Genetic variants of CYP19 (aromatase) and breast cancer risk


The effect of a SNP in exon 10 of CYP19 on tumor mRNA levels and splice variants were studied and correlated with clinical parameters and risk of breast cancer. In the vast majority of breast cancers, the estrogen levels modulate the tumor growth and depend on the activity of CYP19. Patients (n=481) and controls (n=236) were genotyped by T-tracks in a single sequencing reaction (SSR). The frequency of TT genotypes was significantly higher in patients versus controls (P=0.007) particularly among those with stage III and IV disease (P=0.004) and with tumors larger than 5 cm (P=0.001). A significant association between presence of the T allele and the level of aromatase mRNA in the tumors was observed (P=0.018), as well as with a switch from adipose promoter to ovary promoter (P=0.004). Previously, we reported a rare polymorphic allele of CYP19 (repeat (TTTA)12) to be significantly more frequent in breast cancer patients than in controls. Here we describe another polymorphism, a C–T substitution in exon 10 of the CYP19 gene which is in strong linkage disequilibrium with the (TTTA)n polymorphism but with higher frequency of the variant allele. Our data suggest that the T-allele of the CYP19 gene is associated with a ‘high activity’ phenotype.


The age specific incidence rate of breast cancer in women rises until menopause, levels off and then rises again at a much lower rate indicating a possible hormonal influence over the disease risk (Thomas et al., 1997). The role of estrogen in the etiology of human breast cancer is further supported by risk factors such as early age at menarche and late menopause, high serum or urine estrogen levels in breast cancer patients as well as low levels of sex hormone binding protein leading to high bioavailability of unbound estradiol (Hankinson et al., 1998). Approximately 50% of all breast cancers are sensitive to hormone treatment indicating a basic role for estrogen in the progression of the disease, as well. In postmenopausal women, local aromatization of androgens to estrogens is the main source of estradiol in the tumor and surrounding tissue (Pike et al., 1993; Simpson et al., 1994). These local conversions of estrogens as well as the rate-limiting step of conversion of androstenedione to estrone are catalyzed by the aromatase cytochrome P450 complex (Means et al., 1989; Burak et al., 1997). Clinical findings in patients with aromatase deficiency as well as in a knockout mouse (ArKo) lacking functional aromatase, illustrate the role of this enzyme for the development of reproductive organs and mammary glands in female and for the regulation of gonadotropins in both male and female (Harada et al., 1992; Fisher et al., 1998). The gene (CYP19) is regulated in a different manner in different tissues with hormonally controlled promoters like the gonadal type promoter or adipose stroma cell type promoter (Mahendroo et al., 1993; Harada et al., 1993). The markedly increased aromatase expression in transformed neoplastic cells may be partly a consequence of using different tissue specific hormonally regulated promoters. Genetic polymorphisms of CYP19 may be involved in other mechanisms, such as mRNA stabilization, enhancing of transcription or posttranslational regulation of expression. Previously, we reported a polymorphic allele of CYP19 (repeat (TTTA)12) to be significantly more frequent in breast cancer patients than in controls (Kristensen et al., 1998a). Sourdaine et al. (1994) described several mutations in exons 3, 7 and 10 of the aromatase gene in tumors from breast cancer patients. Since the mutation in exon 10 is in the 3′ untranslated region of the mRNA with putative impact on the mRNA stability or regulation of termination of translation, we designed experiments to verify whether it has influence on RNA levels of aromatase and relevant clinical parameters.


A sequence variant caused by a point mutation in exon 10, the 3′ untranslated region of CYP19, previously described in tumor tissue (Sourdaine et al., 1994), was analysed using leukocyte DNA from 399 individuals (163 breast cancer patients and 236 controls) and was shown to be a germline polymorphism. Both the breast cancer patients and the control group population were in Hardy-Weinberg equilibrium with respect to the polymorphism. In addition, genotyping was performed on tumor DNA from 318 breast cancer patients, of which a subset of 156 had been analysed for CYP19 mRNA expression in the tumors. No difference in heterozygosity at the aromatase locus (15q) was observed between patients genotyped on leukocyte DNA and those genotyped on tumor DNA corrected for stage. Nonetheless, despite the infrequent LOH on chromosome 15 in breast cancer, to evaluate a possible bias caused by LOH in the tumor we have genotyped both blood and tumor DNA from 24 patients all heterozygous for the polymorphism without detecting any LOH.

A higher proportion of the breast cancer patients were carriers of the T allele compared to the control group (55.3 vs 46.6%) (P=0.002). There were significantly more breast cancer patients with TT genotype than controls (31.2 vs 22.5%) (P=0.007) (Table 1). A list of odds ratios for the different genotypes in the breast cancer and the control group as well as in the different groups according to stage of disease are given in Table 2. The odds ratio of developing breast cancer for TT carriers vs CC+CT carriers is 1.51 (95% CI, 1.04–2.16). When various clinical and histopathological parameters were analysed, patients presenting with stage III and IV disease had significantly more frequent TT genotypes compared to controls (P=0.004) (Table 1), OR 2.09 (95% CI, 1.29–3.36) (Table 2).

Table 1 Genotype distribution (number (frequency)) of the polymorphism in exon 10 of CYP19. Relation to stage of the disease and tumor size
Table 2 Odds ratios of developing breast cancer for the different genotypes in the 3′ untranslated region of exon 10 of CYP19

In order to separate tumors expressing factors for rapid local growth from tumors with metastatic phenotype, we focused on those subjects whose stage of disease was considered high due to large tumor size. The observed difference in the allele distribution of CYP19 became even higher when patients with high stage of the disease due to large tumor size alone and no distant metastases were only included (P=0.002), OR 3.12 (95% CI, 1.80–5.30). When tumor size was analysed in relation to genotypes, patients with tumors larger than 5 cm were found to be significantly more often TT carriers than control individuals (P=0.001) (Table 1). A significant difference between the genotypes of patients with tumors larger than 5 cm vs patients with smaller tumors was also observed (P=0.045).

The previously described rare A1 allele containing the longest repeat (TTTA)12, has been found to be significantly more frequent in breast cancer patients than in controls (3.6 vs 1.6%), indicating that carriers may have an increased risk of developing breast cancer, OR 2.42 (95% CI, 1.03–5.80) (Kristensen et al., 1998a). Using a subset of the same samples of patients (leukocyte DNA from patient case-series 1, see Materials and methods) we could demonstrate that this polymorphism is in strong linkage disequilibrium with the more frequent polymorphism in exon 10 described here (P<0.0001) (Table 3).

Table 3 Association between two polymorphisms in the CYP19 (aromatase) gene, tetranucleotide repeat in intron 4. The repeat length in the different alleles were: A1 (TTTA)12, A2 (TTTA)11, A3 (TTTA)9, A4 (TTTA)8, A5 (TTTA)7

In order to explore the possible difference in expression of a C and T allele of CYP19 we analysed the mRNA levels and the alternative switch of promoter in the tumors of 156 of the genotyped patients. Alternative use of promoters, followed by alternative splicing of a different untranslated exon 1 and providing a tissue specific regulation of gene expression of CYP19 has been repeatedly reported. In healthy adipose tissues, the most often used promoter results in the expression of exon 1.4 (Mahendroo et al., 1993) also abbreviated as 1b (Harada et al., 1993). In tumor tissues, however, a switch to ovary specific exon 1, abbreviated as 1c (Harada et al., 1993) or exon II (Mahendroo et al., 1993) has been observed to be significantly more frequent than in healthy tissue (Harada et al., 1995). In 99 of the tumor samples studied here, the sole or the predominantly expressed exon 1 was either the adipose tissue specific exon 1b or the ovary tissue specific exon 1c. In 39 of the tumor samples there was no detectable CYP19 mRNA and in 18 the use of promoter was ambiguous (in ten cases there was equal amount of transcripts containing 1b and 1c, in four cases there was expression from a different promoter (1d), and in four cases the detection of promoter failed). These cases were excluded from further analysis. Quantitation of CYP19 mRNA showed a wide range of expression.

When comparing the genotypes in patients with levels of CYP19 mRNA (CYP19 mRNA/β-actin, above and below the median level) patients with the TT genotype had significantly more often levels of CYP19 mRNA above the median (P=0.018) (Table 4). The presence of TT genotype was also associated with the alternative use of promoter. Transcripts containing the atypical ovary specific exon 1 (abbreviated 1c or exon II) were significantly more often found in the tumors of individuals with a TT genotype (P=0.004). This switch of promoters was strongly associated with elevated mRNA as 44/69 (64%) of the tumors where the adipose tissue promoter was used had mRNA levels below median, compared to only 6/30 (20%) of those with ovarian promoter. High levels of mRNA expression (above median) were strongly associated with the expression of the ovary promoter (24/30) (P<0.001).

Table 4 Genotype distribution (number (frequency)) of the polymorphism in CYP19, exon 10 in relation to mRNA levels and alternative use of promoter


The polymorphism in exon 10 may possibly enhance the transcription from the ovary promoter and may thus lead to elevated mRNA levels, or it may contribute to regulation independently, by resulting in differently stable ovary specific or adipose specific transcripts. The switch of promoters was strongly associated with elevated mRNA levels and may be responsible for the observed upregulation of transcription of CYP19. Computer modeling (Walter et al., 1994) suggested two different folding patterns for the 3′ end of the CYP19 transcript, depending on the nucleotide in the polymorphic position, C or T. The C–T change in this proposed computer model destroys an AT base pair and shortens the affected helix by two base pairs, since the neighboring T-A base pair is destabilized by the mutation. This may lead to weakening or destroying of this motif and may influence mRNA stability. Alternatively, the polymorphism may contribute to different responsiveness of the ovary specific promoter to cAMP dependent regulation (Zhao et al., 1997; Michael et al., 1997) or the regulation of the adipose specific promoter by TNF-α or class I cytokines, such as IL-6 18. However, we could not exclude that the genetic polymorphism in exon 10 of CYP19 described by us may be in linkage disequilibrium with another, unknown genetic variant, which may be the functionally responsible one. Since the polymorphism is in the vicinity of the UAG (19 nucleotides downstream of termination of translation), one may speculate that a possible secondary structure of the transcript might influence both the stability of the transcript and the regulation of translation termination. Toda et al. (1989) described alternative RNA processing using different poly(A) signals of aromatase mRNA. It is under current investigation whether the polymorphism in exon 10 described here is associated with these poly(A) variants. Our data indirectly suggest that the T-allele of the CYP19 gene is associated with a ‘high activity’ phenotype, since we observe a strong association of the genotype to mRNA levels and large tumor size. High activity of aromatase has been directly related to mammary tumor cell proliferation (Durgam et al., 1995). It has been shown that insertion of the proviral MMTV (mouse mammary tumor virus) within exon 10 in the 3′ untr of CYP19, leads to overexpression of the gene, resulting in mammary tumor cell proliferation in a gene dosage dependent manner, highly potentiated by androstenedione – the principal substrate of aromatase, precursor of estradiol (Durgam et al., 1995). This overexpression may be due to the disruption of a negative regulatory element by integration of the MMTV in the 3′ untr of CYP19 (aromatase) (Durgam et al., 1995) of the mouse. Here we describe a polymorphism in the same area of the human CYP19, the distribution of which significantly differs between breast cancer patients and controls. Since the frequency of the polymorphic T allele was particularly high among patients presenting with high stage disease and with tumors larger than 5 cm and was significantly associated with mRNA levels as well as a switch from the normally used adipose promoter to ovary promoter, individuals homozygous for this allele may have accelerated production of tissue estrogen and therefore higher risk for developing tumors with rapid local growth.


DNA samples from a total of 481 breast cancer patients were genotyped. Genotyping was performed on three different case-series: (1) Leukocyte DNA from 163 patients, admitted to the Norwegian Radium Hospital, mean age at diagnosis 57 (27–89), where all tumor stages were observed (stadium I, II, III and IV), (2) 240 blood and tumor DNA samples from a consecutive series of breast cancer patients admitted to the City Hospital of Oslo (Ullevaal), mean age at diagnosis 65 (28–91), where mainly stage I and II disease were observed, (3) tumor DNA from a selected series of 78 patients, admitted to the Haukeland Hospital, Bergen, with stage III and IV disease with a large tumor size, mean age at diagnosis 64 (32–85). As controls leukocyte DNA from a series of 236 Norwegian healthy female individuals obtained through the Norwegian Population Registry as a population based series of women aged 20–44, living in the Oslo area were used for genotyping.


A modified automated DNA sequencing with a fluorescent labeled primer was used to genotype PCR products in a single sequencing reaction (SSR) (Kristensen et al., 1998b). Single sequencing reactions analysed on an automated DNA sequencer can be used as a universal tool for screening for all types of known mutations in various genes. Compared to analysis by full sequencing, four times more samples can be analysed per gel in considerably shorter time. In many circumstances one is not interested in complete DNA sequences, but only in differences between them. In these cases, single sequencing reactions have been used for many years, e.g. to screen recombinant vectors for a given insert and for other purposes. We have shown that single sequencing reactions performed on PCR products, with subsequent analysis on an ALF Express DNA sequencer, can be as informative, sensitive and accurate as complete sequencing reactions for the genotyping of known point mutations in human DNA (Kristensen et al., 1998b). This is achieved by using the mutated Taq polymerase (Thermosequence (Amersham-Pharmacia) or AmpliTaqFS (Perkin Elmer)), in cycle sequencing reactions with one single dideoxyanalog, chosen to correspond to the base found in one of the two alleles. In the example shown in Figure 1 there is either a C or a T in the polymorphic position in the DNA. In a T-track sequencing reaction a T/T individual will have a normally sized T in this position (contribution to the signal from both alleles), whereas a person with a C/T genotype will have a T with half the normal size (contribution from only one allele) and a C/C individual will have no signal in that position in the T-track. Analysing the mutation in exon 10 of the human aromatase gene we have shown that an unambiguous genotype could be elucidated in more than 90% of the analysed samples (Kristensen et al., 1998b). This simplified sequencing-based genotyping method makes it possible to analyse a high number of samples per sequencing gel, without loss of sensitivity and accuracy. A 367 bp long fragment of the gene (CYP19) was amplified from leukocyte DNA using primers as described previously (Sourdaine et al., 1994). The polymorphic site to be investigated was situated 75 nucleotides from the 5′ end of the sequencing primer. After data collection single T-tracks were analysed using the AlleleLinks package (both from Pharmacia Biotech) (Figure 1). The peak table was then exported to a text file and imported into Microsoft Excel for further analysis.

Figure 1

Single T-tracks using the AlleleLinks package (Pharmacia Biotech). The polymorphic site is in position 75. Single track analysis consisted of first identifying peaks, followed by sizing the products using two of the invariant sites (marked with filled dots) as internal standards, and then measuring the relative area under the relevant peaks by setting the area under one of the invariant neighboring T-peaks to 1. The peak table was then exported to a text file and imported into Microsoft Excel for further analysis. TT homozygous genotype is presented in lanes 27, 29, 31–34, 36, heterozygous CT genotype in lane 30, homozygous CC is presented in lanes 28 and 35. An ambiguous result was obtained in lane 26, subsequently genotyped as CC

Quantitation and analysis of alternative usage of exons 1 of aromatase mRNA

RNA was prepared from fresh frozen tissues as described previously (Harada et al., 1993). Quantitative analysis of aromatase mRNA was performed by competitive reverse transcription – PCR, using human aromatase cDNA as an internal standard as described previously (Harada et al., 1993, 1995). A CYP19 specific sense fluorescence labeled primer and internal standard specific antisense primer were used and the fluorescent PCR products were electrophoresed in a 2% agarose gel and analysed with a Gene Skanner 362 Fluorescence Fragment Analyzer (ABI Perkin Elmer). The amount of aromatase mRNA was calculated from the peak areas of the fluorescent products by the internal standard method and corrected after quantitation of β-actin mRNA. The utilization of alternative exons 1 of the aromatase gene was investigated by RT–PCR using sense primers specific for exons 1a, 1b, 1c, 1d and a fluorescent labeled antisense primer, specific for exon 2 according to the previously described protocol (Harada et al., 1993).

Statistical analysis

Statistical evaluation was performed using the Pearson Chi square test with Yates' correction. P values less than 0.05 were considered statistically significant. Odds ratios with 95% confidence intervals were calculated where appropriate by standard method for retrospective case control studies.


  1. Burak Jr WE, Quinn AL, Farrar WB and Brueggemeier RW . 1997 Breast Cancer Research and Treatment 44: 57–64

  2. Durgam VR, Easton JA, Surya R and Tekmal RR . 1995 Biochimica et Biophysica Acta 1263: 89–92

  3. Fisher CR, Graves KH, Parlow AF and Simpson ER . 1998 Proc Natl Acad Sci USA 95: 6965–6970

  4. Hankinson SE, Willet WC, Manson JE, Colditz GA, Hunter DJ, Spiegelman D, Barbieri RL and Speizer FE . 1998 J Nat Cancer Inst 90: 1292–1299

  5. Harada N, Ogawa H, Shozu M and Yamada K . 1992 Am J Hum Genet 51: 666–672

  6. Harada N, Utsumi T and Takagi Y . 1993 Proc Natl Acad Sci USA 90: 11312–11316

  7. Harada N, Utsumi T and Tagaki Y . 1995 Pharmacogenetics 5: 59–64

  8. Kristensen VN, Andersen TI, Lindblom A, Nesland J, Olsen A and Børresen-Dale AL . 1998a Pharmacogenetics 8: 43–48

  9. Kristensen T, Nedelcheva Kristensen V and Børresen-Dale AL . 1998b BioTechniques 24: 832–835

  10. Mahendroo MS, Mendelson CR and Simpson ER . 1993 J Biol Chem 268: 19463–19470

  11. Means GD, Mahendroo MS, Corbin CJ, Mathis JM, Powell FE, Mendelson CR and Simpson ER . 1989 J Biol Chem 264: 19386–19391

  12. Michael MD, Michael LF and Simpson ER . 1997 Mol Cell Endocrinol 134: 147–156

  13. Pike MC, Spicer DV, Dagmoush L and Press MF . 1993 Epidemiol Rev 15: 17–35

  14. Simpson ER, Mahendroo MS, Nichols JE and Bulun SE . 1994 Int J Fertil 39: 75–83

  15. Sourdaine P, Parker MG, Telford J and Miller WR . 1994 J Mol Endocrinol 13: 331–337

  16. Thomas HV, Key TJ, Allen DS, Moore JW, Dowsett M, Fentiman IS and Wang DY . 1997 Brit J Cancer 75: 1075–1079

  17. Toda K, Terashima M, Mitsuuchi Y, Yamazaki Y, Yokoyama Y, Nojima S, Ushiro H, Maeda T, Yamamoto Y, Sagara Y and Shizuta Y . 1989 FEBS Lett 247: 371–376

  18. Walter AE, Turner J, Kim MH, Lyttle P, Mueller P, Mathews DH and Zuker M . 1994 Proc Natl Acad Sci USA 91: 9218–9222

  19. Zhao Y, Agarwal VR, Mendelson CR and Simson ER . 1997 J Steroid Biochem Mol Biol 61: 203–210

Download references


This work was supported by the Norwegian Cancer Society and the Norwegian Council for Science and Humanities. The authors would like to thank Professor Zuker, Institute for Biomedical Computing, Washington University for his comments on the RNA folding model.

Author information



Corresponding author

Correspondence to Anne-Lise Børresen-Dale.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Kristensen, V., Harada, N., Yoshimura, N. et al. Genetic variants of CYP19 (aromatase) and breast cancer risk. Oncogene 19, 1329–1333 (2000).

Download citation


  • CYP19 (aromatase)
  • estrogen metabolism
  • polymorphism
  • breast cancer risk

Further reading