E-cigarette risk assessment based on biomarkers: good approach but important problems in the analysis

 

By Dr Farsalinos

Just today, I saw an analysis presented at the website of the University of Otago in New Zealand, presenting the case of using biomarkers of exposure as a method of performing a risk assessment analysis for e-cigarettes relative to smoking. The analysis is authored by Prof Wilson, Dr Gartner and Prof Edwards, and uses currently available evidence from published studies and poster presentations on biomarkers of exposure.

First, I should say that this is a very good idea. We need to find feasible ways of performing such a risk assessment, since it will take years to have cohort studies with longitudinal follow-up to measure the risk from e-cigarette use in comparison to smoking. However, I have noticed some important problems in their analysis, which I present herein.

The analysis was focused on carbon monoxide, urine NNAL, acrolein biomarkers (mainly 3-HPMA) and cardiovascular biomarkers from one acute-response study. There are many interpretational problems when comparing the levels of biomarkers with smokers only, because the authors did not consider that some biomarkers are also detected in non-smokers, because of exposure from sources unrelated to smoking.

 

One of the biggest studies evaluating acrolein metabolites in urine of smokers vs. non-smokers was the NHANES 2005-2006 analysis published in the journal Environmental Health Perspectives. They evaluated 2 metabolites, 3-HPMA (same as used in the Wilson et al. analysis) and CEMA. Both biomarkers were detected in non-smokers, and the levels were 20% and 39% that of smokers. Another study by the group of Stephen Hecht (a well-known researcher on biomarkers of smoking exposure) found again that smoking cessation (without the use of any alternative product resulted in levels of 3-HPMA that were 24% that of smokers.

As mentioned in an excellent review on acrolein, it is ubiquitously present in (cooked) foods and in the environment and is formed from carbohydrates, vegetable oils and animal fats, amino acids during heating of foods, and by combustion of petroleum fuels and biodiesel. So, there are plenty of sources beyond smoking, and this is why its biomarkers are detected both in vapers and in non-smokers.

Therefore, the analysis by Wilson et al. should have concluded that as far as acrolein is concerned, the risk for disease among e-cigarette users is identical to non-smokers, and the relative-to-smoking risk is 0%. The results clearly show no acrolein exposure from e-cigarettes.

For NNAL, although some studies are showing detectable levels in non-smokers, I will accept Stephen Hecht’s position that it is not detectable unless there is environmental tobacco exposure. It is true that it is a biomarker which is very tobacco-specific. Total NNAL is a measure of NNK exposure (one of the major tobacco-specific nitrosamines). The only source in e-cigarettes is the minute amounts present in nicotine (since it is extracted from tobacco). Thus, first of all it is extremely important to use pharma grade (USP or Eur.Ph. grade) nicotine in e-liquids. Furthermore, it is extremely important to exclude dual use, as well as environmental tobacco exposure. We have already published a study showing that the levels of NNK are extremely low in e-liquids, and no more nitrosamines are produced during to evaporation process. The levels of total NNAL found in the Hecht study were about 1.5% the levels of smokers. It is important to mention that it takes weeks for NNAL levels to be reduced, because of the prolonged half-life of this metabolite. Thus, if someone is a recent quitter, the levels will still be lower compared to smoking but not as low as they will found after several months of smoking abstinence. Additionally, finding levels of 1.5% that of smoking does not mean that the risk of e-cigarette use is 1.5% that of smoking. Instead, the risk of e-cigarette use is 1.5% of the attributable-to-NNAL proportion of the risk of smoking. As mentioned above, NNAL is a measure of NNK exposure, so we need to find the risk from smoking that is attributed to NNK (and accept a strong correlation between level of NNK exposure and levels of NNAL in urine). Then, we should apply the following equation:

ECrisk (relative to smoking) = (% of smoking risk attributed to NNK) x 1.5%

Based on the Cancer Risk Indices for tobacco cigarette smoke compounds presented in the study by Fowles & Dybing (table 2), the proportion of smoking risk attributed to NNK exposure is about 0.80% of the total smoking risk. Thus, the e-cigarette risk attributed to NNK exposure is 0.012% of the total risk from smoking. This is orders of magnitude lower than the 5% residual risk for e-cigarettes estimated by Public Health England and the Royal College of Physicians. Of course, you need to add all other emissions and associated risks to generate the total level of relative risk, but considering that some emissions are completely absent from e-cigarettes (like CO, aromatic hydrocarbons etc), I do not expect the relative risk to be more than 5% (most likely, it will be far below that).

 

The study evaluating inflammatory mediators and endothelial function (flow-mediated dilatation-FMD, Carnevale et al.) after acute exposure should be excluded from the analysis. First, no biomarker of cardiovascular disease has any prognostic value when measured after an acute challenge or intervention. I challenge anyone to show me 1 study showing that the acute response in any of these biomarkers was ever found to have a prognostic value.

This is extremely important to understand. For example, the guidelines on FMD measurements clearly state that before the measurements subjects should abstain from the use of any stimulants (like nicotine, caffeine and alcohol) for at least 4-6 hours. Food, exercise, environmental temperature, emotional stress, noise, phase of menstrual cycle and medications are other factors which affect FMD measurements. This is because the stimulation of the sympathetic nervous system is biasing the measurements and does not provide a correct estimate of resting function. It is the resting measurements that have a prognostic value in cardiovascular disease. Of note, caffeine (which is not a risk factor for cardiovascular disease) and exercise (which is beneficial for cardiovascular health) have similar acute effects on endothelial function and oxidative stress as e-cigarette use (see here, here, here and here). To show how misleading the measurement of acute effects are, it is worth noting that although using NRTs acutely increases arterial stiffness (study here), quitting smoking with NRTs reduces arterial stiffness in resting conditions (study here). It is clear that the acute effects are irrelevant, and someone should test under resting conditions smokers vs. e-cigarette users vs. quitters without using any alternative product. The latter group is very important because we know some biomarkers (like CRP, fibrinogen and WBCs) take years after smoking cessation before being fully reversed.  

 

In conclusion, I think the approach by Wilson et al. to propose a risk assessment analysis based on biomarkers of exposure is realistic and appropriate. However, the interpretation of their preliminary analysis, as mentioned in the university blog, has some important flaws which need to be addressed in order to generate accurate conclusions about the relative risk of e-cigarettes compared to smoking.

 

 

Background Image

Header Color

:

Content Color

: