Thursday, January 21, 2016

A Sensory Analysis of Marijuana Volatiles: Not Such Great S***

The chemistry behind the scent of marijuana is a compelling topic for a number of reasons. First off, it’s a big technical challenge. Pot consists of an extraordinarily complex mixture of volatile compounds and sorting them out is a big job. (Paradoxically, the main psychoactive ingredient tetrahydrocannabinol is odorless.) Secondly, it is of forensic interest: law enforcement seeks to detect the scent while smugglers try to cover it up. Finally, legalization has spurred increased interest in the aromatic qualities of pot as a consumer product. Here in Colorado, it has been fascinating to observe the industry grapple with branding and struggle to formulate edibles that consumers find acceptable.

So I was psyched to find a new paper in PLoS ONE titled “Characterizing the smell of marijuana by odor impact of volatile compounds: an application of simultaneous chemical and sensory analysis.” It is written by a pair of researchers at Iowa State, who use the latest methods of sampling (solid phase micro-extraction) and chemical analysis (multidimensional GC-MS) to characterize the concentration of various volatile molecules in samples of fresh marijuana. They also bring a sensory element to their analysis: they compare chemical concentrations to odor threshold data available in the literature, and have someone sniff and evaluate the various molecules as they emerge from the gas chromatograph. So far, so good.

Based on their results, authors Somchai Rice and Jacek Koziel conclude that when it comes to characterizing the smell of marijuana, “more attention should be focused on highly odorous compounds typically present in low concentrations.” (They mention nonanal, decanol, o-cymene, and benzaldehyde.) This, in principle, is a reasonable and potentially useful conclusion. Unfortunately, it is limited by the paper’s shortcomings in exposition and experimental design.

Rice and Koziel claim to add 200 new molecules to the list of previously known marijuana volatiles. But take a closer look at how they phrase their claim:
Over 200 compounds are being added to the list of what is currently known to be emitted from illicitly packaged marijuana.
They refer to “illicitly packaged marijuana” because they based their analysis on three samples of pot (of unidentified strain), all obtained from the evidence room of the Iowa Division of Criminal Investigation. One of the samples was a gram of pot analyzed along with the plastic baggie it was zipped up in. Another consisted of approximately 50 kilos of pot stuffed into a “US military-style duffel bag.” Thus the chemicals analyzed by Rice and Koziel are not exclusively marijuana-based—they include molecules off-gassing from the packaging.
In this research, the authors are not differentiating between VOC emitted from marijuana samples and VOC emitted from packaging.
Indeed. So Rice and Koziel have now muddied the scientific literature on pot volatiles with who knows how many irrelevant chemicals. It will be up to future researchers to sort through the mess they have created.

What about Rice and Koziel’s conclusion that certain chemicals found in low concentration may have a major impact on the smell of pot? It’s a reasonable idea. Perfume chemists have long known that the most common molecule in a mixture is not necessarily the smelliest. A highly potent odor molecule can impact the overall scent if present even in trace quantities. Despite Rice and Koziel repeatedly patting themselves on the back for being the first to apply the concept to marijuana, one might say that it is the first rule of chemosensory analysis.

The authors use the concept of Odor Activity Value (OAV) to make their point. OAV is an attempt to relate a chemical’s concentration to its sensory impact. One calculates OAV by dividing the chemical’s concentration in the sample by its olfactory threshold concentration (i.e., the lowest limit of detectability to the human nose). If a chemical is present at less than its threshold concentration (OAV < 1.00) it is unlikely to contribute to a mixture’s smell. Rice and Koziel use OAVs to identify potentially important odor components of pot. This is fine as a first pass through the data, but when the authors use OAVs in statistical analyses, they stretch the concept almost beyond its limits.

Why? Because every smell molecule has its own concentration-intensity curve. For every step increase in concentration of molecule A, for example, its odor intensity might increase dramatically. For molecule B, in contrast, it might take many step increases in concentration before its odor intensity is noticeably stronger. Therefore, samples of A and B, set at the same OAV, could have very different odor intensities. This makes OAVs useful as a first pass to identify the smelly ingredients in a mixture, but it is perilous to use OAVs as a measure of comparative odor impact.

Finally, this paper is poorly written and would have benefited from closer editorial attention (yes, I’m talking to you, John Glendinning). The introduction wanders all over the map. It begins by mentioning odor as probable cause for search and seizure, but the ridiculous S1-Table is a random grab-bag of U.S. legal cases which completely omits the fact that at least nineteen states have search and seizure rules for marijuana based on the “in plain smell” doctrine. The intro then discusses analytical techniques, veers into a consideration of drug dogs and scent training, and then into the subject of human olfactory abilities. Like wow, man, everything is connected to everything else, you know?

The chemical and sensory analysis of marijuana scent is an increasingly important topic, and while Rice and Koziel have made a preliminary effort I expect that much better work will be done in the near future.

The study discussed here is “Characterizing the smell of marijuana by odor impact of volatile compounds: an application of simultaneous chemical and sensory analysis,” by Somchai Rice & Jacek A. Koziel, published in PLoS ONE 10(12):e0144160, 2015.

1 comment:

Peter Apps said...

A very sound critique. In addition to the shortcomings that you list, and several others, the main problem is that they have based their "identifications" on mass spectral matches, with the use of an unspecified number of un-named retention standards. So by any reasonable standards they have not actually identified very much. As an example; in Table S2 "4-methyldecane" elutes before "2-methylpentane".

So much for peer review.