CYP17A1 +27T>C

This polymorphism is located in the cytochrome P450 17A1 promoter region but its effect on steroid hormone biosynthesis is inconclusive⁵⁶ (Figure 1). Inappreciable differences in percent MD across genotypes were observed in four studies restricted to 94 to 268 premenopausal women and regardless of age (< 50 years versus > 50 years), study design (mammograms from prospectively enrolled healthy women versus pre-diagnosis mammograms from breast cancer cases), and ethnicity^{39, 41, 42, 46}. Non-significant associations were also observed among samples of 171 to 1,260 postmenopausal women irrespective of current^{39, 41} or past or never^{39, 53} PMH use. The data, therefore, provide little support for a role of this polymorphism in MD or in breast cancer⁵⁷.

Open in a separate window

Figure 1

Pathways of Steroid Hormone Synthesis, Metabolism and Sensitivity of Tissues

Steroid hormone biosynthesis. The cytochrome p450 (CYP) 17A1 enzyme catalyzes the conversion of pregnenolone and progesterone to the hormones, dehydroepidandrosterone (DHEA) and androstenedione, respectively, which are further metabolized to estrone (E1) and 17β-estradiol (E2) by the CYP19A1 (aromatase) enzyme. Androgen conversion to estrogen in adipose tissue by CYP19A1 is an important source of bioactive endogenous estrogens among postmenopausal women.The hydroxysteroid dehydrogenase (HSD) enzyme HSD3β1 catalyzes the interconversion of pregnenolone and progesterone and of DHEA and androstenedione, while HSD17β1 catalyzes the interconversions of androstenedione and testosterone and of E1 and E2. Testosterone and E2 bind to the serum sex hormone binding globulin (SHBG), which regulates the bioavailability of these hormones in target cells. A. Estrogen catabolism. The CYP1 family of enzymes are phase I enzymes that function in the oxidative metabolic activation and deactivation of compounds including several steroid hormones. E1 and E2 preferentially undergo 2-hydroxylation (2-OH-E1 or -E2) by the CYP1A1 and 1A2 enzymes, whereas E1 and E2 preferentially undergo 4-hydroxylation (4-OH-E1 or -E2) by CYP1B1. Catecholestrogens are deactivated by catechol-O-methyltransferase (COMT). UDP-glucuronosyltransferase (UGT) 1A1 is a detoxifying enzyme, which converts endogenous substrates including E1 or E2 and their metabolites to inactive hydrophilic glucuronides. B. Progesterone catabolism. The aldoketo reductase family 1, member C4 (AKR1C4) enzyme catalyzes the conversion of progesterone and androstenedione to their corresponding alcohols. C. Gene transcription. Progesterone, testosterone and E2 bind to their respective nuclear receptor proteins, progesterone receptor (PR), androgen receptor (AR) and estrogen receptor alpha (ERα), and activate genes with corresponding responsive elements. The steroid receptor coactivator protein, amplified in breast-1 (AIB1) interacts with ERαin a ligand-dependent manner to increase estrogen-dependent transcription.

COMT Val¹⁵⁸Met

The COMT enzyme inactivates the bioactive 2- and 4- hydroxy catechol estrogens by methylation⁵⁸ (Figure 1). The Val¹⁵⁸Met polymorphism decreases enzyme activity⁵⁸ and is the most widely studied polymorphism with MD. Among healthy premenopausal women^{41, 42, 46}, carriers of two copies of the Met allele had higher MD in one study of Caucasians participating in a case-control study of breast cancer⁴¹, and lower MD in two studies of women enrolled from a screening mammography clinic^{42, 46}. The latter two studies comprised mixed ethnicities including Asians and, although adjustment was made for ethnicity, this and the younger age distribution compared to the study by Haiman et al⁴¹ may account for the differences among relatively healthy premenopausal women. The remaining studies of COMT Val¹⁵⁸Met among premenopausal women included breast cancer patients whose screening mammograms were retrieved within the five years preceding diagnosis³⁹ or whose mammograms were performed after breast cancer diagnosis but prior to treatment³⁸ showed only minor differences in MD across COMT genotypes, and the study by Takata et al³⁸ did not evaluate healthy controls separately from cases by menopausal status.

Two studies which enrolled 67³⁹ and 1,260⁵³ postmenopausal women who never used PMH or used PMH in the past showed no particular pattern of percent MD across COMT genotypes. Investigations of postmenopausal women that did not differentiate among hormonal regimens but which adjusted for PMH use reported either no significant differences of MD across COMT genotypes^{42, 50} or significantly lower MD among carriers of both Met alleles⁴¹. The two studies^{41, 53} closest in comparability due to similar age distributions and study populations (healthy women participating in a nested case-control or prospective cohort study) had disparate findings, which could be attributable to the three fold difference in sample sizes, the statistical adjustment for PMH use⁴¹ versus sample restriction to past or never PMH use⁵³, or to method of mammogram assessments (computer assisted^{39, 41} versus visual assessment by three radiologists using the Boyd six-category scale⁵³). Statistical adjustment for PMH use or menopausal status without evaluation of the assumption that the effect is constant across strata of these variables results in pooled estimates that can conceal important stratum-specific effects. Also, although both MD assessment methods are shown to predict risk of breast cancer¹⁰, classification of MD using the Boyd categorical 6-point scale can overestimate risk of breast cancer among women aged >50 years compared to the computer-assisted threshold assessment method ⁷. Thus, an ideal comparison of results by genotype necessarily relies on similar methodology. In summary, although the COMT Val¹⁵⁸Met polymorphism does not appear to be a strong risk factor for breast cancer⁵⁹ regardless of menopausal status⁵⁷, the data suggest it may influence premenopausal MD.

ESR1 XbaI and PvuII

Estrogen receptor alpha is a nuclear receptor that mediates estrogen or other steroid hormone action by regulating gene transcription (Figure 1). The two polymorphisms in intron 1 of ESR1 (known as XbaI and PvuII) have no known functional effect on gene expression^{57, 60}, although intronic polymorphisms may affect the post transcriptional processing of gene products and affect receptor binding affinity.

Three studies examined the XbaI polymorphism. The European Prospective Investigation into Cancer and Nutrition (EPIC) cohorts analyzed cross-sectional estimates of MD across XbaI genotypes from 620 postmenopausal women in the Netherlands (EPIC-Prospect) who never used PMH⁴⁷ and 1,286 postmenopausal women in the United Kingdom (EPIC-Norfolk) who were past or never users of PMH⁵³. While the former study⁴⁷ reported a significantly higher MD with increasing copies of the x variant allele, the latter study⁵³ reported no appreciable difference in percent MD across genotypes. The greater sample size in the latter study⁵³ is a strength, although MD assessments were subjective and differed from the computer-assisted method used in the EPIC-Prospect study. In a third study⁵², the computer-assisted method was used for all MD readings among mostly (~70%) postmenopausal women participating in the combined EPIC Prospect and Norfolk cohorts. In this study, the densities of 781 women who never used PMH were compared at baseline and at 5 years of follow-up across XbaI genotypes, thereby reflecting a natural change in MD during this time period. In addition, mammographic densities were compared across XbaI genotypes among 795 women who never used PMH at baseline but who subsequently initiated PMH use during the 5 years of follow-up, thereby reflecting the influence of PMH use during this period. The results suggested that the women who carried at least one x allele had significantly attenuated declines in MD if they took PMH⁵². The PvuII polymorphism, which is in linkage disequilibrium with the XbaI polymorphism, was also studied in the two aforementioned investigations and shows similar results^{47, 52}. Confirmation of these findings by independent investigators would suggest that PMH use interacts with the XbaI and PvuII polymorphisms to moderate the natural decline in MD associated with the postmenopausal years, possibly to influence breast cancer risk, but this remains to be determined. The PvuII p allele and the XbaI x allele have each been associated with breast cancer^61–63.

PGR PROGINS complex

Two (Val⁶⁶⁰Leu and His⁷⁷⁰His) of three polymorphisms in the progesterone receptor (PGR) gene are in complete linkage disequilibrium and together all three are called the PROGINS complex⁶⁴ (Figure 1): they may affect ligand binding and gene transcription⁶⁵. In the EPIC-Prospect⁴⁷ and EPIC-Norfolk⁵³ studies, neither Val⁶⁶⁰Leu or His⁷⁷⁰His were associated with cross-sectional estimates of MD among postmenopausal women who had not used or who were past users of PMH⁵³, nor with longitudinal change in MD among women who used PMH⁵². Each copy of the Val⁶⁶⁰Leu allele was reported to increase breast cancer risk by a modest 8% (P = 0.05)⁶⁶.

IGFBP-3 −202C>A

This polymorphism is located in the promoter region of the gene and may increase serum IGFBP-3 levels⁶⁷. It has been investigated by five groups. Examination of 215 healthy premenopausal women participating in a prospective study of breast cancer in the United Kingdom⁴⁸ and 741 healthy premenopausal women attending a mammography clinic in Quebec, Canada⁵⁴ showed no influence of the genotype on MD. A third study among 206 healthy premenopausal women attending an outpatient clinic in Toronto, Canada showed a significantly higher MD among women with two copies of the A allele⁴⁵. Each study used a computer-assisted method to assess MD. However, none of the studies defined the ethnic composition of their sample of women or adjusted for ethnicity. The potential for population stratification among a potentially multi-ethnic sample residing in Toronto⁴⁵ with the UK⁴⁸ and French Canadian⁵⁴ participants warrants further exploration. In addition, postmenopausal women who never used PMH or used PMH in the past^{45, 48} showed no appreciable difference in percent MD across genotypes of this polymorphism. Also, two additional studies, the Australian twin study of 307 postmenopausal women⁵⁰ and the Nurses’ Health Study of 917 postmenopausal controls that participated in a case-control study of breast cancer³⁰ found no difference in MD across genotypes following adjustment for PMH use. In summary, the data suggest this polymorphism is not associated with MD regardless of menopausal status, although the C allele has been associated with lower circulating levels of IGFBP-3⁶⁸ and marginally significant decreased breast cancer risk⁶⁶.

IGF-I rs1520220 C>G and rs6220 A>G

The rs1520220 C>G polymorphism is located in intron 3, while rs6220 A>G is located in the 3′ UTR; neither polymorphism’s function is known. Both were evaluated in two studies of premenopausal women comprising large samples of healthy women attending a mammography clinic (746 in Quebec, Canada⁵⁴ and 1,928 in the Netherlands³²). Only the Canadian study observed a significantly higher percent MD among women who carried two copies of the rs1520220 G allele⁵⁴ and both studies observed a marginally significantly higher percent MD with two copies of the rs6220 G allele^{32, 54}. Both studies^{32, 54} also showed higher serum IGF-I with each polymorphism, although significance was attained with two copies of the rs6220 only³². Rs6220 A>G has also been associated with increased breast cancer risk⁶⁹.

A third study among healthy (204 premenopausal and 917 postmenopausal) women in the Nurses’ Health Study observed significantly lower percent MD among those with both copies of the rs1520220 G allele, but no apparent difference was noted by menopausal status³⁰. The evidence thus far does not support a consistent relation of rs1520220 C>G with MD or with breast cancer risk^{70, 71}, although rs6220 A>G may be a potential candidate for premenopausal MD.

HSD3β1 Asn³⁶⁷Thr

One seemingly consistent finding among genes in the steroid hormone biosynthesis pathway (Figure 1) is the lower level of MD among Caucasian carriers of the Asn³⁶⁷Thr variant of hydroxysteroid dehydrogenase 3β1 (HSD3β1). This polymorphism has no known function. A cross-sectional study of 242 (~50% postmenopausal) Caucasian breast cancer patients with pre-diagnosis mammograms found a significantly lower percent MD among women with one or more copies of the Thr allele compared to no copies³⁹. This finding was confirmed in a study of 457 healthy Caucasian women (~67% postmenopausal) from 207 sisterhoods that also reported a significantly lower MD from one or more copies of the Thr allele⁵⁰. Analyses restricted to 176 premenopausal women with two copies of the Thr allele displayed a non-significantly higher MD in the former study³⁹, but a marginally significantly lower MD in the latter study of 150 premenopausal women⁵⁰. Potential reasons for the discrepancy could be variability from small sample sizes within subgroup analyses, the comparison of mammographic densities obtained retrospectively from women who were recruited as breast cancer cases with those of a selected group of twin-pairs, and the slightly lower age among the breast cancer cases compared to the twin-pairs, suggesting an enrichment of younger, higher-risk women in this³⁹ study. Although lower MD is seen for the Thr allele among Caucasian women after adjusting for menopausal status, the associations are inconsistent among premenopausal women, and not significant among postmenopausal women. The relevance of this polymorphism to MD, therefore, remains unclear. The association between the Asn³⁶⁷Thr variant and breast cancer risk has not been studied.

CYP19A1 (TTTA)n

The cytochrome P450 19A1 gene, also known as aromatase, contains a microsatellite repeat polymorphism that may alter the splice site during post-translational protein modification⁷²; however no functional effect of the polymorphism on estrogen biosynthesis is known⁷³ (Figure 1). Two studies of 486⁴¹ and 550³¹ Caucasian women found no statistical association with MD among carriers of the 10 repeat allele compared to carriers without this allele in models that adjusted for age, BMI and menopausal status^{31, 41}. In the same studies, carriers of the 12 repeat allele had marginally significantly lower MD levels than women without this allele in one³¹ but not in the other⁴¹ investigation. Both were case-control studies of breast cancer and used the computer-assisted thresholding method of MD assessment among controls. Differences in SNP associations with MD may be from chance or from subtle ancestral differences among the Caucasian samples. Additional analyses of haplotypes that captured variation across the entire CYP19A1 gene with MD were not statistically significant³¹. Thus, the association of the CYP19A1 12 repeat allele with MD is inconclusive, and does not appear to be associated with breast cancer^{57, 74, 75}.

CYP19A1 +27(TCT)+/- and +268T>C

CYP19A1 +27(TCT)+/- is an insertion/deletion polymorphism in intron 4 and is in linkage disequilibrium with CYP19A1 (TTTA)n. CYP19A1 +268T>C resides in exon 10 of the 3′ untranslated region (UTR) of the gene, but has no known function (Figure 1). Neither polymorphism was significantly associated with percent MD among 550 Caucasian women participating as controls in a case-control study of breast cancer in the United States regardless of menopausal status or PMH use among postmenopausal women³¹ or among 1,286 postmenopausal women of unreported ethnicity participating in a prospective cohort study in the United Kingdom who were past or never users of PMH⁵³. A third study among 486 Caucasian controls in the United States also did not observe a significant association of the +268T>C polymorphism following adjustment for age and BMI and irrespective of menopausal status or use of PMH among postmenopausal women⁴¹. In the absence of additional data, these polymorphisms do not appear to be associated with MD nor appear to be breast cancer risk factors^{74, 75}.

AR CAG repeat polymorphism

Two studies investigated this polymorphism in the androgen receptor gene (Figure 1). Longer transcripts may produce less active receptor^{76, 77}. MD did not vary by number of AR CAG repeats (≥ 22) among 538 (~73% postmenopausal) healthy Caucasian women participating as controls in a study of breast cancer⁴¹ or by ≥ 21 AR CAG repeats among 246 (~50% postmenopausal) Caucasian women with breast cancer whose screening mammograms were retrieved within the five years preceding their cancer diagnosis⁴⁰. The former study⁴¹ did not observe an effect on percent MD from PMH use, although women in the latter study⁴⁰ with the longer less active AR allele had higher MD if they also took a combination formulation of estrogen plus progesterone PMH. The associations with MD may reflect a large number of comparisons in the data of relatively small numbers of women⁴⁰. Moreover, the AR polymorphism is inconsistently associated with breast cancer risk^{78, 79}.

CYP1A2 −154A>C (CYP1A1*1F)

This polymorphism may induce enzyme activity⁸² to lower bioactive estrogen concentrations and, hypothetically, MD. A marginally significantly lower MD was observed overall among 268 pre- and 60 postmenopausal women screened in a mammography clinic, and particularly among premenopausal women⁴⁶, whereas a statistically significantly higher MD was reported among control subjects from a case control study of breast cancer in 97 pre- and 153 postmenopausal women³⁸. Interestingly, both studies sampled from among female residents in Hawaii. The former study⁴⁶ composed healthy women with an entry (baseline) mammogram, while the latter study³⁸ incorporated healthy women participating in two nested case-control studies nested within the Multiethnic Cohort Study who each had an average of 2.5 mammograms in order to reduce intra-individual variability in MD measurements. Given similarities in computerized mammographic assessments and adjustment with similar covariates, the findings from the latter study³⁸ may be more reliable due to increased precision of the measure. With only two studies, the influence of the polymorphism with MD is inconclusive, as is its association with breast cancer in the Shanghai Breast Cancer Study⁸³, although decreased risk has been observed among predominantly postmenopausal women in the Multiethnic Cohort Study⁵⁹.