Caffeine, Melatonin and Other Contaminants Found in US CBD Products

A new study of 21 American CBD products, currently in preprint, has found evidence that many products are mislabeled, contaminated, and deliberately adulterated.

The study, which was shared with Analytical Cannabis ahead of peer review, adds to the growing body of evidence that many CBD products are misrepresented by bad actors in the industry.

Around half of the products studied were affected in some way

Before the CBD products were even analyzed, two were already deemed to be misleading; both CBD items displayed the words “FDA Approved” on their packaging, yet non-prescription CBD products have never been endorsed by the FDA.

Following the packaging inspection, all 21 products were studied further using ultra-high pressure liquid chromatography (UHPLC) coupled high resolution accurate mass spectrometry (HRAM).

Although determining the absolute quantities of cannabinoids present in each product was not within the initial scope of work, study author Dr Ben Orsburn notes in his preprint paper that most of the products did not show any quantifiable levels of THC in their resultant spectra. However, four products did contain THC above the limit of detection. From examining the area under each peak, it is believed that some of these products may be over the federally mandated maximum THC allowance in such products.

In total, 11 of the 21 commercially available products appeared to be affected by some sort of misrepresentation, contamination, or possible adulteration.

The most common contaminant detected, erucamide, is a surfactant commonly used in manufacturing, which is already a well-characterized leachable contaminant in pharmaceutical products.

Other substances detected are more concerning. Polyethylene glycols N14 and N15 were identified “with high confidence” in two samples. The presence and safety of such polyethylene glycols (PEGs) has been questioned before in relation to cannabis vape products, where the PEGs are liable to degrade into more harmful compounds when vaporized.

The over-the-counter sleep supplement melatonin was found in two samples, and the stimulant caffeine was found in one. Valpromide, an antiepileptic and antipsychotic drug marketed in some European countries, was also detected.

“[The data] is so convincing; it’s 100 percent sequence coverage for this prescription medication – and I don’t know why that makes sense for it to be there,” study author Dr Ben Orsburn told Analytical Cannabis.

“I mean, if you’re marketing a CBD product for relaxation and helping you sleep and you put a bunch of melatonin in it, then that makes sense for why that’s there. So, it seems a little more nefarious that here is the potential that somebody is just doing a little chemistry experiment and not really thinking about it too much,” said Orsburn.

Possible fentanyl analog found in one product

Most concerningly, the non-targeted screening appears to have detected some amounts of fluorofentanyl, a synthetic opioid and Schedule I compound, in one mail order product obtained for the study.

“When I found something that looked like fentanyl, I was like, ‘OK, stop,’” Orsburn recalled.

“You need to have special licenses to even get the positive controls to test [for fentanyl],” he explained. “We looked into that, and we actually tried to obtain more of those products so that we could send it onto somebody who did, but we have not been successful in re-obtaining this mail order product.”

“So, I can’t say that it’s absolutely true, but my gosh, it looks really convincing.”

Orsburn also told Analytical Cannabis that some of the mismatched fluorofentanyl ions might indicate the presence of a second fluorofentanyl isomer, 4-fluorofentanyl.

“What I don’t really go into in the paper – because I just thought it was kind of out of scope and really hard to visualize – is that I’ve got a lot of mismatched fragment ions that don’t make a lot of sense there in terms of the proposed 3-fluorofentanyl. But a lot of those fragment ions that show up in figure four [in the preprint paper] actually also match 4-fluorofentanyl. So, if it was this [fluorofentanyl], and it was a mix of these two isotopes, then the score of this molecule actually goes up dramatically.”

“Obviously, that’s really scary,” he stressed.“It’s just crazy; it doesn’t belong there.”

What is non-targeted screening?

The analysis technique used in this screening study, non-targeted screening, is not seen so commonly in the cannabis industry, but it is a mainstay in the pharmaceuticals sector and in toxicology analysis. Its appeal as an analytical tool hinges on its ability to identify hundreds or even thousands of compounds in a single sample.

“[With] targeted mass spectrometry when you’re looking for a pesticide, the instrument is only looking for the mass of that pesticide and two known fragment ions that would come from that pesticide,” Orsburn explained.

“In non-targeted screening, you actually just open up the whole mass range and you look at everything. Anything that ionizes and is visible, you create a peak for. You compare the peaks of everything you see, and you look at the things that are quantitatively very different in, for example, one product versus all the other products. And then when you look at that one ion that is clearly different between everything else, then you can focus in on the question of ‘what is that ion?’”

In Orsburn’s study, the resulting peaks from each sample run were searched against six different spectral libraries. No database is fully comprehensive, so this served to increase the likelihood of finding a match, but also as a way to draw increased confidence in a result if the same match was found across different complementary libraries. These matches were verified once more by manually comparing the experimental data and the relevant library spectra.

Where does the industry go from here?

Such findings make for concerning reading. But for the good actors in the industry, there may not be so much to worry about. Following good manufacturing practice (GMP) and all local regulations should likely be enough to avoid problems like this, Orsburn believes.

“Accidents happen, and sometimes you don’t control your whole supply chain – maybe you’re a producer and you’re purchasing a lot of material from other farms – there I do think that non-targeted screening would be really valuable,” he said.

“I think that, on its own, just following good manufacturing practices and not cutting corners, I don’t think that we’re going to see problems like this.”

Orsburn also told Analytical Cannabis that he is currently working on a way to make all of his data publicly available for other interested scientists to view.

“I’m a big open data advocate; so one of the things that we did make clear in this preprint is that I’m working on a way right now to make all of these data files available – it’s just really big.”

“But I will do my best to make all of this publicly available, so that anybody else can check my work. And I would be thrilled to find out that this isn’t really fluorofentanyl.”