Evidence for an intensity-dependent interaction of NAT2 acetylation genotype and cigarette smoking in the Spanish Bladder Cancer Study
1Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Rockville, Maryland 20852, USA.
2Centre for Research in Environmental Epidemiology, Municipal Institute of Medical Research (IMIM), Barcelona, Spain.
3Department of Social Medicine, Medical School, University of Crete, Heraklion, Greece.
4Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Rockville, Maryland 20852, USA.
5Hormone and Reproductive Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Rockville, Maryland 20852, USA.
6Universidad de Oviedo, Oviedo, Spain.
7Department of Pharmacology & Toxicology, James Graham Brown Cancer Center, University of Louisville School of Medicine, Louisville, KY 40292, USA.
8Unidad de Investigación, Hospital Universitario de Canarias, La Laguna, Spain.
9Consorci Hospitalari Parc Taulí, Sabadell, Spain.
10University Pompeu Fabra, Barcelona, Spain.
11Hospital General de Elche, Elche, Spain.
*Corresponding author. Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Rockville, Maryland 20852, USA. E-mail: lubinj{at}mail.nih.gov
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
Background The N-acetyltransferase 2 (NAT2) enzyme detoxifies aromatic amines, an important class of carcinogens in tobacco smoke. Slow acetylation phenotype individuals have reduced detoxification capacity compared with those with a rapid/intermediate phenotype. Analysis of the Spanish Bladder Cancer Study found an odds ratio (OR) for slow acetylators relative to rapid/intermediate acetylators of 0.9 in never-smokers and 1.6 in ever-smokers, a 1.8-fold enhancement in smokers. Evidence indicates that acetylation is an exposure-dependent process, and thus the magnitude of the interaction may also depend on exposure level.
Methods We extend a comprehensive three-parameter linear-exponential model for the excess odds ratio (EOR) for smoking to include effects of NAT2 status, and reanalyse smoking and NAT2 status for the bladder cancer data.
Results We show that variations in smoking risk with NAT2 status result from interactions with smoking intensity (cigarettes per day) and not total pack-years of exposure. In addition, the relative increase in smoking risk in NAT2 slo acetylators increases with smoking intensity.
Conclusions Analyses reveal an enhanced effect for smoking intensity and bladder cancer in NAT2 slow acetylators which increases with intensity.
Keywords Bladder neoplasms, smoking, dose response, interaction
Accepted 23 February 2007
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
The N-acetyltransferase 2 (NAT2) enzyme detoxifies aromatic amines, an important class of carcinogens in tobacco smoke. Individuals with a slow acetylation phenotype have reduced detoxification capacity compared with those with a rapid or intermediate phenotype.1,2 Analysis from the Spanish Bladder Cancer Study (SBCS) found an odds ratio (OR) for slow acetylators relative to rapid/intermediate acetylators of 0.9 in never-smokers and 1.6 in ever-smokers, a 1.8-fold enhancement in smokers. Acetylation is an exposure-dependent process, and therefore the magnitude of the interaction for NAT2 status and smoking may also depend on exposure level. The previous analysis of SBCS data did not report increased ORs for slow acetylators with increasing smoking intensity (cigarettes smoked per day). We re-examine those data using a new comprehensive model for smoking, which increases power for identifying patterns of association.
In studies of bladder cancer, odds ratios (OR) increase with smoking intensity, but can level off and even decline at high intensities.3 Standard approaches are problematic for evaluating intensity effects, since they are based on joint ORs for smoking duration and intensity, which can lead to confounding of intensity effects by total exposure. For example, in a typical logistic regression the intensity parameter represents the effect per unit intensity (i.e. ln(OR) per cigarette per day) at a fixed duration. The same is true for the duration parameter but at a fixed intensity. A comparison of ORs at two different intensities therefore reflect not only the different intensities but also different total pack-years since duration is fixed. For example, at 30-years duration, ORs for 20 and 30 cigarettes per day incorporate effects of different total exposures, i.e. 30 and 45 pack-years, respectively. Thus, the intensity parameter does not represent a ‘pure’ intensity effect but embeds an additional effect of total pack-years. To address this limitation, investigators developed a three-parameter model for the excess odds ratio (EOR) for lung cancer based on total pack-years and intensity.4 The model describes the EOR/pack-year in terms of the delivery rate of exposure, i.e. for a given pack-years, risk from exposure delivered at low intensity compared with high intensity. Below 15–20 cigarettes per day, there is a direct exposure rate effect, whereby the EOR/pack-year increases with increasing intensity, i.e. for fixed pack-years increasing intensity (or decreasing duration) increases risk. Above 20 cigarettes per day, there is an inverse exposure rate effect, whereby the EOR/pack-year decreases with increasing intensity, i.e. for fixed pack-years increasing intensity (or decreasing duration) decreases risk.
We extend the model to include NAT2 status and reanalyse the SBCS data to consider (i) whether enhanced effects in NAT2 slow acetylators result from an interaction with smoking intensity or total pack-years or both and (ii) whether the interaction is exposure-dependent.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
Study data
The SBCS was a hospital-based case-control study conducted in 18 hospitals from five areas in Spain between 1998 and 2001.2,5 Cases were incident, histologically confirmed carcinomas of the urinary bladder between ages 21–80 years. Controls were selected from patients admitted to participating hospitals for diagnoses unrelated to exposures of interest, and frequency matched on age, sex, ethnicity and region. The study enrolled 1150 cases and 1149 controls. Collection of blood or buccal cell samples, DNA extraction, genotype assays and specification of NAT2 rapid, intermediate and slow acetylator status were described previously.
Never-smokers were subjects who smoked fewer than 100 cigarettes in their lifetime.
The study was approved by the applicable institutional review boards.
Models
We extend the linear-exponential model for total pack-years of smoking (d) and mean cigarettes smoked per day (n) to include NAT2 phenotype (s, where s = 1 and s = 0 denote slow and rapid/intermediate acetylators, respectively). We define I intensity categories and ni, i = 1,...,I, indicator variables, where ni = 1 for intensity n within the i-th category and zero otherwise. The model is
|
| (1) |
) is the smoking adjusted OR for slow acetylators compared with rapid/intermediate acetylators. Within the i-th intensity category, ORs are linear in d (i.e. OR = 1 + 
id), where i is the EOR/pack-year. Linearity of the ORs by pack-years is demonstrated below. For calculating estimates and 95% confidence intervals (CIs), 
replaces i to avoid range restrictions. Model (1) is multiplicative in NAT2 phenotype and smoking. We evaluate an interaction of NAT2 status with pack-years, and hence departures from a multiplicative interaction, by including the factor exp(
s), namely, |
|
) represents the proportional effect for slow acetylators on the EOR/pack-year. We evaluate an interaction with smoking intensity by including exp(is), namely, |
|
i) represents the intensity-specific proportional effect for slow acetylators.
For a model with continuous intensity, we first factor out i from model (1) to obtain
|
|
|
| (2) |
1 ln(n) + 2 ln(n)2} and ß is the EOR/pack-year at g(n) = 1. This form for g(·) provided the best fit compared with a variety of alternative choices.4 We evaluate interactions by replacing ßdg(n) with ßdgs(n), ßsdg(n) or ßsdgs(n), where subscript ‘s’ denotes a separate parameter (ß) or set of parameters (1 and 2) for each phenotype. The difference in deviance relative to model (2) (or relative to ßsdgs(n)) provides a likelihood ratio test of homogeneity of the pack-year and/or intensity effects over NAT2 status. Degrees of freedom equal the difference in numbers of parameters. Note that this is a shorthand notation, where for example ßsdgs(n) represents 
i Ki(s) ßsdgs(n), with Ki(s) taking value one if s = i and zero otherwise. We fit models using Epicure,6 adjusting for age, sex, region, ever employed in a high-risk occupation, and total fruit and vegetable intake. After deleting subjects with incomplete data, analyses include 931 cases and 978 controls.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
Figure 1 shows ORs for bladder cancer by categories of pack-years and intensity, with CIs omitted for clarity. Within intensity categories, ORs increase with pack-years. The fitted linear models suggest greater increased risks with pack-years for slow acetylators compared with rapid/intermediate acetylators. In smokers of 40+ cigarettes per day, the OR for the highest pack-years category in slow acetylators deviates markedly from ORs at lower pack-years, and strongly influences the fitted line. We therefore repeat analyses omitting 27 cases and 30 controls with 120 or more pack-years of exposure. The slope for slow acetylators (dot line) increases substantially when we limit smoking to under 120 pack-years. The figure reveals three important features: (i) trends are approximately linear within intensity categories, (ii) slopes vary with intensity category, and (iii) the relative distance between slopes increase with intensity, suggesting an intensity-dependent interaction with NAT2 phenotype.
|
Using model (1), a comparison of models M0 and M1 in Table 1 rejects homogeneity of slopes across intensity categories (P < 0.001), with EOR/pack-year estimates decreasing with increasing intensity. Adjusted for smoking (model M0), OR = 1.46 (95% CI 1.2–1.8) for slow acetylators. Model M2 indicates no significant effect of NAT2 status in never-smokers (OR = 1.13, 95% CI 0.7–1.8), but an enhanced effect for slow acetylators, i.e. exp(
) = 1.43 (95% CI 0.7–2.7) times the EOR/pack-year for rapid/intermediate acetylators. However, a multiplicative model M1 is not rejected (P = 0.25). Parameter estimates for model M3 show progressively greater effects for slow acetylators with intensity categories 1.08, 1.30, 1.77 and 2.19, respectively, although homogeneity (M3 and M1) is not rejected (P = 0.35). For subjects smoking under 120 pack-years, the interaction of NAT2 status and smoking is suggestive (P = 0.15) (M2 and M1, results not shown), and the proportional effects for slow acetylators for model M3 are 1.13, 1.33, 1.82 and 3.74, respectively, with P = 0.06 and 0.05 for the three (M3 and M2) and four (M3 and M1) degrees of freedom tests of homogeneity of NAT2 phenotype, respectively.
|
Figure 2 shows EOR/pack-year estimates (
i) for an expanded number of intensity categories by NAT2 status for all subjects (upper panels) and for subjects smoking under 120 pack-years (lower panels). Model (2) closely corresponds to category-specific EOR/pack-year estimates, and suggests a greater risk for slow acetylators. There are limited data under 10–15 cigarettes per day and patterns are very uncertain.
|
Comparisons of deviances of model (2) with ßdgs(n), ßsdg(n) and ßsdgs(n) suggest that variations of smoking risk with NAT2 status result from interactions with smoking intensity, and not total pack-years. In the restricted data, the decrease in deviance compared with ßdg(n) (larger decreases denote improved fit) is 6.1 on two degrees of freedom for ßdgs(n) (P = 0.05, homogeneity of gs(·) is rejected), while the deviance decrease is 2.1 on one degree of freedom for ßsdg(n) (P = 0.15, homogeneity of ßs is not rejected). Relative to the ‘full’ model, ßsdgs(n), the deviance increase (larger increases denote degraded fit) is 7.6 for ßsdg(n) (P = 0.02, homogeneity of gs(
) is rejected) and is 3.6 for ßdgs(n) (P = 0.06, homogeneity of ßs is nearly rejected).
Heterogeneity of gs() indicates that the relative difference between curves vary with intensity. For example, at 20, 40 and 60 cigarettes per day, the fitted EOR/pack-year estimates for slow acetylators are 1.41(=0.179/0.127), 1.91(=0.086/0.045) and 2.64(=0.044/0.017) times the fitted estimates for rapid/intermediate acetylators (Figure 2, upper panels, cross symbol), respectively, although homogeneity is not rejected (P = 0.37). In the restricted data, the relative effects of the fitted estimates for slow acetylators at 20, 40 and 60 cigarettes per day are 1.37(=0.176/0.129), 3.50(=0.112/0.032) and 9.94(=0.078/0.008), and homogeneity is rejected (P = 0.02) (Figure 2, lower panels, cross symbol).
In our data set, 83% of smokers with known tobacco type report smoking black tobacco. In addition, ORs of bladder cancer relative never-smokers are similarly elevated for black-tobacco-only smokers, black-and blond-tobacco smokers, and smokers of unknown tobacco type. We combine these subgroups as known or likely black-tobacco smokers. Since only 76 cases and 106 controls smoked blond tobacco exclusively, we are unable to evaluate model (2) comprehensively by type of tobacco. However, using a multiplicative factor for the EOR in model (2) based on an indicator variable, with one denoting known or likely black-tobacco smokers and zero otherwise, the EOR/pack-years is increased by 1.75 (95% CI 1.1–2.9) for black-tobacco smokers relative to blond-tobacco smokers. Including indicator variables for type of tobacco by NAT2 status, the effect on the EOR of slow acetylators relative to rapid/intermediate acetylators is 1.51 (95% CI 0.8–2.8) in black-tobacco smokers and 1.28 (95% CI 0.4–3.9) in blond-tobacco smokers, although homogeneity of these estimates is not rejected (P = 0.73).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
Our analysis of the SBCS reveals an inverse exposure rate effect (reduced potency) above
15 cigarettes per day, i.e. for fixed total pack-years increasing intensity (or decreasing duration) decreases risk, which is similar to the intensity pattern for lung cancer.4 The results confirm a previous analysis2 that the effect of NAT2 phenotype on urinary bladder cancer risk is dependent upon smoking intensity, with a greater relative effect in slow acetylators at higher smoking intensity. This finding is consistent with the recent report of Gu et al.7 that the effect of NAT2 phenotype on urinary bladder cancer risk increased in heavy smokers compared with light smokers. There was also suggestive evidence that the increased effect in slow acetylators may be greater in black-tobacco smokers. The results are in keeping with our current understanding that NAT2 detoxifies aromatic amine carcinogens such as 4-aminobiphenyl (4-ABP) present in cigarette smoke. As such, a higher biologically toxic dose of aromatic amine carcinogen would be expected in slow acetylators compared with rapid/intermediate acetylators following equivalent smoking intensities. Further, the intensity related effects observed in these studies suggest a certain aromatic amine dose that must be achieved in order for NAT2 to effectively compete with other higher affinity enzymatic pathways (e.g. glucuronidation)8–10, thus enabling individuals with rapid/intermediate acetylator phenotype to more effectively detoxify aromatic amine carcinogens in cigarette smoke than those with the slow acetylator phenotype. In contrast, Vineis et al.11 reported results from a cross-sectional study of healthy individuals that 4-ABP haemoglobin adducts were higher in slow than in rapid/intermediate acetylators at low or non-detectable nicotine–cotinine levels, whereas the difference between slow and rapid/intermediate acetylators was less evident at increasing nicotine-cotinine levels. This suggested that NAT2 detoxification of the aromatic amines present in cigarette smoke could become saturated at higher smoking intensity, thus minimizing the beneficial NAT2 detoxification present in individuals with the rapid/intermediate NAT2 acetylator phenotype. However, determinants of haemoglobin adduct levels in peripheral blood may not be the same as factors that influence the formation of 4-ABP DNA adducts in urothelial tissue, the target site for bladder carcinogenesis.
Interestingly, while the effect of tobacco on risk of bladder cancer appears to differ by NAT2 acetylation status, particularly at high tobacco intensity, we observe that the reduced potency effect occurs in both rapid/intermediate and slow acetylators. In other words, for intensities above 15 cigarettes per day, there is a decreasing EOR/pack-year with increasing intensity, a pattern consistent with reduced exposure potency or a ‘wasted’ exposure effect, which is apparent in both NAT2 acetylation groups (Figure 2). The EOR/pack-year patterns likely reflect both biological phenomena, as well as influences of nicotine satiation whereby carcinogenic yield per cigarette decreases with increasing intensity as smokers seek to maintain addiction-sufficient nicotine levels resulting in cigarettes per day increasingly overestimating internal exposure rate.
Differential effects of acetylation status by amount smoked have been suggested previously,7 but to our knowledge have not been evaluated formally. Our approach provides a framework for evaluating smoking and NAT2 interactions, and suggests that variations in smoking risk by NAT2 status result from an interaction with smoking intensity and not total pack-years. Smoking less than 15 cigarettes per day carries relatively little additional risk for NAT2 slow acetylators, while smoking more than 15–20 cigarettes per day results in increasing effects on the EOR/pack-year for bladder cancer, compared with rapid/intermediate acetylators. However, for less than 10 cigarettes per day, patterns for EOR/pack-year are uncertain, since there are few subjects (62 cases and 111 controls) and the range of pack-years is limited (median and interquartile range are 8.4 and 4.3–9.5 pack-years, respectively). In summary, our analysis provides evidence of an enhanced effect for smoking intensity and bladder cancer in NAT2 slow acetylators that increases with intensity.
|
Acknowledgement |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
This research was supported by the Intramural Research Program of the National Cancer Institute, National Institutes of Health. This project was also partly funded by the Spanish Ministry of Health (GO3/174, PI061614, CO3/09, and C03/10), the Fundació Marató TV3 and the European Union (BMH4-98-3243).
Conflict of interest: None declared.
Key Messages
|
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementReferences |
|---|
1 Vineis P, Marinelli D, Autrup H, Brockmoller J, Cascorbi I, Daly AK, et al. Current smoking, occupation, N-acetyltransferase-2 and bladder cancer: a pooled analysis of genotype-based studies. Cancer Epidemiol Biomarkers Prev (2001) 10:1249–52.
2 Garcia-Closas M, Malats N, Silverman D, Dosemeci M, Kogevinas M, Hein DW, et al. NAT2 slow acetylation, GSTM1 null genotype, and risk of bladder cancer: results from the Spanish Bladder Cancer Study and meta-analyses. Lancet (2005) 366:649–59.[CrossRef][Web of Science][Medline]
3 Vineis P, Kogevinas M, Simonato L, Brennan P, Boffetta P. 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-Rev in Mutat Res (2000) 463:103–10.
4 Lubin JH, Caporaso N. Cigarette smoking and lung cancer: modeling total exposure and intensity. Cancer Epidemiol Biomarkers Prev (2006) 15:517–23.
5 Samanic C, Kogevinas M, Dosemeci M, Malats N, Real FX, Garcia-Closas M, et al. Smoking and bladder cancer in Spain: effects of tobacco type, timing, environmental tobacco smoke and gender. Cancer Epidemiol Biomarkers Prev (2006) 15:1348–54.
6 Preston DL, Lubin JH, Pierce DA, McConney ME. Epicure User's Guide (2006) Seattle, Washington, USA: HiroSoft International Corporation.
7 Gu J, Liang D, Wang YF, Lu C, Wc XF. Effects of N-acetyl transferase 1 and 2 polymorphisms on bladder cancer risk in Caucasians. Mutat Res-Genetic Toxicol Environ Mutagen (2005) 581:97–104.[CrossRef]
8 Babu SR, Lakshmi VM, Huang GPW, Zenser TV, Davis BB. Glucuronide conjugates of 4-aminobiphenyl and its N-hydroxy metabolites–pH stability and synthesis by human and dog liver. Biochem Pharmacol (1996) 51:1679–85.[CrossRef][Web of Science][Medline]
9 Green MD, King CD, Mojarrabi B, Mackenzie PI, Tephly TR. Glucuronidation of amines and other xenobiotics catalyzed by expressed human UDP-glucuronosyltransferase 1A3. Drug Metab Dispos (1998) 26:507–12.
10 Al Zoughool M, Talaska G. High-performance liquid chromatography method for determination of N-glucuronidation of 4-aminobiphenyl by mouse, rat, and human liver microsomes. Anal Biochem (2005) 340:352–58.[CrossRef][Web of Science][Medline]
11 Vineis P, Bartsch H, Caporaso N, Harrington AM, Kadlubar FF, Landi MT, et al. Genetically based N-acetyltransferase metabolic polymorphism and low-level environmental exposure to carcinogens. Nature (1994) 369:154–56.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Baris, M. R. Karagas, C. Verrill, A. Johnson, A. S. Andrew, C. J. Marsit, M. Schwenn, J. S. Colt, S. Cherala, C. Samanic, et al. A Case-Control Study of Smoking and Bladder Cancer Risk: Emergent Patterns Over Time J Natl Cancer Inst, November 18, 2009; 101(22): 1553 - 1561. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. Lubin, M. Purdue, K. Kelsey, Z.-F. Zhang, D. Winn, Q. Wei, R. Talamini, N. Szeszenia-Dabrowska, E. M. Sturgis, E. Smith, et al. Total Exposure and Exposure Rate Effects for Alcohol and Smoking and Risk of Head and Neck Cancer: A Pooled Analysis of Case-Control Studies Am. J. Epidemiol., October 15, 2009; 170(8): 937 - 947. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Lubin, J. Virtamo, S. J. Weinstein, and D. Albanes Cigarette Smoking and Cancer: Intensity Patterns in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study in Finnish Men Am. J. Epidemiol., April 15, 2008; 167(8): 970 - 975. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. Lubin, N. Caporaso, D. K. Hatsukami, A. M. Joseph, and S. S. Hecht The Association of a Tobacco-Specific Biomarker and Cigarette Consumption and Its Dependence on Host Characteristics Cancer Epidemiol. Biomarkers Prev., September 1, 2007; 16(9): 1852 - 1857. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Rothman, M. Garcia-Closas, and D. W Hein Commentary: Reflections on G. M. Lower and colleagues' 1979 study associating slow acetylator phenotype with urinary bladder cancer: meta-analysis, historical refinements of the hypothesis, and lessons learned Int. J. Epidemiol., February 1, 2007; 36(1): 23 - 28. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||