During the past several years, a variety of new tobacco products and nicotine delivery devices have been introduced. These include ENDS, heated tobacco products such as Philip Morris’ IQOS and British American Tobacco’s Glo, and oral nicotine delivery products such as Zyn, On! and Velo. Of these, ENDS, in particular electronic cigarettes are the most widely used. Most but not all public health researchers are of the opinion that these new products are generally less harmful than conventional tobacco products. The extent to which they could reduce harm is unknown, largely because some adverse health effects, in particular cancer and chronic lung disease, take many years to develop. Therefore, thorough epidemiological studies have not been possible. Furthermore, dual use of ENDS with combusted cigarettes is commonplace, and the extent of reduced exposure and potential harm in ENDS users who continue to smoke cigarettes is difficult to assess. (Goniewicz et al., 2018; Borland et al., 2019; Piper et al., 2019; Smith et al., 2021) In short term studies of acute effects of novel products such as ENDS, (Hajek et al., 2017; St Helen et al., 2020a) it is important to determine recent use of tobacco products as well as the extent of dual use of both products. For exposure assessment, self-reported use of particular products can be useful, but they are of limited utility for obtaining quantitative data. In this regard, biomarkers of exposure are useful. Biomarkers would be especially important to assess recent dual use in clinical trials of novel products and for epidemiological studies of health effects. Specific biomarkers have not been identified for e-cigarettes and other ENDS, because the substances in these products are nicotine, solvents used to generate the aerosols (propylene glycol and glycerol), and flavoring compounds which are found in conventional tobacco products and have dietary sources as well. (Schick et al., 2017) Pyrolysis reactions transform components of the e-liquids into various products during aerosol formation, but as yet no pyrolysis products unique to ENDS have been identified that could serve as selective biomarkers. Consequently, other than self-reports, which are of limited value, the only viable approach is to use biomarkers specific to tobacco products to identify and estimate the extent of their use in people using ENDS.

Tobacco contains a number of pyridine alkaloids other than the major alkaloid nicotine. (Schmeltz and Hoffmann, 1977; Rodgman and Perfetti, 2013) These minor alkaloids, which include anabasine, anatabine, anatalline, and nicotelline (Figure 1) are present in cigarette tobacco in concentrations ranging from about 1 to 1000 μg/g, compared to concentrations of about 15 mg/g for nicotine. (Jacob et al., 2013; Lisko et al., 2013) Although nicotine in nearly all e-liquids used in ENDS is derived from tobacco, the nicotine in most (but not all) products has been purified sufficiently that minor alkaloid concentrations are low compared to the amounts present in tobacco. (Palazzolo et al., 2019; Jacob et al., 2020) Consequently, minor alkaloids may be used as biomarkers for the use of conventional tobacco products in people using ENDS. (Berlin et al., 2019; Jacob et al., 2020) In this article we describe two new methods for quantitation of nicotine-related minor tobacco alkaloids in urine that can be used in this approach. One of these methods also measures concentrations of the tobacco-specific carcinogen metabolite 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL). The goal of our studies was to develop methods to simultaneously quantify multiple biomarkers useful in studies of dual use of ENDS and combusted cigarettes.

The first method is based on our published LC-MS/MS method for nicotelline, (Jacob et al., 2013) an alkaloid that we have proposed as a biomarker to distinguish ENDS use from use of combusted cigarettes. (Jacob et al., 2020) Since little if any nicotelline is excreted unchanged, and the metabolites identified so far are N-oxides (Figure 2), the method involves treating urine with titanium trichloride to convert the N-oxides back to nicotelline (Figure 3) that can be readily extracted and measured. Chromatography and mass spectrometry parameters were modified to include other minor tobacco alkaloid analytes. The second LC-MS/MS method utilizes a derivatization with hexanoic anhydride, developed for the carcinogen biomarker NNAL to enhance sensitivity. (Jacob et al., 2008) This derivatization also converts the secondary amine alkaloids anabasine, anatabine, and anatalline into amides, which results in improved chromatography, and allows simultaneous determination of these alkaloids with NNAL. (Figure 4) An advantage of these new methods is simultaneous determination of multiple biomarkers that have a wide range of biological half-lives, ranging from 2–3 h for nicotelline, to more than 10 days for NNAL. This can be important if measures of long-term exposure and recent exposure are desired. Another advantage is higher sensitivity (lower limits of quantitation) than previously reported methods, thus facilitating low-level exposure assessment. These advantages of the two new methods should make them especially useful in studies of dual use of ENDS and conventional tobacco products. We also introduce the alkaloid anatalline as a new, highly selective biomarker for tobacco exposure.

Two methods for minor tobacco alkaloids were developed, with the goal of simultaneously measuring urine concentrations of multiple analytes of interest. These include the established biomarkers anabasine and anatabine, and nicotelline, that we proposed as a biomarker for the particulate matter derived from tobacco smoke, (Jacob et al., 2013) anatalline, a little-studied tobacco alkaloid that we are developing as a new biomarker, and NNAL, a well-established biomarker for the tobacco-specific nitrosamine NNK (Hecht, 2002).

Method 1 is based on an LC-MS/MS method we developed for nicotelline. Since little if any nicotelline is excreted in urine unchanged, and the only metabolites characterized so far are N-oxides, the method involves treating urine with titanium trichloride, which reduces the N-oxides back to nicotelline that can be readily quantitated. (Jacob et al., 2013) (Figure 3). We have modified this method to include anabasine, anatabine, anatalline, and a lactam metabolite of anatalline as analytes. Nicotelline is highly selective for tobacco, and was undetectable or present at very low concentrations in 70 e-liquids that we analyzed. (Jacob et al., 2020) Therefore, we proposed that nicotelline could be used as a biomarker for combusted tobacco use in people using e-cigarettes. (Jacob et al., 2020) Nicotelline has a short half-life, 2–3 h, and is useful for detecting recent tobacco use. (Jacob et al., 2013) But nicotelline concentrations are undetectable in 12–24 h after tobacco use ceases. The tobacco-specific nitrosamine metabolite NNAL has also been used as a biomarker for tobacco use in ENDS users, but NNAL has a very long half-life, >10 days (Hecht et al., 1999; Goniewicz et al., 2009) and it can be measured in urine for several weeks after tobacco cessation. Consequently, biomarkers with half-lives longer than nicotelline, but shorter than NNAL, such as anabasine, anatabine, and anatalline would be also useful in studies of the short-term effects of switching from combusted cigarettes to e-cigarettes.

The minor alkaloids anabasine and anatabine have been used as biomarkers for tobacco use in people using nicotine-containing medications for tobacco cessation. (Jacob et al., 2002; Suh-Lailam et al., 2014; Sinclair et al., 2020) They have also been used as biomarkers for tobacco use in people using ENDS. (Berlin et al., 2019) Anabasine and anatabine have half-lives of about 16 and 10 h, respectively, and can detect tobacco use for a few days following tobacco cessation. (Jacob et al., 1999) Therefore, they are complementary to nicotelline (t ½ = 2 h) and NNAL (t ½ > 10 days). (Benowitz et al., 2020) However, in contrast to nicotelline, anabasine and anatabine have been found in e-liquids, sometimes in concentrations as high as in cigarette tobacco normalized to nicotine, (Palazzolo et al., 2019; Jacob et al., 2020) which may limit their utility as biomarkers selective for tobacco use. Anatalline is another minor alkaloid that we are developing as a biomarker. Like nicotelline, (Jacob et al., 2020) and in contrast to anabasine and anatabine, anatalline was undetectable or present at very low concentrations in 70 e-liquids that we analyzed. (Table 4). Interestingly, nicotelline does not appear to be naturally occurring (probably not biosynthesized) in the tobacco plant, and is mainly formed from anatalline by pyrolysis and oxidation in burning tobacco. This was demonstrated by adding anatalline to a non-tobacco plant material, oregano, preparing a “cigarette” from this, combusting, collecting and analyzing the smoke. In parallel, an oregano “cigarette” without anatalline was prepared and combusted. Nicotelline was detected in the smoke from the oregano “cigarette” spiked with anatalline, but not in the smoke from the “cigarette” without anatalline. (Jacob et al., 2013) This demonstrated that nicotelline can be formed from anatalline during combustion conditions, and that neither nicotelline nor anatalline are likely to be formed by combustion/pyrolysis of organic materials. The half-life of anatalline appears to be similar to that of anabasine or somewhat longer, which is apparent from the data presented in Figure 8, but additional studies will be required to determine its half-life. Therefore, we propose that anatalline would be a more selective biomarker than anabasine and anatabine for tobacco use in ENDS users, but otherwise would have similar attributes.

Method 1 uses the same sample prep as our published LC-MS/MS method for nicotelline. (Jacob et al., 2013) Chromatography and mass spectrometry parameters were modified to include anabasine, anatabine and anatalline. Data on precision, accuracy, and limits of quantitation for the method are presented in Table 2. Representative selected reaction monitoring (SRM) chromatograms are in Figure 5. This method was used to obtain data on concentrations of anabasine, anatabine, anatalline, an anatalline metabolite, and nicotelline in cigarette smokers and in non-smokers urine, summarized in Table 5. Concentrations of the widely used biomarkers for nicotine exposure, cotinine and trans-3’-hydroxycotinine are included for comparison, since concentrations, especially cotinine, have been used for many years as biomarkers of tobacco exposure, and can serve as an index of the extent of tobacco and/or nicotine product use. Applications of the minor alkaloids in studies of dual use of ENDS and conventional tobacco products would likely include cotinine concentrations as well to assess overall nicotine product use. We believe that this is the first published data on concentrations of anatalline and its metabolite in human urine.

Method 2 is based on a LC-MS/MS method we developed for the tobacco-specific nitrosamine metabolite NNAL. (Jacob et al., 2008) It involves treating extracts with hexanoic anhydride to give an ester derivative, that facilitates sample clean up via extraction with non-polar solvents, resulting in increased sensitivity compared to analyses with underivatized NNAL. Hexanoic andydride converts anabasine, anatabine and anatalline to hexanoic acid amides that likewise facilitates clean up of extracts and allows simultaneous determination along with NNAL. (Figure 4) Data on precision, accuracy, and limits of quantitation for the method are presented in Table 3. Representative selected reaction monitoring (SRM) chromatograms are in Figure 6. We think that it is interesting to note that in all SRM chromatograms from smokers’ urine extracts a peak with a retention time of about 0.45 min longer than the anatalline peak is observed, not found in chromatograms from non-smokers urine spiked with the anatalline standard. In Method 1 chromatograms, SRM chromatograms from smokers’ urine extracts a partially resolved peak with a shorter retention time than the anatalline is observed, not found in chromatograms from non-smokers spiked with the anatalline standard. (Figures 5–7). These peaks are clearly derived from a substance inhaled in cigarette smoke, possibly an isomeric alkaloid, and chromatographic separation is needed to accurately quantify anatalline. We also observed a peak partially resolved from anatalline in SRM chromatograms from a cigarette tobacco extract. (Jacob et al., 2013)

Method 2 has been used in a crossover study of dual users of e-cigarettes and combusted cigarettes. (St Helen et al., 2020b) The participants used e-cigarettes or smoked combusted cigarettes in separate 2-day study blocks, followed by a third 2-day block when they abstained from the use of any nicotine product, enforced by the study being carried out on a research ward with no access to nicotine-containing products. Urine samples were collected and analyzed for biomarker concentrations (Figure 8). These data are from a subset of participants (n = 19) in which the 2-day abstinence block immediately followed the e-cigarette block. For the five days prior to the e-cigarette block, participants were instructed to use their usual e-cigarette product and not use other tobacco or nicotine-containing products. However, since they were outpatients, compliance could not be enforced. We used data from this study, in which concentrations of anabasine, anatabine, anatalline, and NNAL were determined using Method 2, and nicotelline was available from previous analyses using a published method, (Jacob et al., 2013) to illustrate the attributes of the various biomarkers in terms of their different rates of elimination (Figure 8). Nicotelline, with a half-life of 2–3 h detects recent combusted tobacco use, and concentrations were near or below the limit of quantitation (LOQ) during the inpatient e-cigarette and abstinence blocks. Anatabine (t_1/2 = 10 h) was measurable during the e-cigarette block but not the abstinence block. Anabasine, anatalline, and NNAL were measurable in both the e-cigarette and abstinence blocks because their half-lives are too long to fall below the LOQ during the course of this study. (Benowitz et al., 2020) NNAL, due to its very long (>10 days) half-life can be detected for several weeks following tobacco cessation. Depending on the goals of a particular study, confirmation of short-term or long-term tobacco cessation may be desirable. NNAL would be the most useful for population studies in which any tobacco use in the past 2–3 months is of interest. The minor alkaloids are more relevant for assessment of short-term cigarette smoking, such as looking for point prevalence of smoking in smoking cessation trials. Also, the high sensitivity of Method 2 extends the time frame of applicability of anabasine and anatabine in which these biomarkers can be measured following tobacco cessation, and makes low-level exposure assessment possible. The lower limits of quantitation for anabasine and anatabine in Method 2 are 0.03 ng/ml. For comparison, the LLOQs reported in Wei et al. for a method used in large, population-scale studies including the National Health and Nutrition Examination Survey (NHANES) and the Population Assessment of Tobacco and Health (PATH) study are an order of magnitude higher, 0.5 and 0.4 ng/ml, respectively. (Wei et al., 2014) The LLOQ of NNAL in Method 2, 0.003 ng/ml, is sufficient for measuring exposure in cigarette and cigar smokers and smokeless tobacco users, since this is well below the cutpoint of 0.010–0.040 ng/ml for distinguishing active use from passive exposure. (Benowitz et al., 2020) For low-levels of secondhand smoke exposure, a more sensitive method may be advantageous, such as the method we reported that uses a larger urine volume but the same sample prep as Method 2, and has a LLOQ of 0.00025 ng/ml. (Jacob et al., 2008)

The goal of our studies was to develop methods to simultaneously quantify multiple biomarkers useful in studies of dual use of ENDS and combusted cigarettes.

Both of the methods we describe include anabasine, anatabine, and anatalline as analytes. Anabasine and anatabine have been used for a number of years as biomarkers to distinguish the use of nicotine-containing medications from the use of conventional tobacco products, and continue to be used in large population studies, such as the National Health and Nutrition Examination Survey (NHANES) and the Population Assessment of Tobacco and Health (PATH) study. (Wei et al., 2014) They have also been used to distinguish e-cigarette use from combusted cigarette use. (Berlin et al., 2019) In this report, we introduce anatalline as a new biomarker, with a rate of elimination similar to anabasine and anatabine, based on preliminary data, as illustrated in Figure 8. We suggest that anatalline has similar attributes to anabasine and anatabine, but unlike those two biomarkers it has not been found to any significant extent in e-liquids (Table 4 ) and therefore should be more selective for the use of conventional tobacco products.

The two methods differ in that Method 1 also measures nicotelline, and Method 2 also measures NNAL. Method 1 cannot measure NNAL because the titanium trichloride reagent used to reduce nicotelline N-oxides decomposes NNAL. Method 2 uses derivatization with hexanoic anhydride to enhance sensitivity for NNAL and also enhances the sensitivity for anabasine, anatabine, and anatalline. Nicotelline cannot be measured with Method 2, because little if any is excreted unchanged, and the only known metabolites are N-oxides, and reduction of these with titanium trichloride to nicotelline is required for sensitive quantitation. (Jacob et al., 2013) Another difference between the methods is that Method 2 employs a deconjugation step using β-glucuronidase, since about 50% of NNAL is excreted as glucronides. Since nicotine and cotinine are N-conjugated (pyridine nitrogen), the possibility exists that anabasine, anatabine, and anatalline might likewise be N-conjugated. To address this possibility, we analyzed a pooled smokers’ urine with and without deconjugation using Method 2. The data are presented in Table 6. Anabasine and anatabine were excreted as glucuronides to the extent of 20 and 31%, respectively, but interestingly glucuronidation of anatalline was not detected. As expected, NNAL was about 50% conjugated.

Method 1 also measures a lactam metabolite (Figure 1) of anatalline, which is reported for the first time. The lactam metabolite of nicotine, cotinine is a valuable biomarker for various reasons, including a longer half-life than nicotine. (Benowitz et al., 2020) By analogy, the lactam metabolite of anatalline might likewise be a useful biomarker. Consequently, we postulated the existence of this metabolite, and found that it indeed it is excreted in urine of smokers (Table 5). But, it proved to be a difficult analyte, perhaps due to its polarity and inefficient extraction limiting method sensitivity. This is reflected in the relatively high LOQ (0.5 ng/ml compared to the other analytes and the correlation coefficient of the standard curve. However, due to its potential attributes, further studies of this metabolite including efforts to develop a more sensitive method may be warranted.

Concerning the relative merits of these two new methods, the choice may depend on which analytes are most important for a particular study. Method 1 simultaneously measures nicotelline, anabasine, anatabine, and anatalline. Method 1 would be most appropriate if a measure of recent cigarette smoking, within 24 h, was desired, which would be provided by nicotelline, with a half-life of 2–3 h. This method also detects smoking occurring over the past several days, from anatabine, anabasine, and anatalline, with half-lives ranging from 10 h to 16 or more hours. Anabasine, anatabine, and anatalline may be useful for confirming tobacco cessation over this time frame, such as studies of the effects of short-term switching from combusted cigarettes to e-cigarettes. Method 2 simultaneously measures anabasine, anatabine, anatalline, and NNAL. Method 2 would be most appropriate if the goal were to detect, and measure the extent of smoking occurring over several weeks, which would be provided by NNAL, which has a half-life in excess of 10 days. Method 2 will also detect smoking occurring within several days, from concentrations of anabasine, anatabine, and anatalline. Method 2 would also be most appropriate for low-level exposure assessment, because it is more sensitive than Method 1 by a factor of about 3 for anabasine, anatabine, anatalline (Tables 2, 3).

The value of multiple tobacco biomarkers with a range of elimination rates is discussed in a recent publication, including a figure illustrating the time course for concentrations to fall below the LLOQ. (Benowitz et al., 2020)

In conclusion, two new methods for tobacco biomarkers have been developed, that can be applied to studies of dual users of ENDS and conventional tobacco products. Advantages include simultaneous determination of multiple analytes, and improved sensitivity compared to previous methods that could be useful for low-level exposures. We also introduce a new biomarker, anatalline. Anatalline has similar attributes to anabasine and anatabine, and may have greater specificity for tobacco than those two alkaloids.