Too High? Too Low? The Stats Say So

What differentiates light from heavy cannabis use?
 

The definition of light versus heavy use is subjective to the research question being asked. Our study assumed that the pharmacological effect was driven by cannabinoid receptor activation/occupancy. We designated users who have steady-state THC concentrations that maintain pharmacological activity as “heavy users.” This is a scenario most similar to the chronic dosing of medications. We designated people who use regularly but are not chronically dosing to achieve steady-state effects as “light users.”

What makes 11-COOH-THC so useful for measuring cannabis use?
 

Although THC concentrations are historically measured to determine use, it is not a very reliable biomarker because concentrations readily decline shortly after intake (and there is often a lack of information in a clinical setting regarding the time since last use). 11-COOH-THC is significantly more robust when an objective measurement is needed. It has a terminal half-life of ~18 hours, allowing a wider window for detection and quantification. In addition, because of its long half-life, concentrations are less susceptible to dramatic changes. 11-COOH-THC also accumulates in regular users, so a higher 11-COOH-THC concentration could relate to higher dose and frequency. We chose our concentration cutoff value by using a parent-metabolite pharmacokinetic model and Monte Carlo simulations to model THC and 11-COOH-THC concentrations for different usage patterns. Using the resultant concentration distribution profiles for a mixed user group with daily and 3x daily use, we generated precision recall curves to determine a cutoff value that would distinguish heavy from light users with 80 percent precision.

What are the limitations to this approach?
 

There are a few. First, the approach requires an understanding of the clinical population being studied and the general use patterns of the individuals. In an ideal scenario, a researcher would want to use population-specific use patterns. We found that, in a scenario where 70 percent of users in the simulated population were light users and 30 percent were heavy users, a higher cutoff value was needed to achieve the 80 percent precision rate. Likewise, a lower cutoff was adequate if the population consisted predominantly of heavy users (“true responder” group). Second, although our modeling approach is quite robust in that it simulates the systemically available dose, accounting for percent THC and bioavailability, the current method would require additional validation for different dosing routes and formulations. However, these are all components that can be refined in the model and validated in the future (pending available data). Third, a clear definition of the biological effect size would be ideal in designing studies that are adequately powered. This may not be known a priori, especially if the biological mechanism of the effect has not been discerned.  Can you anticipate any reluctance on the part of cannabis researchers to adopt this methodology?
We believe that accurate determination of the pharmacological effects of cannabis requires an understanding of the dose-response relationship. However, application of our objective classification method requires upfront work to understand population use patterns, effect size of the biological effect being measured, and a basic understanding of PK/PD principles. This may lead to some initial reluctance to embrace this methodology until population use is better understood in clinical populations of interest. However, we demonstrate in the paper that the benefits of recruiting fewer subjects and yielding higher power will inevitably make cannabis research less resource-intensive.

Given the subjective nature of self-reporting, should we expect an element of inaccuracy in studies of this kind?
 

Yes, there will likely be some inherent inaccuracies with self-reporting. In our study cohort, research staff left the room and were blinded to the response. Despite efforts to minimize the perception of any stigma by the research team, a significant proportion of participants still misrepresented their cannabis use. We found that there were nearly equivalent rates of false positives (self-reported non-users with detectable cannabinoids; 17 percent) and false negatives (self-reported daily users without cannabinoids; 14 percent). In the latter case, participants may overestimate their use frequency and report daily use even when they are using less frequently.