by Carl V Phillips
I received an inquiry about diacetyl from a reporter in the e-cigarette press and though I would share my response more widely. Most readers will know that diacetyl is a food additive that produces a strong buttery flavor, and that it is the most controversial among the several controversial flavoring agents in e-cigarettes. The concern is that high-dose inhalation exposure is believed to produce the horrible acute lung disease, bronchiolitis obliterans, commonly known as “popcorn lung” because the cases have been found in workers in flavored popcorn factories and at least one consumer who was huffing the outgassing from microwaved popcorn. (I say “believed” because all of these victims inhaled a cocktail of chemicals, not just the one molecule, so it is consistent with the evidence that the outcome depends on the multiple exposures and is even possible that diacetyl is not a necessary part of the mix.)
My regular readers might also recall that elsewhere I have referred to the diacetyl controversy as an own goal. It was vapers and e-cigarette advocates who introduced the worries about this molecule, which is quite reasonable, but then managed to build the level of consternation well above what was suggested by the evidence. The most important bit of evidence is that after many millions of person-years of vaping, some unknown but nontrivial portion of which included diacetyl exposure, there has not been a reported case of a vaper getting bronchiolitis obliterans. Those who follow the anti-vaping propaganda will no doubt know that the anti-THR liars have picked up on the consternation as their latest claim — “Vaping causes popcorn lung!!!” Never mind that it hasn’t done so.
Anyway, the science is complicated and there are no simple answers. The evidence is consistent with diacetyl in e-cigarette liquid being hazardous, and with it being the most hazardous single characteristic among e-cigarettes (there is nothing that we know poses a measurable risk for any serious disease). The evidence is also consistent with this exposure causing no risk at the relevant quantities.
I balked at answering the series of set-piece questions in the reporter’s query. I was not comfortable answering some of them because they embed premises in the questions that any answer would implicitly endorse. Others were difficult technical questions that no one, as far as I know, has the expertise to answer, or that would require an entire paper to provide a scientifically legitimate answer. So instead I offered the following response:
The framing of the discussion surrounding diacetyl in the vaper community is misleading in several ways. First, it is extremely difficult to estimate risks from consumer exposures based on occupational epidemiology results, which generally involve extreme levels of exposure and a cocktail of different exposures rather than just a single chemical. It is far more difficult to estimate real risks based on artificial toxicology experiments. Both extrapolations involve heroic guesses. One such guess is about the dose-response, a concept that seems to be widely understood among those discussing the issue. But other guesses are even more difficult than that, including how different exposures interact (no one is being exposed to just one molecule, either vaping or in an occupational setting), and how particular subpopulations might react. Fortunately we know from real-world experience that widespread e-cigarette use, for a period of a several years, has not caused any detectable outbreaks of any serious acute disease including the one that is popularly attributed to diacetyl exposure. That is the most solid evidence we have.
However, inhalation of anything other than clean air may cause risk. Eliminating one molecule, or ten, from vapor is not going to change that. We know enough from vapor chemistry and epidemiology to be reassured that for normal vaping, there is no known major hazard. But we do not know nearly enough about the interaction of airway biology and this concentration of exposure over decades (even for the very benign chemicals, like the carriers) that we can be confident we have identified potential harm pathways. Obviously any risks are dwarfed by those of smoking; those of us who have made educated guesses about the risk put it in the range of 1/100th that from smoking. But though the “we just don’t know anything!” rhetoric we have to deal with is clearly absurd, there is genuine uncertainty. It would be surprising if the risk were as high as 3% that from smoking and astonishing if it was 5%. But given the complexity of biology it is possible that long-run exposure causes that much risk or even more, particularly for more harmful combinations of liquid chemistry, hardware, and behavior. (Note that we do not really know which of those are more harmful, though we can make some guesses — we can just be sure that some are worse than others.)
We can expect that some particular chemicals and technologies (e.g., being able to overheat) will be eliminated from the mix as the products evolve. But we will never really know if this made any difference for health outcomes. Nor will we know if any alterable small risk remains — the epidemiology is simply incapable of detecting small risks in a population of ex-smokers. Anyone choosing to vape should recognize this and make their decision based on it. Anyone who is very worried about what they are inhaling should probably just switch to snus or NRT, which are almost certainly slightly lower risk than vaping, rather than fall into the trap of believing that playing whack-a-mole with individual chemicals will change that.