I am somewhat dismayed by the claims made, given the circumstances that are required to generate alarming results.
On the one hand, a sealed product contains high levels of benzoic acid, but does not generate any measurable levels of benzene. User fillable devices contain only much lower levels.
Having identified a maximum level of 2mg/ml in the user fillable devices, rather than attempting to generate some data using this level, which might have some real world value, the authors use 9mg/ml. No justification for this is provided.
The power levels used also appear to have been arbitrarily assigned. It is accepted in the vaping community, and well documented in the literature, that users will not inhale 'dry hits'. The authors have previously documented the generation of high levels of toxicants during dry hit conditions, including a wide range of respiratory irritants. Given the nature of this potential exposure it should not be a surprise that users would avoid such conditions. However, rather than attempting to understand the use of these products, or to determine a threshold beyond which a dry hit occurs (which would enable a worst case model to be used), the authors have taken the view that because high power levels are available, they are relevant.
It does not take much critical thinking to expose the weakness of this approach.
For instance, my toaster has settings from 1 to 8. Experience tells me that, for my favored bread and style of toast, a setting of 3 is ideal. If a researcher arbitrarily decided that, since a setting of 6 is available, this should be used (alongside arbitrarily creating a bread recipe of their own) to model my exposure to PAHs, acrylamide and other thermal breakdown products of bread they would be alarmed. However, since the toast would be effectively cremated, and I would not eat it, it would not represent my exposure at all.
The situation is little different here. A significant reduction in biomarkers of exposure to toxicants is demonstrated in users of electronic cigarettes (e.g https://doi.org/10.1093/n... and https://doi.org/10.7326/M... ), indicating that, as actually used, toxicant generation is low.
The existence of such data should be considered when designing and interpreting in vitro results.
Millions of people continue to expose themselves to the vast array of harmful chemicals found in cigarette smoke, and an increasing number of them believe, incorrectly, that vaping offers little benefit if they switch. Researchers have an obligation to provide information that can help people make informed choices, not raise the spectre of hypothetical risks that are conceptually and demonstrably unlikely.