Association of urinary biomarkers of tobacco exposure with lung cancer risk in African American and White cigarette smokers in the Southern Community Cohort Study

Abstract

Background:

After accounting for smoking history lung cancer incidence is greater in African Americans than Whites. In the Multiethnic Cohort, total nicotine equivalents (TNE) are higher in African Americans than Whites at similar reported cigarettes per day (CPD). Greater toxicant uptake per cigarette may contribute to the greater lung cancer risk of African Americans.

Methods:

In a nested case-control lung cancer study within the Southern Community Cohort, smoking-related biomarkers were measured in 259 cases and 503 controls (40% White, 56% African American). TNE, the trans-3-hydroxycotinine:cotinine ratio, 4-(methylnitrosamino)-1–3-(pyridyl)-1-butanol (NNAL), mercapturic acid metabolites of volatile organic compounds, phenanthrene metabolites, cadmium, and (Z)-7-(1R,2R,3R,5S)-3,5-dihydroxy-2-[(E,3S)-3-hydroxyoct-1-enyl]cyclopenyl]hept-5-enoic acid (8-iso-PGF2α) were quantified in urine. Unconditional logistic regression was used to estimate the odds ratios and 95% confidence intervals for each biomarker and lung cancer risk.

Results:

TNE, NNAL and cadmium were higher in cases than controls (adjusted for age, race, sex, body mass index (BMI) and CPD). Among cases, these levels were higher in African Americans compared to Whites. After accounting for age, sex, BMI and pack-years, a one-SD increase in log-TNE (OR=1.30; 95% CI: 1.10–1.54) and log-NNAL (OR=1.27; 95% CI: 1.03–1.58 with TNE adjustment) were associated with lung cancer risk. In this study, where NNAL concentration is relatively high, the association for log-TNE was attenuated after adjustment for log-NNAL.

Conclusion:

Smoking-related biomarkers provide additional information for lung cancer risk in smokers beyond smoking pack-years.

Impact:

Urinary NNAL, TNE and cadmium concentrations in current smokers, particularly African American smokers, may be useful for predicting lung cancer risk.

Introduction:

Lung cancer is the second most common cancer in men and women and the number one cause of cancer deaths in the United States (1, 2). While cigarette smoking is the primary risk factor for this disease, smoking-related lung cancer risk is not equal across populations (3). After accounting for population variations in self-reported smoking history, African American smokers have a greater incidence of lung cancer than White smokers (4). To better understand factors contributing to this racial disparity, the Multiethnic Cohort (MEC) study examined differences in nicotine metabolism and biomarkers of internal smoking dose and tobacco carcinogens across racial and ethnic groups (5, 6). Consistent with the direction of their increased risk of lung cancer, relative to Whites, African Americans had higher urinary concentrations of total nicotine equivalents (TNE) per cigarette smoked. TNE, a biomarker of internal smoking dose, which is the sum of urinary nicotine and its major metabolites, accounts for greater than 85% of the nicotine uptake (7, 8). African American smokers also excreted higher levels of several tobacco carcinogen biomarkers per cigarette smoked (5, 9–11). Among these is total 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL), a metabolite of the tobacco specific lung carcinogen, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (10). Urinary cadmium (Cd) levels were higher in African American smokers compared to Whites in the MEC, after adjustment for TNE and smoking duration (11). These data provide support for the hypothesis that the increased risk of lung cancer observed among African Americans relative to Whites is due in part to the greater uptake of carcinogens per cigarette smoked.

A key determinant of internal smoking dose is cytochrome P450 2A6 (CYP2A6)-catalyzed nicotine metabolism (12). In most smokers, CYP2A6 catalyzed 5'-oxidation, which leads to the formation of cotinine, accounts for 70 to 80% of nicotine metabolism. Cotinine is further metabolized to 3'-hydroxycotinine (3-HCOT), a reaction almost exclusively catalyzed by CYP2A6 (12). Several studies have confirmed the relationship between CYP2A6 activity (measured as the ratio of 3-HCOT to cotinine) or CYP2A6 genotype to the number of cigarettes smoked and the uptake of nicotine per cigarette (12, 13). In the MEC, CYP2A6 activity, and in the Shanghai Cohort Study and the Singapore Chinese Health Study CYP2A6 genotype was related to smoking dose and lung cancer risk (6, 14–16).

CYP2A6 genetic variants coding for low or no activity enzyme are more common in populations of Asian descent (12). A significant percentage of Japanese Americans have relatively low levels of CYP2A6 activity, resulting in lower nicotine intake, lower TNE levels, and a lower risk of lung cancer in this group relative to populations with fewer low activity individuals, such as Whites and African Americans (12, 17). A number or reduced activity CYP2A6 variants occur uniquely in individuals of African ancestry, and smokers who carry these variants metabolize nicotine less efficiently (18–21). In the MEC, African American smokers with reduced activity CYP2A6 alleles, have lower urinary concentrations of TNE and NNAL than African Americans who do not carry these alleles (17). However, as a group, African American smokers in the MEC have higher concentrations of TNE and NNAL than do Whites (10, 22).

Here we utilize a subset of Southern Community Cohort Study (SCCS) data, from a nested case-control study of lung cancer, to confirm our MEC findings that TNE levels and other smoking-related biomarkers are higher in African Americans than Whites at similar reported cigarettes per day (CPD) as well as evaluate whether urinary biomarker for internal smoking dose (TNE) and several other tobacco-related biomarkers to lung cancer risk.

Methods

Study population

Study participants are from a case-control study of lung cancer nested within the SCCS. This cohort was designed to evaluate disparities in cancer and other chronic diseases in primarily African American and Whites from the US South (23). Details for SCCS and its recruitment have been previously published (23, 24). In brief, recruitment occurred between March 2002 and September 2009; 85,806 adults aged 40–79 years residing in 12 southern states of the U.S. (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia) were enrolled (23–25). Two-thirds of the participants self-reported their race as African American and the remainder predominantly as non-Hispanic White. Recruitment primarily took place at community health centers (~85%), institutions, which provide primary health services in medically underserved areas. Approximately 15% of participants were recruited from mailings to an age-, sex-, and race-stratified random sample of the general population.

Approximately 20.8% of cohort participants provided baseline spot urine samples. Among this subcohort, 43% were current smokers (smoking at least 100 cigarettes in their lifetime) at the time of urine collection. Cases were ascertained via linkage with the 12 state cancer registries in the study area and/or the National Death Index mortality records (25). Among current smoking participants with available urine samples, 264 incident lung cancers (International Classification of Diseases-Oncology codes C340–C349) occurred during study follow-up (end of follow-up December 31, 2015). Two controls for every case were selected at random among a subcohort of SCCS current smoking participants with urine samples, matched on age (± 2 years), sex, self-reported race and recruitment site. The study includes 259 lung cancer cases and 503 controls after exclusion for participants missing smoking history (n=2), data for TNE or CYP2A6 enzymatic activity (n=21) or BMI (n=7). This study was conducted in accordance with recognized ethical guidelines (e.g., Declaration of Helsinki, CIOMS, Belmont Report, U.S. Common Rule). Approval was obtained by the Institutional Review Boards of the Vanderbilt University, University of Hawaii, and University of Southern California. Written informed consent was obtained from all study participants.

Epidemiologic data

Trained study interviewers administered a computer-assisted personal interviews to collect baseline data on participants’ demographic characteristics and potential cancer risk factors, including personal and family medical history and tobacco use history. Cigarette smokers were defined as those who reported smoking at least 100 cigarettes in their lifetime. Current smokers were defined as those who answered “Yes” to “Do you smoke now?”. Menthol smokers were defined as those who answered “Yes” to “Are the cigarettes you usually smoke menthol?” Pack-years of smoking were computed by multiplying packs (reported CPD by 20) by the number of years smoked (derived from the age reported starting and quitting smoking for former smokers, or the age starting smoking and current age for current smokers).

Urinary Biomarkers of smoking dose and nicotine metabolism.

The urinary biomarkers of nicotine metabolism and tobacco smoke toxicants were quantified at the University of Minnesota as previously described: TNE (the molar sum of nicotine N-oxide, total nicotine, total cotinine, and total 3-HCOT)(26), total NNAL(27), (Z)-7-[1R,2R,3R,5S)-3,5-dihydroxy-2-[(E,3S)-3-hydroxyoct-1-enyl] cyclopentyl]hept-5-enoic acid (8-iso-PGF2α), a biomarker of oxidative damage(28), cyanoethylmercapturic acid (CEMA), S-phenylmercapturic acid (SPMA), 3-hydroxypropyl mercapturic acid (3-HPMA) and 2-hydroxypropyl mercapturic acid (2-HPMA) (29–32), phenanthrene tetraols (PheT),1-. 2- and 3-hydroxyphenanthrene (1-PheOH, 2-PheOH and 3-PheOH)(9, 27, 33) and Cd(11). “Total” refers to the compound and its glucuronide conjugate(s). CEMA is a metabolite of acrylonitrile (a toxicant found in tobacco smoke but at very low concentrations in the general human environment) (34). SPMA, 3-HPMA and 2-HPMA are metabolites of benzene, acrolein, and propylene oxide, respectively. Internal laboratory quality control (QC) urinary samples from smokers were used, 1 to 5 QC samples were included per 96 well plate. The coefficients of variation for the biomarker levels of QC samples from smokers ranged from 3.2 % for cotinine to 12% for 1-PheOH and 2-HPMA (Supplemental Table 1).

CYP2A6 enzymatic activity was quantified as the urinary ratio of total 3-HCOT/cotinine(22). Creatinine was analyzed using a colorimetric microplate assay (CRE34-K01) purchased from Eagle Bioscience (https://eaglebio.com/product/creatinine-microplate-assay-kit/).

Statistical methods.

Descriptive statistics were generated for cases and controls, separately. To examine racial differences in smoking-related biomarkers (TNE, CYP2A6 enzymatic activity ratio, total NNAL, CEMA, SPMA, 3-HPMA, 2-HPMA, PheT, PheOHs, Cd and 8-iso-PGF_2α) covariate-adjusted geometric means were computed for each racial group at the mean covariate vector. All biomarkers were transformed by the natural log. Values presented in the tables are back-transformed to their natural scale as geometric means (and 95% confidence intervals) for ease of interpretation. We used two multivariable linear models: Model 1) adjusted for the following predictors: age, sex, BMI (log transformed), CPD (log transformed) and racial group (White, African American, other; when not stratified by race); and Model 2) additionally adjusted for TNE, except that for TNE, which was adjusted for CYP2A6 activity. These adjustments were done to assess whether associations remained independent of internal smoking dose (TNE) or nicotine metabolism (CYP2A6), respectively. Models were run both with and without creatinine adjustments. A creatinine adjustment may help to normalize variation across urinary metabolite concentrations due to spot urine collection. However, it is important to note that creatinine levels are also affected by several other factors including sex, muscle mass, age, and racial group. The Third National Health and Nutrition Examination Survey (NHANES) found that African Americans have almost 2-fold higher creatinine levels than Whites (35). We confirmed these differences in the MEC data and demonstrated that expressing urinary biomarker concentrations per mg creatinine may lead to an inflated denominator resulting in the artificial appearance that African Americans have lower smoking biomarkers levels (10). Thus, our primary findings are unadjusted for creatinine. Analyses within racial group were conducted with and without adjustment for creatinine.

To estimate the odds ratios and 95% confidence intervals for each smoking-related biomarker and risk of lung cancer, we conducted unconditional logistic regression. Here, the model was adjusted for age, sex, race, BMI (log transformed) and self-reported pack-years (Model 1). Model 1 was further adjusted for urinary TNE (Model 2) to assess the independent effects of internal smoking dose beyond that provided by self-reported measures of smoking. For TNE, Model 2 was adjusted for the ratio of urinary total 3-HCOT:cotinine. To compare the lung cancer associations across the various smoking-related biomarkers, we standardized each biomarker by dividing each log-transformed biomarker by its respective log-transformed SD (Supplemental Table 2).

To account for multiple testing, we used the approach proposed by Cheverud(36). The correlation matrix of all 12 log-transformed biomarkers (TNE, Total NNAL, total 3HCOT / cotinine, CEMA, 2-HPMA, 3-HPMA, SPMA, 8-isoPGF2α, Cd, PheT, Ratio PheT/sum 1,2,3-PheOH, Sum 1-,2-, and 3-PheOH) indicated that there were three effective number of independent tests. Thus, a Bonferroni corrected p-value of 0.017 (p=0.05/3) was used. The correlation across biomarkers can be found in Supplemental Table 3. Furthermore, each Table present the adaptive FDR p-values as an alternative. To account for potential reduction in smoking due to lower lung function or early symptoms, we conducted a sensitivity analysis restricting to cases diagnosed >2 years of urine collection.

In addition, conditional logistic regression was run among 239 sets (1 to 2 matching; n=239 cases and 478 controls); associations were consistent with those from the unconditional logistic regression. Therefore, to retain all 761 study participants, the results for unconditional regression, adjusting for match covariates were presented.

Data availability

Data are available to researchers who submit a proposal via the SCCS online submission system (https://www.southerncommunitystudy.org/for-researchers.html).

Results

The characteristics of the 259 lung cancer cases and 503 controls from the SCCS who were current smokers at baseline, included in this study are presented in Table 1. As a result of matching criteria, the distribution of age, race and sex were similar across the cases and controls. The lung cancer cases and their corresponding controls had a median age of cohort enrollment of 53 and 54, respectively. Approximately half of the participants were female; 56% identified as African American, and 40% as White, the remaining 4% belonging to other racial and ethnic groups. In comparison to controls, lung cancer cases had lower BMI and reported smoking a greater number of CPD and greater pack-years.

Table 1.

Characteristics of Southern Community Cohort study (SCCS) participants at time of enrollment (baseline) by lung cancer case status

	Cases				Controls
	n	Median	Q1a	Q3	n	Median	Q1a	Q3
Age at Enrollment (years)	259	54	49	60	503	53	48	59
Age at lung cancer diagnosis (years)	259	58	54	65	0
BMIb ((kg/m2)	259	25.2	22.2	29.1	503	27	23	31.3
Cigarettes per day	259	18	10	20	503	10	7	20
Packyears a	258	30	16.5	47	503	20.5	10.8	36
Creatinine (mg/dL)	259	101	50.8	162	503	117	64	176
Sex		Percent				Percent
Female	130	50.2%			250	49.7%
Male	129	49.8%			253	50.3%
Race/Ethnicity
White	104	40.2%			200	39.8%
African American	145	56.0%			281	55.9%
Othersd	10	3.9%			22	4.4%
Menthol Smokers
Yes	128	49.4%			287	57.2%
No	131	50.6%			215	42.8%
Histologic Type
Non-Small Cell Lung Cancer	194	74.9%
Adenocarcinoma	100	38.6%
Large cell lung cancer	8	3.1%
Squamous cell carcinoma	57	22.0%
Other	29	11.2%
Small Cell Lung Cancer	41	15.8%
Not-otherwise Specified	8	3.1%
Histology Missing	16	6.2%

^aQ1, Q3 are the 25^th and 75^th percentile.

^bBMI, body mass index

^cPackyear data missing for one participant

^dOthers includes Latinos, American Indian/Alaskan Natives, Other and mixed racial groups.

There were some unique differences in characteristics between African Americans and Whites (Supplemental Table 4). The median age of enrollment was younger in African American cases and controls (53 and 52 years) compared to White cases and controls (56 and 55 years). African Americans reported fewer CPD and had lower pack-years (cases: 10 CPD, 21 pack-years; controls: 10 CPD,16.9 pack-years) compared to Whites (cases: 20 CPD, 41.5 pack-years and controls: 20 CPD, 33 pack-years). Additionally, African Americans had a 2-fold higher mean urinary creatinine concentration than Whites. Lastly, African Americans were more likely than Whites (44% vs 32%) to be diagnosed with adenocarcinoma of the lung.

Despite smoking half the number of CPD, after adjustment for age, sex, BMI and CPD, the geometric mean TNE level for African Americans in the study were modestly higher than those of Whites (67.1 vs 58.9 nmol/ml, p=0.05, Supplemental Table 5). However, among controls the geometric mean TNE levels (adjusted for CPD, age, sex, BMI, and CYP2A6 activity) were not different between African Americans and Whites (55.7 versus 59.4 nmol/ml, p=0.42). Additionally, CEMA, 3-HPMA, 8-iso-PGF2α, Cd and the sum of the phenanthrene phenols levels (1-, 2-, and 3-PheOH) were significantly higher in African Americans (all and controls-only) than Whites, with or without adjustment for TNE (p’s<0.01; Supplemental Table 5). Biomarker data is not corrected for creatinine unless comparing differences within race. African Americans have higher creatinine levels than Whites (35, 37, 38) and adjusting urinary biomarker concentrations for creatinine when comparing levels across these groups is problematic and may misrepresent the data.

The geometric mean biomarker concentrations, stratified by cases and controls, are presented in Table 2. The level of TNE in lung cancer cases was significantly higher than in controls (71.0 vs 60.3 nmol/mL, p=0.009). The difference was modestly attenuated when adjusted for CYP2A6 activity (67.3 versus 59.2 nmol/ml, p=0.020). Among the other biomarkers, total NNAL was significantly higher in cases compared to controls (p=0.005). After adjustment for TNE (model 2), NNAL remained higher in cases (1.91 pmol/ml NNAL versus 1.69 pmol/ml, p=0.02).

Table 2.

Geometric means of cigarette smoking-related urinary biomarkers by lung cancer status ^a

	Cases (n=259)	Controls (n=503)	P-valueb	Adaptive FDR p-value
	Geometric means, (95% CI)	Geometric means, (95% CI)
TNE
Model 1	71.0 (62.3–80.8)	60.3 (54.1–67.2)	0.009	0.06
Model 1 + CYP2A6 activity	67.3 (59.8–75.6)	59.2 (53.7–65.2)	0.02	0.07
CYP2A6 enzymatic activity
Model 1	5.39 (4.73–6.14)	4.97 (4.46–5.54)	0.20	0.47
Model 2	5.24 (4.66–5.89)	5.06 (4.59–5.58)	0.53	0.53
Total NNAL
Model 1	1.99 (1.73–2.3)	1.65 (1.46–1.85)	0.005	0.06
Model 2	1.91 (1.71–2.13)	1.69 (1.54–1.85)	0.02	0.07
CEMA
Model 1	666 (578–769)	595 (529–670)	0.096	0.34
Model 2	633 (575–697)	614 (568–665)	0.51	0.51
3-HPMA
Model 1	4212 (3529–5027)	3749 (3267–4341)	0.17	0.47
Model 2	3977 (3488–4534)	3886 (3485–4332)	0.71	0.71
2-HPMA
Model 1	356 (304–417)	350 (307–400)	0.83	0.89
Model 2	341 (299–389)	360 (323–401)	0.38	0.42
SPMA
Model 1	9.80 (7.99–12.0)	9.22 (7.79–10.9)	0.53	0.74
Model 2	9.25 (7.84–10.9)	9.56 (8.33–11.0)	0.68	0.68
8-isoPGF2α
Model 1	1.22 (1.05–1.41)	1.22 (1.08–1.37)	0.98	0.98
Model 2	1.17 (1.03–1.33)	1.24 (1.12–1.38)	0.33	0.40
Cadmium
Model 1	0.724 (0.63–0.83)	0.63 (0.56–0.71)	0.04	0.19
Model 2	0.696 (0.62–0.78)	0.64 (0.59–0.71)	0.16	0.25
PheT
Model 1	2.33 (1.98–2.73)	2.33 (2.04–2.66)	>0.99	>0.99
Model 2	2.22 (1.95–2.53)	2.40 (2.15–2.67)	0.23	0.32
Ratio PheT/sum 1,2,3-PheOH c
Model 1	0.594 (0.54–0.66)	0.58 (0.53–0.63)	0.63	0.74
Model 2	0.590 (0.53–0.65)	0.58 (0.54–0.63)	0.81	0.81
Sum 1-,2-, and 3-PheOH c
Model 1	3.86 (3.43–4.35)	4.01 (3.63–4.43)	0.51	0.74
Model 2	3.73 (3.41–4.09)	4.09 (3.8–4.41)	0.03	0.08
1-PheOH c
Model 1	1.51 (1.34–1.71)	1.56 (1.41–1.73)	0.62	0.74
Model 2	1.46 (1.33–1.61)	1.59 (1.47–1.72)	0.07	0.13
2-PheOH c
Model 1	0.787 (0.7–0.89)	0.814 (0.74–0.9)	0.56	0.74
Model 2	0.762 (0.69–0.84)	0.830 (0.77–0.9)	0.06	0.13
3-PheOH c
Model 1	1.46 (1.29–1.66)	1.54 (1.39–1.71)	0.35	0.70
Model 2	1.41 (1.29–1.55)	1.58 (1.46–1.71)	0.01	0.07

Abbreviations. CI, confidence intervals; TNE, total nicotine equivalents; NNAL, 4-(methylnitrosamino)-1–3-pyridyl)-1-butanol; CEMA, cyanoethylmercapturic acid, 3-HPMA, 3-hydroxypropylmercapturic acid; 2-HPMA, 2-hydroxypropylmercapturic acid, SPMA, S-phenylmercapturic acid; Cd, cadmium; 8-iso-PGF_2α, (Z)-7-(1R,2R,3R,5S)-3,5-dihydroxy-2-[(E,3S)-3-hydroxyoct-1-enyl]cyclopenyl]hept-5-enoic acid, PheT, phenanthrene tetraol; 1-,2-,3-PheOH, 1-, 2- and 3-hydroxyphenanthrene; PheT/PheOH, a proposed biomarker of metabolic activation of polycyclic aromatic hydrocarbons (PAH).

^aTNE are expressed in nmol/ml, the concentrations of all other biomarkers are in pmol/mlModel 1: Adjusted for age, sex (male/female), race/ethnicity (White, African American, other), log BMI, log CPDModel 2: Model 1 + TNECYP2A6 enzymatic activity = ratio of total trans 3’-hydroxycotinine (3-HCOT)/cotinine

^bComparisons between cases and controls

^cCases, n=255 and Controls, n=501 for PheT ratio, 1-,2-, and 3-PheOH

After stratification of the study population by race, the biomarker concentrations were compared between cases and controls separately for African Americans and for Whites with adjustment for creatinine, as well as age, sex, BMI, and CPD (Supplemental Table 6). TNE levels were significantly higher in cases compared to controls for both African Americans (79.2 vs 61.0 nmol/ml, p<0.0001) and Whites (71.9 vs 56.0 nmol/ml, p<0.0001); adjustment for CYP2A6 activity had very little effect. Cd levels in both racial groups and NNAL levels in African Americans were higher in cases than controls, with TNE adjustment (Model 2, Supplemental Table 6). In Whites, 3-HPMA and 8-iso-PGF_2α were modestly higher in cases compared to controls (both p’s=0.04).

The effect of menthol cigarette use on TNE, NNAL and CEMA levels are presented in Table 3. After adjustment for age, sex, BMI, and CPD, geometric mean. NNAL levels were significantly lower in menthol cigarette users (1.52 pmol/ml) compared to non-menthol cigarette users (1.95 pmol/ml, p= 0.002). The difference remained significant after further adjusting for TNE, and was observed in African American but not White smokers (model 2). Among all study participants and African American smokers CEMA levels were non-significantly lower in menthol versus non-menthol cigarette users. None of the other biomarkers varied between menthol and non-menthol cigarette smokers, for either African Americans or Whites.

Table 3.

Geometric means of urinary total nicotine equivalents (TNE), NNAL, CEMA by menthol use^a

	Overall				African Americans					Whites
	no menthol (n=346)	yes menthol (n=415)				no menthol (n=85)	yes menthol (n=340)				no menthol (n=239)	yes menthol (n=65)
	Geometric means (95% CI)	Geometric means (95% CI)	p	Adaptive FDR p		Geometric means (95% CI)	Geometric means (95% CI)	p	Adaptive FDR p		Geometric means (95% CI)	Geometric means (95% CI)	p	Adaptive FDR p
TNE
Model 1	67.3 (59.8–75.8)	59.1 (51.9–67.2)	0.075	0.075		75.7 (63.5–90.3)	63.7 (58.4–69.5)	0.08	0.08		65.0 (58.4–72.3)	58.5 (47.5–72.1)	0.37	0.37
Model 1 + log CYP2A6 activity	64.7 (58.1–72.0)	58.1 (51.7–65.2)	0.10	0.12		74.1 (63.1–87.0)	63.4 (58.6–68.7)	0.09	0.12		65.8 (59.9–72.2)	60.6 (50.5–72.8)	0.43	0.43
Total NNAL
Model 1	1.95 (1.72–2.22)	1.52 (1.32–1.75)	0.002	0.003		2.24 (1.86–2.68)	1.66 (1.52–1.82)	0.004	0.004		1.85 (1.64–2.09)	1.54 (1.21–1.95)	0.17	0.17
Model 2	1.88 (1.71–2.08)	1.61 (1.45–1.8)	0.01	0.03		2.05 (1.78–2.35)	1.65 (1.54–1.77)	0.006	0.03		1.81 (1.64–1.99)	1.69 (1.4–2.04)	0.54	0.54
CEMA
Model 1	667 (585–759)	559 (485–644)	0.03	0.03		834 (691–1005)	675 (616–741)	0.05	0.05		582 (516–656)	476 (376–602)	0.13	0.13
Model 2	639 (586–697)	598 (544–657)	0.22	0.22		754 (663–859)	669 (628–714)	0.10	0.12		565 (522–611)	534 (458–623)	0.52	0.52

^aTNE are expressed in nmol/ml, and NNAL and CEMA in pmol/ml

Model 1: adjusted for age, sex (male/female), log BMI, log CPD, and additional adjusted for ethnicity when participants are not stratified by race

Model 2: Model 1 + TNE

The relationship of urinary biomarker levels to lung cancer risk are presented in Table 4. Among the biomarkers measured, TNE (OR=1.30; 95% CI: 1.10–1.54) and NNAL (OR=1.28; 95 CI: 1.08–1.50) were both significantly associated with lung cancer risk after adjustment for age, sex, race/ethnicity, BMI, and pack-years (model 1). Additional adjustment for CYP2A6 activity had no impact on the association of TNE with lung cancer risk. However, additional adjustment for log-NNAL resulted in a null association (OR=1.19; 95% CI: 0.91–1.55; p=0.20). The association of NNAL with lung cancer risk remained significant after adjustment for TNE (OR=1.27; 95% CI: 1.03–1.58; model 2). The CYP2A6 enzymatic activity ratio was not associated with lung cancer risk with or without adjustment for TNE. The effect for TNE and NNAL remained statistically significant for adenocarcinoma (n=100, OR=1.38; 95% CI: 1.05=1.81) but not in squamous cell carcinoma (n=57) and small cell lung cancer (n=41) (p’s>0.22; Supplemental Table 7). Associations were consistent even after restricting to cases diagnosed >2 years after study entry.

Table 4:

Association of cigarette smoking-related urinary biomarkers with lung cancer risk

Biomarkersa (SD of the respective log-biomarker)	Cases	Controls	Model 1b				Model 2c,d
			OR	(95%CI)	P	Adaptive FDR p	OR	(95%CI)	P	Adaptive FDR p
TNE (0.83)	258	503	1.30	(1.10–1.54)	0.002	0.03	1.31	(1.08–1.58) e	0.006	0.06
Total NNAL (0.89)	258	503	1.28	(1.08–1.50)	0.004	0.03	1.27	(1.03–1.58)	0.03	0.08
CYP2A6 activity (total 3HCOT / cotinine) (0.80)	258	503	1.11	(0.95–1.3)	0.20	0.43	1.04	(0.88–1.25)	0.63	0.63
CEMA (0.91)	258	503	1.17	(1.00–1.38)	0.06	0.18	1.09	(0.85–1.40)	0.48	0.48
2-HPMA (0.97)	258	503	1.02	(0.87–1.19)	0.83	0.83	0.90	(0.74–1.09)	0.28	0.33
3-HPMA (1.11)	258	503	1.14	(0.97–1.34)	0.11	0.29	1.05	(0.84–1.30)	0.69	0.69
SPMA (1.25)	258	503	1.05	(0.90–1.23)	0.54	0.69	0.94	(0.78–1.14)	0.54	0.54
8-isoPGF2α (0.91)	258	503	1.00	(0.85–1.17)	0.95	0.95	0.90	(0.75–1.08)	0.24	0.33
Cd (0.87)	258	503	1.18	(1.00–1.38)	0.05	0.18	1.11	(0.91–1.35)	0.30	0.33
PheT (1.00)	258	503	1.00	(0.86–1.18)	0.97	0.97	0.87	(0.72–1.06)	0.17	0.27
Ratio PheT/sum 1,2,3-PheOH (0.62)	254	501	1.04	(0.89–1.22)	0.63	0.69	1.015	(0.86–1.19)	0.86	0.86
Sum 1-,2-, and 3-PheOH (0.75)	254	501	0.95	(0.81–1.12)	0.55	0.69	0.78	(0.63–0.96)	0.02	0.07
1-PheOH (0.76)	254	501	0.96	(0.82–1.13)	0.64	0.69	0.81	(0.66–1.00)	0.05	0.09
2-PheOH (0.74)	254	501	0.96	(0.82–1.13)	0.63	0.69	0.81	(0.66–0.99)	0.04	0.09
3-PheOH (0.79)	254	501	0.93	(0.79–1.09)	0.38	0.69	0.74	(0.60–0.93)	0.01	0.06

^aAll urinary biomarkers were standardized using log transformation and dividing the individual value by the overall population SD of the log biomarker and therefore the OR corresponds to a per one-unit SD change in log biomarker level.

^bModel 1 includes age at urine collection, sex (male/female), race/ethnicity (White, African American, Other), log-BMI, pack-years

^cModel 2: Model 1 + TNE

^dFor TNE log CYP2A6 activity is adjusted for in model 2

^eFor TNE an additional adjustment for log-NNAL (OR=1.19; 95% CI: 0.91–1.55; p=0.20).

Abbreviations. CI, confidence intervals; TNE, total nicotine equivalents; NNAL, 4-(methylnitrosamino)-1–3-pyridyl)-1-butanol; CEMA, cyanoethylmercapturic acid, 3-HPMA, 3-hydroxypropylmercapturic acid; 2-HPMA, 2-hydroxypropylmercapturic acid, SPMA, S-phenylmercapturic acid; Cd, cadmium; 8-iso-PGF_2α, (Z)-7-(1R,2R,3R,5S)-3,5-dihydroxy-2-[(E,3S)-3-hydroxyoct-1-enyl]cyclopenyl]hept-5-enoic acid, PheT, phenanthrene tetraol; 1-,2-,3-PheOH, 1-, 2- and 3-hydroxyphenanthrene; PheT/PheOH, a proposed biomarker of metabolic activation of polycyclic aromatic hydrocarbons (PAH).

After stratification by race, TNE (OR=1.47; 95% CI: 1.12–1.92) and NNAL (OR=1.53; 95% CI: 1.10–2.13) were associated with lung cancer risk in African Americans (Table 5, model 1). Following additional adjustment for creatinine, the associations for TNE and NNAL with lung cancer risk increased (OR=2.1; 95% CI: 1.47–3.00 and OR=1.74; 95% CI: 1.23–2.45, respectively; model 2). In addition, Cd was associated with lung cancer risk after creatinine adjustment (OR=1.97; 95% CI: 1.34–2.91). Among Whites, after creatinine adjustment, TNE (OR=2.66; 95% CI: 1.61–4.41) and Cd (OR=2.32; 95% CI: 1.50–3.60), but not NNAL (OR=1.43; 95% CI: 1.01–2.03), were significantly associated with the risk of lung cancer. For each of these biomarkers, in African Americans and Whites the magnitude of the increase in the odds ratio with creatinine adjustment was relatively large. For example, the ORs for TNE with lung cancer risk in Whites was 2.66 (p<0.0001) with creatinine adjustment compared to no statistically significant association without this adjustment (Table 5, model 1 vs 2).

Table 5:

Association of cigarette smoking-related urinary biomarkers and lung cancer risk stratified by racial group

Biomarkersa	cases	controls	Model 1b				Model 2c
			OR	(95% CI)	p	Adaptive FDR p	OR	(95% CI)	p	Adaptive FDR p
African American
TNE d	144	281	1.47	(1.12–1.92)	0.0054	0.0750	2.10	(1.47–3.00)	<.0001	0.001
Total NNAL	144	281	1.53	(1.10–2.13)	0.01	0.08	1.74	(1.23–2.45)	0.0016	0.007
CYP2A6 activity (total 3HCOT / cotinine) c	144	281	1.13	(0.90–1.43)	0.30	0.60	c
CEMA	144	281	1.14	(0.8–1.61)	0.48	0.60	1.33	(0.91–1.93)	0.14	0.36
2-HPMA	144	281	0.93	(0.71–1.22)	0.61	0.70	1.12	(0.82–1.52)	0.48	0.62
3-HPMA	144	281	1.15	(0.83–1.59)	0.40	0.60	1.30	(0.93–1.84)	0.13	0.36
SPMA	144	281	0.98	(0.75–1.27)	0.87	0.92	1.13	(0.85–1.51)	0.40	0.62
8-isoPGF2α	144	281	0.90	(0.68–1.18)	0.44	0.60	1.14	(0.81–1.61)	0.44	0.62
Cd	144	281	1.21	(0.90–1.61)	0.20	0.60	1.97	(1.34–2.91)	0.0006	0.004
PheT	144	281	0.99	(0.76–1.28)	0.92	0.92	1.21	(0.89–1.64)	0.22	0.41
Ratio PheT/sum 1,2,3-PheOH	141	279	1.09	(0.88–1.36)	0.44	0.60	1.16	(0.92–1.46)	0.21	0.41
Sum 1-,2-, and 3-PheOH	141	279	0.84	(0.63–1.12)	0.24	0.60	1.02	(0.73–1.42)	0.92	0.92
1-PheOH	141	279	0.86	(0.66–1.12)	0.27	0.60	1.02	(0.75–1.39)	0.88	0.88
2-PheOH	141	279	0.88	(0.66–1.17)	0.37	0.60	1.07	(0.77–1.48)	0.68	0.79
3-PheOH	141	279	0.81	(0.60–1.08)	0.15	0.60	0.94	(0.68–1.31)	0.73	0.79
Whites
TNE d	104	200	1.15	(0.85–1.55)	0.36	0.79	2.66	(1.61–4.41)	0.0001	0.0007
Total NNAL	104	200	1.09	(0.8–1.47)	0.60	0.79	1.43	(1.01–2.03)	0.05	0.06
CYP2A6 activity (total 3HCOT / cotinine)	104	200	1.01	(0.75–1.35)	0.97	0.97
CEMA	104	200	1.08	(0.73–1.57)	0.71	0.79	1.58	(1.01–2.46)	0.04	0.06
2-HPMA	104	200	0.95	(0.71–1.27)	0.73	0.79	1.31	(0.93–1.84)	0.12	0.12
3-HPMA	104	200	1.06	(0.77–1.46)	0.71	0.79	1.48	(1.02–2.14)	0.04	0.06
SPMA	104	200	0.95	(0.70–1.29)	0.74	0.79	1.23	(0.88–1.74)	0.23	0.23
8-isoPGF2α	104	200	0.95	(0.73–1.24)	0.72	0.79	1.41	(1.00–1.98)	0.05	0.06
Cd	104	200	1.10	(0.82–1.48)	0.51	0.79	2.32	(1.50–3.60)	0.0002	0.0007
PheT	104	200	0.75	(0.55–1.04)	0.085	0.48	1.18	(0.77–1.80)	0.45	0.45
Ratio PheT/sum 1,2,3-PheOH	103	200	0.92	(0.71–1.19)	0.50	0.79	1.04	(0.79–1.37)	0.80	0.80
Sum 1-,2-, and 3-PheOH	103	200	0.75	(0.53–1.06)	0.10	0.48	1.21	(0.76–1.93)	0.42	0.42
1-PheOH	103	200	0.78	(0.55–1.10)	0.16	0.48	1.27	(0.81–2.00)	0.31	0.31
2-PheOH	103	200	0.79	(0.57–1.09)	0.15	0.48	1.18	(0.79–1.76)	0.43	0.43
3-PheOH	103	200	0.72	(0.50–1.03)	0.07	0.48	1.10	(0.69–1.74)	0.69	0.69

^aAll urinary biomarkers were standardized using log transformation and dividing the individual value by the overall population SD of the log biomarker and therefore the OR corresponds to a per one-unit SD change in log biomarker level. The SD of the log-biomarker can be found in Supplemental Table 2

^bModel 1 is the same as Model 2 in Table 4, it includes age at urine collection, sex (male/female), log-BMI, Pkyrs and TNE

^cModel 2: Model 1 + log creatinine, CYP2A6 is not adjusted for creatinine as it is a ratio

^dTNE is also adjusted for log CYP2A6 activity

Abbreviations. CI, confidence intervals; TNE, total nicotine equivalents; NNAL, 4-(methylnitrosamino)-1–3-pyridyl)-1-butanol; CEMA, cyanoethylmercapturic acid, 3-HPMA, 3-hydroxypropylmercapturic acid; 2-HPMA, 2-hydroxypropylmercapturic acid, SPMA, S-phenylmercapturic acid; Cd, cadmium; 8-iso-PGF_2α, (Z)-7-(1R,2R,3R,5S)-3,5-dihydroxy-2-[(E,3S)-3-hydroxyoct-1-enyl]cyclopenyl]hept-5-enoic acid, PheT, phenanthrene tetraol; 1-,2-,3-PheOH, 1-, 2- and 3-hydroxyphenanthrene; PheT/PheOH, a proposed biomarker of metabolic activation of polycyclic aromatic hydrocarbons (PAH).

Discussion

In this nested case-control study of smoking-related biomarkers and lung cancer in smokers, after adjustment for CPD, urinary levels of TNE and total NNAL were significantly higher in cases. Moreover, both TNE and total NNAL were associated with an increased risk of lung cancer even after adjustment for pack-years of smoking and other risk factors. After stratification by race/ethnicity, and additional adjustment for creatinine, we found that TNE, NNAL and Cd were associated with the lung cancer risk of African American smokers and TNE and Cd were associated in White smokers. TNE and NNAL levels were higher in African Americans compared to Whites in the overall study population, but not among controls. These data support the hypothesis that the reported higher risk of African American for lung cancer is, at least in part, due to a higher internal smoking dose (as measured by TNE) per CPD, and therefore higher carcinogen exposure for African American smokers relative to White smokers (4, 17).

The key advantage of TNE over other biomarkers of smoking-related toxicant exposure is that it is not a single metabolite or the toxicant itself, it is the sum of nicotine and 5 metabolites and is therefore not significantly affected by metabolism. Smoking frequency and intensity are driven by an individual’s desire for nicotine. Nicotine is not a carcinogen, but total nicotine consumption of a smoker is an excellent biomarker of total smoking dose and captures a smoker’s exposure to a myriad of tobacco carcinogens and toxicants that can contribute to lung cancer development, including mutagenesis, inflammation, epigenetic alternations, and oxidative stress (3). Among the biomarkers in this study only NNAL is tobacco specific, but the others are all significantly elevated in smokers compared to non-smokers (5,11,28,29,33,34).

Metabolism and exposure to the smoking-related toxicants that produces the biomarker will vary by individual and by population. Here, we assessed whether urinary smoking-related biomarkers are associated with lung cancer risk, even after accounting for self-reported pack-years, across and within two populations. Biomarker differences might provide clues for which carcinogen may contribute to cancer. For instance, here we found that NNAL, a metabolite of NNK, a known lung carcinogen, was associated with lung cancer risk, even after accounting for TNE. This association was stronger in African Americans, which aligns with their higher risk of lung cancer. However, with the current study design (i.e. nested case-control) and as these biomarkers are highly correlated (where the first three eigenvalue explains 75% of the variance) and are measured metabolites of the tobacco toxicants or carcinogens, not the effect carcinogen dose, any associated biomarkers should be interpreted as potential markers of lung cancer risk and not causal. As such, in African Americans the data presented here support the possible use of total NNAL levels in addition to TNE to predict lung cancer risk. Cd appears to provides additional information for lung cancer risk in both African Americans and Whites.

CYP2A6 activity influences TNE levels by its effect on nicotine metabolism and smoking intensity (12). Lower CYP2A6 activity can lead to a decreased uptake of nicotine and tobacco constituents, and in a subcohort of current smokers in the MEC, African American smokers who carry reduced activity alleles (CYP2A6*9 and CYP2A6*17) excreted lower levels of TNE (17). The frequency of these alleles in African Americans is about 20%. However, the frequency of the deletion allele CYP2A6*4 is <2% compared to almost 20% in JA. In the MEC, adjusting for CYP2A6 activity (the ratio of 3-HCOT to cotinine) had a significant effect on average TNE levels of Japanese Americans smokers, but at most a modest effect on the average TNE levels in African American smokers (17). Similarly, adjusting for CYP2A6 activity in the SCCS smokers in this study had little or no effect on TNE levels.

In a previous prospective analysis conducted in the MEC, the urinary ratio of total 3-HCOT to cotinine was associated with lung cancer risk after adjusting for self-reported smoking history and TNE (6, 14). However, in this nested case-control study, CYP2A6 activity was not associated with lung cancer risk. This result was not surprising since the MEC analyses included a significant number of Japanese Americans, who have a high prevalence of CYP2A6 variants (4, 17). There are no Japanese Americans in the SCCS.

Greater than 70% of African American smokers use menthol cigarettes. Smoking menthol cigarettes is associated with increased smoking initiation by adolescents and reduced smoking cessation among adults(39, 40). Previously, a large case-control study in a different cohort found no association between menthol use and an increased risk of lung cancer (41). An earlier SCCS analysis, found that menthol cigarette-users smoked, on average, 1.6–1.8 fewer CPD than non-menthol cigarette smokers, regardless of race, and that menthol use was inversely associated with the risk of lung cancer (OR=0.65, 95% CI: 0.47–0.90) (25). Additionally, in the SCCS, there was no association of menthol cigarette use with an increased risk of cardiovascular disease (42). Consistent with these prior data from the SCCS, we found that TNE and NNAL levels were lower in menthol, compared to non-menthol, cigarette-users. African American menthol users, but not Whites had lower NNAL levels than non-users. Previously, Muscat et al. (43) reported that after adjusting for CPD, there was no difference in NNAL levels by menthol status in African Americans or Whites. Therefore, while menthol cigarette use may facilitate the uptake of smoking, it does not contribute to the higher smoking intensity, and thereby the higher tobacco related disease risk of African American smokers.

It is unclear what factors or conditions may be driving the higher internal smoking dose per cigarette found in African American smokers in both the SCCS and the MEC study. Prior studies have found that social determinants of health, such as socioeconomic status, discrimination or unsafe neighborhoods, can lead to stress and smoking disparities, including delayed cessation and increase use of tobacco products and/or smoking intensity (44). Additional efforts to examine these factors and smoking intensity are ongoing.

The present study had a few limitations. Urine for the measurement of biomarkers was only available for a single time point. Additionally, this was a spot urine sample as opposed to first morning, overnight or 24-hour urine. Constituent levels in a spot urine sample are dependent in part on the time between last cigarette smoked and urine collection. To help correct for this we also measured creatinine. However, creatinine adjustment resulted in an increase in the OR for lung cancer risk in relation to TNE and NNAL in Whites and African Americans. It is unclear what drives this effect, creatinine may reflect variations by race/ethnicity, sex, time of urine collection, overall health, and other unknown factors (35, 38, 45). More study is required to understand what these variables are and how they influence the association with lung cancer risk. Another difference between these two studies is that the SCCS has fewer years of follow-up; the average time since urine collection and lung cancer diagnosis is approximately 4 years, whereas in the MEC it was approximately 12 years. However, when we removed cases diagnosed within 2 years from study entry, the results did not change, suggesting that these associations less likely reflect smoking behaviors due to lung function and / or early symptoms. Lastly, while we utilized a Bonferroni corrected p-value to account for the effective number of independent tests (n=3, identified from eigenvalues from a correlation matrix of all 12 biomarkers), some findings may be spurious from multiple testing.

This study had a number of strengths, including the nested case-control study design from a well-characterized prospective study of African Americans and Whites. Another strength is the use of TNE, which is minimally affected by individual differences in nicotine metabolism, as a measure of internal smoking and carcinogen dose. The quantitation of multiple tobacco-related biomarkers and evaluation of their relationship with lung cancer risk is also a strength. Many of these biomarkers are not specific to tobacco exposure and may in part reflect other detrimental environmental exposures that may influence susceptibility to lung cancer particularly in this cohort with oversampling of participants of lower socioeconomic backgrounds.

In conclusion, our study found that even after accounting for age, sex, and smoking history as measured by self-reported pack-years, TNE, a marker of smoking carcinogen dose, is an additional risk factor for lung cancer in smokers beyond self-reported pack-years. NNAL and Cd, even with adjustment for TNE were also associated with lung cancer. Quantifying urinary NNAL, Cd and TNE concentrations in current smokers may provide additional information for lung cancer risk prediction in current smokers. This may be particularly useful for African American current smokers as this and prior studies have shown that TNE levels (which measures internal smoking dose and is highly correlated with smoking carcinogens and toxicants) better correlate with disease risk than self-reported smoking history (4, 12, 46). Additionally, and not unexpectedly, menthol use does not increase nicotine and carcinogen uptake.