Secondary Logo

Journal Logo

Original Article

Cigarette Smoking and Lung Cancer

Modeling Effect Modification of Total Exposure and Intensity

Lubin, Jay H.*; Caporaso, Neil; Wichmann, H Erich; Schaffrath-Rosario, Angelika§; Alavanja, Michael C. R.

Author Information
doi: 10.1097/EDE.0b013e31812717fe
  • Free

Abstract

Understanding the relationship of lung cancer and cigarette smoking and modifiers of that relationship provides a framework for evaluating the molecular basis of carcinogenesis, including activation and detoxification of tobacco carcinogens and DNA repair. A more detailed understanding of etiology may improve risk assessment and public health decision-making.1 Specifically, in studies of lung cancer and bladder cancer, odds ratios (ORs) increase with smoking intensity (cigarettes smoked per day), but then often level off at high intensities.2 This pattern has been described but not adequately quantified.

Models that incorporate smoking duration and intensity are problematic for exploring this pattern due to changing total exposures. For example, in a logistic model, the intensity parameter represents the ln(OR) per cigarette per day at a fixed duration. A similar interpretation applies for the duration parameter at a fixed intensity. Because duration is fixed, ORs at 2 different intensities therefore reflect not only the different intensities but also different total pack-years. For 30 years' duration, ORs at 20 and 30 cigarettes per day embed effects of different total exposures, ie, 30 and 45 pack-years, respectively. Thus, the intensity parameter does not represent a “pure” intensity effect, but includes the effect of pack-years. In contrast, ORs for 20 and 30 cigarettes per day obtained from a model that includes pack-years and intensity reflect on intensity effect unconfounded by total exposure.

Using data on never, current, and recent (within 5 years) former smokers, investigators have developed an excess odds ratio (EOR) model for lung cancer. The model is linear in pack-years and exponential in the logarithm of intensity and its square, which quantifies the effects of intensity on the EOR per pack-year.3 The model isolates the intensity effects for fixed total pack-years, thus enabling the comparison of ORs for total exposure delivered at low intensity (for long duration) and at high intensity (for short duration). Below 15–20 cigarettes per day, there is a direct exposure rate effect (increased potency or exposure enhancement effect), whereby the EOR per pack-year increases with increasing intensity, ie, for equal pack-years, increasing intensity (decreasing duration) increases risk. Above 20 cigarettes per day, there is an inverse exposure rate effect (reduced potency or wasted exposure effect), whereby the EOR per pack-year decreases with increasing intensity, ie, for equal pack-years increasing intensity decreases risk.

Using pooled data from 2 large studies of lung cancer, we evaluated variations of the linear pack-years effect and exponential intensity effects by years since smoking cessation, attained age, age started smoking, sex, inhalation and type of cigarette (filter or nonfilter/mixed) smoked.

METHODS

Studies of Smoking and Lung Cancer

The European Smoking and Health Study was a hospital-based case-control study of lung cancer conducted between 1976 and 1980 at 7 European centers (Glasgow, Hamburg, Heidelberg, Vienna, Paris, Milan, and Rome).4,5 The study enrolled 7,804 cases and 15,207 controls, frequency-matched by age, sex, and center. We excluded 712 subjects who smoked cigars or pipes, 199 subjects who started smoking after age 40, and 36 subjects with missing data. The basic dataset included 4988 cases (4336 men and 652 women) and 8389 controls (7142 men and 1247 women), who were age 50 to 75 years. Participants were never-smokers, current smokers, or former smokers who stopped smoking within 5 years of enrollment.

The German Radon Study was a population-based lung cancer case-control study conducted between 1990 and 1997 in 23 regions of Germany.6–8 The study enrolled 4071 cases age 75 years and younger and 4628 controls, frequency-matched on age, sex, and region. We excluded smokers who started smoking after age 40, smoked cigars or pipes, or had temporarily stopped smoking for more than 5 years, leaving 3212 cases (2589 men and 614 women) and 3809 controls (2932 men and 877 women). The basic dataset includes 2256 cases (1796 men and 460 women) and 2157 controls (1512 men and 645 women) age 50 to 75, who were never-smokers, current smokers, or former smokers who stopped smoking within 5 years of enrollment. Initial analysis indicated that the inclusion of 158 men (70 cases and 88 controls) who smoked up to 8 cigarettes per day resulted in models that slightly underestimated the EOR per pack-year at moderate and high intensities. Although inference is unaffected, we omitted these subjects.

We imposed the age restriction to reduce the potential impact of any genetic cancer predisposition in younger cases or diagnostic ambiguity in elderly cases, and omitted cigar or pipe smokers to allow consistent analyses of cigarette exposure. Both studies were approved by appropriate internal review boards.

Models

We defined I intensity categories and indicator variables ni, i = 1, …, I, where ni = 1 for intensities within the ith category and zero otherwise and fit the model

where d is pack-years. Within category i, ORs are linear in pack-years (ie, OR = 1 + γid). The slope γi defines the EOR per pack-year. For calculating estimates, exp(

) replaces γi. Factoring out γ1, model (1) becomes

Thus, a natural modeling for continuous intensity, n, is

where g(.) represents intensity effects and β represents the EOR per pack-year at g(n) = 1. We set g(n) = exp{φ1 ln(n) + φ2 ln(n)2}. This form provided a better fit than g(n) = exp{φ1 ln(n) + φ2n} or g(n) = exp{φ1n + φ2n2}. Adding ln(n) × n, ln(n)3 or n3 did not further improve fit. The parameters φ1 and φ2 define the modulation of the EOR per pack-year with intensity, and their relative size identifies the maximum EOR per pack-year, βg(nmax), where for φ1 > 0 and φ2 < 0 nmax = exp(−φ1/2φ2). The first derivative of the EOR per pack-year function, ∂[β g(n)]/∂n = β g(n){φ1 + 2φ2 ln(n)}ln, describes changes in the function with intensity.

We evaluated effect modification by graphing models (1) and (2) within categories of the factor; we tested statistical significance using variants of the EOR in model (2), namely, β d gf(n), βfd g(n), and βfd gf(n), where “f” denotes a separate parameter (β) or set of parameters (φ1 and φ2) for each level of the factor. The difference of deviances for nested models defines a test of homogeneity of the pack-year and intensity effects across levels of the factor. Degrees of freedom equal the difference in numbers of parameters.

We evaluated homogeneity over study, s, using βs×fd gs×f(n), where “s×f” denotes all levels of s and f. Similarly, “s,f” denotes separate parameters for levels of f and one additional parameter for study effect. For all factors, intensity effects are homogeneous across studies, ie, inclusion of gs×f(n) or gf(n) provides comparable fit. We therefore present results based on βs×f gf(n). For the figures, we summarized models by setting the study variable to values +1 (ESHS) or −1 (GRS).

We used the binary outcome module in Epicure (HiroSoft International Corp., Seattle, WA), with stratification on center/region, sex and age (five levels; 50–54, …, 70+).

RESULTS

Among never-smokers and cigarette-only smokers, percentages of ever-smokers for male and female cases and controls were similar in both studies. The German study had fewer current smokers, whereas smokers in the European study consumed at higher intensities for longer durations (Table 1).

T1-19
TABLE 1:
Summary of Smoking Data for Case-Control Studies by Sex

Model for Pack-Years and Smoking Intensity

Applying the analytic methods used previously with the European study, we computed ORs for the German study by categories of pack-years and intensity relative to never-smokers. ORs by pack-years are approximately linear within intensity categories. Slope estimates (γi's) vary with intensity. As seen below, the fitted model (2) closely conforms to the γi estimates. Study-specific estimates of EOR per pack-year (β) differ significantly (P < 0.01), whereas homogeneity of φ1 and φ2 over study is not rejected (P = 0.17), indicating similar intensity patterns (Table 2). Estimates of β are 0.0047 for the European study and 0.0146 for the German study, and summary estimates of φ1 and φ2 are 2.86 and −0.495, respectively.

T2-19
TABLE 2:
Results for Modeling Excess Odd Ratios (EOR) of Lung Cancer Per Pack-Year and Smoking Intensity by Categories of Other Factors: Pooled Data From the European Smoking and Heath Study (ESHS) and German Radon Study (GRS)

Evaluation of Effect Modification

Table 2 shows estimates for the βs×fd gf(n) model, maxima for the EOR functions, deviance changes relative to the study-specific βsd g(n) model, and P values for 4 tests of interaction for pack-years and 3 tests of interaction for intensity and f (see table footnote).

We added former smokers to the basic dataset to evaluate smoking cessation. Effect modification by time since smoking cessation derives from an interaction with intensity and not pack-years. The intensity and cessation interaction is statistically significant, regardless of the adjustment for the pack-years and cessation interaction (P = 0.01, comparing βs×fd gf(n) with βs×fd g(n); P = 0.04, comparing βs, fd gf(n) with βs, fd g(n); and P < 0.01, comparing βsd gf(n) with βsd g(n)). Adjusting for the intensity and cessation interaction, the pack-years and cessation interaction is not statistically significant (P = 0.14, comparing βs×fd gf(n) with βsd gf(n); and P = 0.78, comparing βs, fd gf(n) with βsd gf(n)). The model βs gf(n) closely describes the intensity-specific EOR per pack-year estimates (γi) within smoking cessation categories (Fig. 1). First derivatives suggest that differences in EOR per pack-year functions result from a reduced modulation with intensity by increasing time since cessation (Fig. 1D).

F1-19
FIGURE 1.:
(A–C) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within categories of years since cessation of smoking and the fitted EOR per pack-year based on the model βs gf(n) (solid line) and the pointwise 95% prediction interval (dashed line). Results are summarized over study. (D) first derivatives of the fitted models.

We added subjects under age 50 to the basic dataset to assess age. The intensity and age interaction is significant, after adjusting for the pack-years and age interaction (Table 2). Controlling for the intensity and age interaction, pack-year effects are statistically homogeneous across age groups. As with cessation, differences in the EOR per pack-year functions (Fig. 2) result from differential modulation with intensity by age, primarily a reduced variation in the youngest age group.

F2-19
FIGURE 2.:
(A–D) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within categories of attained age and the fitted EOR per pack-year based on the model βs gf(n) (solid line) and the pointwise 95% prediction interval (dash line). Results are summarized over study. (E) First derivatives of the fitted models.

Interactions between the age at which smoking began and intensity or pack-years are each statistically significant after controlling for the other interaction. Parameter estimates and fitted EOR per pack-year maxima are different for people who start smoking at young ages, particularly at very low intensities of smoking (Fig. 3).

F3-19
FIGURE 3.:
(A–D) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within categories of age started smoking and the fitted EOR per pack-year based on the model βs, f gf(n) (solid line) and the pointwise 95% prediction interval (dashed line). Results are summarized over study. (E) First derivatives of the fitted models.

Differences in the EOR for smoking by sex result from variations in total pack-years rather than intensity. Controlling for the interaction of pack-years and sex, the interaction of intensity and sex was not statistically significant (P = 0.30), while the interaction of pack-years and sex was significant (Fig. 4). Notably, the interaction by pack-years and sex differs by study (Table 2: χ2(1) = 47.1–38.5 = 8.6 comparing βs×fd g(n) with βs, fd g(n), P < 0.01). Estimates from the model βs×fd g(n) areβESHS,male = 0.0121, βESHS,female = 0.0062, βGRS,male = 0.0507, βGRS,famale = 0.0109, φ1 = 2.43 and φ2 = −0.432. Relative pack-year effects for men and women are 2.0 in the European study and 4.7 in the German study.

F4-19
FIGURE 4.:
(A, B) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within sex and the fitted EOR per pack-year based on the model βs×f g(n) (solid line) and the pointwise 95% prediction interval (dashed line). Results are summarized over study. (C) First derivatives of the fitted models.

Effect modification by frequency of inhalation (data only from the European study) and depth of inhalation results from variations with intensity and not pack-years. The frequency of inhalation and intensity interaction is statistically significant (P < 0.01), even after controlling for the interaction of frequency and pack-years (P = 0.03). Controlling for intensity, the interaction of pack-year and frequency of inhalation is not significant (P = 0.78). Results are similar for depth of inhalation. Differences in the EOR per pack-year by inhalation pattern derive from a reduced modulation with intensity in less vigorous inhalers (Fig. 5). More frequent and deeper inhalers have higher EOR per pack-year maxima that occur at lower intensities.

F5-19
FIGURE 5.:
(A–C) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within categories of depth of inhalation and the fitted EOR per pack-year based on the model βs gf(n) (solid line) and the pointwise 95% prediction interval (dashed line). Results are summarized over study. (D) First derivatives of the fitted models.

The interaction of intensity and type of cigarette (filter-only or nonfilter-only/mixed) is statistically significant (P < 0.1), although p-values diminish after controlling for the interaction of pack-years and type of cigarette (P = 0.12 using βs×f and P = 0.13 using βs, f). Controlling for the interaction of intensity and type of cigarette, the interaction of pack-years and type of cigarette is not significant. Analysis suggests that effect modification with type of cigarette results from variation with smoking intensity (Fig. 6).

F6-19
FIGURE 6.:
(A, B) Estimated excess odds ratio (EOR) per pack-year for categories of smoking intensity (square symbol) within categories of type of cigarette smoked and the fitted EOR per pack-year based on the model βs gf(n) (solid line) and the pointwise 95% prediction interval (dashed line). Results are summarized over study. (C) First derivatives of the fitted models.

Finally, first-derivative plots showed that effect modifiers have their greatest impact on the EOR per pack-year function under 20–25 cigarettes per day.

DISCUSSION

Variations in smoking-related ORs with time since smoking cessation, age, frequency and depth of inhalation, and type of cigarette derive from interactions with smoking intensity, rather than pack-years. In contrast, variations in smoking-related ORs with sex result from interactions with total pack-years, while smoking intensity effects are similar in men and women. Variations in ORs with the age at which smoking began result from interactions with both intensity and pack-years.

The modeling of the EOR per pack-year suggests enhanced carcinogenic potency of cigarettes at low intensities and a relatively reduced potency at moderate and high intensities, both overall and within levels of effect modifiers. The latter pattern parallels relationships reported in biomarker studies of smoking.2,9,10 Polycyclic aromatic hydrocarbons (PAH), many of which are known to be carcinogenic, occur in tobacco smoke, and after metabolic activation they form DNA and protein adducts.11 Individuals exposed at environmental levels have higher DNA-adduct levels in white blood cells per unit exposure to PAHs than coke oven workers exposed at high levels, suggesting reduced carcinogenic potency at high exposures. Similarly, among never-smokers and current smokers, the ratio of serum carboxyhemoglobin to number of cigarettes smoked per day decreased with increasing smoking intensity.12 Studies of PAHs and serum carboxyhemoglobin have limitations for evaluating tobacco effects, since they can arise from nontobacco sources. In contrast, 4-(methylnitrosamino)–(3-pyridyl)-1-butanone (NNK) is a tobacco-specific carcinogen, and levels of its metabolites are markers of tobacco effects.13,14 Using data from by Joseph et al,15 an ongoing analysis finds a decline in the ratio of NNK metabolites to urinary cotinine (a marker of tobacco exposure) with increasing cotinine levels, which is consistent with decreasing potency (J. H. Lubin and S. S. Hecht, personal communications). Intensity patterns may thus reflect biologic phenomena, with enhanced potency resulting from reduced DNA repair capacity,16,17 and with reduced potency resulting from an increased repair capacity in heavy tobacco users,18–21 saturation of activation pathways,9–11 or an increased induction of detoxification enzymes.22

Although patterns and effect modifications of the EOR per pack-year may reflect biologic phenomena, these patterns may also reflect influences of nicotine satiation, whereby carcinogenic yield per cigarette decreases with increasing intensity as smokers seek to maintain addiction-sufficient nicotine levels, such that the number of cigarettes per day increasingly overestimates the internal exposure rate. However, in the European study, there was no evidence of a relationship between frequency or depth of inhalation and intensity after controlling for total pack-years.3 In contrast, a study of 190 smokers did find increased plasma cotinine and nicotine levels with increased intensity, and a marginally significant (P = 0.08) decline in “nicotine boost”, ie, an increase in blood plasma nicotine per cigarette.23

We evaluated the potential influence of nicotine satiation in 2 ways, and found that, while cigarettes smoked per day may overestimate internal exposure rate at higher intensities, it is unlikely that nicotine satiation fully explains the inverse exposure rate pattern and the complex patterns of effect modification. We first analyzed effect modification within categories of inhalation, assuming that “overestimation” of internal exposure rate by cigarettes per day was greater in smokers who moderately or deeply inhale than in smokers who slightly inhale. The patterns of effect modification by age, age started smoking, sex and smoking cessation were similar in the 2 inhalation groups.

We next conducted a sensitivity analysis to evaluate the effects of overestimation of internal exposure rate by cigarettes per day, using cotinine as a marker of internal exposure rate. In smokers, cotinine levels increase approximately linearly with smoking intensity up to about 20–30 cigarettes per day.12,15,24–30 At higher intensities, cotinine has been variously shown to increase without diminution,12,24,27,29,30 to increase but at a diminished rate,24,25,28 or to increase until there is a leveling and possibly even a decline.15,24,26 For the sensitivity analysis, we created an adjusted intensity variable, nadj, then recomputed pack-years and refitted models. We assumed that observed intensity n accurately characterized internal exposure rate up to no cigarettes per day, ie, the regression of nadj on n increased from zero with slope one. We considered 2 adjustment schemes. One scheme specified a piecewise linear relationship with a reduced slope above no, namely, nadj = n for n ≤ no and nadj = no + (nno) × (1 − K), for n > no, where K is the reduction fraction. A second scheme applied a proportionality factor that increasingly deviates from one. The regression equation was nadj = n for n ≤ no and nadj = no + (n − no) exp{ln(0.5) × (n − no)/T} for n > no, where T is analogous to a “half-life” and represents the number of cigarettes per day that results in a 50% adjustment. Above no, nadj increases to a maximum then declines. For example, with no = 25 and T = 20, values n = 30, 45 and 80 cigarettes per day adjust to nadj = 25 + (30–25) × 0.50.25 = 29.2, 25 + (45–25) × 0.51.0 = 35.0 and 25 + (80–25) × 0.52.75 = 33.2 cigarettes per day, respectively.

We evaluated no = 20, 25, 30 with K = 0.3, 0.7 for scheme 1 and T = 20, 30, 40 for scheme 2. We found only minimal changes in statistical inference on effect modification, due to the high proportion of light and moderate smokers in the data, with 76% of smokers consuming 25 or fewer cigarettes per day, and 86% smoking 30 or fewer. For various no and T or K, the adjusted estimates of EOR per pack-year within intensity categories increased (since observed cigarettes per day reflects an “overestimate” of internal exposure rate). However, the shape of the EOR function remained largely unchanged, since simultaneously nadj for higher intensities (x-axis values) shifted towards lower intensities. Nevertheless, since ranges for intensities are reduced, power to detect variations declines.

Our analysis of effect modification finds that ORs for lung cancer decline with time since cessation of smoking.1 More specifically, intensity plays a major role in determining risk in active smokers (by modulating the EOR per pack-year), but a more limited role in long-term former smokers (Fig. 1). Thus, factors related to reduced potency at high intensities have diminished impact with longer time since carcinogenic challenge, a pattern consistent with greater DNA repair capacity in current, as compared with former or never smokers, and in heavier smokers.20

Variations in smoking-related ORs with frequency and depth of inhalation and type of cigarette are mediated through smoking intensity. Smokers who never or only slightly inhale, or who smoke only filtered cigarettes, incur less intensity-dependent modulation of the EOR per pack-year compared with more vigorous inhalers or nonfilter/mixed smokers. This implies that, for equal total pack-years, the level of lung cancer risk is relatively less influenced by intensity in infrequent or slight inhalers, and relatively more influenced by intensity in frequent or deep inhalers.

There is debate about the magnitude of ORs for lung cancer by sex for comparable cigarette exposure.31–42 Evidence suggests that women and men differ in their expression of phase I and II enzymes, NNK metabolism, DNA repair capacities, and growth and hormonal factors,19,43,44 but also suggests comparable susceptibility.45 In our data, intensity effects are similar for men and women, suggesting that delivery and processing of tobacco carcinogens are comparable in the 2 sexes. However, estimates of EOR per pack-year are higher in men than in women, indicating differential consequences of total exposure or duration of exposure.

Patterns of EORs per pack-year differ for subjects younger than age 50, consistent with differential risk at younger ages.46 However, the number of subjects under age 50 and the range for pack-years are limited and results are therefore uncertain.

In the pooled data, 1675 smokers (854 cases and 821 controls) started smoking before age 15 years, and 574 smokers (287 cases and 287 controls) started before age 13 years. Figure 3 suggests that smoking at low intensities may be particularly deleterious for those who start smoking in their preteen and early teen years.

The fitted EOR per pack-year maximum is 0.91 for the German study, which is 3 times the 0.29 value for the European study. In the 7 European centers maxima range from 0.11 to 0.54, with a value of 0.16 for Hamburg and Heidelberg centers combined, all lower than the value for the German study. These differences occur in both men and women. Reasons for the differences are not clear, particularly since smoking exposures are greater in the European study (and in Hamburg and Heidelberg) than in the German study. Lung cancer rates vary widely throughout Europe, with male rates declining from the mid-1970s through 1990s, depending on country, and female rates increasing, except in the United Kingdom and Ireland.47–49 The European study was hospital-based, and an over-representation of smokers in control subjects could explain the difference. However, results did not materially change when we accounted for concurrent diseases among the controls.

In summary, we find that variations of lung cancer risk with time cessation of smoking, age, frequency and depth of inhalation and type of cigarette smoked are mediated through smoking intensity, rather than total pack-years. In contrast, differences in lung cancer risk by sex are mediated through total exposure, with intensity effects similar by sex. Implications of these findings relative to the molecular basis of smoking risk need further elucidation.

ACKNOWLEDGMENTS

We thank Montserrat Garcia-Closas of the Division of Cancer Epidemiology and Genetics, National Cancer Institute, for useful discussions on this and related topics.

REFERENCES

1. U.S. Department of Health and Human Services. The Health Consequences of Smoking: A Report of the Surgeon General. ISBN 0-16-051576-2. 2004. Washington, DC: 20402, U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, Superintendent of Documents, U.S. Government Printing Office.
2. Vineis P, Kogevinas M, Simonato L, et al. Levelling-off of the risk of lung and bladder cancer in heavy smokers: an analysis based on multicentric case-control studies and a metabolic interpretation. Mutat Res. 2000;463:103–110.
3. Lubin JH, Caporaso N. Cigarette smoking and lung cancer: modeling total exposure and intensity. Cancer Epidemiol Biomarkers Prev. 2006;15:517–523.
4. Lubin JH, Blot WJ, Berrino F, et al. Patterns of lung cancer risk according to type of cigarette smoked. Int J Cancer. 1984;33:569–576.
5. Lubin JH, Blot WJ. Assessment of lung cancer risk factors by histologic category. J Natl Cancer Inst. 1984;73:383–389.
6. Wichmann HE, Rosario AS, Heid IM, et al. Increased lung cancer risk due to residential radon in a pooled and extended analysis of studies in Germany. Health Phys. 2005;88:71–79.
7. Kreuzer M, Heinrich J, Wolke G, et al. Residential radon and risk of lung cancer in Eastern Germany. Epidemiology 2003;14:559–568.
8. Kreienbrock L, Kreuzer M, Gerken M, et al. Case-control study on lung cancer and residential radon in western Germany. Am J Epidemiol. 2001;13:42–52.
9. Lewtas J, Walsh D, Williams R, et al. Air pollution exposure DNA adduct dosimetry in humans and rodents: evidence for non-linearity at high doses. Mutat Res. 1997;378:51–63.
10. Lutz WK. Dose-response relationships in chemical carcinogenesis: superposition of different mechanisms of action, resulting in linear-nonlinear curves, practical thresholds, J-shapes. Mutat Res. 1998;405:117–124.
11. Phillips DH. Smoking-related DNA and protein adducts in human tissues. Carcinogen 2002;23:1979–2004.
12. Law MR, Morris JK, Watt HC, et al. The dose-response relationship between cigarette consumption, biochemical markers and risk of lung cancer. Br J Cancer. 1997;75:1690–1693.
13. Carmella SG, Akerkar SA, Richie JP, et al. Intraindividual and interindividual differences in metabolites of the tobacco-specific lung carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) in smokers urine. Cancer Epidemiol Biomarkers Prev. 1995;4:635–62.
14. Anderson KE, Kliris J, Murphy L, et al. Metabolites of a tobacco-specific lung carcinogen in nonsmoking casino patrons. Cancer Epidemiol Biomarkers Prev. 2003;12:1544–1546.
15. Joseph AM, Hecht SS, Murphy SE, et al. Relationships between cigarette consumption and biomarkers of tobacco toxin exposure. Cancer Epidemiol Biomarkers Prev. 2005;14:2963–2968.
16. Neumann AS, Sturgis EM, Wei QY. Nucleotide excision repair as a marker for susceptibility to tobacco-related cancers: a review of molecular epidemiological studies. Mol Carcinog. 2005;42:65–92.
17. Wei QY, Spitz MR. The role of DNA repair capacity in susceptibility to lung cancer: a review. Cancer Metastasis Rev. 1997;16:295–307.
18. Wei QY, Cheng L, Amos CI, et al. Repair of tobacco carcinogen-induced DNA adducts and lung cancer risk: a molecular epidemiologic study. J Natl Cancer Inst. 2000;92:1764–1772.
19. Spitz MR, Wei QY, Dong Q, et al. Genetic susceptibility to lung cancer: The role of DNA damage and repair. Cancer Epidemiol Biomarkers Prev. 2003;12:689–698.
20. Shen HB, Spitz MR, Qiao YW, et al. Smoking, DNA repair capacity and risk of nonsmall cell lung cancer. Int J Cancer. 2003;107:84–88.
21. Hung RJ, Brennan P, Canzian F, et al. Large-scale investigation of base excision repair genetic polymorphisms and lung cancer risk in a multicenter study. J Natl Cancer Inst. 2005;97:567–576.
22. Gu J, Liang D, Wang YF, et al. Effects of N-acetyl transferase 1 and 2 polymorphisms on bladder cancer risk in Caucasians. Mutat Res. 2005;581:97–104.
23. Patterson F, Benowitz N, Shields P, et al. Individual differences in nicotine intake per cigarette. Cancer Epidemiol Biomarkers Prev. 2003;12:468–471.
24. Blackford AL, Yang G, Hernandez-Avila M, et al. Cotinine Concentration in Smokers from Different Countries: Relationship with Amount Smoked and Cigarette Type. Cancer Epidemiol Biomarkers Prev. 2006;15:1799–1804.
25. Campuzano JC, Hernandez-Avila M, Jaakkola MS, et al. Determinants of salivary cotinine levels among current smokers in Mexico. Nicotine Tob Res. 2004;6:997–1008.
26. Etter JF, Perneger TV. Measurement of self reported active exposure to cigarette smoke. J Epidemiol Community Health. 2001;55:674–680.
27. Lewis SJ, Cherry NM, Niven RM, et al. Cotinine levels and self-reported smoking status in patients attending a bronchoscopy clinic. Biomarkers. 2003;8:218–228.
28. O'Connor RJ, Giovino GA, Kozlowski LT, et al. Changes in nicotine intake and cigarette use over time in two nationally representative cross-sectional samples of smokers. Am J Epidemiol. 2006;164:750–759.
29. Olivieri M, Poli A, Zuccaro P, et al. Tobacco smoke exposure and serum cotinine in a random sample of adults living in Verona, Italy. Arch Environ Health. 2002;57:355–359.
30. Richie JP, Carmella SG, Muscat JE, et al. Differences in the urinary metabolites of the tobacco-specific lung carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone in black and white smokers. Cancer Epidemiol Biomarkers Prev. 1997;6:783–790.
31. Henschke CI, Miettinen OS. Women's susceptibility to tobacco carcinogens. Lung Cancer. 2004;43:1–5.
32. Risch HA, Miller AB. Re: Sex, smoking, and cancer: a reappraisal. J Natl Cancer Inst. 2002;94:308.
33. Zang EA. Re: Sex, smoking, and cancer: a reappraisal. J Natl Cancer Inst. 2002;94:308–309.
34. Perneger TV. Sex, smoking, and cancer: a reappraisal. J Natl Cancer Inst. 2001;93:1600–1602.
35. Tang DL, Rundle A, Warburton D, et al. Associations between both genetic and environmental biomarkers and lung cancer: evidence of a greater risk of lung cancer in women smokers. Carcinogen 1998;19:1949–1953.
36. Mcduffie HH. Are female smokers at higher risk for lung-cancer than male smokers—a case-control analysis by histologic type. Am J Epidemiol. 1994;140:185–186.
37. Wilcox AJ. Are female smokers at higher risk for lung-cancer than male smokers—a case-control analysis by histologic type. Am J Epidemiol. 1994;140:186.
38. Hoover DR. Are female smokers at higher risk for lung-cancer than male smokers—a case-control analysis by histologic type. Am J Epidemiol. 1994;140:186–187.
39. Risch HA, Howe GR, Jain K, et al. Are female smokers at higher risk for lung-cancer than male smokers - a case-control analysis by histologic type—reply. Am J Epidemiol. 1994;140:187–188.
40. Solomon L, Astemborski J, Warren D, et al. Are female smokers at higher risk for lung-cancer than male smokers—a case-control analysis by histologic type—reply. Am J Epidemiol. 1994;140:189–190.
41. Risch HA, Howe GR, Jain M, et al. Are female smokers at higher risk for lung-cancer than male smokers - a case-control analysis by histologic type. Am J Epidemiol. 1993;138:281–293.
42. Bain C, Feskanich D, Speizer FE, et al. Lung cancer rates in men and women with comparable histories of smoking. J Natl Cancer Inst. 2004;96:826–834.
43. Hill CE, Affatato AA, Wolfe KJ, et al. Gender differences in genetic damage induced by the tobacco-specific nitrosamine NNKand the influence of the Thr241Met polymorphism in the XRCC3 gene. Environ Mol Mutagen. 2005;46:22–29.
44. Patel JD. Lung cancer in women. J Clin Oncol. 2005;23:3212–3218.
45. Marrogi AJ, Mechanic LE, Welsh JA, et al. TP53 mutation spectrum in lung cancer is not different in women and men. Cancer Epidemiol Biomarkers Prev. 2005;14:1031–1033.
46. Liu G, Zhou W, Christiani DC. Molecular epidemiology of non-small cell lung cancer. Semin Resp Crit Care Med. 2005;26:265–272.
47. Brennan P, Bray I. Recent trends and future directions for lung cancer mortality in Europe. Br J Cancer. 2002;87:43–48.
48. Tyczynski JE, Bray F, Parkin DM. Lung cancer in Europe in 2000: epidemiology, prevention, and early detection. Lancet Oncol. 2003;4:45–55.
49. Janssen-Heijnen MLG, Coebergh JWW. The changing epidemiology of lung cancer in Europe. Lung Cancer. 2003;41:245–258.
© 2007 Lippincott Williams & Wilkins, Inc.