Steroid Sulfatase Expression Is an Independent Predictor of Recurrence in Human Breast Cancer¹

Recently, as part of a new therapeutic strategy for breast cancers, several groups of compounds (1, 2, 3, 4, 5, 6, 7, 8) have been designed to inhibit estrone sulfatase. Human breast cancer tissue can hydrolyze E1S³ to estrone (9, 10, 11), and estrogen levels are 10 times higher in the breast tumors of postmenopausal women than in plasma from the same individuals (12, 13). This concentration gradient is not entirely due to estradiol uptake and binding to the ER, because tissue estradiol correlates poorly with ER levels (14). Local synthesis of estrogen from circulating precursors via estrone sulfatase (9, 10, 11) and/or aromatase (9, 15, 16) pathways might therefore be responsible. The E1S plasma concentration is 10–20 times higher than that of estrone (10, 17), and its half-life is much longer (18). Thus, it is likely that E1S in plasma can act as a reservoir for the formation of unconjugated estrogens by the action of estrone sulfatase. Treatment of postmenopausal breast cancer patients with aromatase inhibitors, such as aminoglutethimide or 4-hydroxyandrostenedione, results in only a modest reduction in plasma estrogen concentrations (19, 20). Furthermore, plasma E1S concentrations remain relatively high (400–1000 pg/ml) after the administration of the new aromatase inhibitors CGS 16949A, fadrozole, or vorozole (21, 22, 23). Hydrolysis of E1S by estrone sulfatase is the most likely source of plasma estrogens, and the level of sulfatase activity in human breast tumors is much higher than that of aromatase (9). Its biological relevance is therefore of interest.

STS appears to be a single enzyme that hydrolyzes several sulfated steroids such as E1S, DHEA-S, and cholesterol sulfate (24, 25), although it has been reported that a distinct form might act on E1S (26, 27). Investigations of the physicochemical or kinetic characteristics of STS in microsomal preparations of breast tumors or breast cancer cells have been inconclusive in this regard, with evidence for both the same and distinct enzymes being responsible for the catalysis of E1S and DHEA-S (27, 28, 29). However, a recent study showed that STS DNA-transfected COS-1 cells hydrolyze not only E1S but also DHEA-S (30). Although E1S itself is unable to bind to the ER and stimulate a biological response, additional in vitro studies with MCF-7 cells have demonstrated that it is converted to estradiol (27, 31) and can cause an increase in progesterone receptor levels (32) and pS2 protein (33) and stimulate the production of cathepsin D, an estrogen-inducible protein (34, 35). Very recently, (p-O-sulfamoyl)-N-tetradecanoyl tyramine, an estrone sulfatase inhibitor, was demonstrated to completely inhibit sulfatase activity and the proliferation of MCF-7 cells (8). Androstenediol can bind to the ER and stimulate ER-positive breast cancer cells (36) and dimethylbenzanthracene-induced mammary tumors in rats (37). In postmenopausal women, isotopic infusion studies have revealed that almost 90% of androstenediol originates from DHEA-S, which is initially converted to dehydroepiandrosterone or androsterone sulfate (38). Therefore, STS could be active not only in the formation of estrone from E1S, but also in the synthesis of androstenediol from DHEA-S.

At present, little is known about the biological and/or clinical significance of intratumoral estrogen and/or estrogenic steroid synthesis. We have therefore investigated the relationship between the tissue levels of STS mRNA and clinicopathological status in a series of human breast cancers. We report here that patients with breast cancers containing high levels of STS mRNA have significantly shorter disease-free survival as compared with those with low levels of STS mRNA.

Materials for this study were obtained from 97 patients with primary breast carcinomas who underwent curative operations at Fujita Health University Hospital, Marumo Hospital, and Tochigi Cancer Center between 1990 and 1994. The average age of the patients was 52.2 ± 9.5 years (mean ± SD; range, 33–77 years). Fifty patients received adjuvant chemotherapy, and 65 patients were given adjuvant endocrine therapy. Disease recurrence was documented on the basis of physical examination, radiological and laboratory tests, and/or other relevant diagnostic procedures. The median follow-up period was 42.5 months (range, 22–94 months) for patients still alive with high STS mRNA levels and 50.5 months (range, 31–95 months) for those with low STS mRNA levels.

A total of 97 tumors were classified by pathologists according to the WHO scheme for typing breast tumors (39). Histologically, there was 1 case each of ductal carcinoma in situ, mucinous carcinoma, and invasive lobular carcinoma and 94 cases of invasive ductal carcinomas, of which 52 cases were node negative, and 45 cases were node positive. Each tumor, except for one ductal carcinoma in situ and one invasive lobular carcinoma, was graded in parallel according to the criteria described by Bloom and Richardson (40). Tumor size was measured at surgery by the operating physicians. Immediately after surgical removal, the specimens were frozen in liquid nitrogen and then stored at −80°C until use. ERs were assayed by means of the dextran-coated charcoal method with a cutoff value of 5 fmol/mg protein (41). Fifty-seven cases were ER positive, 37 cases were ER negative, and 3 cases were unknown.

Frozen tissues were homogenized in 5 m guanidine thiocyanate containing 5 mm sodium citrate and 0.5% sodium sarcosyl, and total RNA fractions were prepared from the homogenates as described by Chirgwin et al. (42). The RNA concentration was determined from the spectrophotometric absorption at 260 nm.

Quantitative analysis of STS mRNA in the RNA fractions was carried out by RT-PCR using a fluorescent primer in the presence of an internal standard RNA. In brief, oligonucleotides of antisense primer STS-2R (5′-AGGGTCTGGGTGTGTCTGTC-3′) for reverse transcription and antisense primer STS-2R (5′-AGGGTCTGGGTGTGTCTGTC-3′) and sense primer STS-1F (5′-ACTGCAACGCCTACTTAAATG-3′) for PCR were synthesized. Sense primer STS-1F for PCR was labeled with FAM (Perkin-Elmer Corp.), a fluorescent dye, after conjugation with Aminlink 2. The coding sequence between the two PCR primer sites is interrupted by an 18-kbp intron in the STS gene. To prepare the internal standard RNA, modified human STS cDNA was constructed by inserting a 64-bp fragment of AluI-digested pUC119-DNA between the two PCR primer sites. The internal standard RNA was synthesized in vitro with T7 RNA polymerase using the modified STS cDNA as a template, purified on an anion exchange column (Qiagen), and then quantitated from the absorbance at 260 nm. Total RNA (1–2 μg) and the internal standard RNA (2 amol) were subjected to reverse transcription with 5 units of RAV-2 reverse transcriptase (Takara Shuzo, Kyoto, Japan) and specific antisense primer STS-2R at 42°C for 40 min. The resulting cDNAs were amplified by PCR using fluorescent dye-labeled primer STS-1F and primer STS-2R. The PCR conditions were denaturation at 94°C for 20 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s for 23 cycles. The fluorescent PCR products were analyzed on a 2% agarose gel with a Gene Scanner 362 Fluorescent Fragment Analyzer (Perkin-Elmer Corp.). The amount of STS mRNA was calculated from the peak areas of the fluorescent products by the internal standard method.

Statistical analyses were carried out using SAS-REL.6.12 software. Mean levels of STS mRNA were compared using the Mann-Whitney U test or the Kruskal-Wallis test. The χ² test was performed for the contingency table analyses. Spearman’s correlation coefficient was used to investigate the correlation among the different clinicopathological variables. Relapse-free and overall survival curves were generated using the method of Kaplan-Meier (43). Survival comparisons were made using the log-rank test and proportional hazards (Cox) multiple regression (44). The event considered in our analysis of relapse-free survival was first recurrence of disease. Overall survival refers to survival with or without recurrence of the disease. Relapse-free survival and overall survival were calculated from the date of first surgery to the date of clinical or pathological relapse or death. The cutoff for significance was P = 0.05.

Examples of typical RT-PCR results are shown in Fig. 1. Fig. 2 shows the distribution of the STS mRNA levels in the 97 breast tissues. The distribution was not normal, and the median value was 728.0 amol/ng RNA (range, 0–11,778 amol/ng RNA).

STS mRNA levels in breast cancer were examined for association with other clinicopathological parameters. The results are summarized in Table 1. The levels of STS mRNA in node-positive tumors were significantly (P = 0.033) higher than those in node-negative tumors. There were also significantly (P = 0.029) higher STS mRNA levels in patients who experienced a recurrence during the follow-up period (mean, 40.8 months; median, 39 months) compared with those with no evidence of further disease (mean, 49.2 months; median, 48 months). There were no other significant correlations.

Because the distribution of STS mRNA levels was approximately log normal, we used the log₁₀ of the value (after setting all 0s to 1) as a continuous variable and examined its association with relapse-free and overall survival. For all patients, higher STS mRNA levels predicted shorter relapse-free survival (P = 0.028; relative risk, 2.847; 95% CI, 1.122–7.227/log-unit increase), but the same relation between STS mRNA levels and overall survival did not appear (P = 0.389).

One common strategy for analyzing continuous variables is to convert them into categorical variables by grouping patients into two or more groups. Most prognostic factors are usually considered as dichotomized, discontinuous variables, but there are no reference values for STS mRNA levels in human breast tumors. For the purpose of statistical evaluation, the patients were divided into two groups of low and high STS mRNA expression, and the cutoff point was selected to give optimal separation between low and high risks of relapse, according to the method of Tandon et al. (45). As shown in Fig. 3, a considerable range of cutoff values (880–1740 amol/ng RNA) gave a statistically significant separation of disease-free survival possibilities. The optimal cutoff point was 1240 amol/ng RNA, giving the minimum P of 0.002 by the log-rank test, which resulted in 28.9% (28 of 97 patients) of the patients being grouped as having high STS mRNA levels in the present series. The generalized Wilcoxon test also gave the same optimal value (data not shown).

Relations between dichotomized STS mRNA levels and other clinicopathological parameters of prognosis and adjuvant therapy were then determined (Table 2). There was no significant association with age, menopausal status, tumor size, nodal status, histological grade, ER status, adjuvant chemotherapy, or adjuvant endocrine therapy. Although the levels of STS mRNA were significantly higher in tumors showing lymph node metastasis than in those without nodal involvement, the node-positive incidence did not differ between the patients with high and low STS mRNA levels.

A total of 20 patients had recurrent disease, and 15 had died at the time of the analyses. The first recurrence sites are shown in Table 3. Overall, 39.3% (11 of 28 patients) of patients with high STS expression exhibited relapse as compared to 13.0% (9 of 69 patients) of those in the low STS expression group. Relapse-free and overall survival curves are shown in Fig. 4. High levels of STS mRNA expression were associated with a significantly (P = 0.002) increased risk of recurrence. There was no association between expression and overall survival.

In an observational study of this type, the patients were not enrolled in a specific protocol. Therefore, we examined treatment history as it related to survival because of the possibility that patients with low STS mRNA levels received less therapy as a group than patients with high STS levels, thus having a potential impact on the comparative survival of the two groups. In this study, 51.5% of patients received adjuvant chemotherapy, and 67.0% were given adjuvant endocrine therapy. The numbers did not differ significantly between the low STS level and high STS level groups (49.3 and 69.6% of the low STS level group and 57.1 and 60.7% of the high STS level group, respectively). The mean STS mRNA levels were also comparable for adjuvant chemotherapy and endocrine therapy (P = 0.415 and 0.278, respectively), indicating that access to treatment was not the determining factor in relapse-free survival. Because the group that received chemotherapy included more advanced cases, the relapse-free survival and overall survival in the chemotherapy group were actually worse than those in the nontreated group in univariate analysis (P = 0.002 and 0.007, respectively). On the other hand, the group that received endocrine therapy consisted almost entirely of ER-positive cases, which generally have a favorable prognosis, and their relapse-free survival and overall survival were actually better than those in the nontreated group in univariate analysis (P = 0.025 and 0.115, respectively). Taken together, these observations indicate that treatment history was not a factor in survival differences between low STS level and high STS level mRNA groups.

On univariate analysis, lymph node involvement (P = 0.001), STS mRNA level (P = 0.002), adjuvant chemotherapy (P = 0.002), adjuvant endocrine therapy (P = 0.025), and ER status (P = 0.023 on the basis of 94 patients, because ER status was not available for 3 patients) were significant prognostic factors for relapse-free survival in 97 breast cancer patients. To assess whether STS mRNA expression was independently prognostic, multivariate analysis was conducted. All variables except endocrine therapy, which significantly correlated with ER status (Spearman’s correlation coefficient, 0.523; P = 0.0001), were taken into account through a stepwise analysis. The model gave nodal status, STS mRNA level, and ER status as independent prognostic factors (Table 4). Nodal status came first (P = 0.002; relative risk, 5.712; 95% CI, 1.883–17.325), followed by STS mRNA level (P = 0.008; relative risk, 3.320; 95% CI, 1.368–8.060) and ER status (P = 0.009; relative risk, 3.372; 95% CI, 1.360–8.363), whereas tumor size, histological grade, and adjuvant chemotherapy were not significant.

The sulfatase pathway in human breast cancer has been well documented (9, 10, 11). There have been few reports of STS mRNA levels in human breast cancers thus far. In the present investigation, they ranged from 0 to 11778 amol/ng RNA and were much higher than the levels of aromatase (mean, 4.53 ± 0.66 amol/ng RNA) measured previously (16). This result is consistent with reports of a greater turnover rate for sulfatase than aromatase in vitro (9). Selcer et al. (8) found that estrone sulfatase inhibitors were effective for depressing the proliferation of estrogen-dependent MCF-7 cells in the presence of E1S as the only source of estrogen. Furthermore, the levels of inhibition of cell proliferation by sulfatase inhibitors are positively correlated with their potency to inhibit the enzyme activity (8). Taking these findings into consideration, we can hypothesize that considerable levels of estrogens may be synthesized by STS from the large circulating reservoir of E1S in breast tumors and act on the carcinoma cells in an autocrine or paracrine fashion.

The present investigation focused on the association of STS expression with clinicopathological factors and overall and relapse-free survival. It provided evidence that STS expression in human breast cancer is a useful prognostic marker for the identification of high- and low-risk patients. Such a good correlation between STS mRNA expression as a continuous or dichotomous variable and relapse-free survival has not been reported previously. Multivariate analysis further showed that the expression of STS mRNA is an independent prognostic indicator in predicting relapse-free survival. Because the relative impact of prognostic factors on relapse-free survival reflects their respective roles in tumor biology, we can speculate that STS may be important in the progression of human breast cancer. The results suggest that in breast cancer tissue expressing high levels of STS, tumor cells that escape surgical removal may grow faster; therefore, the patients may relapse earlier. Recently, considerable efforts have been directed toward the development of the potential application of potent sulfatase inhibitors (1, 2, 3, 4, 5, 6, 7, 8) in women with endocrine-dependent breast tumors. This might be particularly important for breast tumors expressing high levels of STS.

Although we found a good correlation between STS mRNA expression and relapse-free survival, Evans et al. (11) previously found no link between time to relapse and estrone sulfatase activity. The difference in analytical methods used in the two studies, with analyses of STS mRNA levels in the former and enzymatic STS activities in the latter, may have contributed to this inconsistency. In this context, it is of interest that Evans et al. (46) found no correlation between the levels of STS mRNA detected by the RT-PCR method and the estrone sulfatase activities in breast tissues, although this was in contrast to the findings of Pasqualini et al. (47). Evans et al. (46) demonstrated that the sensitivity for detection was lower for activity versus mRNA by RT-PCR, and that the activity was lower in normal breast cells than in breast cancer cells. Taking such a situation into consideration, STS activity might be more susceptible to being confounded by the ratio of neoplastic cells to stromal cells in each biopsy. Another possible explanation is that STS might be active in the conversion of DHEA-S to androstenediol, which could stimulate breast cancer cells, judging from the fact that androstenediol was present in breast tissues (15) and showed estrogenic action in an in vitro study (36).

In our series, prognostic significance for STS mRNA expression was observed in terms of relapse-free survival but not overall survival. A larger series of observation might provide additional data in terms of overall survival.

In conclusion, this is the first report that STS mRNA expression is well correlated with disease-free survival and is an independent prognostic factor in human breast cancer. The results support the hypothesis that estrogens that are locally produced in human breast cancer tissue may play an important role in tumor progression, leading to metastasis. Although the exact significance of intratumoral STS in breast cancers remains unclear, the administration of sulfatase inhibitors to breast cancer patients with high levels of STS mRNA might be an additional treatment option, along with aromatase inhibition as an endocrine therapy.