enjoyed the May episode of When Things Go Wrong, Lab Testing: Could Your Results be Wrong? If you missed the webinar, you are in luck because you can view it here and read the summary of our discussion below.

As an industry, we engage in a decent amount of finger pointing and blaming when it comes to popular lab testing issues such as lab shopping and THC inflation. Several arguments discredit the labs, the very group that can reveal the problems with data. Some believe lab shopping exists solely because labs are incentivizing the use of their services by reporting higher THC results.  The panelists encouraged us to look at this issue from a new perspective by first questioning the way we are performing flower potency testing. 

Generally, cannabinoid concentration testing is performed in flower by analyzing milligrams of a single replicate and the resulting values are used to label a batch of 10 – 100 pounds or more. This methodology is truly not fit for the purpose of labeling. Performing the test in this way does not tell you anything about the inherent variance within a batch after harvest. Additionally, as cannabis is processed and mechanically handled (e.g., drying, trimming, tumbling), trichomes are lost and the variance between buds increases, which is a problem if flower is tested prior to processing and packaging. 

Another topic we discussed in this month’s webinar episode is sampling. It is important to remember that test results most accurately represent the sample received. These test results may be used to represent the larger batch in which the sample originated if a statistically relevant sampling model is used and effective compositing and homogenization of samples is performed prior to sub-sampling for analysis.

To elucidate variances in THC content between buds from the same harvest and explain the need for representative sampling, Jini from Modern Canna Labs told us about an experiment they did with two 3.5g containers of flower buds in which they analyzed about 10 buds from each container. The THC concentration of each bud ranged from 17 to 23%. If you are not sampling properly, the results will be skewed. But more so, cannabis flower is one of the most heterogenous sample types received by labs. No two buds are the same, even pulled from the same area of the plant

The panelists also discussed a common scenario of sending two samples to different labs and getting different results. Typically, these samples are being sent as buds, which we explained are not consistent. It was noted that in order to prepare flower samples for proficiency testing the plant material is thoroughly grinded, homogenized, filtered, and sieved. Rarely is flower tested for compliance or sold to consumers in this form, as it is typically sold as heterogeneous buds. However, this is the only way to have homogenous starting material for flower proficiency tests. Knowing all this, how could we expect two samples of buds from the same harvest to yield consistent results?

We also discussed the issue of THC content being incorrectly correlated to quality (i.e., higher THC % is worth more). Imagine if we did this with wines, assuming the highest alcohol content wines were the highest quality and we should pay more for them. When THC is one of the only standardized label requirements across jurisdictions, that becomes the only thing customers correlate with quality.

Other topics to consider as part of this multi-faceted lab testing issue are Certified Reference Material (CRM) shopping, measurement uncertainty and shelf stability. We discussed there are some CRMs that can skew cannabinoid concentrations and some labs may select CRMs that bias high. Measurement uncertainty, the degree of uncertainty in an analytical result commonly denoted as a plus or minus value, is in every step of the testing process. When you weigh the sample, add solvent, prepare dilutions, use different instruments, run calibrations, integrate peaks, etc., you are introducing some level of error at each of these steps, creating compounded error. Further, shelf stability testing is essential because test results will change over time as a product sits on the shelf.

What can be done to address these testing issues? The panelists noted that standards now exist and we need to use them. ASTM International, AOAC International, and ISO all have cannabis specific consensus standards available to the industry today. Also, we have more data now that we can use to our advantage to improve regulations. For example, using all the data that shows high bud to bud variance within the same harvest, some countries have a 20% permitted variance for THC levels on labels for flower. This degree of variance may not be appropriate for homogenous products, such as distillate or plain chocolate, but it seems to be more appropriate for flower than 10%. Lastly, we need to work towards fit for purpose potency testing.

Questions from the Audience

1. What immediate actions can ethical testing labs and regulators take to stop potency inflation by unethical testing labs?

Require the use of standards, change the way we do potency testing so it is fit for purpose, and as an industry, stop using THC content as the only way to determine value and quality of cannabis. Further, as Kelly pointed out – you can and should report questionable activity. For example – A2LA’s complaint form is easy to use and can be anonymous. Link here

2.What would you recommend as a core set of regulations for states/provinces to prevent potency inflation and enforce accurate reporting of potency?

The Canadian regulations are not perfect, but they are a common model regulatory framework. Also, using the U.S. Pharmacopeia and European Pharmacopeia monographs would be a good start.

3. As a former state regulator/current ISO assessor, it is clear that ISO accreditation alone is not enough. What could replace this regimen so that ISO can return to being a voluntary accreditation instead of giving regulators and customers a false sense of security?

ISO accreditation is equivalent to having a driver’s license. Just because you have a driver’s license does not mean you do not engage in reckless driving; however, it is required as a baseline. We believe ISO accreditation is a good baseline requirement to standardize some of the key operational elements of a laboratory, but you are right, it is not enough. We recommend additional certification programs, such as ASTM’s CANNQ/HEMPQ program, that dive deeper than an ISO audit and is based on the consensus standard D8244 – Guide for Analytical Laboratory Operations supporting the cannabis/hemp industry. Also, we recommend the required use of international consensus standards for sampling, preparing samples for analysis, test methods, and more.

4. Are there valid reference standards (i.e., CRMs) for cannabinoids, especially THC and CBD, that are approved/accepted by the FDA or other regulatory bodies such as Health Canada and/or EMA?

The FDA does not approve CRMs. However, the FDA expects companies to have validated processes for the selection, purchasing, testing, storage, and use of CRMs. Credible CRM providers include governmental agencies such as the U.S. National Institute of Standards and Technology (NIST), the National Institute of Biological Standards and Control (NIBSC) in the U.K., and the National Research Council (NRC) of Canada. There are several private companies that provide CRMs as well. We recommend choosing a CRM provider that is ISO 17034 accredited. Also, we recommend having primary and secondary CRM sources that challenge each other.

5. Have any of you done CRM studies? Looking at variances in CRMs between lots or between ISO CRM manufacturers?

ISO accredited labs should have Quality Control Charts to monitor variances of the CRMs they use across lots. Studies between CRMs from different providers have been performed. We would recommend using governmental agency CRM providers such as NIST and the NRC for primary CRM sources. 

6. Could we use the Orange Photonics Lab3 instrument as a standard for cannabinoid testing, doesn’t it have less human error?

We prefer to avoid endorsing specific commercial products. The issue is that variance is summative of all process steps, not just the instrument being used. Sample preparation, weighing, pipetting, etc. all come with inherent variances, i.e., measurement uncertainty. Also, spectroscopic detection is not specific enough for accurate cannabinoid concentration testing in our opinion.

7. What relative percent difference between results from two different labs would you recommend for regulators performing off-the-shelf/secret shopper investigations of potency inflation samples? 

We would recommend 20% permitted variance for flower, which is what is used in Australia, given the heterogeneity of cannabis buds.

8. Can we do stability and shelf-life testing on products without active ingredients (e.g., cannabinoids) infused to get a more accurate picture of our product life? Or does the addition of these active ingredients cause significant variance in those numbers?

You certainly can, however, it is best practice to perform stability studies on the actual products you will be selling and labeling with a shelf life. We encourage you to look at ASTM D8309-21 which is based on global pharmaceutical and food best practices for guidance on stability testing of cannabis products.

9. In our state we only have two labs applying for testing so we expect a huge backlog for final testing. How do we refine our processes for QA when we may have a pitfall on final testing to validate our processes?

That is a frustrating reality and part of the growing pains we have seen in new marketplaces. At a high level, this can be thought of as a supply chain risk that should be proactively mitigated. Our recommendation is first to ensure you have a robust risk management program in your production facility – quality management systems and adherence to Good Manufacturing Practices. There are many cost effective internal tests (e.g. microbial, environmental swabbing, etc.) that you can collect during production and drying to mitigate the risk of a ‘surprise’ failure. Beyond that, working with your regulators to voice your concerns and offering solutions for them to consider (that are within the statutory frameworks they have to work within) is a great way to build a healthy relationship with your regulators who likely share your concerns but may be struggling to come up with viable solutions as well.

10. Major differences between third-party/ISO accredited testing labs vs. internal testing labs for manufactures. Why advise startup companies/labs to invest in method matching to other ‘big name’ laboratories while there is still a lack of regulated standards?

Fortunately there are a handful of approved test methods – both from the US Pharmacopeia, ASTM International and AOAC International. Internal labs are also able to participate in Proficient Testing programs which allow you to have confidence in the results of your internal labs testing protocols – and their accuracy and precision. Having your own internal data can be incredibly valuable for process checks that can alert you to potential failures and surprises before it is too late (i.e. final product testing by the third party).