We analyzed biomarker concentrations by tobacco use for adult NHANES participants from 1999-2012. NHANES is a health and examination survey that uses a complex multistage design to obtain a nationally representative sample of the US civilian non-institutionalized population (20). NHANES has been conducted on a continuous basis by the National Center for Health Statistics (NCHS) since 1999 and surveys approximately 10,000 participants of all ages in each two-year cycle. Survey participants complete health interviews in their homes that include a cigarette smoking history questionnaire for adults aged 20 years and older. Participants then complete an additional questionnaire on recent tobacco use including smokeless tobacco in a Mobile Examination Center (MEC), where they also receive a medical examination that includes the collection of biospecimens such as urine and blood.

We analyzed biomarker concentrations among the 38,024 adults aged 20 years and older who participated in NHANES between 1999 and 2012. We excluded 736 survey participants who reported use of tobacco or nicotine products other than cigarettes or chewing tobacco or snuff (i.e., cigars, snuff, pipes, nicotine replacement therapy products or any product containing nicotine) during the past five days as well as 5318 participants who did not provide information on past five day tobacco use. We then categorized study participants into four mutually exclusive groups based on their reported cigarette and smokeless tobacco use: 1) “non-tobacco users” reported having smoked fewer than 100 cigarettes in their lives and not having used cigarettes, chewing tobacco, or snuff in the past five days, 2) “smokeless tobacco users” reported using chewing tobacco or snuff in the past five days and currently not using cigarettes at all (228 smokeless tobacco users who reported being former cigarette smokers, having smoked at least 100 cigarettes but currently not smoking at all, were excluded from the analysis for cadmium due to its long biologic half-life, which can be upwards of 10 years (21)), 3) “cigarette smokers” reported having smoked at least 100 cigarettes in their lives and currently smoking every day or some days and not having used chewing tobacco of snuff in the past five days, and 4) “dual cigarette and smokeless tobacco users” reported having smoked at least 100 cigarettes in their lives, currently smoking every day or some days, and having used chewing tobacco or snuff in the past five days. We did not include former cigarette smokers who had not used chewing tobacco of snuff in the past five days in the analysis. The analysis included a total of 23,684 participants.

The biomarkers of exposure selected for this analysis were chosen due to their relevance to tobacco exposure and health outcomes. Cotinine is the primary proximate metabolite of nicotine (22, 23). NNAL is a metabolite of the TSNA 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), which has been identified as a known human carcinogen by the International Agency on Research on Cancer (24, 25). NNK itself is formed from the nitrosation of nicotine or the related alkaloid pseudooxynicotine (26). Lead, cadmium, mercury, and arsenic are elements known to have toxic effects that can be found in tobacco products as well as in other environmental sources (27-30). CYMA is a metabolite of acrylonitrile and a selective biomarker of exposure to smoke (31). Urinary arsenic concentrations were available for 2003-2012 NHANES participants, urinary NNAL concentrations were available for 2007-2012 NHANES participants, and urinary CYMA concentrations were available for 2005-2006 and 2011-2012 survey participants. Data for other biomarkers were available from 1999-2012.

The analytical methods used to obtain these data are available in NHANES documentation (20). Serum cotinine was measured by an isotope dilution - liquid chromatography / atmospheric pressure chemical ionization tandem mass spectrometry process. The half-life of cotinine is 15-20 hours, and its availability in blood, urine, and saliva makes it a commonly used biomarker of recent nicotine exposure (23, 32). Urinary total NNAL was measured using liquid chromatography linked to tandem mass spectrometry. The half-life of NNAL has been estimated to be 10–18 days (33). Blood cadmium, lead, and total mercury were measured using inductively coupled plasma mass spectrometry. Urinary total arsenic was measured using liquid chromatography coupled to plasma dynamic reaction cell mass spectrometry. Urinary CYMA was measured using liquid chromotography coupled with electro spray tandem mass spectrometry. For concentrations below the limit of detection (LOD), a value equal to the LOD divided by the square root of two was used in analysis.

NHANES participants reported information on sex, age, race/ethnicity, and educational attainment. Race/ethnicity was subsequently categorized as non-Hispanic White, non-Hispanic Black, Mexican-American, other Hispanic, and other race including multi-racial. Educational attainment was categorized as less than high school graduate or equivalent, high school graduate or equivalent, and more than high school graduate or equivalent. Body mass index (BMI) for survey participants was calculated as kg/m² from their measured height and weight as a continuous variable.

Demographic and tobacco use variables were characterized using means for continuous variables and percentages for categorical variables. Biomarker concentrations were log-transformed for the analysis to minimize the effects of skewness in the data on estimates, and geometric means of observed biomarker concentrations by tobacco use category were calculated. Univariate and multivariate linear regression analysis were also used to analyze the relationship between biomarkers of exposure and tobacco use category, adjusting for sex, age, race/ethnicity, educational attainment, and BMI. Geometric mean ratios and 95% confidence intervals (CIs), from these analyses were calculated by exponentiating the estimated coefficients and their standard errors. Geometric mean biomarker concentrations were also calculated for cotinine and NNAL by two-year NHANES survey cycle and plotted by time. Box plots were also created to show the unweighted distribution of cotinine and NNAL concentrations for smokeless tobacco users who reported having used chewing tobacco or snuff on 1-2, 3-4, or 5 of the past days; the small number of NHANES participants who reported using both products in the past five days (n=4) were excluded from this particular analysis. All analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC), and all figures were constructed using R version 3.0.2 (R Core Team, Vienna, Austria). Analyses were conducted using the MEC sample weights with the exception of analyses for arsenic, which were conducted with environmental subsample weights, and analyses for CYMA, which were conducted with the 2005-2006 volatile organic compounds subsample weights and 2011-2102 smoking subsample weights. Analyses were conducted taking into account the complex survey design information on survey strata and primary sampling units provided by NCHS.

Table 1 presents demographic and tobacco use information for the NHANES study participants according to tobacco use status. Of the 23,684 individuals, 488 were exclusive smokeless tobacco users, 6791 were exclusive cigarette smokers, 92 were dual cigarette and smokeless tobacco users, and 16,313 were non-tobacco users. Of the smokeless tobacco users, 309 individuals reported using chewing tobacco, 175 reported using snuff , and 4 reported using both chewing tobacco and snuff in the past five days (data not shown). Smokeless tobacco and dual users were overwhelmingly male, at 94.7% (95% CI = 92.1%, 97.2%) and 99.4% (98.6%-100.0%) respectively. Dual users tended to be younger than members of other tobacco use groups with a mean age of 33.1 years (95% CI = 30.3, 35.8). Smokeless tobacco and dual users were also more likely to be non-Hispanic whites than members of other tobacco use groups at 88.7% (95% CI = 85.3%-92.1%) and 94.2% (90.1%-98.2%) respectively. The estimated mean number of cigarettes that dual users smoked on days that they smoked cigarettes in the past five days was less than the estimated mean for cigarette smokers, at 11.9 compared with 14.8 cigarettes, but the difference was not statistically significant (p = 0.071). There was a slight difference in the average number of days that cigarette smokers and dual users had smoked cigarettes in the past five days at 4.6 and 4.2 days respectively (p = 0.042). Smokeless tobacco users tended to have consistently used smokeless tobacco in the past five days, with an average of 4.2 days (95% CI = 4.1-4.4) having used chewing tobacco for chewing tobacco users and an average of 4.3 days (95% CI = 4.1-4.5) having used snuff for snuff users.

Table 2 presents geometric mean biomarker concentrations by tobacco use status. Mean serum cotinine concentrations were higher for smokeless tobacco users (178.9 ng/mL, 95% CI = 145.5, 220.0) than for cigarette smokers (130.6 ng/mL, 95% CI = 122.3, 139.6). Cotinine concentrations for dual users (184.1 ng/mL, 95% CI = 132.4, 256.0) were similar to concentrations for smokeless tobacco users. Mean urinary NNAL concentrations were higher for smokeless tobacco users (583.0 pg/mg creatinine, 95% CI = 445.2, 763.5) and dual users (430.3 pg/ mg creatinine, 95% CI = 284.8, 650.1) than for cigarette smokers (217.6 pg/mg creatinine, 95% CI = 193.0, 245.2). NNAL concentrations were also consistently higher for smokeless tobacco users compared with cigarette smokers when analyzed in terms of concentration categories. For example, 37.5% (95% CI = 28.6, 46.4%) of smokeless tobacco users had NNAL concentrations below 500 pg/mg creatinine and 62.2% (95% CI = 54.6%, 69.7%) had concentrations below 1000 pg/mg creatinine. In contrast, 74.2% (95% CI = 70.4%, 78.1%) of smokers had NNAL concentrations below 500 pg/mg creatinine and 94.1% (95% CI = 92.4%, 95.8%) had concentrations below 1000 pg/mg creatinine. Mean NNAL concentrations were generally comparable for chewing tobacco (402.3 pg/mg creatinine, 95% CI = 294.3, 549.9) and snuff (463.7 pg/mg creatinine, 95% CI = 315.6, 681.2) users. Exclusion of the relatively small proportion of current cigarette smokers who reported not having smoked cigarettes in the past five days (n=301 of 6791) in sensitivity analysis produced similar results. For example, the geometric mean concentration of cotinine for the remaining smokers was 156.7 ng/mL (95% CI = 150.3, 163.4) and the mean concentration for NNAL was 247.3 pg/mg creatinine (95% CI = 225.4, 271.3).

Mean concentrations of blood lead were higher among smokeless tobacco users (1.76 μg/L, 95% CI = 1.62, 1.91), dual users (1.76 μg/L, 95% CI = 1.55, 2.00), and cigarette smokers (1.76 μg/L, 95% CI = 1.71, 1.81) compared with non-tobacco users (1.18 μg/L, 95% CI = 1.16, 1.21). Mean concentrations of blood cadmium, blood mercury, and urinary arsenic were not elevated among smokeless tobacco users compared with non-tobacco users.

Mean CYMA concentrations were higher among cigarette smokers (117.3 ng/mg creatinine, 95% CI = 103.1, 133.4) and dual users (35.4 ng/mg creatinine, 95% CI = 2.1, 606.8) but not among smokeless tobacco users (2.21 ng/mg creatinine, 95% CI = 1.11, 4.39) compared with non-tobacco users (1.47 ng/mg creatinine, 95% CI = 1.37, 1.58).

Table 3 presents results from multivariate regression analyses conducted to analyze whether tobacco use status was associated with higher biomarker concentrations, adjusting for demographic and socioeconomic factors.

After adjustment for age, sex, race/ethnicity, education, and BMI, smokeless tobacco users, cigarette smokers, and dual users had increased geometric mean ratios for serum cotinine compared with non-tobacco users. Smokeless tobacco users also had increased geometric mean ratios compared with cigarette smokers (p = 0.04). Smokeless tobacco users, cigarette smokers, and dual users also had increased geometric mean ratios for urinary NNAL compared with non-tobacco users, and smokeless tobacco users and dual users had increased geometric mean ratios compared with cigarette smokers.

Smokeless tobacco users, along with cigarette smokers and dual users, had increased geometric mean ratios for blood lead compared with non-tobacco users. Smokeless tobacco users did not have increased geometric mean ratios for any of the other biomarkers.

Figure 1 presents geometric mean serum cotinine and urinary NNAL concentrations for cigarette smokers and smokeless tobacco users over time. Cotinine concentrations for smokers and smokeless users were relatively consistent over time, although estimates for the smaller number of smokeless tobacco users showed more variability. Tests of trend for cotinine concentrations produced a p-value of 0.895 for smokers and 0.403 for smokeless tobacco users. Mean NNAL concentrations for smokers were relatively consistent from 2007-2008 to 2011-2012 but declined dramatically for smokeless tobacco users from a geometric mean of 1013.7 pg/mg creatinine (95% CI = 738.9, 1390.8, n = 81) in 2007-2008 to 328.6 pg/mg creatinine (95% CI = 164.7, 655.6, n = 53) in 2011-2012. Tests of trend for NNAL concentrations produced a p-value of 0.943 for smokers and 0.003 for smokeless tobacco users.

Figure 2 presents boxplots showing the distribution of cotinine and NNAL concentrations for chewing tobacco and snuff users by frequency of use in terms of the number of days that they had used the product in the past five days. The figure shows that concentrations consistently increased with number of days of use. Tests of trend for the association between biomarker concentrations and days using the product produced p-values less than 0.0001 for chewing tobacco and snuff for cotinine and equal to 0.003 for chewing tobacco and 0.03 for snuff for NNAL.