Endocrine Therapy of Advanced Disease: Analysis and Implications of the Existing Data¹

Endocrine therapy remains an important approach to the treatment of breast cancer, both in metastatic disease and in the adjuvant setting. Hormonal therapy is particularly useful because it is not only efficacious but is also associated with minimal toxicity and excellent quality of life. In recent years, a number of advances have been made in endocrine therapy in both the metastatic and the adjuvant setting.

In the premenopausal setting, estrogen stimulates breast cancer growth. Reduction of estrogen levels by ovarian removal, ovarian irradiation, or the use of a LHRH³ analogue can shrink metastatic disease in about 30% of unselected premenopausal women with metastatic breast cancer. It is also well known that women whose tumor is positive for ERs and/or PgRs have a higher rate of response to any type of endocrine therapy than those whose tumors have low or negligible ER and PgR levels. Women whose tumors are either ER or PgR positive have RRs of 30% or 40%, respectively, to hormonal manipulation, whereas those whose tumors are both ER and PgR positive have RRs as high as 70%. Women whose tumors are both ER and PgR negative have RRs to endocrine therapy of <5% (1, 2).

In premenopausal women with metastatic disease, ovarian ablation by surgery or radiation was historically the first line of endocrine manipulation to be tried for the treatment of metastatic disease (3). In the late 1970s and early 1980s, however, it was demonstrated that tamoxifen, already known to be effective in postmenopausal women with metastatic disease, was also quite active in premenopausal women with widespread disease (4). It was also demonstrated in several small studies (5, 6, 7) and in one small meta-analysis (8) that tamoxifen was approximately equivalent to ovarian ablation as a treatment for metastatic disease in the premenopausal setting. Thus, by the late 1980s and early 1990s, the approach to treatment of premenopausal women with metastatic disease involved the use of tamoxifen until maximal benefit had been achieved and disease once again progressed, followed by ovarian ablation by surgery or radiation, followed by therapies felt most suitable for postmenopausal endocrine-responsive women.

In the early 1990s, however, the LHRH analogues, particularly goserelin (Zoladex), became available. These drugs were shown to be effective in reducing estrogen levels to below postmenopausal levels within 21–28 days in >90% of premenopausal women (9). Phase II trials in 228 women showed a RR of 36.4% and a response duration of 44 weeks. Phase III trials, in particular one large study, showed LHRH analogues to be approximately equivalent to surgical ovarian ablation in women with metastatic disease (10). These data motivated some physicians to move toward the use of LHRH analogues as first-line endocrine therapy in premenopausal women, followed by tamoxifen.

Most recently, several small clinical trials have examined the role of a LHRH analogue plus tamoxifen versus a LHRH analogue alone in the treatment of premenopausal women with metastatic disease (11, 12, 13, 14); these trials have been combined in a small meta-analysis (15). This meta-analysis has shown an increase in objective RR of 31%, an improvement in progression-free survival (hazard ratio = 0.70; P = 0.001), and an improved survival (hazard ratio = 0.78; P = 0.02) in women who received the combination in comparison with the LHRH analogue alone. There are some caveats to be considered in the interpretation of this data. In particular, in three of the four studies included in the meta-analysis, there was no formal crossover of patients who received the LHRH agonist alone to tamoxifen as second-line therapy. In fact, what one would like to compare would be the two drugs used concurrently versus their use sequentially, and this was not done in three of the four studies. In addition, the toxicity data were not collected in any detail, and no quality of life data comparing the combination with the use of a LHRH analogue alone exist. Nonetheless, these data suggest that the use of a LHRH agonist plus tamoxifen could result in an improved clinical outcome, in comparison with the use of a LHRH agonist followed by tamoxifen, in the premenopausal metastatic setting. The current options for the premenopausal metastatic cascade are shown in Fig. 1.

It is worth noting that there are no data to support the use of aromatase inhibitors in women who are still premenopausal. It is believed that aromatase inhibitors are unable to overcome the action of ovarian aromatase and may therefore result in a surge of ovarian aromatase activity, producing even higher estrogen levels in the premenopausal woman. Thus, clinicians are strongly advised against the use of an aromatase inhibitor unless they are certain that the patient is postmenopausal. It is also interesting to note that some 20% of premenopausal women may still have premenopausal hormone levels even though they become amenorrheic after chemotherapy (16). Thus, it may be worthwhile to test the hormonal milieu in perimenopausal patients and in women who have seemingly become postmenopausal after chemotherapy before considering the use of aromatase inhibitors.

In postmenopausal women, estrogen levels are already low, but they remain sufficient to stimulate breast cancer cell growth. Androgens produced in the female body by the adrenal glands are transformed by the enzyme aromatase, which exists in fat, muscle, brain, and many other tissues, into estrogens. Aromatase inhibitors can reduce the levels of estrogen thus produced by blocking the aromatase enzyme. The prototype of the aromatase inhibitors, aminoglutethimide, is actually a poor aromatase inhibitor in that it is not particularly selective. It also blocks 11-, 17-, and 21-hydroxylation of the steroidal pathways in the adrenal gland, thus reducing levels of glucocorticoids and mineralocorticoids. As a result, aminoglutethimide has many side effects. It requires the coadministration of hydrocortisone and/or fludrocortisone, which in turn cause side effects of their own. The newer, so-called third-generation aromatase inhibitors are extremely specific and hence have comparatively few side effects. The aromatase inhibitors are generally divided into nonsteroidal and steroidal types. The major nonsteroidal types available are anastrozole (Arimidex) and letrozole (Femara). The steroidal aromatase inhibitor on the market today is exemestane (Aromasin). The nonsteroidal aromatase inhibitors anastrozole and letrozole bind to the heme part of the enzyme, and this binding is reversible. The steroidal aromatase inhibitors, on the other hand, bind to the substrate binding site of the enzyme. This binding is irreversible; thus, these drugs are often referred to as aromatase inactivators. This irreversible binding could in theory provide some additional benefit, but that benefit remains to be proven in clinical trials.

The aromatase inhibitors were initially studied in comparison with megestrol acetate and aminoglutethimide in women whose breast cancer had progressed after therapy with tamoxifen. Three large trials of aromatase inhibitors compared with megestrol acetate, (17, 18, 19, 20) and two large trials of aromatase inhibitors compared with aminoglutethimide (500 mg/daily without added cortisone; Refs. 21, 22, 23, 24, 25) took place. As shown in Table 1, the aromatase inhibitors were superior to megestrol acetate for several or all end points in most of these trials and to aminoglutethimide for most end points in both trials.

After this, the aromatase inhibitors were compared with tamoxifen as first-line therapy in two large trials in women with metastatic disease, who either had never received tamoxifen for metastatic or adjuvant disease or had completed adjuvant tamoxifen more than 6 or 12 months previously. In the large trial of anastrozole versus tamoxifen as first-line hormonal therapy, the trial was conducted in two components—a United States/Canadian study that randomized 353 women (23), and a European study that randomized 668 women (24). The planned end points were TTP, RR, and tolerability. The trial was designed for a pooled analysis of all 1021 patients.

Table 2 shows the results of the two trials presented separately. In the United States/Canadian trial, there was a statistically significant difference in favor of anastrozole for the end point of a combination of PR and/or CR, plus stable disease for prolonged time periods (NC; P = 0.009). For the anastrozole arm compared with the tamoxifen arm, TTP was also significantly better at (11 versus 5.6 months; P = 0.05). When the results were pooled as planned, there was still a small but significant difference in TTP between the anastrozole and tamoxifen arms (8.5 versus 7 months; P = 0.02), but no significant difference in CR, PR, and NC. The incidence of venous thromboembolic disease was somewhat lower in both subtrials and in the pooled data in favor of anastrozole, whereas hot flushes were similar in one substudy and greater in the anastrozole arm in the other. Gastrointestinal disturbances were slightly more frequent in the tamoxifen arm, as was lethargy. No survival difference has been shown in this study as yet, although many deaths have likely occurred. Thus, it is probable that there will be no survival benefit for anastrozole in comparison with tamoxifen in the metastatic setting.

The second large trial was a double-blind, double-dummy, randomized study with a crossover comparing letrozole (2.5 mg/day) with tamoxifen (20 mg/day) in women whose disease had progressed. Those women could not have had any prior hormonal therapy for metastatic disease. Prior adjuvant hormonal therapy with tamoxifen was permitted, but it must have been completed at least 6 months before study entry. Here the results are clearer (Table 3): TTP, TTF, objective RR, and CB were all significantly better in favor of letrozole (25).

In addition (Fig. 2), letrozole was superior in TTP in a variety of subgroups including the intent-to-treat group, the groups previously treated and not previously treated with adjuvant tamoxifen, receptor-positive women, receptor-unknown women, women with soft tissue disease, and women with bone metastases, as well as those with visceral metastases. Hot flushes, nausea, and alopecia/hair thinning were relatively similar between the two treatment arms, as were thromboembolic events. It is of interest that the rate of thromboembolic events for tamoxifen reported in this trial is less than half of that collected in a similar group of patients in the anastrozole versus tamoxifen trials. The rate of thromboembolic events in the letrozole versus tamoxifen trial may have been underestimated in general.

Most recently, OS rates comparing letrozole with tamoxifen in first-line hormonal therapy have become available (Fig. 3). Whereas women on letrozole had a trend toward improved survival during the first 2 years, now 4–5 years of follow-up of the entire study population shows a median OS for letrozole of 34 months versus 30 months for tamoxifen [P = 0.530 (log-rank test); P = 0.0790 (Wilcoxon test)]. The Wilcoxon test puts more emphasis on the early part of the curve and thus is closer to reaching conventional significance; still, neither result is significant (25).

Table 4 shows pooled results in all three randomized Phase III trials of tamoxifen versus an aromatase inhibitor. The boxing around the outcome measures indicates a significantly better response for the aromatase inhibitor than for tamoxifen in the trial under which it is listed. The exemestane versus tamoxifen data are from a randomized Phase II trial of only 120 women (26, 27). This trial has enrolled some 370 patients and has closed to accrual, with its projected sample size completed. Full results are not as yet available.

Currently, two Phase III trials comparing third-generation aromatase inhibitors head-to-head in postmenopausal women who have failed tamoxifen are ongoing or have been completed. A study comparing anastrozole with letrozole was presented by Dr. Carston Rose at the 2002 annual meeting of the ASCO (28). In this open-label study, 650 women with measurable or assessable lesions were randomized to receive anastrozole or letrozole. End points included TTP, RR, TTF, and OS. Table 5 shows the results, with overall RR being slightly better for letrozole than anastrozole (hazard ratio = 1.7; P = 0.014), but CB being similar (hazard ratio = 1.24; P = 0.218). In this study 52% of the patients were estrogen unknown. In known receptor positive patients, the response rate was similar for the two drugs. There was no difference in TTP, TTF, duration of response, or duration of CB. Although a full presentation of this study was not made, there was some skepticism that the study clearly showed any difference between letrozole and anastrozole in the metastatic setting.

There are now data showing that exemestane can produce responses in women with progressive disease after treatment with other aromatase inhibitors. Lonning et al. (29) showed an objective RR of about 12% and a CB rate of about 28% in such women treated with exemestane.

Fulvestrant (Faslodex, ICI 182,780), first believed to be a pure antiestrogen, but now better known as an ER down-regulator, is now available for therapy at least in the United States, if not in Canada. Fulvestrant has been in development for a number of years but was held back because of lack of availability of an oral formulation. Fulvestrant has been most commonly studied as an i.m. injection of 250 mg once monthly, but there has been concern that this dosing level may not be quite adequate. Loading with 250 mg every 2 weeks for 3–4 injections, followed by dosing at 250 mg/monthly, results in steady-state kinetics much more quickly than does monthly dosing.⁴

Initial interest in fulvestrant was raised when a series of 19 women whose disease had relapsed after previous response to tamoxifen for advanced breast cancer or after more than 2 years of adjuvant therapy with tamoxifen were treated with fulvestrant in an early Phase II trial. Although there were no CRs, seven women had PRs, and six had stable disease for protracted time periods, for a CB rate of 69% (30). Because this study showed clear antiestrogenic effects and lack of cross-resistance to tamoxifen, with high RRs and long duration of responses in tamoxifen failures, and because there were no major local or systemic safety or toxicity issues with the use of this drug, the drug was felt to be appropriate for further investigation.

Subsequently, two large Phase III clinical trials of fulvestrant compared with anastrozole were opened in North America and in the ROW. The North American trial (0021) was a double-blinded study, whereas that for the ROW (0020) was open label. A total of 851 postmenopausal patients with advanced breast cancer who had relapsed or progressed after endocrine therapy were studied in these trials (31). Combined data from these two studies, which once again were designed to be analyzed together, showed no difference in TTP (5.5 months for fulvestrant versus 4.1 months for anastrozole) and no significant difference in TTF (4.6 months for fulvestrant versus 3.6 months for anastrozole). The trend to benefit for fulvestrant in terms of TTP and TTF, however, was greater in the North American (32) than in the ROW subtrial (33) but was not significant for the pooled data. Adverse effects, including vasodilation, injection site pain, and nausea, were virtually identical in the two treatment arms. Withdrawals due to drug-related adverse events were 0.5% in the fulvestrant arm and 1% in the anastrozole arm. The incidence of thromboembolic events, weight gain, and vaginitis were low for both treatments.

Thus, fulvestrant represents the first in a new class of drugs that bind, block, and degrade the ER. Fulvestrant offers a new treatment option for postmenopausal women with advanced breast cancer. It is the only antiestrogen to have demonstrated a CB in over 40% of patients whose disease has progressed or relapsed after previous tamoxifen therapy. It is at least as effective as the third-generation aromatase inhibitors in women who have previously received tamoxifen. Both Phase III trials of fulvestrant versus anastrozole demonstrate at least equivalence and perhaps benefit over anastrozole for all major efficacy end points. Fulvestrant, despite its i.m. route of administration, is well tolerated, with few adverse events.

Currently, we are awaiting publication of the large randomized trial (0025) of fulvestrant versus tamoxifen in first-line therapy. The primary end point for this study will be TTP with secondary end points including RR, duration of response, TTF, time to death, tolerability, and quality of life.

Thus, today we may well begin the treatment of metastatic breast cancer in postmenopausal women with a different endocrine cascade. Whereas before we would have started with tamoxifen followed by a third-generation aromatase inhibitor and then by megestrol acetate, today we would likely begin with an aromatase inhibitor, followed by a series of other hormonal therapies. We do not currently know whether tamoxifen or fulvestrant is better after disease progresses following treatment with an aromatase inhibitor. What is the role of megestrol acetate in this sequence? In addition, we now have considerable data showing that exemestane can produce responses in women who have had progressive disease after treatment with other aromatase inhibitors. Should an aromatase inactivator be used after an aromatase inhibitor instead of moving to a drug of another class?

In a recent study, the efficacy of tamoxifen as second-line therapy following anastrozole was assessed in postmenopausal women with advanced breast cancer: 57% of patients who received tamoxifen subsequent to anastrozole had a CB of more than 24 weeks; of these, 50% or more maintained this benefit for a duration of 12 months or longer. Of the patients who received anastrozole as second-line therapy following tamoxifen, 43.5% showed a CB for a similar duration of time (34). A retrospective analysis has also shown that women with advanced breast cancer who progress on fulvestrant remain sensitive to subsequent treatment with anastrozole and letrozole (35). Thus, whereas there are good data to suggest that for women who have progressed on tamoxifen, fulvestrant is an adequate treatment, as the aromatase inhibitors move to first-line therapy, we will require data on the efficacy of fulvestrant after aromatase inhibitors to determine the most appropriate sequence of hormonal therapy. Currently, agents such as megestrol acetate have been relegated to fourth- and fifth-line therapies, and agents such as estrogen have been nearly forgotten. Considering that aromatase inhibitors produce a very low estrogen environment, it will be important to explore the role of fulvestrant, an estrogen down-regulator, and determine whether this may be more or less useful in this setting than in other approaches. Perhaps drugs such as estrogen should now be reexamined in this environment.

In summary, the endocrine cascade in postmenopausal women remains somewhat undetermined. Probably it is fair today to start with tamoxifen, followed by an aromatase inhibitor, or with an aromatase inhibitor followed by tamoxifen—in either case followed by fulvestrant as a third-line therapy. Megestrol acetate and high-dose estrogen may remain fourth- and fifth-line options. Hopefully, trials will be carried out to better clarify the optimal sequences that should be used.

Over the last few years, there have been considerable data suggesting that women whose breast cancers are ER/PgR positive but overexpress the HER2/neu oncogene are less responsive to endocrine therapy, particularly tamoxifen (36, 37, 38, 39, 40, 41). Data on this matter are contradictory, however (42, 43).⁵ In addition, recent data obtained in the neoadjuvant setting have suggested that women whose tumors are ER/PgR positive but also overexpress neu/erbB1 are much more likely to respond to the aromatase inhibitor letrozole than to tamoxifen (44). Once again, there have been conflicting reports (45). Now data are about to be published from Love and his colleagues regarding premenopausal women in Vietnam randomized to receive ovarian ablation and tamoxifen in the adjuvant setting versus the same endocrine therapy delayed to the time of recurrence. The women whose tumors were ER/PgR positive benefited highly from adjuvant as opposed to delayed ovarian ablation and tamoxifen; those whose tumors also overexpressed HER2/neu showed even more benefit in comparison with those who received delayed therapy (46). Thus, the response in relation to HER2/neu in premenopausal women to ovarian ablation and tamoxifen is the opposite of what has been reported with tamoxifen alone in many other settings. That is, women who overexpressed HER2/neu are more likely to respond to endocrine therapy in this study, whereas other studies have suggested that they are less likely to respond (47).

It seems that this matter requires further clarification (48). First, the inconsistent findings may relate to incorrect measurements of ER, PgR, and HER2/neu. When careful measurements of ER/PgR are done, there is actually quite a small proportion of ER/PgR-positive women who also overexpress HER2/neu. More careful work in a variety of clinical trials settings will hopefully clarify this issue.

Endocrine therapy for women with metastatic breast cancer has become more complex but even more rewarding than previously. In the last 30 years, we have developed a large number of agents that were not available before, including the antiestrogen tamoxifen, the LHRH agonists, the aromatase inhibitors, and now the new ER down-regulator fulvestrant (Faslodex). We must continue to explore the optimal usage of these compounds over future months and years to provide the best benefit for patients in this setting. Further exploration of the role of other predictive measures, such as HER2/neu overexpression, in the selection of optimal endocrine will constitute an important part of these future studies.

Dr. Aman Buzdar: Ideally, we would like to see head-on comparative trials to ask the question whether one aromatase inhibitor is better than the other. If you look at objective RRs, it seems the difference is in how the control arm behaved in these studies.

Dr. Kathleen I. Pritchard: I’m not convinced there are clear data to delineate between them at the present time. The drug we know least about is exemestane, but there are patients who have failed other aromatase inhibitors who will have some clinical response to exemestane. It’s unclear whether the data on the other aromatase inhibitors simply haven’t been collected in a systematic way, or whether exemestane is different when used after the other aromatase inhibitors.

Dr. James Ingle: There is no prospective study, but there are the data presented at ASCO going from steroidal to the nonsteroidal, or the other way around, and seeing some responses.

Dr. Pritchard: In small numbers of patients, there are data to suggest that a response can be seen in almost all of these settings with changing to a hormone of another type or to the other of these two subclasses of aromatase inhibitors. How selective those data are and what they really mean are unclear right now.

Dr. Kent Osborne: If you have a tamoxifen-treated patient with HER2 overexpression, in whom tamoxifen is having some agonist activity, what happens when you switch her to an aromatase inhibitor? The tumor is going to continue to grow for a while. It takes 2 months to eliminate the tamoxifen from the tumor because of its half-life. All of the receptors in the tumor are saturated by tamoxifen for a minimum of 2 months, and you cannot even detect ER by a ligand binding assay. During that time, an aromatase inhibitor is not working at all; you are simply seeing a tamoxifen withdrawal response when you treat the patient with an aromatase inhibitor. I think the second-line aromatase inhibitor studies that looked at HER2 as a predictive marker were confounded by this phenomenon.