Biomarkers of environmental tobacco smoke exposure.

Biomarkers are desirable for quantitating human exposure to environmental tobacco smoke (ETS) and for predicting potential health risks for exposed individuals. A number of biomarkers of ETS have been proposed. At present cotinine, measured in blood, saliva, or urine, appears to be the most specific and the most sensitive biomarker. In nonsmokers with significant exposure to ETS, cotinine levels in the body are derived primarily from tobacco smoke, can be measured with extremely high sensitivity, and reflect exposure to a variety of types of cigarettes independent of machine-determined yield. Under conditions of sustained exposure to ETS (i.e., over hours or days), cotinine levels reflect exposure to other components of ETS. Supporting the validity of cotinine as a biomarker, cotinine levels have been positively correlated to the risks of some ETS-related health complications in children who are not cigarette smokers.

A biomarker is desirable in quantitating systemic exposure of nonsmokers to constituents of environmental tobacco smoke (ETS). Self-report measures such as hours per day exposed to ETS by nonsmokers are likely to be imprecise indicators of intake of tobacco smoke because of variations in the number of cigarettes smoked, proximity of nonsmokers to smokers, room ventilation, and other environmental characteristics, as well as individual differences in sensitivity to and concern about adverse effects of ETS. The optimal assessment of exposure to tobacco smoke would be by analysis of the concentrations of a component of smoke, a biologic marker or biomarker, in body fluids of an exposed individual. Two broad questions must be considered in assessing the validity of a biomarker of tobacco smoke exposure. The first is, How well does the concentration of a marker chemical in the air reflect exposure to toxic constituents of smoke that are of concern with respect to health? The second is, How well does a concentration of a particular chemical in a biologic fluid reflect an individual's intake of that chemical (or a related chemical) from tobacco smoke?
The National Research Council (1) has proposed criteria for a valid marker of ETS in the air as follows. The markers a) should be unique or nearly unique for ETS so that other sources are minor in comparison, b) should be easily detectable, c) should be emitted at similar rates for a variety of tobacco products, and a should have a fairly constant ratio to other ETS components of interest under a range of environmental conditions encountered. Furthermore, the validity of a biomarker depends on the accuracy of the biologic fluid measurement in quantitating the intake of the marker chemical, which in turn may be influenced by individual differences in rates or patterns of metabolism or excretion, the presence of other sources (such as diet) of the chemical, and sensitivity and specificity of the analytic methods used to measure the chemical.
Other issues of interest in assessing the risks of exposure to ETS are how well the biomarker indicates long-term exposure to ETS as well as whether a biomarker predicts the likelihood of ETS-related disease and if so how well.

Potential Biomarkers of Environmental Tobacco Smoke
A variety of biomarkers of tobacco smoke exposure are proposed ( Table 1). The measurement of cotinine concentrations in biologic fluids has been used most widely by scientists to evaluate ETS exposure because cotinine reflects exposure to nicotine, which is almost specific to tobacco. Chemicals in tobacco smoke such as carbon monoxide or cyanide (the latter metabolized in the body to thiocyanate) can be measured in blood. However, the levels of these chemicals are nonspecific, i.e., there are significant sources of carbon monoxide and cyanide, including the body's own metabolism, other than ETS. Thus, these markers are both nonspecific and insensitive markers of ETS exposure. Other markers that have been proposed to quantitate tobacco exposure include adducts of 4-aminobiphenyl to hemoglobin in red blood cells (2)(3)(4), adducts of benzo[a]pyrene and other potential carcinogens to DNA in white blood cells (5)(6)(7)(8), adducts of polycyclic aromatic hydrocarbons (PAHs) to plasma albumin (9), urinary excretion of nicotine-derived nitrosoamines (10), urinary hydroxyproline or n-nitrosoproline excretion (11), and urinary mutagenicity (2,12). Also, solanesol has recently been proposed as perhaps the best marker of particle exposure in the air (13,14). Unfortunately, solanesol is extensively metabolized in people and levels are quite low, making quantitation difficult. Levels of 4-aminobiphenyl-hemoglobin (4-ABP-Hb) adducts have been shown to be elevated in nonsmoking ETS-exposed adults compared to non-ETS-exposed adults (2)(3)(4). Adduct levels showed a significant dose response with increasing history of ETS exposure in a study by Hammond et al. (3). However, 4-ABP-Hb levels in nonexposed nonsmokers were substantial, and there was considerable overlap between levels for exposed and nonexposed nonsmokers.
Similar findings were reported for PAH adducts with DNA or albumin. A study by Mooney et al. (15) in which these biomarkers were followed before and Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 after subjects stopped smoking for 14 months illustrated the problem of nonspecificity for both 4-ABP and PAH adducts. Levels of 4-ABP-Hb and PAH-DNA adducts dedined to an average of only 25 and 48%, respectively, of the baseline smoking value after subjects stopped smoking for 8 months. Thus, residual levels were still substantial, which made detection of ETS exposure difficult ( Figure 1). In contrast, blood cotinine levels fell to an average of 2% that observed during baseline smoking, a level low enough to be able to detect nicotine exposure from ETS. Nicotine-derived nitrosoamines such as 4-(methylnitrosoamino)-1-(3-pyridyl)-1butanone (NNK) are specific for tobacco exposure and are metabolized to a butanol metabolite (4-(methylnitrosoamino)-1-(3pyridyl)-1-butanol (NNAL), and its glucuronide (NNAL-GLUC) (10). Urine levels of NNAL + NNAL-GLUC are elevated in nonsmokers exposed to ETS, and in one small study the correlation between urine NNAL plus its glucuronide with urine cotinine concentration was quite strong (r= 0.89). The assay for NNAL is technically demanding; as yet very few subjects have been studied using this biomarker, so its general utility in exposure assessment is difficult to assess.
Overall, it appears that although a number of markers may reflect exposure to particular components of tobacco smoke in active smokers, most of the measures are too nonspecific (i.e., high baseline values even in nonexposed nonsmokers or environmental sources other than tobacco smoke) and/or insensitive (i.e., the increment due to ETS exposure is small compared to baseline values) for use in quantitation of levels of smoke exposure to which most nonsmokers are exposed. At present, cotinine appears to be the most specific and most sensitive biomarker for exposure to nicotine from ETS. A limitation of using cotinine, which is discussed in "Cotinine as a Biomarker," is that cotinine indicates ongoing exposure but not long-term exposure to ETS.

Cotinine as a Biomarker
Cotinine, the major proximate metabolite of nicotine, has been widely used as a biomarker of tobacco exposure (16,17). Plasma cotinine concentrations correlate better to various measures of biologic effects of cigarette smoking than does selfreported cigarettes per day (18,19).
Cotinine concentrations in plasma, urine, and saliva of nonsmokers have been used to assess population exposure to ETS to develop risk estimates for lung cancer related to ETS exposure (20,21). The prevalence of significant ETS exposure in control (reportedly unexposed) groups has been estimated based on cotinine measurements and used to adjust lung cancer risk estimates upward for comparison with actual unexposed controls (1,22,23).
The validity of using cotinine as a biomarker for ETS exposure has been questioned (24,25). Concerns include a) the concentration of nicotine in the air is not a good marker of other constituents of ETS because the ratio of nicotine to other ETS components is highly variable and depends on such factors within a space as surfaces, ventilation rate, sampling duration, time since smoking, and air distribution patterns; b) the ratio of nicotine emission to respirable suspended particles (RSP) emission is not constant across a wide range of cigarettes; c) exposure to nicotine vapor in the absence of other ETS components can occur; d) there is no standard method for determining nicotine or its metabolites in biologic fluids; e) interindividual differences in rates and patterns of nicotine and cotinine metabolism make the use of nicotine or cotinine as biomarkers of limited utility, and f) dietary nicotine exposure may confound low-level determinations of nicotine and cotinine in biologic fluids. A detailed discussion of these criticisms and why cotinine is a valid biomarker of ETS  Figure 1. Biomarkers in blood following smoking cessation. Abbreviations: PAH-DNA, polycyclic carbon-DNA adducts; 4-ABP-Hb, 4-aminobiphenyl-hemoglobin adducts; B, baseline (precessatio tion); SD, standard deviation; 1 Ow, 8m and 1 4m = 10 weeks, 8 and 14 months after smoking cessa of subjects with data were cotinine, n=40; PAH-DNA, n=40; 4-ABP-Hb, n= 16. The SD for c ng/ml at baseline, and averaged 0.7 ng/ml after cessation; the SD for PAH-DNA was 8.8 adductm at baseline, and averaged 5.2 after cessation; the SD for 4-ABP-Hb was 8.7 nM/mHb at baseline after cessation. Adapted from Mooney et al. (15). exposure has been published recently (26). The following section is a summary of that discussion.

Exposure to Nicotine from Environmental Tobacco Smoke
Nicotine is a chemical found in all tobacco products. The tobacco in manufactured cigarettes contains between 6 and 12 mg of nicotine (27). Seventy-five percent or more of nicotine emitted from a cigarette is emitted into the air as sidestream smoke, which contributes substantially to ETS (1,22,23). The amount of nicotine in sidestream smoke, when normalized for the generation of tar, is similar for different brands of cigarette, independent of nominal nicotine yield (22,23,28). For example, Rickert et al. (28) found an average tar-tonicotine ratio of 5.9 in the sidestream smoke of 15 different brands of cigarettes, with a relatively low degree of variability (coefficient of variation 12.2%). They found no difference in the sidestream smoke tar-to-nicotine ratio when comparing ventilated and nonventilated cigarettes.
Because sidestream smoke is not identical to ETS, it is desirable to examine the composition of ETS generated by different brands of cigarettes. Leaderer and Hammond (29) confirmed in chamber studies that the air RSP-to-nicotine ratio generated by smokers smoking 10 different brands of U.S. cigarettes with widely differing machine-determined yields is similar. The average RSP-to-nicotine ratio was 14.1 (coefficient of variation 13.4%). The ratio was similar for filtered and nonfiltered cigarettes. The results of the: (28) and Leaderer and Han studies are consistent, and ther son to believe that the compos stream smoke generated by ot marketed cigarettes, which a nicotine content, should be dif ever, if future cigarettes cont. tine but generate similar amc and other combustion prod duced by currently marketed has been proposed to make ci addictive (30), ETS nicotii expected to underestimate other tobacco combustion proc In mainstream smoketaken in by the smoker-nic tained in particles composed and other nicotinelike alkalo most of the nicotine leaves et phase and becomes part of ti vapor phase (29,31,32). Nic( is breathed in through the no, and inhaled into the lungs by Nicotine is extremely soluble is highly extracted from ET respiratory tree (33).
Levels of nicotine and othei ETS decay at different rates that the ratio of nicotine to tuents of ETS may differ at var time after generation of the 35). Another source of varia RSP-to-nicotine ratio is du ground level of partides arising other than ETS. Thus, when particle levels for ETS decline background particle concentral tially influence the ratio of RS At low concentrations of EI 4-ABP-Hb becomes very large. For this reason, measur-(n= 16) ing the slope of the regression line between air nicotine concentration and respirable particle concentration is the best way to assess the degree of correlation between air RSP and nicotine (29). In addition, when air samples are collected over the time interval of a typical human exposure, i.e., over hours or days, RSP-to-nicotine ratios are much less variable compared with spot 1Ow 8m 14m or brief measurements. Leaderer and Hammond (29) showed an RSP-to-nicotine slope of 9.8 for 47 home air samples samc aromatic hydro-pled over several days. The correlation coef-)n smoking condi-ficient between RSP and nicotine was 0.8.
ition. The number Similar ratios for RSP to nicotine were otinine was 114 reported by Miesner et al. (36) in works/1 08 nucleotides place samples. Thus, time-averaged ratios of and averaged 3.6 nicotine to RSP and presumably other ETS constituents are relatively consistent (21). When a person is exposed to ETS over time, Rickert et al. the intake of nicotine reflects exposure to nmond (29) other constituents of ETS.
re is little rea-Cotinine as a Biomarker sition of side-.her currently for Intake of Nicotine ire of similar The presence of cotinine in a biologic fluid Tferent. How-indicates exposure to nicotine. There is ain less nicosome individual variation in the quantita-)unts of RSP tive relationship between cotinine levels in lucts as pro-the blood, saliva, or urine and the intake of cigarettes, as nicotine. This is because different people igarettes non-convert different percentages of nicotine to ne would be cotinine (usual range 55-92%) and exposure to because different people metabolize cotiducts.
nine at different rates (usual range of coti--i.e., smoke nine clearance, 19-75 ml/min) (37). The otine is con-relationship between nicotine and cotinine of tar, water, can be expressed mathematically as follows oids. In  where COT= cotinine and NIC= nicotine. At steady state, Generation rate of COT = elimination rate of COT = CLCOT X CBSS [2] where CLCOT is the clearance of cotinine and CB,, is the steady-state blood cotinine concentration.
Therefore, combining Equations 1 and2: Intake rate of NIC CLCOT X CBS5 % Conv NIC to COT [3] Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 Or rearranging: Intake rate of NIC % Conv NIC to COT) BS [4] K x CBss In adult smokers, the conversion factor (K) that on average converts a blood level of cotinine to a daily intake of nicotine has been estimated to be 0.08 mn/24hr (range 0.05-0.10, coefficient of variation 21.9%) (37). Thus, a cotinine level of 300 ng/ml, a typical value for a smoker, corresponds to a daily intake of 24 mg nicotine. Clearance data are similar for smokers and nonsmokers (38), so the K factor is expected to be similar in nonsmokers.
Cotinine levels only approximate nicotine intake because of variability in the conversion factor. However, the degree of variability in the conversion factor (coefficient of variation=21.9%) is not particularly great compared with variability in the clearance of most other drugs and is much less than the degree of variability typically observed for pharmacodynamic parameters (39). Even with this inevitable degree of imprecision, cotinine levels in large groups of subjects would be expected to accurately reflect average group exposure to nicotine from ETS.
Because nicotine from ETS is absorbed into the bloodstream and cotinine is generated from nicotine in the liver and released into the bloodstream, blood levels of cotinine should most closely reflect the dose of nicotine absorbed from ETS. However, most studies of ETS exposure have used saliva or urine concentrations of cotinine because these samples are easier to obtain. Saliva and blood cotinine levels are highly correlated, with a saliva-to-blood ratio of 1. 1 to 1.4 (40,41). Therefore, saliva and blood cotinine levels can be used interchangeably. Urine concentrations are also highly correlated with blood concentrations, with urine levels about 6 times higher than those for blood.

Typical Levels of Cotinine from ETS Exposure and Estimation of Corresponding Dose of Nicotine
Cotinine levels in people exposed to ETS have been studied by many research groups (for reviews, see 1,22,23). Most studies found increasing levels of cotinine with increasing levels of self-reported ETS exposure.
Nonsmoking subjects in various studies represent several different populations. The data of Jarvis et al. (42) on adults attending cardiovascular disease clinics at a London hospital are an example of estimating daily nicotine intake from ETS (42). Using urine concentrations of 7.7 and 1.6 ng/ml and the equations described previously, the estimated daily intake of nicotine by nonsmokers was 100 and 20 pg for those reporting exposure and no exposure to ETS, respectively. Extreme ETS exposure is likely to occur in pubs and bars where smoking is common and ventilation is often poor. Among 42 nonsmoking bar staff in London and Birmingham, England, the median saliva cotinine concentration was 7.95 ng/ml (SD 6.1) and ranged from 2.2 to 31.3 ng/ml (43). Using Equation 4, the median nicotine intake is estimated to be 630 pg/day. The maximal nicotine intake, corresponding to a saliva cotinine concentration of 31.3 ng/ml, is estimated to be 2.5 mg/day (nicotine intake equivalent to actively smoking 2.5 cigarettes per day).

Air Levels of Nicotine from Environmental Tobacco Smoke and Predicted Cotinine Levels in Biologic Fluids
The theoretical relationship between air levels of nicotine and cotinine levels in the urine of nonsmokers can be described and is useful in understanding potential sources of variability in that relationship. Assume a workplace level of nicotine in the air due to ETS of 20 pg/m3 (44). The dose of nicotine inhaled is equal to the product of air concentration and ventilation rate. A typical ventilation rate for an adult during light activity is 1 m3/hr. Thus, the intake of nicotine would be about 20 pg/hr. About 71% of nicotine that is inhaled is absorbed (33), so the systemic dose of nicotine is estimated to be about 14 pg/hr. Assuming an 8-hr workplace exposure, this would be equivalent to 112 pg/day. Using the equations described previously results in this level of intake producing an average urine cotinine concentration of 8.6 ng/ml, a value consistent with that measured in nonsmokers exposed to ETS. Air nicotine levels measured by Hammond et al. (45) over 9 hr at 11 Massachusetts office worksites that allowed smoking indicated a median level of 8.6 pg/m3 (45). The absorption of nicotine from this level of exposure over 9 hr is predicted to be 55 pg, resulting in an average urine cotinine concentration of 4.0 ng/ml. For perspective, in office workplaces that banned smoking, the median air nicotine level was 0.3 pg/m3. Individual variability may exist in the factors that determine the relationship between air levels of nicotine and urine cotinine concentrations. Potential sources of variability include respiratory ventilation rate (e.g., higher minute ventilation with higher work levels), extent of pulmonary retention of nicotine by the lung, timing of the sample collection versus time of exposure, sources of exposure other than that under study, percent metabolic conversion of nicotine to cotinine, total and renal clearance of cotinine, and urine flow rate. It is expected, therefore, that urine cotinine would yield only an approximate estimation of air nicotine levels.
The relationship between ambient air nicotine levels and cotinine levels in the urine or saliva of nonsmokers has been reported in three studies. Studies by Marbury (46) and Henderson (47) involving children in the home and one study by Coultas (44) of adults in the workplace found a reasonably strong correlation between ambient air nicotine and urine cotinine concentrations (correlation coefficients, r= 0.81, Marbury; r= 0.68, Henderson; r= 0.60, Coultas). These correlations are probably as high as can be expected, given the sources of variability in nicotine uptake and metabolism. In view of the multiple potential sources of individual variability, the Marbury, Henderson, and Coultas studies support the predictive value of urine cotinine concentration as a biomarker of ETS-derived nicotine exposure.

Nicotine in Food as a Source of Cotinine
Several foods contain small amounts of nicotine (48)(49)(50). It has been suggested that nicotine from food might falsely indicate exposure to ETS (24,25). Davis et al. (49) estimated that average daily consumption of tomatoes, potatoes, cauliflower, and black tea together could result in a daily intake of 8.8 pg nicotine. They estimated, based on a maximum consumption of all of these particular foods on the same day, that a person could ingest as much as 99.9 pg of nicotine per day from food. It should be noted that more than 50% of intake in the Davis study was based on drinking black tea; this study and others have shown that some black teas contain no nicotine. It is not known how much nicotine is contained in the typical tea consumed by most Americans.
Also of note, the nicotine intake from tea reported by Davis et al. (49) necessitates drinking about 4 qt (3,840 ml) per day. Consumption of such large volumes of fluid would result in a urine output much greater than the 1000 ml these authors assume in their prediction of urine cotinine concentration. A larger urine volume would substantially reduce concentrations of cotinine in the urine below the Davis et al. estimate of 6.2 ng/ml.
Average nicotine intake (i.e., absorbed dose) from significant ETS plus dietary exposure is about 80 pg. Repace (51) used average American vegetable consumption data, which included 27 g of tomatoes and 75 g of potatoes, to estimate daily nicotine intake from the diet of 0.7 pg/day. As noted previously, Davis et al. calculated that daily consumption of tomatoes, potatoes, cauliflower, and black tea might result in a daily intake of 8.8 pg nicotine. Even in the latter case, the expected intake from a diet rich in nicotine-containing food is only 10% of the total nicotine exposure experienced by a person with significant ETS exposure. Conversely, an intake of 8.8 pg nicotine per day from food would be expected to yield a steady-state urine cotinine level of less than 0.7 ng/ml, which is well below the level indicating significant ETS exposure.
The impact of tea drinking on serum cotinine levels of nonsmokers in Scotland has been studied explicitly (52). No effect on plasma cotinine levels was observed with consumption levels of up to 10 cups or more of tea per day. In contrast, the same study showed a robust relationship between self-reported ETS exposure and plasma cotinine levels. Thus, nicotine in tea appears to contribute little to cotinine levels in most people and would be insignificant compared with nicotine levels from exposure from ETS. I conclude that although food is a source of low-level nicotine exposure, for most people it represents an insignificant exposure compared with exposure to ETS and is likely to inflate population estimates of ETS nicotine exposure very little.
Nicotine Emissions from the Environment as a Source of Human Exposure Nicotine from ETS deposits on room surfaces, such as walls and carpets, and may contaminate house dust. Nicotine emissions from surfaces or dust in the air may result in measurable levels of nicotine in the air that persist after the last cigarette was smoked in a room (53). Similarly, nicotine can be emitted from the clothes of smokers even when they are not smoking in the room. Thus, it has been suggested that nonsmokers may be exposed to nicotine not only through direct exposure to ETS but also by exposure to air in a room in which smoking has occurred in the past or to air in a room shared by people whose clothes have been contaminated by tobacco smoke (53). Although concentrations of nicotine have been measured in the air under these conditions, levels of nicotine are quite low compared to those from ETS. For example, house dust in a nonsmoker's home in which cigarettes were smoked on one occasion was reported to contribute to nicotine levels in the air of 0.2 to 0.7 pg/m3 over the next few days (53). Using calculations described previously assuming a ventilation rate of 1 m3/hr, and assuming an 8-hr exposure would produce a daily nicotine intake of 1.1 to 4.0 pg and result in a urine cotinine concentration of 0.1 to 0.3 ng/ml. These values are trivial compared to those derived from ETS exposure.
Health and Other Biologic Effects That Validate Cotinine as a Biomarker of ETS Exposure A significant relationship between biologic effects of ETS and cotinine levels of biologic fluids would further support cotinine as a quantitative marker of ETS exposure. Several studies support this concept. Matsunga et al. (54) studied the effects of ETS exposure on the metabolic clearance of theophylline, a drug whose metabolism is known to be increased in nonsmokers by the presence of cigarette smoke. In 14 nonsmokers, significant correlations were found between plasma cotinine (r= 0.72) or urinary cotinine (r= 0.79) and the clearance of theophylline. Strachan et al. (55) studied 736 seven-year-old school children and found a positive correlation between the quintile of salivary cotinine levels and the risk of middle ear effusion. In another study of a group of 770 children, Strachan et al. (56) reported a significant inverse correlation between salivary cotinine and various tests of lung function. Similar results were reported among a group of 2,511 children in a study by Cook et al. (57), and Rylander et al. (58) found that the risk of wheezing bronchitis in children 18 months of age or younger increased as urinary cotinine excretion increased. Finally, and perhaps most relevant to concerns about occupational exposure in adults, Tunstall-Pedoe et al. (59) found a gradient of risk of diagnosed coronary heart disease that increased with increasing serum cotinine in nonsmokers. Thus, several different biologic effects of ETS were shown to be quantitatively related to cotinine levels, supporting the idea that cotinine levels reflect ETS exposure and effects.

Conclusion
To summarize, the National Research Council criteria for a valid marker of ETS exposure include: a) should be unique or nearly unique for ETS so that other sources are minor in comparison; b) should be easily detectable; c) should be emitted at similar rates for a variety of tobacco products; and 4a should have a fairly constant ratio to other ETS components of interest under a range of environmental conditions encountered (1). Nicotine in the air and measurement of its metabolite cotinine in biologic fluids meet these criteria reasonably well. Interindividual variability exists among any set of biologic measurements. Such variability may limit the value of predictions based on measurements in individuals, but this variability is compensated for in studies of large numbers of subjects, as in epidemiologic studies. Supporting this conclusion is the observation that cotinine levels in nonsmokers have been positively correlated with the risks of some ETS-related health complications in children. The evidence presented in this review indicates that cotinine levels provide valid and quantitative measures of average ongoing human ETS exposure over time. The main limitation in using cotinine is that it does not provide a measure of long-term ETS exposure, a measure that would be most useful for epidemiologic studies.
Other biomarkers, particularly 4-ABP-Hb adducts, PAH-DNA or albumin adducts, and urine levels of metabolites of the nicotine-derived nitrosoamine NNK more directly reflect exposure to carcinogens in tobacco smoke than cotinine. Some of these biomarkers are better markers of long-term exposure, but available data indicate that they lack adequate sensitivity or specificity to be useful as quantitative biomarkers of ETS exposure. Whether levels of these biomarkers are useful in predicting disease risk remains to be determined.