Prevalence and Spectrum of p53 Mutations Associated with Smoking in Breast Cancer¹

Breast cancer is the second leading cause of cancer-related deaths in American women (1). Epidemiological evidence suggests that breast cancer development is influenced by a variety of environmental and behavioral factors. The majority of breast cancer risk factors identified to date are reproductive or hormonal factors that increase estrogen exposure; however, these generally demonstrate only modest associations with breast cancer (2).

Cigarette smoking is a well-known risk factor for several cancers, including those of the lung (3), head and neck (4, 5), and bladder (6). Some studies suggest that female smokers sustain more genetic damage (7, 8, 9) and are at greater risk of developing lung or bladder cancer (19, 20, 12) compared with males exposed to comparable quantities of tobacco smoke. Although smoking is a major risk factor for some cancers, the relationship between smoking and breast cancer risk has remained controversial despite considerable research. Most epidemiological studies on smoking and breast cancer have demonstrated a weak positive (13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27) or null association (28, 29, 30, 31, 32), whereas others detected a weak inverse association (33, 34, 35, 36, 37).

Among positive studies, increases in breast cancer risk were seen among subsets of women, including those who started smoking at an early age (13, 24, 25), those who were heavy smokers and/or who had smoked for a long time (13, 14, 15, 16, 17), postmenopausal former smokers (19, 20, 21), or those with a high-risk carcinogen-metabolizing enzyme genotype (21). An association between passive tobacco smoke exposure and the occurrence of breast cancer has also been observed in several studies (24, 26, 32). Furthermore, exclusion of passive smokers from the reference group has helped to reveal positive associations between active smoking and breast cancer (22, 24, 26). Some epidemiological studies have found weak protective effects of smoking for either sporadic breast cancer (33, 34, 35, 36) or in carriers of BRCA1 or BRCA2 mutations (37). This putative protective effect is thought to be mediated by the antiestrogenicity of tobacco smoke, which is suggested by studies demonstrating that women who smoke have lower urinary estrogen levels, decreased bone density, and earlier age at menopause than do nonsmokers (38, 39, 40, 41, 42). Cigarette smoking has also been linked to shorter survival among breast cancer patients, although the basis for this effect is unclear (43, 44).

In contrast to the epidemiological studies, results of in vitro, animal, and human biomarker studies strongly suggest that breast tissue is a target for the carcinogenic effects of tobacco smoke. Tobacco smoke contains a mixture of highly mutagenic compounds, including PAHs,³ aromatic amines, and N-nitrosamines, that act in the initiation and/or promotion of neoplasia (45). The tobacco smoke PAH, BaP, is one of the most potent carcinogenic compounds in vivo and in vitro (46) and is a major constituent of tobacco smoke (47). The tobacco-specific nitrosamine NNK [4-(methylnitrososamino)-1-(3-pyridyl)-1-butanone], is a potent lung carcinogen in animals (47), and metabolites of NNK have been detected in nonsmoking women exposed to passive cigarette smoke (48). The PAHs are lipophilic, are stored in adipose tissue (49), are metabolized and activated by human mammary epithelial cells (50), and cause mammary tumors in rodents (51). Human mammary lipid extracts have been shown to induce DNA damage in breast epithelial cells (52), and tobacco-related mutagens have been isolated from the breast fluid of smokers (53). Furthermore, BaP-like DNA adducts have been identified in histologically normal breast tissues of breast cancer patients (54). These observations provide evidence that tobacco smoke constituents reach breast tissue and suggest their potential for involvement in human breast carcinogenesis.

Mutations in the p53 tumor suppressor gene have been found in 15–30% of breast cancers (55, 56, 57) and are thought to be associated with poor clinical prognosis (58, 59, 60, 61, 62, 63, 64, 65). The p53 protein functions as a transcription factor and regulates cell proliferation, DNA repair, differentiation, and apoptosis (66). Studies have demonstrated that certain mutagenic carcinogens leave their “fingerprint” on the DNA in the form of specific patterns of mutation in the p53 gene, revealing important clues for disease etiology (55). Examples of p53 mutational fingerprints associated with specific carcinogen exposures include the high proportion of G:C→T:A transversions in the lung tumors of smokers, the high frequency of G:C→T:A transversions at p53 codon 249 in hepatocellular carcinomas associated with aflatoxin exposure, and CC→TT mutations in skin cancers associated with exposure to UV radiation (55). Previous studies have suggested that the p53 mutational spectrum in breast cancer bears some similarity to both lung cancer and colon cancer, exhibiting a substantial proportion of G:C→T:A transversions, similar to lung cancer (55, 56, 67), but also a high proportion of G:C→A:T transitions at CpG sites as well as deletions and insertions, as in colon cancer (55, 56). These studies, however, did not specifically assess the effects of tobacco smoke exposure on the pattern of p53 mutations in breast cancer.

This report extends the work of previous studies by providing data on the link between cigarette smoking and p53 mutational status and spectrum within breast cancer. Analysis of a population-based series of breast cancer cases from the CBCS (68) demonstrates that smoking is associated with the p53 mutational status of breast tumors and that differences in the p53 mutational spectrum between smokers and nonsmokers are consistent with those found in lung cancer, thus implicating smoking in breast carcinogenesis.

The CBCS is a population-based, case-control study of breast cancer. Participants include women 20–74 years of age residing in 24 contiguous counties of central and eastern North Carolina (68). Women with a first diagnosis of invasive breast cancer between 1993 and 1996 (Phase 1 of the CBCS) were identified by the North Carolina Central Cancer Registry through a rapid case ascertainment system. Women diagnosed before age 50 and African-American women were oversampled to ensure that they comprised roughly half the study sample. Additional details of the study design are described elsewhere (19, 68, 69). All aspects of this research were approved by the University of North Carolina School of Medicine Institutional Review Board. A total of 861 breast cancer cases were eligible for and consented to participate in Phase 1 of the CBCS. Epidemiological risk factor information, including smoking history, was obtained from questionnaires that were administered to participants in their homes by trained nurse-interviewers (19). Smoking status (or ever smoker) was defined on the basis of lifetime exposure to at least 100 cigarettes. Current smokers were women who smoked at the time of diagnosis, whereas former smokers quit smoking any time before diagnosis (19). Clinical data and information on tumor characteristics were obtained from medical records or direct histopathological review of tumor tissue.

Formalin-fixed, paraffin-embedded tumor blocks were obtained from pathology departments at participating hospitals for 798 of the 861 breast cancer cases. Tumors were sectioned as described previously (70) and underwent standardized histopathological review by the study pathologist (J. Geradts). The invasive cancer area was selectively dissected, and DNA lysates were prepared for molecular analyses by a proteinase K extraction method as described previously (71, 72). In the present study, we analyzed the p53 mutational status of breast tumors from 456 of the 798 cases for whom tumor blocks were available. Of the 342 cases not evaluated, 114 were determined by histopathological review to have insufficient tumor volume for molecular analysis, 24 tumors could not be assessed because of poor DNA quality, and the remaining 204 tumors have not yet been evaluated.

Mutations in exons 4–8 of the p53 gene were evaluated by a screening algorithm incorporating SSCP analysis and manual DNA sequencing (71, 72). Briefly, PCR amplification was carried out for individual exons 5–8, whereas exon 4 was amplified as two overlapping segments. SSCP analysis of each exon was conducted at both 25°C and 4°C to increase the sensitivity of mutation detection. Samples exhibiting abnormal band migration at either temperature were sequenced in the forward and reverse DNA strands. Mutations were confirmed by sequencing of a second, separately amplified PCR product to rule out the possibility of artifactual mutations (73). Fifty randomly selected samples that exhibited no abnormalities on SSCP analysis were also sequenced, and in no case did we detect mutations.

Archived data on p53 mutations in breast cancer were downloaded from the most recent version (June 2001) of the IARC p53 Mutation Database.⁴ The p53 information in IARC selected for comparison with CBCS data was from primary breast tumors from females obtained through surgery or biopsy; cell line data were excluded from analysis. Only mutations in exons 5–8 and the flanking introns were evaluated, whereas exon 4 data were excluded because of the incomplete nature of this information in the IARC database.

p53 mutations were evaluated for prevalence and type. Using SAS software (SAS Institute, Cary, NC), we used χ² statistics and logistic regressions to measure the association between p53 mutations and smoking status or other characteristics.

Tumor blocks were obtained for 798 of 861 (93%) breast cancer patients who consented to participate in Phase 1 of the CBCS. Of these, 456 tumors having sufficient quantity and quality of tumor tissue for molecular analysis were evaluated for p53 mutations. Slightly more than half (57.5%) of the 456 breast cancer patients were younger than 50 years of age, and 38.2% were African American, consistent with the sampling scheme for the study (Table 1). The majority of patients had stage I or II disease. Patients evaluated for p53 mutations had somewhat higher clinical stage (38.9% stage 1, 48.0% stage II; P = 0.06) and were more likely to be lymph node positive (41.2%; P = 0.008) than those who were not evaluated (n = 405; 46.1% stage 1, 43.9% stage II; 32.2% node positive). The 456 cases who were evaluated for p53 mutations did not differ from those who were not evaluated with regard to smoking status (P = 0.63).

Former smokers were slightly older, whereas current smokers were somewhat younger and were more frequently white than were never smokers; however, these differences were not statistically significant (Table 1). The breast tumors of ever smokers were not significantly different from those of never smokers with respect to stage, size, node status, ER or PR status, or histological characteristics.

A total of 108 p53 mutations were identified among the 456 breast tumors evaluated, with 1 mutation detected per tumor, for a prevalence of 24%. Of the 108 mutations, 77 (71%) were point mutations and 31 (29%) were deletions or insertions ranging in size from 1 to 15 bp (Table 2). The complete list of mutations identified is given in Table 3. Point mutations occurring at hotspot codons 175, 248, 249, 273, and 282 comprised 35.1% of all point mutations. p53 mutations were significantly associated with stage II disease or greater, larger tumor size, ER and PR negativity, higher mitotic index, poor degree of differentiation, marked nuclear atypia, and high tumor grade, but not with lymph node positivity (Table 4).

The prevalence of p53 mutations was higher in premenopausal breast cancer cases (27.5%) than in postmenopausal cases (19.9%; P = 0.05) and was also slightly higher in African-American women (26%) than in white women (21%; P = 0.22); however, the difference was not statistically significant. When stratified by stage, the prevalence of p53 mutations was identical (20%) in African-American and white women with stage 3 and 4 disease and slightly, but not significantly higher in African-American than in white women with stage 1 (17.3 and 14.4%, respectively; P = 0.63) and stage 2 disease (34.2 and 27.1%, respectively; P = 0.30). Overall, the p53 mutational spectrum in African Americans was similar to that in whites. Although somewhat more G:C→T:A transversions and A:T→G:C transitions were observed in blacks (12.5 and 6.7%, respectively) than in whites (6.0 and 3.3%, respectively), these differences were not significant (P = 0.30 for G:C→T:A and P = 0.26 for A:T→G:C, compared with all other mutations).

To determine whether the mutational spectrum in the CBCS was consistent with the cumulative p53 data collected for breast cancer, we compared the distribution of p53 mutation types among the primary breast tumors of the CBCS with those listed for breast cancer in the most recent, updated version (2001) of the IARC p53 Mutation Database. This comparison was restricted to exons 5–8 of p53 because only partial data exist in this database for other exons, including exon 4. As shown in Fig. 1, the spectrum of 94 mutations in exons 5–8 in the CBCS was not statistically different from that observed among 562 breast tumor mutations in the IARC p53 database (P = 0.35).

A higher prevalence of p53 mutations was found in the breast tumors of current smokers (36.5%; P = 0.02) than in the tumors of never smokers (23.6%; Table 5). Somewhat fewer mutations were found among former (16.2%) than never smokers, but the difference was not statistically significant (P = 0.09). Transversion mutations occurred more frequently in the breast tumors of both current (8.2%; P = 0.03) and former smokers (4.9%; P = 0.05) compared with never smokers (3.1%). After adjustment for age, race, stage, tumor size, menopausal status, and family history of breast cancer, current smokers were significantly more likely to harbor any p53 mutations (OR, 2.11; 95% CI, 1.17–3.78), p53 transversions (OR, 3.37; 95% CI, 1.03–11.06), and G:C→T:A transversions (OR, 10.53; 95% CI, 1.77–62.55) compared with never smokers (Table 5). Former smokers were also somewhat more likely to harbor G:C→T:A transversions compared with never smokers (OR, 2.43; 95% CI, 0.37–15.73), although this difference was not significant. Further adjustment for ER and PR status, combined tumor grade, histological characteristics, alcohol consumption, body mass index, age at menarche, age at first full-term pregnancy, parity, oral contraceptive use, benign breast biopsy, or education level did not alter the associations of p53 mutation with current or former smoking (data not shown).

The distribution of p53 mutation types in the 108 mutation-positive tumors, according to smoking status, are shown in Fig. 2. The p53 mutational spectra revealed a significantly greater proportion of G:C→T:A transversions in the breast tumors of current smokers (16.1% of all mutations; P = 0.04) compared with never smokers (3.7%). The proportion of G:C→T:A transversions observed in former smokers (13.0%) was similar to that in current smokers. The majority of G:C→T:A transversions (75.0%) in current and former smokers occurred on the nontranscribed DNA strand. Several additional shifts in the mutational spectrum previously noted in the lung tumors of smokers also were observed in the breast tumors of current smokers relative to never smokers, although these were not statistically significant. For example, among current smokers, 9.7% of mutations were A:T→G:C transitions compared with 3.7% in never smokers (P = 0.27), 25.8% of mutations were G:C→A:T transitions at CpG sites compared with 33.3% in never smokers (P = 0.47), and 22.6% of mutations were deletions and insertions compared with 31.5% in never smokers (P = 0.38).

Current and former smokers differed significantly in their tobacco smoke exposure characteristics: 74.1% of current smokers versus 35.9% of former smokers reported smoking for >20 years (P = 0.001), and three times as many current smokers (29.4%) as former smokers (10.6%) began smoking at ≤15 years of age (P = 0.001). Whereas the duration of tobacco smoke exposure was not correlated with the prevalence of p53 mutations overall in current and former smokers, we observed significantly more G:C→A:T transitions at non-CpG sites (26.9% of mutations; P = 0.05) and slightly, but not significantly, more G:C→T:A transversions (19.2% of mutations; P = 0.38) among cases who smoked >20 years compared with those who smoked for <20 years (7.1% G:C→A:T at non-CpG; 10.7% G:C→T:A). Among former smokers, the prevalence of p53 mutations was lower than among current smokers and varied with time since quitting. Compared with current smokers (36.5%), the mutation prevalence was 26.3% among former smokers who quit within 1 year of diagnosis, 10.5% among those who quit between 1 and 5 years before diagnosis, and 12.9% among those who quit >5 years before diagnosis (P = 0.001, Mantel-Haenzel χ²).

A previous report from the CBCS suggested that smoking was associated with an increased risk of breast cancer among postmenopausal former smokers (19). Examination of the p53 mutational spectra revealed a marginally significant increase in the proportion of G:C→T:A transversions (15.2% of mutations) among postmenopausal women compared with premenopausal women (4.8%; P = 0.06). Premenopausal and postmenopausal women were also evaluated according to smoking status, although the numbers of mutations in each strata were small. Among premenopausal cases, G:C→T:A transversions were found only in current smokers (3 of 20 mutations, or 15% of total mutations). Among postmenopausal cases, G:C→T:A transversions comprised 2 of 22 (9.1%) mutations in never smokers, 3 of 13 (23.1%) mutations in former smokers, and 2 of 11 (18.2%) mutations in current smokers (P = 0.51).

In the present study, we showed that the p53 mutational spectrum obtained in the CBCS, a large population-based series of primary invasive breast tumors, is similar to the mutational spectrum represented for breast cancer in the IARC p53 Mutation Database. Furthermore, we demonstrated that cigarette smoking is associated with both the prevalence and spectrum of p53 mutations in breast tumors. Past studies suggested that the p53 mutational pattern in breast cancer exhibited some similarity to the spectra observed in certain smoking-associated malignancies, particularly lung cancer (56, 67, 74, 75, 76, 77, 78, 79); however, associations with smoking were not evaluated. In the CBCS, the mutational spectrum in the breast tumors of smokers was characterized by a statistically significant increase in the proportion of G:C→T:A transversions, most of which occurred on the nontranscribed DNA strand. Also among smokers, there was some suggestion of increased A:T→G:C transitions and decreased G:C→A:T transitions at CpG sites, as has been observed previously in the lung tumors of smokers (67), although these differences were not statistically significant. The increased prevalence of G:C→T:A mutations among smokers supports the mutagenic, DNA-damaging effect of cigarette smoking in breast tissue and thus provides further evidence implicating tobacco smoke exposure in human breast carcinogenesis.

The breast tumors of patients who were long-term (>20 year) ever smokers had the most pronounced smoking-associated mutational pattern, with a significantly higher proportion of G:C→A:T transitions at non-CpG sites and slightly, but not significantly, more G:C→T:A transversions than for patients who smoked for <20 years. Mutational changes at guanine bases, such as G:C→T:A transversions and G:C→A:T transitions at non-CpG sites, are consistent with exposure to metabolically activated carcinogens, including BaP and N-nitrosamines in tobacco smoke (46). In fact, studies by Denissenko et al. (80) have demonstrated the preferential formation of BaP adducts at guanine bases in the hotspot codons of p53 that are frequently mutated in lung cancer. Increased A:T→G:C mutations have also been found in the lung tumors of smokers, but have not been attributed to a particular component of tobacco smoke (8, 81).

Breast cancer patients who smoked at the time of diagnosis (current smokers) were significantly more likely to have p53 mutation-positive cancer compared with never smokers, consistent with previous studies that reported elevated levels of p53 mutations among smokers with cancers of the lung, bladder, esophagus, and head and neck (55, 67, 74, 82). Our results are also consistent with the study of Gammon et al. (83), who found that p53 protein overexpression in breast tumors detected by immunohistochemistry was associated with current cigarette smoking. The increased prevalence of p53 mutation-positive breast cancer among current smokers in our study was not attributable to their diagnosis with more advanced cancers: current smokers were clinically similar to never smokers, and adjustment for stage, tumor size, node status, and other potential confounders such as ER status or tumor grade, did not appreciably alter the association of p53 mutation with smoking status.

In contrast to current smokers, former smokers overall were less likely than never smokers to have p53 mutation-positive breast cancer. Additionally, among the former smokers, the prevalence of p53 mutations varied with time since quitting, being highest among recent quitters who stopped smoking within the year of their cancer diagnosis. The occurrence of p53 mutations in breast tumors, therefore, appears to be most closely related to tobacco smoke exposure around the time of cancer diagnosis. Despite their lower prevalence of p53 mutations, former smokers exhibited a mutational spectrum similar to that of current smokers, characterized primarily by an increased proportion of G:C→T:A transversions. Therefore, ever (current and former) smokers were most clearly distinguished from never smokers by G:C→T:A transversions.

That current and former smoking appears to have opposing effects on the p53 mutational status of breast tumors is not surprising in light of the conflicting epidemiological literature (13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37) and may lend biological support to the theory that smoking has dual carcinogenic and antiestrogenic effects on breast tissue that vary with the characteristics of exposure. Several prior studies, including the CBCS, suggested that some subsets of postmenopausal women may be at increased risk of breast cancer attributable to smoking (19, 20, 21). Interestingly, we found a marginally significant increase in the proportion of G:C→T:A transversions in p53 among postmenopausal women compared with premenopausal women. Although the numbers were too small to be statistically significant, the postmenopausal former and current smokers exhibited the highest proportion of G:C→T:A transversions.

Although the exact mechanisms responsible for the observed association between smoking and p53 mutational status of breast cancers are unclear, several possibilities in addition to the genotoxicity of cigarette smoke seem plausible. An induction of DNA repair could contribute to the decreased mutation level among former smokers (84). Alternatively, modulation of p53 mutational status might occur through the tumor promotional (85, 86, 87) or antiestrogenic activity of tobacco smoke or possible effects on ER. Cigarette smoking increases the metabolism of estradiol (38), and recent reports demonstrate the existence of regulatory cross-talk between p53 and ER-mediated signaling pathways (88, 89, 90, 91, 92) and between ER and the aryl hydrocarbon receptor, both of which bind tobacco smoke constituents (93, 94, 95). Such inter-regulation might underlie the association between the p53 mutation positivity of breast tumors and loss of ER expression observed in this and other studies (59, 61, 63, 64, 96, 97, 98, 99). Additional research is clearly needed to decipher the mechanisms through which smoking modifies the p53 mutational status of breast cancer.

In addition to the etiological implications of our work, our results also have potential clinical significance. Cigarette smoking during the period immediately preceding breast cancer diagnosis was associated with an increase in p53 mutation-positive breast cancer. This raises the possibility that active smoking could, through its effect on p53 status, be associated with the development of more clinically aggressive breast tumors. Previous studies have suggested that p53 mutations are an independent prognostic factor associated with poorer survival in node-negative and -positive breast cancer patients (59, 61, 63, 64, 99). Although the breast cancer cases in our study were not followed to ascertain survival, p53 mutation positivity was significantly associated with clinical or histological features generally thought to be markers of poor prognosis, including larger tumor size, stage II disease or greater, ER and PR negativity, high mitotic index, and high tumor grade. In addition, current smoking was associated with increased prevalence of p53 mutations across all stages, grades, and hormone receptor expression subgroups of breast cancer. Several reports have attributed increased breast cancer mortality among smokers to a delay in seeking care, resulting in their diagnosis with more advanced cancers (43, 44). However, our findings suggest that the difference in mortality may have a biological basis.

The relationship between tobacco smoking and breast cancer is of major public health and clinical importance. Our data indicate that cigarette smoking may influence the pattern and presence of p53 mutations in breast tumors, which potentially affects the aggressiveness of the tumor. We observed a positive association between p53 mutation-bearing breast tumors and current smoking status and found that the p53 mutational fingerprint in breast cancers of smokers compared with nonsmokers was consistent with that seen in lung cancer. Because cigarette smoking is a modifiable risk factor, the prevention or cessation of smoking may reduce both the genetic damage associated with cigarette smoking and the incidence of more clinically aggressive, p53 mutation-positive breast cancers.

We thank the staff and participants of the Carolina Breast Cancer Study for invaluable contributions to the study.