Genetic Polymorphisms of the CYP19A1 Gene and Breast Cancer Survival

Estrogens stimulate breast cell division and play a crucial role in the pathogenesis and progression of breast cancer. Women with high levels of serum estrogens have an increased risk of breast cancer (1). Estrogen concentrations are 10 to 50 times higher in malignant breast tissue than in plasma (2), and estrogen in breast tissue may act locally to promote the growth of breast carcinomas (3).

During and after menopause, a woman's endogenous estrogen production converts from a predominantly ovarian source to a peripheral source. Aromatase, encoded by the CYP19A1 gene, is the main enzyme that catalyzes the final and rate-limiting step of estrogen biosynthesis, aromatization of androstenedione and testosterone to estrone and estradiol, respectively. A direct effect of aromatase on in situ estrogen synthesis in the breast has been reported (4). Elevated levels of aromatase expression have been observed in breast tumors relative to normal breast tissue (5). This evidence indicates a potential role for the CYP19A1 gene in the development and progression of breast cancer. The importance of aromatase in the pathogenesis of breast cancer has also clearly been shown in a clinical setting, as inhibitors of this enzyme have been regularly used in the treatment of postmenopausal breast cancer (6). A recent study suggested that aromatase inhibitors might be more effective than modulators of the estrogen receptor in slowing tumor progression (7).

To date, a number of studies have been done to evaluate genetic polymorphisms in the CYP19A1 gene in relation to breast cancer risk (8-13). However, only one study has examined the association of a particular genetic marker, the (TTTA) repeat in intron 4 in this gene, with breast cancer survival (14) and found a significant association between longer repeat length and improved survival. In the present study, we comprehensively evaluated CYP19A1 genetic variants in relation to breast cancer survival in a large cohort of patients recruited as part of a population-based case-control study, the Shanghai Breast Cancer Study (15).

In the Shanghai Breast Cancer Study, breast cancer patients were identified through a rapid case ascertainment system, supplemented by the Shanghai Cancer Registry, a population-based tumor registry. A total of 1,602 patients with a primary breast cancer diagnosis were identified between August 1996 and March 1998. Of them, 1,459 were recruited into the study with a response rate of 91% (16). Of the 1,459 breast cancer patients, 4 subjects were excluded from the survival study due to lack of adequate follow-up information. A peripheral blood sample (10 mL from each woman) was obtained from 1,193 patients, 82% of the 1,455 study participants in the survival study. The blood samples were processed within 6 hours of collection and stored at −70°C until the relevant bioassays were conducted. Medical charts were reviewed to abstract information on cancer diagnosis, tumor-node-metastasis (TNM) stage, estrogen receptor and progesterone receptor status, and cancer treatment. Pathologic slides for all cases were reviewed independently by two senior pathologists to confirm the cancer diagnosis.

Patients were followed until July 2005 for cancer recurrence and mortality with a combination of two active follow-up surveys and record linkage to death certificates kept by the Vital Statistics Unit of the Shanghai Center for Disease Control and Prevention (17). The median follow-up time for the cohort was 7.1 years. Through interview of patients, or next of kin for deceased patients, we obtained information on disease progression, recurrence, quality of life, and cause of death (if deceased). Of the 1,455 eligible patients, 1,378 were followed-up via in-person contact or by phone (active follow-up) at least once during the two follow-ups. Among them, 266 deaths were identified, 237 from breast cancer, 26 from other diseases, and 3 from unclear causes. Survival status for the 77 participants who were not actively followed was established by linkage to the death registry, and 47 deaths were identified with all but one due to breast cancer. The remaining 30 patients had no match in the death registry and were assumed to be still living on December 2004, 6 months before our last search of the vital statistics registry, to allow for a possible delay of entry of death certificates into the registry. Breast cancer relapse information was not available for these 30 patients and the one person who died from other diseases. Including the three women who lacked detailed information on cause of death through active follow-up, 34 women in total were excluded from the disease-free survival analysis. The study was approved by the institutional review boards of all participating institutes. Informed consent was obtained from each participant.

The genetic markers of the CYP19A1 gene included in this study were selected based on a report from the Multiethnic Cohort Study (10). Based on haplotype analyses of 74 densely spaced single-nucleotide polymorphisms (SNP) across the gene, four haplotype blocks (blocks 1-4) and 19 haplotype tagging SNPs were identified in a Japanese population. Because of the similarity in linkage disequilibrium patterns between Chinese and Japanese populations (18), we used these 19 haplotype tagging SNPs in the present study.

Genomic DNA was extracted from buffy coats with a Puregene DNA Purification Kit (Gentra Systems, Minneapolis, MN) following the protocol of the manufacturer. SNP rs1004984 was genotyped with a high-throughput Masscode assay at BioServe Biotechnologies, Ltd. (Laurel, MD). The nonsynonymous SNP, rs700519, was genotyped by PCR-RFLP and confirmed by direct sequencing with BigDye Terminator Chemistry on an ABI 3700 (ABI, Applied Biosystems, Foster City, CA). Genotyping of the other 17 SNPs was done by running the 5′nuclease TaqMan allelic discrimination assay using an ABI 7900 (ABI). Details about assays, primers, probes, and procedures are available on request.

Laboratory staff were blind to the identity of the subjects. Quality control samples were included in the genotyping assays. Each 96-well plate of genomic DNA contained multiple quality controls, including one water, two samples of CEPH 1347-02, two known duplicates, and two blinded duplicates. The average agreement of the genotypes for these 19 markers determined for the blinded quality control samples was 98.7%.

Included in the present study were 1,136 patients with both survival status and genotype information. The primary outcomes were disease-free survival and overall survival. The endpoints included cancer recurrence/metastasis or death due to breast cancer for the analysis of disease-free survival, and death from any cause for the analysis of overall survival. Survival time was calculated as the time from cancer diagnosis until the occurrence of the study endpoints, censoring at the date of last contact or noncancer death (for disease-free survival only). The Kaplan-Meier method was used to estimate the survival function, and differences in survival across groups defined by genotypes were examined using the log-rank test. The Cox regression model was employed to compute hazard ratios (HR).

The program PHASE, which is based on a Bayesian statistical model (19), was used to reconstruct haplotypes within each block of the CYP19A1 gene. The association of haplotypes and breast cancer survival was evaluated using a Cox model by treating each haplotype as a continuous variable with probabilistically assigned values, as described by Zaykin et al. (20). All statistical analyses were done with SAS version 9.1 (SAS Institute, Cary, NC) and all tests were based on two-sided probability.

Table 1 shows selected demographic factors and TNM stages for breast cancer and the corresponding 5-year overall survival rate. The descriptive characteristics were similar for the entire cohort of 1,455 breast cancer patients and the 1,136 subjects who were included in the present study. Most of the patients received surgery, adjuvant chemotherapy, and tamoxifen treatment. As expected, TNM stage was the major prognostic factor for survival.

Genotype and allele distributions for the 19 SNPs are summarized in Table 2. With the exception of rs6493494, rs12907866, and rs2414096, all other SNPs were consistent with the Hardy-Weinberg equilibrium distribution (P > 0.05). Table 3 and Fig. 1 show the results of association analyses for the genetic polymorphisms and breast cancer survival. Significant associations were observed for disease-free survival for each of the five SNPs located in haplotype block 2. Patients homozygous for the minor alleles had reduced 5-year disease-free survival rates compared with those carrying the major allele (69.1% versus 77.1%, 56.9% versus 76.7%, 68.9% versus 76.7%, 70.2% versus 76.5%, and 69.4% versus 76.4% for hCV1664178, rs12900137, rs730154, rs936306, and rs1902586, respectively). The age-adjusted HRs and 95% confidence intervals (95% CI) were 1.5 (1.1-2.1), 2.1 (1.2-3.6), 1.5 (1.1-2.0), 1.4 (1.0-2.0), and 1.4 (1.0-2.0) for hCV1664178, rs12900137, rs730154, rs936306, and rs1902586, respectively. The nonsynonymous SNP, rs700519 (Arg264Cys), located in haplotype block 4, was also associated with breast cancer survival. The age-adjusted HR (95% CI) for the Cys/Cys (T/T) genotype was 2.2 (1.2-4.1) for overall survival and 2.1 (1.1-3.9) for disease-free survival compared with those carrying the major allele, Arg (C). The Kaplan-Meier survival curves presented in Fig. 1 show apparent associations of disease-free survival after breast cancer diagnosis with the genotypes defined by these six SNPs. The distributions of the genotypes of these SNPs were similar across TNM stage and estrogen receptor/progesterone receptor status. Additional adjustment for these tumor characteristics did not materially change the genotype-survival associations (data not shown). The observed associations were similar between tamoxifen users and nonusers (data not shown). Analysis stratified by menopause status indicated that the above associations were mainly evident in premenopausal women. No association was observed in postmenopausal women (Table 4). No clear association was observed for the other 10 SNPs in relation to either overall survival or disease-free survival (data not shown).

Haplotype analyses were conducted to evaluate the combined effect of the SNPs located within each haplotype block (Table 5). The genotype distributions of three SNPs, rs6493494, rs12907866, and rs2414096, deviated from Hardy-Weinberg equilibrium (Table 2) and, thus, they were eliminated from the haplotype reconstruction. The frequencies of major haplotypes are similar to those observed in a Japanese population (10). An association was observed for the haplotype CCCTA (minor alleles for hCV1664178, rs12900137, rs730154, rs936306, and rs1902586 in block 2) for disease-free survival with an age-adjusted HR of 1.9 (95% CI, 1.1-3.3). The association, however, was not statistically significant for overall survival. Analysis stratified by menopause status indicated that the association of haplotype CCCTA with survival was mainly evident among premenopausal women, with HRs being 2.3 (95% CI, 1.1-4.6) for overall survival and 2.8 (95% CI, 1.5-5.2) for disease-free survival. No association was observed among postmenopausal women. The haplotype AAGC in block 4 was associated with both overall survival (HR, 1.7; 95% CI, 0.9-3.5) and disease-free survival (HR, 2.6; 95% CI, 1.2-5.6) among premenopausal women. We did not find any associations of other haplotypes with either disease-free survival or overall survival. Including the three SNPs that deviated from the Hardy-Weinberg equilibrium did not affect the association of the haplotypes with breast cancer survival described above (data not shown).

This is one of the first studies to evaluate the associations of CYP19A1 genetic polymorphisms with breast cancer survival. We have identified associations between all of the five SNPs located in haplotype block 2 and survival. Women who are homozygous for the minor allele at any of the five SNPs located in block 2 have a decreased rate of disease-free survival compared with those carrying the major allele. Haplotype analyses confirmed the above associations. Haplotype CCCTA (all minor alleles of the five SNPs in block 2) was associated with decreased disease-free survival. We also found associations of the minor allele homozygote of rs700519 (Cys264Arg) with both overall survival and disease-free survival. These findings are biologically plausible, given the pivotal role of the CYP19A1 gene in estrogen metabolism, its potential role in tumor growth and progression, and the potential functional significance of these polymorphisms.

It is widely accepted that estrogen plays an important role in the growth and progression of human breast cancer by disrupting the normal balance between cell differentiation and proliferation (21). Aromatase catalyzes the biosynthesis of estrogen in the adipose tissues through the conversion of androgens (22). Currently, aromatase inhibitors are regularly used in the treatment of postmenopausal breast cancer (6). Thus, it is biologically plausible that the polymorphisms in the CYP19A1 gene, which encodes the aromatase, may be associated with breast cancer survival. In our study, the association was observed in premenopausal women only. The reasons for this specific association, however, were unclear. Menopause status for our study subjects was recorded at the time of cancer diagnosis, and it is likely that the majority of the premenopausal women included in our study had menopause not long after cancer diagnosis due to breast cancer adjuvant therapies. The treatment-induced menopause suddenly cuts off the supply of ovary-synthesized estrogen to breast cancer cells that initially grow in a high-estrogen environment (premenopausal breast tissues). These cancers may have a greater need of estrogen for their growth than those developed in postmenopausal women, and after menopause, most of the estrogens were synthesized in adipose tissues by aromatase. Therefore, it is conceivable that aromatase activity (and thus CYP19A1 gene variants) may be more closely related to breast cancer survival in premenopausal than in postmenopausal women.

The CYP19A1 gene, located at 15q21.2, is composed of nine coding exons (II-X) covering ∼30 kb and at least 10 different untranslated first exons that are regulated by tissue-specific promoters (23). CYP19A1 gene expression has been suggested to play a role in neoplastic proliferation in human breast carcinomas. Elevated levels of CYP19A1 mRNA have been observed in breast cancer tissue as compared with normal breast tissue (5). The increased expression of CYP19A1 in breast cancer tissues was associated with a switch in the promoter utilization. Adipose tissue, in general, including the adipose tissue of a normal breast, maintains low levels of aromatase expression primarily via promoter I.4, whereas promoters I.3, I.7, and II are used only minimally (24). However, in breast cancer, activities of the latter three promoters are strikingly increased (5, 24).

Promoter I.7, recently cloned by Sebastian et al. (25), is located ∼36 kb upstream of the coding region. It is unique in that it is a GATA-2-regulated endothelial promoter and its usage is correlated with the extent of angiogenesis in breast cancer tissue (25). Interestingly, angiogenesis has been postulated to be one of the mechanisms for resistance to hormone therapy (26). Consistent with the potential role of this promoter in breast cancer pathology, we found an association between minor alleles of all five SNPs located around this promoter (in block 2) and decreased disease-free survival of breast cancer, both by individual locus analysis and by haplotype analysis. To our knowledge, our study is the first to identify the association of these polymorphisms around promoter I.7 and breast cancer survival. In line with our results, the haplotype CCCTA was also found to be associated with an elevated risk of breast cancer with an odds ratio of 1.2 (95% CI, 1.1-1.4) in the Multiethnic Cohort Study (10).

Promoters II and I.3 are in block 4, where we genotyped six SNPs. Association was observed at the SNP rs700519. The T/T (Cys/Cys) genotype was associated with a lower survival rate for both overall survival and disease-free survival. This SNP results in amino acid changes from Arg to Cys, which may lead to a difference in enzyme activity. Although no reports are available about the relationship between this polymorphism and breast cancer progression, several studies have reported a relationship between the Cys allele and increased breast cancer risk, with odds ratios (95% CIs) of 2.4 (1.0-5.5) in Hawaiians (10), 1.4 (1.1-1.9) in Japanese women (10), and 1.5 (1.1-2.2) in Korean women (27), although no association was observed in Caucasians (28) or another Japanese population (29). We did not find any apparent associations with breast cancer survival through haplotype analyses for the SNPs located in block 4, nor did the Multiethnic Cohort Study for breast cancer risk (10).

This is the first study to comprehensively evaluate the association of CYP19A1 genetic polymorphisms with breast cancer survival. The study hypotheses were based on sound biologic plausibility: the critical role of the CYP19A1 gene in estrogen biosynthesis. Selection bias was minimized given the population-based study design, high response rate, and high follow-up rate. In addition, >98% of the patients in this study belong to a single ethnic group; thus, the potential confounding effect due to ethnicity is limited. The large sample size and long follow-up period provided high statistical power. A potential concern for the study is that as relapse and cause of death information was collected on the basis of self-reports or death certificate information, some errors may exist. However, these errors are likely to be random, which is likely to attenuate the association observed in the study. Three tagging SNPs selected initially for the study were excluded from the data analyses because of their deviation from Hardy-Weinberg equilibrium. This may affect the accuracy of the study in predicting common haplotypes. However, the results from our study about the association between CYP19A1 genotypes and breast cancer survival should not be affected by these three SNPs. The haplotype tagging SNPs were selected based on the extensive work conducted by Haimen et al. (10). Using the latest data from the HapMap project, a panel of 38 tagging SNPs was required to capture all 170 common variants for the CYP19A1 gene and its adjacent region with pairwise r² ≥ 0.80. Among our 19 genotyped SNPs, 15 SNPs were included in the HapMap project and they can capture 91 of the 170 common variants with r² ≥ 0.80. These data provide additional assurance that the total SNP panel used in our study should be able to capture most of the common variants.

In summary, this study provides evidence for associations of breast cancer survival with SNPs located in haplotype block 2 and the nonsynonymous SNP in the CYP19A1 gene. These results are novel and, if confirmed, may have significant clinical implications for breast cancer treatment.

We thank Dr. Qi Dai, Dr. Fan Jin, and Jia-Rong Cheng for their contributions in coordinating data and specimen collection in Shanghai; Qing Wang and Regina Courtney for technique assistance in genotyping assays; Bethanie Hull for technical assistance in the preparation of this manuscript; Dr. Hui Cai for statistical consultation in data analysis; and Dr. Christopher A. Haiman from the University of Southern California for sharing the information for primers and probes of several SNPs.