Hormone replacement therapy dependent changes in breast cancer-related gene expression in breast tissue of healthy postmenopausal women

Abstract

Risk assessment of future breast cancer risk through exposure to sex steroids currently relies on clinical scorings such as mammographic density. Knowledge about the gene expression patterns in existing breast cancer tumors may be used to identify risk factors in the breast tissue of women still free of cancer. The differential effects of estradiol, estradiol together with gestagens, or tibolone on breast cancer-related gene expression in normal breast tissue samples taken from postmenopausal women may be used to identify gene expression profiles associated with a higher breast cancer risk. Breast tissue samples were taken from 33 healthy postmenopausal women both before and after a six month treatment with either 2mg micronized estradiol [E2], 2mg micronized estradiol and 1mg norethisterone acetate [E2+NETA], 2.5mg tibolone [T] or [no HRT]. Except for [E2], which was only given to women after hysterectomy, the allocation to each of the three groups was randomized. The expression of 102 mRNAs and 46 microRNAs putatively involved in breast cancer was prospectively determined in the biopsies of 6 women receiving [no HRT], 5 women receiving [E2], 5 women receiving [E2+NETA], and 6 receiving [T]. Using epithelial and endothelial markers genes, non-representative biopsies from 11 women were eliminated. Treatment of postmenopausal women with [E2+NETA] resulted in the highest number of differentially (p<0.05) regulated genes (16.2%) compared to baseline, followed by [E2] (10.1%) and [T] (4.7%). Among genes that were significantly down-regulated by [E2+NETA] ranked estrogen-receptor-1 (ESR1, p=0.019) and androgen receptor (AR, p=0.019), whereas CYP1B1, a gene encoding an estrogen-metabolizing enzyme, was significantly up-regulated (p=0.016). Mammary cells triggered by [E2+NETA] and [E2] adjust for steroidogenic up-regulation through down-regulation of the estrogen-receptor pathway. In this prospective study, prolonged administration of [E2+NETA] and to a lesser extent of [E2] but not [T] were associated in otherwise healthy breast tissue with a change in the expression of genes putatively involved in breast cancer. Our data suggest that normal mammary cells triggered by [E2+NETA] adjust for steroidogenic up-regulation through down-regulation of the estrogen-receptor pathway. This feasibility study provides the basis for whole genome analyses to identify novel markers involved in increased breast cancer risk.

Figure 1

Flow diagram of cases during recruitment and breast tissue sampling.

Figure 2

Confirmation of the volunteers' adherence to treatment. A: both the individual and the mean changes in FSH concentration (in IU/L, with the 95% confidence interval) at the onset and at the end of treatment with each of the four treatment modalities are depicted. In one participant the level of FSH rose from 71.4 IU/L to 77.6 IU/L during six month treatment with [T]. This difference was considered to be within the normal variability of the FSH level. B: both the individual and the mean changes in SHBG concentration (in nmol/L, with the 95% confidence interval) at the onset and at the end of treatment with each of the four treatment modalities are depicted. In one individual the level of SHBG failed to drop substantially during the six month intake of [T]. The participant with lack of SHBG change is different from the one with a slight rise in the FSH level (Figure 2A) and both were therefore not considered as being non‐compliant.

Figure 3

Supervised hierarchical clustering to evaluate similarity among treatment groups. Expression levels of markers expressed differently (p<0.05) among treatment groups. Marker names, shown on the right side of the clustering diagram, were normalized by Spearman rank correlation from −1.0 to 1.0. Blue squares in the cluster diagram indicate a positive relative transcript expression (0–1.0), yellow squares a negative relative expression (−1.0 to 0) and black squares depict a relative expression of zero. Each row depicts a single gene, each column a single case. The dendrogram on top present the relatedness of the profiles of individual cases and the dendrogram on the left the relatedness of the individual genes and miRNAs in this clustering. The longer the dendrogram arm, the greater the difference in between individual cases and genes within a cluster. Cases are color‐ and number‐coded according to the treatment group: (i) gray [20]; [no HRT] before therapy and black [21], [no HRT] after 6 months. (ii) pink [10], [E2] before therapy and red [11], [E2] after 6 months. (iii) light green [30], [E2+NETA] before therapy and dark green [31], [E2+NETA] after 6 months. (iv) light blue [40], [T] before therapy and dark blue [41], [T] after 6 months.

Figure 4

Box‐Whisker plots of gene markers significantly differentially expressed after HRT. mRNA expression data of TFF1, STMN1, C2CD4A and NCAM1 in 22 biopsies taken before start of therapy [no HRT] were compared with those of the biopsies trichotomized according the various treatment groups. The box‐plot shows the five statistics (lower whisker is 5% minimum, lower box part is the 25th percentile, solid line in box presents the median, upper box part is 75th percentile, and upper whisker is 95% maximum). Significance levels (*) are relative to the expression levels measured in the control group [no HRT]. PFDR adjusted, 10%<0.05*.

Figure 5

Individual reactions in the change in mRNA levels of nuclear receptors in relation to the hormonal treatment. mRNA expression levels of the nuclear receptors AR, ESR1, ESR2 and PGR in 22 biopsies taken before start of therapy were compared with those measured in biopsies of the same individuals after treatment. T‐test statistics (2‐sided p) are indicated in the graphs.

Figure 6

Pathways of steroid hormone synthesis, metabolism and tissue sensitivity to HRT (Kelemen et al., 2008; Pasqualini and Chetrite, 2005; Masson et al., 2010) MiRNA's predicted to target steroidogenic or metabolizing enzymes are encased in blue. Genes measured as either up‐ or down‐regulated in our current study after a specific hormone replacement therapy are marked by colored circles. The effects exerted by [E2+NETA] are depicted in green, the effects exerted by [E2] in red, and the effects exerted by [T] in gray. a) Steroid hormone biosynthesis: Cytochrome P450 17A1 (CYP17A1) 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 CYP19A1 (aromatase). Androgen conversion to estrogen in adipose tissue by CYP19A1 is an important source of bioactive endogenous estrogens among postmenopausal women. Hydroxysteroid dehydrogenase 3ß1 (HSD3ß1) catalyzes the interconversion of pregnenolone and progesterone and of DHEA and androstenedione, whereas HSD17β1 catalyzes the conversion between androstenedione or testosterone and E1 or E2, respectively. The availability of HSD17β1 is regulated by a number of miRNAs, which are inhibited both by [E2+NETA] and [E2]. b) Estrogen metabolism: The ‘sulfatase pathway’ converts stored estrogen sulfates into the bioactive unconjugated E1 and sulfotransferases convert estrogens into the biologically inactive estrogen sulfates. The CYP‐family of enzymes consists of a cluster of enzymes that function in the oxidative metabolic activation and deactivation of compounds including several steroid hormones. E1 and E2 undergo 2‐hydroxylation by the CYP1A1, CYP1A2 and CYP3A4 enzymes and 4‐hydroxylation by CYP1B1. Catecholestrogens are deactivated by catechol‐O‐methyltransferase (COMT). UDP‐glucuronosyltransferase 1A1 (UGT1A1), crystallin zeta, quinone reductase (CRYZ), glutathione S‐transferase alpha 1 (GSTA1) and sulfotransferases (SULT1/2) are detoxifying enzymes that convert endogenous substrates to inactive metabolites. CRYZ is inhibited by miRNA‐186 and ‐21, which are negatively regulated by [E2+NETA], potentially resulting in an accumulation of catechol estrogen quinines, which are potential initiators of breast cancer. c) Progesterone metabolism: Aldo–ketoreductase family 1 member C4 (AKR1C4) catalyzes the conversion of progesterone and androstenedione to their corresponding alcohols. d) Gene transcription: Progesterone, testosterone and E2 bind to their respective nuclear receptor proteins, progesterone receptor (PGR), androgen receptor (AR) and estrogen receptor alpha (ESR1), and activate genes with corresponding responsive elements resulting in gene expression of genes with such responsive elements. The expression of PGRA/B is up‐regulated by [E2], whereas those of ESR1 and of AR are down‐regulated by [E2+NETA]. e) Growth factor receptors and cell division: Two genes encoding membrane bound receptors, EGFR (alternatively symbolized as HER1) and ERBB2 (HER2), both considered being breast cancer oncogenes, are down‐regulated by [E2+NETA], whereas a number of kinases and cell‐division cycle genes, such as CDK1, CDC20 and CCNE2, are up‐regulated by [E2+NETA] or, in the case of CCNE1 by [E2]. In contrast, CTSD, which encodes cathepsin D and which is often used as a breast cancer tumor marker (Cavalieri et al., 2006), is down‐regulated by [E2+NETA]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)