Measurement of Sex Steroid Hormones in Breast Adipocytes: Methods and Implications

Understanding the actions of sex steroid hormones in mammary carcinogenesis is critical for developing improved methods of preventing and treating breast cancer, but the lack of validated methods for measuring hormones within breast tissues has been limiting. The importance of studying intramammary hormone levels is supported by several observations. First, the doubling of breast cancer risk associated with elevated serum levels of estrogens and androgens seems lower than expected given that hormones represent a key etiologic factor in the genesis of this tumor (1). Second, limited data suggest that steroid hormone levels in breast tissues, including neoplastic, glandular, and adipose tissues, differ dramatically from blood levels, particularly among postmenopausal women (2). Further, within the breast, hormone levels between premenopausal and postmenopausal women are comparable (3-7). The source of high levels of breast tissue estrogens in postmenopausal women has been attributed both to the uptake of circulating hormones and to the local synthesis and metabolism of these steroids in the breast itself (8). A growing body of evidence shows that enzymes involved in hormone metabolism (aromatase, sulfatase, sulfotransferase, 17β-hydroxysteroid dehydrogenase) are expressed and functional in normal and neoplastic breast tissues (8, 9). Thus, the breast, as a site of uptake, biosynthesis, and metabolism of circulating hormones, has the capacity to modulate its microenvironment.

Methodologic challenges have limited studies of breast tissue hormone concentrations (9). Breast tissue varies in relative epithelial, stromal, and fat composition, making it difficult to develop a standardized metric for measuring hormone levels in pulverized specimens and compare results between subjects. In addition, some laboratory assays have suffered from lack of specificity and limited sensitivity. Understanding differences between intracellular and interstitial fluid hormone levels is another important issue. Given that mammary fat has a relatively homogeneous cellular makeup and is an important site for hormone production and/or sequestration, particularly after menopause, we refined and validated methods for measuring steroid hormones in breast adipocytes.

This report presents the results of a proof-of-performance study designed to refine an existing assay procedure and to assess the reproducibility of hormone measurements obtained by the modified assay. Breast adipose tissue specimens were collected from women undergoing surgical resections for a spectrum of pathologic conditions ranging from noninvasive breast lesions to invasive carcinoma. These samples were used to measure levels of androstenedione (A), testosterone (T), estrone (E₁), and estradiol (E₂) in adipocytes.

Anonymized adipose tissue samples that were not required for diagnosis were systematically collected from breast specimens removed for clinical indications. Age and final pathologic diagnosis based on the most severe lesion found in the resection or preceding biopsy were recorded as follows: invasive carcinoma (ductal or lobular), in situ carcinoma (ductal or lobular), atypical hyperplasia (ductal or lobular), and ductal hyperplasia or nonspecific fibrocystic changes. We collected two fat samples from each breast specimen. For specimens containing grossly identifiable invasive carcinoma, we removed one fragment adjacent to the tumor edge and one as far distant from the tumor edge as possible. For specimens that did not contain identifiable invasive carcinoma, we collected one fragment from the center of the specimen and the other from near the margin of the resection. Sampling was done to obtain the highest percentage of pure fat based on gross inspection; areas of adipose tissue intimately intermixed with fibroglandular tissue were avoided. Tissues were frozen at −80°C within 1 h following removal. This project was conducted under an exemption granted by the review boards of participating institutions.

The study was conducted in three steps. In step I, two fat specimens from five women were evaluated to assess the feasibility of extracting and measuring steroid hormones using a modified assay procedure. In step II, we evaluated the reproducibility of the measurements using tissues obtained from an additional five women. In step II, one adipose fragment from each woman was divided into four samples and measured in two batches. Each batch contained two aliquots per subject. In step III, which was restricted to specimens containing invasive cancer, we preliminarily assessed whether hormone levels in adipose tissue differ by proximity to the breast tumor. Specifically, we compared measurements from proximal and distal adipose tissue samples obtained from 10 women undergoing surgical excision of a breast cancer. For each woman in step III, two aliquots were obtained from the proximal and distal fragments. The schema for this component of the study is shown in Fig. 1. For each woman, two extracts were assayed in the same batch, one obtained from the proximal and one from the distal fragment. Samples were kept at −80°C at the laboratory until analyzed.

Quantification of steroids in adipocytes was conducted using a modification of the methods described by Rodbell (10) and O'Brien et al. (11), in which oil was separated from the more dense stromal cells by flotation. Approximately 0.4-g portions of breast adipose tissue were minced and incubated in a 20-mL glass vial with 1 mL of 1 mg/mL collagenase (type 1A; Sigma Chemical Co.) in Krebs Ringer bicarbonate buffer (4% bovine serum albumin; Sigma Chemical Co.) at 37°C for 50 min in a shaking water bath. During enzymatic disaggregation, significant lysis of the adipocytes occurred and oil droplets were formed. The floating oil layer was separated by centrifugation in a microcentrifuge at room temperature for 2 min at 10,000 rpm, and the oil was weighed. A, T, E₁, and E₂ were then quantified by RIA with preceding purification steps. Internal standards (³H-A,³H-T, ³H-E₁, and ³H-E₂) were first added to the oil and incubated at 50°C for 10 min to follow procedural losses of each steroid. The steroids were then extracted from the oil with 80% methanol/H₂O and the extract was evaporated. The residue was then redissolved in 0.4 mL of distilled water and an additional extraction was carried out with 3 mL of hexane/ethyl acetate (3:2). This extraction step was repeated and the solvents were evaporated to dryness and redissolved in 1 mL of isooctane. The isooctane was transferred to a Celite partition chromatography column, and A, T, E₁, and E₂ were eluted sequentially and quantified by RIA as described previously (12-15). Results are expressed per gram of oil.

From the residual aqueous fraction after extraction, a 0.1-mL aliquot was taken to measure estrone sulfate (E₁S) by direct immunoassay using a commercial kit (Diagnostics Systems Laboratories).

We assessed assay reproducibility using a nested, within-person ANOVA model with data transformed to the natural logarithmic scale. Variance components methods were used to estimate the variability in the laboratory measurements. We estimated the variability among subjects (σ²_a), variability between batches for a given subject (σ²_b), and variability associated with different aliquots of the same extracted tissue (σ²) using the SAS procedure Proc VARCOMP (16). From the variance estimates, we computed an estimate of the intraclass correlation coefficient as (σ²_a / [σ²_a + σ²_b + σ²]) and an estimate of the assay coefficient of variation (CV; ref. 17). Geometric means are presented.

Initially, assays were conducted on adipose tissues obtained from five women, of whom four were undergoing reduction mammoplasty and one was undergoing surgical excision of ductal carcinoma in situ. The women ranged in age from 23 to 59 years (mean 44 years), and two were postmenopausal. Two fragments of adipose tissue were removed from the same region of each breast, and the steroid hormones were measured in adipocytes from each fragment.

For all women, levels of A were highest, followed by E₁, T, and E₂. This ranking was the same for samples from premenopausal and postmenopausal women, but absolute levels of A, T, and E₂ in premenopausal samples were approximately double those of postmenopausal samples. Concentrations of E₁ were similar for both groups. The within-subject CVs ranged between 20% and 32% for the four hormones evaluated (Table 1). No E₁S was detectable in the samples.

Next, we evaluated the reproducibility of the laboratory measurements using extracts obtained from a single tissue fragment from five additional women ages 26 to 45 years (mean 36 years). These samples were obtained from a younger group than those in step I. For each woman, a total of four aliquots were obtained from the adipose fragment, two of which were assayed in one batch and the remaining two were assayed in a second batch. The steroid hormones were measured in adipocytes from each fragment.

The hormones in these samples showed a similar pattern, and levels were of similar magnitude as observed in step I, with concentrations of A being highest and E₂ levels lowest (Table 2). The overall CVs were lower than those found initially, ranging from 8% to 19%, with the within batch CVs ranging from 5.1% to 16.7% and the between-batch CVs all less than 10%. The intraclass correlation coefficients were quite high, with values of 80% or greater for all of the analytes measured.

For the final component of the study, adipose fragments distal and proximal to the breast tumor were evaluated. Samples were extracted from adipose tissues in 9 of the 10 women who provided samples and ranged in age from 39 to 84 years, with a mean of 61.3 years. Five women were postmenopausal at the time of surgical excision of the tumor. The pattern of relative concentrations of the different hormones was similar to that found in the previous steps; however, the absolute values were lower. Hormone levels were similar for adipose tissue fragments that were distal versus proximal to the tumor, with the possible exception of E₁, where levels were 10% higher in distal fragments (Table 3).

In this study, the steroid assay method yielded within-batch CVs ranging from 5% to 17%; between-batch CVs ranging from 2% to10%; and intraclass correlation coefficients of 80% to 95% for A, T, E₁, and E₂ in breast adipocytes. Whereas sex steroids may be found in adipocytes in an unesterified form or esterified to fatty acids (lipoidal estrogens), the assay method used to measure the steroid hormones in breast adipocytes in the present study predominantly detected unesterified estrogens, as esterified estrogens are not likely to be removed during the chromatography procedure and require hydrolysis to free the estrogens (18). Also, conjugated steroids, such as E₁S, have low lipid solubility and, thus, would not be expected to be sequestered in breast adipocytes, as confirmed in this study. The present data support the feasibility of measuring unconjugated steroid hormone concentrations in breast adipocytes in epidemiologic and clinical studies.

After menopause, the incidence rates of estrogen receptor–positive breast carcinoma continue to increase despite decreasing serum estrogen levels, suggesting that intramammary hormone synthesis may be an important mechanism in carcinogenesis. The aromatization of androgens to estrogens in peripheral adipose tissue is a primary source of circulating estrogen in the postmenopause, and recent data suggest that extragonadal aromatase expression increases with age following menopause (19). Adipocytes, adipose stromal cells, and vascular cells are the primary cell constituents of adipose tissue, with the largest proportion being adipocytes (20). Within the breast, aromatase expression has been reported in both tumor and adipose tissue (20); furthermore, within the breast adipose, aromatase expression is higher in stromal cells compared with adipocytes, yet estrogen receptor α expression is greater in adipocytes (21, 22).

In the present study, we measured steroid hormones in oil derived from adipocytes. Among the hormones measured, we found that A was most abundant, followed by E₁, T, and E₂, with the latter hormones at concentrations ranging from 2% to 13% of that found for A. This ranking was similar for premenopausal and postmenopausal adipose samples. Given that our data are expressed per gram of oil as opposed to per milliliter of serum, it is difficult to interpret findings with respect to hormone levels in the circulation. However, the relative concentrations of the adipocyte hormones were similar to their ranking in serum. During the follicular and luteal phases of the menstrual cycle, average serum levels of A are 20- to 30-fold higher than those of E₁, whereas average serum T levels are 3- to 8-fold higher than E₂ (23). In the present study, we found that levels of A were approximately 10-fold higher than E₁ in both premenopausal and postmenopausal adipose extracts; for T, levels were about twice that of E₂ in premenopausal samples and more than 7-fold higher in the postmenopausal fat. However, it is notable that A was at least 23 times higher than T in the adipocytes, compared with a 3- to 5-fold difference in female serum. This suggests that the availability of substrate for E₁ formation by aromatase in adipose tissue is high.

Few studies have explored hormone concentrations in breast tissues (8), most of which measured hormones in tumor tissue, tissue surrounding the tumor, and/or distal glandular tissue (3, 4, 6, 24, 25). Analyses of steroid hormones in breast adipocytes have been even more restricted. In one study that used a different purification procedure than ours, but similar RIAs to measure hormone levels (26), the ranking of A, E₁, T, and E₂ concentrations in breast adipose tissue from postmenopausal women was similar to our findings. However, the values were expressed as pmol/g tissue, precluding direct comparisons of absolute levels. Reported absolute values in this study (pmol/g tissue) were as follows: A = 16, E₁ = 0.67, T = 0.62, and E₂ = 0.20. The similarity in ranking between these studies despite differences in methods is encouraging.

Some previous reports have found that promoter switching may increase aromatase activity in tissues adjacent to carcinoma, but we did not find topographical differences in adipose hormone levels in relation to tumor. This could reflect our selection of pure adipose tissue samples based on gross examination, and it is possible that adipose tissue that is located within microscopic distances of tumor as opposed to macroscopically removed might show hormone elevations. Similar findings have been reported in another study that compared tumor and adipose tissue from premenopausal and postmenopausal breast cancer cases (27), where little difference in hormone concentrations between adipose tissue samples at various distances from the breast tumor were observed.

Further, in the previously described study (27), low but measurable quantities of E₁S and estradiol sulfate were found in breast adipose tissue (142 and 59 fmol/g, respectively). Explanations for the contradictory findings regarding the sulfated estrogen are not clear, but may rest in the assay methodology. Sulfated molecules are highly polar; thus, E₁S in the fat should be negligible. Detection of sulfated compounds could represent contamination of adipose tissue with steroids from interstitial fluids or blood, either within the tissue vessels or from hemorrhage into the tissue.

Addressing the importance of local hormone metabolism in mammary carcinogenesis will require future interdisciplinary studies in which epidemiologic factors are combined with detailed pathologic and biological tissue characterizations. Research is needed to develop methods for defining tissue content and for standardizing metrics for expression of steroid hormone concentrations. Advances in this area may permit the improved etiologic understanding of hormonal carcinogenesis, which in turn holds promise for developing new prevention and treatment strategies and better monitoring of therapies.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.