Introduction
Epidemiologic evidence linking cigarette smoking to increased risk for development of breast cancer is mounting (for review [
1,
2]). In addition, findings from both epidemiology and molecular biology indicate that there is differential susceptibility within the population to development of malignant neoplasms following exposure to certain xenobiotics because of polymorphisms in genes that encode metabolizing enzymes.
Previously, we reported a differential effect of
N-acetyltransferase (
NAT) 2 genotype on the association between active and passive smoking and breast cancer risk [
3]. The identification of passive smoking as a breast cancer risk factor, particularly for fast acetylators, implied that heterocyclic aromatic amines (HCAs) are among the responsible carcinogens. HCAs are particularly abundant in side-stream tobacco smoke [
4] and are activated by
O-acetylation catalyzed by NATs [
5]. Because sulfotransferase (SULT)1A1 (encoded by
SULT1A1) is also involved in the metabolism of pro-carcinogens from tobacco smoke, the additional investigation of a common polymorphism in the
SULT1A1 gene might provide deeper insight into the modification of susceptibility to breast cancer.
The SULT1A1 enzyme is generally associated with detoxification of xenobiotic compounds and has been implicated in oestrogen metabolism. However, Glatt and coworkers [
6,
7] showed that several substances can be activated by the conjugation reaction with SULT1A1, among which are pro-carcinogens such as polycyclic aromatic hydrocarbons and HCAs, both of which are present in tobacco smoke [
8,
9].
In contrast to earlier assumptions [
10], there is increasing evidence that the SULT1A1 enzyme apparently does not play an important role in oestrogen metabolism
in vivo. Results from
in vitro studies showed that only the SULT1E1 enzyme is capable of the sulfonation of oestradiol, oestrone and catecholestrogens at physiologically relevant concentrations [
11,
12]. For instance, Adjei and Weinshilboum [
11] showed that K
m values for the sulfonation of oestradiol with SULT1A1 were in the micromolar range, which is clearly above physiological concentrations, whereas the K
m value for SULT1E1 was considerably lower (0.029 ± 0.01 μmol/l).
Large interindividual variations in the biochemical and metabolic properties of the SULT1A1 enzyme have been observed that can partly be explained by a G to A polymorphism at nucleotide 638 (Arg
213His), referred to as the
*2 allele. The
*2 allele has been associated with lower activity and lower thermal stability of the SULT1A1 enzyme [
13,
14], and thus reduced bioactivation of mutagens [
6].
Thus far, three case–control studies that investigated the association between
SULT1A1 genotype and breast cancer risk [
15‐
17] have been reported. Results were not consistent and the effect of smoking was not considered in any of those case–control studies. Saintot and coworkers [
18] recently reported a positive interaction between smoking and the variant allele for
SULT1A1 with respect to breast cancer risk in a case-only study.
We conducted the present study to elucidate the potential role of SULT1A1 genotype alone and in combination with NAT2 genotype as a modifier of susceptibility to breast cancer associated with exposure to tobacco smoke among predominantly premenopausal women.
Results
The women included in the present study, for whom a comprehensive history of active and passive smoking was available, closely resemble the original study population with respect to the distributions of several sociodemographic characteristics and putative risk factors, such as age, family history of breast cancer, body mass index, education level, parity, menopausal status, alcohol consumption, smoking and breastfeeding (data not shown).
Selected characteristics of the present study population are summarized in Table
1. The mean (± standard deviation) age for breast cancer patients was 42.9 ± 5.5 years and that for control individuals was 42.7 ± 5.6 years. The frequency of the
SULT1A1*2 allele was 0.33 among cases and 0.35 among controls (0.32 and 0.34 in the original population). Of cases and control individuals, 52.7% and 57.7%, respectively, were carriers of at least one
SULT1A1*2 allele. The distribution of
SULT1A1 genotypes was in Hardy–Weinberg equilibrium (
P = 0.92 for control individuals,
P = 0.09 for cases).
Table 1
Selected characteristics of breast cancer patients and controls
Mean age (years) | 42.9 | 42.7 |
Mean body mass index | 23.9 | 24.3 |
First-degree family history of breast cancer | 62 (14.8%) | 53 (6.0%) |
Menopausal status | | |
Premenopausal | 325 (77.6%) | 718 (81.2%) |
Postmenopausal | 25 (6.0%) | 49 (5.5%) |
Unknown | 69 (16.5%) | 117 (13.2%) |
SULT1A1 genotype | | |
*1/*1
| 198 (47.3%) | 374 (42.3%) |
*1/*2
| 169 (40.3%) | 403 (45.6%) |
*2/*2
| 52 (12.4%) | 107 (12.1%) |
NAT2 genotypea
| | |
Fast acetylator | 177 (42.2%) | 351 (39.7%) |
Slow acetylator | 242 (57.8%) | 533 (60.3%) |
The overall risk for breast cancer among carriers of the SULT1A1*2 allele was not significantly different from that in women with the SULT1A1*1/*1 genotype (adjusted OR 0.83, 95% CI 0.66–1.06). The distributions of potential risk factors, such as first-degree family history of breast cancer, body mass index, alcohol consumption, menopausal status, parity and breastfeeding, were similar in carriers and noncarriers of the SULT1A1*2 allele. There was also no major effect of SULT1A1 genotype in combination with NAT2 acetylator status on breast cancer risk (data not shown).
We assessed the effect of
SULT1A1 genotype on the association between smoking and breast cancer risk, initially comparing ever active smokers with nonsmokers (i.e. passive-only smokers were included in the reference group). The ORs for variables such as smoking status (current or former active smoker), duration and pack-years of smoking did not differ by
SULT1A1 genotype (data not shown). We then considered a separate category of only passively exposed women, with a reference group comprising women with neither active nor passive cigarette smoke exposure (Table
2). Associations of breast cancer risk with smoking variables were apparent, but the risk estimates were similar for carriers and for noncarriers of the
SULT1A1*2 allele. In the analysis of passive smoking among never active smokers, we observed a tendency toward higher ORs in women with the
SULT1A1*1/
*1 genotype compared with carriers of the
SULT1A1*2 allele (Table
2). The test for interaction between
SULT1A1 genotype and passive smoking was not statistically significant (
P = 0.6).
Table 2
Odds ratios for breast cancer associated with smoking variables stratified by SULT1A1 genotype
Never active | 86 | 159 | 1 (ref) | 88 | 209 | 1 (ref) |
Ever active | 112 | 215 | 0.92 (0.63–1.34) | 133 | 301 | 1.01 (0.73–1.42) |
Never active/passivec, d
| 20 | 56 | 1 (ref) | 19 | 57 | 1 (ref) |
Former active | 44 | 94 | 1.23 (0.64–2.36) | 59 | 157 | 1.11 (0.60–2.06) |
Current active | 68 | 121 | 1.44 (0.76–2.71) | 74 | 144 | 1.51 (0.82–2.78) |
Average cigarettes per dayc, e
|
>0–9 | 42 | 106 | 1.06 (0.55–2.03) | 52 | 154 | 0.99 (0.54–1.84) |
10–19 | 56 | 73 | 2.12 (1.10–4.07) | 60 | 99 | 1.75 (0.94–3.25) |
20+ | 14 | 34 | 1.05 (0.45–2.46) | 20 | 48 | 1.09 (0.51–2.32) |
Duration of smokingc, d
|
1–15 years | 36 | 80 | 1.19 (0.61–2.34) | 46 | 145 | 0.97 (0.52–1.82) |
16+ years | 76 | 135 | 1.44 (0.77–2.69) | 87 | 156 | 1.62 (0.89–2.95) |
Age at first cigarettec, d
|
9–15 years | 21 | 47 | 1.09 (0.51–2.35) | 20 | 63 | 0.84 (0.40–1.79) |
16–18 years | 56 | 109 | 1.33 (0.70–2.52) | 70 | 145 | 1.42 (0.77–2.62) |
19+ years | 35 | 59 | 1.51 (0.75–3.05) | 43 | 93 | 1.42 (0.74–2.72) |
Pack-yearsc, e
|
>0–10 | 59 | 129 | 1.22 (0.65–2.27) | 76 | 191 | 1.15 (0.64–2.07) |
11+ | 53 | 84 | 1.77 (0.91–3.42) | 56 | 110 | 1.43 (0.77–2.67) |
Passive smokingf
|
No | 20 | 57 | 1 (ref) | 19 | 57 | 1 (ref) |
Yes | 66 | 102 | 1.69 (0.89–3.21) | 69 | 152 | 1.40 (0.74–2.64) |
Duration of passive exposure in adulthoodf, g, h
|
1–13 years | 29 | 40 | 1.91 (0.91–4.00) | 33 | 70 | 1.47 (0.72–2.98) |
14+ years | 32 | 47 | 1.66 (0.79–3.48) | 29 | 61 | 1.47 (0.70–3.10) |
Cumulative lifetime exposure (in hours/day-years)f, g, i
|
1–55 | 22 | 49 | 1.13 (0.52–2.38) | 37 | 76 | 1.76 (0.89–3.50) |
56+ | 43 | 52 | 2.12 (1.06–4.25) | 32 | 71 | 1.39 (0.68–2.83) |
We investigated the combined effect of
SULT1A1 and
NAT2 genotype with respect to smoking, and observed elevated ORs associated with passive smoking only (OR 3.23, 95% CI 1.05–9.92) in NAT2 fast acetylators with the
SULT1A1*1/
*1 genotype but not in NAT2 fast acetylators carrying the
SULT1A1*2 allele (Table
3). There was also a difference in OR for 11 or more pack-years of active smoking by
SULT1A1 genotype, but the risk estimates were not significant. The test for interaction between
SULT1A1 genotype and active/passive smoking among NAT2 fast acetylators did not reach statistical significance (
P = 0.4). Among NAT2 slow acetylators, risk estimates for active and passive smoking did not differ by
SULT1A1 genotype.
Table 3
Odds ratios for breast cancer associated with smoking stratified by NAT2 acetylator status and SULT1A1 genotype
NAT2 fast acetylatorc
|
Never active/passive | 6 | 20 | 1 (ref) | 8 | 18 | 1 (ref) | 14 | 38 | 1 (ref) |
Passive only | 30 | 37 | 3.23 (1.05–9.92) | 37 | 59 | 1.28 (0.50–3.31) | 67 | 96 | 1.95 (0.97–3.91) |
1–10 pack-years | 22 | 53 | 1.26 (0.40–3.95) | 30 | 74 | 0.76 (0.29–1.96) | 52 | 127 | 0.98 (0.48–1.98) |
11+ pack-years | 23 | 38 | 2.21 (0.68–7.12) | 20 | 50 | 0.75 (0.27–2.06) | 43 | 88 | 1.22 (0.59–2.54) |
NAT2 slow acetylator |
Never active/passive | 14 | 33 | 1 (ref) | 11 | 39 | 1 (ref) | 25 | 75 | 1 (ref) |
Passive only | 36 | 65 | 1.35 (0.62–2.91) | 32 | 93 | 1.18 (0.53–2.66) | 68 | 158 | 1.25 (0.72–2.16) |
1–10 pack-years | 37 | 76 | 1.24 (0.57–2.69) | 46 | 117 | 1.34 (0.62–2.92) | 83 | 193 | 1.24 (0.72–2.12) |
11+ pack-years | 30 | 46 | 1.83 (0.79–4.26) | 36 | 60 | 2.07 (0.91–4.72) | 66 | 106 | 1.85 (1.04–3.30) |
The results were generally similar when the analysis was restricted to the subgroup of women with premenopausal status, although the confidence intervals were wider. With regard to the differential effect of SULT1A1 and NAT2 genotype, the OR for passive smoking was 2.24 (95% CI 0.68–7.35) in NAT2 fast acetylators with the SULT1A1*1/*1 genotype and 1.03 (95% CI 0.39–2.71) in fast acetylators carrying the SULT1A1*2 allele.
Discussion
Our data do not suggest a strong influence of SULT1A1 genotype alone or in combination with NAT2 on the risk for breast cancer. There is no clear evidence that the SULT1A1 Arg213His single nucleotide polymorphism investigated in this study in itself is an important effect modifier of breast cancer risk associated with active/passive smoking among women up to age 50 years.
Differences in risk estimates for carriers and noncarriers of the SULT1A1*2 allele associated with smoking were apparent among NAT2 fast acetylators but not among slow acetylators. The observed estimates indicated that fast acetylators with the SULT1A1*1/*1 genotype were at higher risk for breast cancer than were carriers of the SULT1A1*2 allele when exposed to tobacco smoke, with a particularly prominent increase in risk for passive smokers versus never active/passive smokers.
We cannot rule out the possibility that the observed risk elevation for
SULT1A1*1/
*1 among fast acetylators was due to chance, because confidence intervals were wide for the combined analysis of genotypes. However, it seems biologically plausible that the combination of
NAT2 and
SULT1A1 'fast' genotypes is unfavourable. Both enzymes have been shown to be capable of bioactivating several pro-carcinogens. NAT2 is thought to play a major role in the activation of
N-hydroxy derivatives of HCAs by
O-acetylation [
5]. The sulfonation of a variety of xenobiotics or their metabolites, such as polycyclic aromatic hydrocarbons, HCAs and aromatic amines, can lead to short-lived conjugates that may react with DNA and other cellular nucleophiles [
23]. Studies that investigated the genotype–phenotype correlation for
SULT1A1 clearly indicate that the
SULT1A1*2 allele is associated with decreased catalytic activity of the respective allozyme as compared with the sulfonation activity of the wild-type SULT1A1 enzyme [
6,
13,
14]. Consequently, in individuals with this genotype combination, reactive metabolites from acetylation and sulfonation might accumulate and lead to greater DNA damage and increase tumourigenesis.
For instance, DNA adducts of 2-amino-3-methylimidazo [4,5-
f]quinoline (IQ) and 2-amino-1-methyl-6-phenylimidazo [4,5-
b]pyridine (PhIP) have been detected in human breast milk [
24]. These HCAs, which are also present in tobacco smoke, have been classified as probably and possibly carcinogenic to humans, respectively [
25]. In mutagenicity assays after heterologous expression of NAT2 and SULT1A1 in
Salmonella typhimurium,
N-hydroxy-IQ was found to be efficiently activated by NAT2 whereas
N-hydroxy-PhIP was specifically activated by the SULT1A1 enzyme [
26]. Likewise, mutagenicity of 2-amino-3-methyl-9
H-pyrido [2,3-
b]indole, another abundant HCA, was strongly enhanced in a
Salmonella typhimurium strain expressing SULT1A1 [
27].
Consistent with a role of greater bioactivation of HCAs associated with
SULT1A1*1 rather than
SULT1A1*2 in carcinogenesis are previous reports of greater risk for breast cancer and prostate cancer associated with intake of well done red meat [
14,
15]. In accord with this notion are the findings of three studies that investigated the formation of DNA adducts of heterocyclic and aromatic amines [
28‐
30]. All of them found a tendency toward a higher capacity for adduct formation for the SULT1A1*1 enzyme as compared with the *2 allozyme, which is in agreement with a more efficient activation of several pro-mutagens by SULT1A1*1 than by the *2 allelic variant reported by Glatt and coworkers [
6].
Two previous studies [
15,
17] reported an increased risk for breast cancer associated with the
SULT1A1*2 allele
per se. Risk estimates were statistically significant only in the study conducted by Zheng and coworkers [
15], a nested case–control study in postmenopausal women, which found an 80% elevated risk for homozygous carriers of the variant allele. We found no association with the
SULT1A1*2 variant although our study had 93% power to detect an OR of equivalent magnitude at a significance level of α = 0.05.
In a recent case-only study, an interaction between the
SULT1A1 polymorphism and tobacco smoke exposure with an OR for interaction of 2.55 (95% CI 1.21–5.36) for current smokers carrying the
*2 allele was found [
18]. Results from our case–control study did not provide an indication for such a strong interaction between
SULT1A1 genotype and smoking. Accordingly, we failed to detect a significant interaction in a case-only analysis of our data, although the power of our study is similar and the precondition of independence between genotype and exposure in the general population was fulfilled. The ORs (95% CIs) for interaction were 1.04 (0.50–2.14) for passive smoking only, 1.34 (0.63–2.86) for former smoking, and 1.15 (0.55–2.39) for current smoking.
We feel confident that our study population is representative of the general German population. The observed allele frequencies for
SULT1A1 are in accordance with previous studies conducted in Caucasian populations [
31] and the
SULT1A1 genotype distribution did not deviate from Hardy–Weinberg equilibrium. Re-contacting the study participants for the telephone interview might have introduced selection bias. However, the participants in the present study closely resemble the original study population with regard to the distributions of relevant characteristics. Also, we do not believe that recall bias is a major concern because smoking was not known to be associated with breast cancer at the time of the interviews, and the correlation of reported active smoking between the original study and the present study is high [
20]. Moreover, previous studies showed that the validity for self-reported active smoking, as well as passive smoke exposure, is high and nondifferential in cases and controls [
32,
33].
Although
in vitro data suggest that SULT1A1 may not play an important role at physiologically relevant oestrogen concentrations [
11,
12], the question regarding whether
SULT1A1 genotype actually has an effect on oestrogen metabolism
in vivo deserves further study. Concerning the expression of SULT1A1 and SULT1E1 enzymes, for instance, there is some controversy in the literature. Falany and coworkers [
34,
35] observed SULT1E1 expression in normal breast epithelial cells, whereas Williams and coworkers [
36] reported that only SULT1A1 was expressed at detectable levels. Because we cannot definitely rule out a potential role of SULT1A1 in the metabolism of oestrogens, we analyzed our data also with respect to use of oral contraceptives and various reproductive factors that may alter exposure to oestrogens. The results did not provide any indication for a modification of breast cancer risk related to oestrogens by
SULT1A1 genotype (data not shown) and corroborate recent evidence indicating that sulfonation of oestrogens catalyzed by SULT1A1 is less relevant in normal breast tissue in physiological conditions [
11,
12].
The inconsistent findings of previous studies, which also considered the possible involvement of the
SULT1A1 gene in other cancer sites (summarized by Glatt and Meinl [
23]), and the broad substrate specificity of the SULT1A1 enzyme indicate the complexity of the issue. Elucidation of the potential effects of
SULT1A1 genotype on a hormone-related cancer, such as breast cancer, is rendered more complicated by the fact that smoking may alter oestrogen levels in the body [
37‐
40]. Moreover, and as suggested by our findings, it is possible that
SULT1A1 genotype only exerts a detectable effect in combination with other genes, not to mention several polymorphic genes that are involved in oestrogen metabolism. Further determinants such as varying levels of enzyme expression or enzyme induction, which cannot easily be assessed in epidemiological studies, might also be of importance. Nevertheless, we cannot exclude that, because of limitations in statistical power, we were unable to detect a potential weak or moderate association or interaction between
SULT1A1 genotype, smoking and breast cancer, independent of
NAT2 genotype.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CL performed the statistical analysis and drafted the manuscript. AR was responsible for the genotyping assays, and contributed to study design and manuscript preparation. SK conducted the re-contacting of study participants in 1999 and participated in the statistical analyses. JCC conceived the study and supervised the project. All authors read and approved the final manuscript.