Earlier meta-analyses suggested that tamoxifen had no benefit in women younger than age 50 (16, 17). Because of the inclusion of women with ER-negative tumors, and because the duration of tamoxifen treatment was usually only 1 or 2 years in these early trials, definitive conclusions could not be drawn. Later meta-analyses indicated that more prolonged treatment (approximately 5 years) results in a significant benefit in women younger than 50 years, as well as in older women, so long as their tumors are ER positive (18). The recently updated meta-analyses confirm these results (Table 43-2) (12). Patients younger than 45 years of age, most of whom are premenopausal, benefit in terms of recurrence and mortality from 5 years of tamoxifen nearly as well as those 55 years of age and older. The benefits found with 5 years of tamoxifen in younger women, along with the lack of benefit with the shorter durations used in earlier studies, strongly suggest that longer treatment is very important in this age group. These data also indicate that the tamoxifen dose used in these studies can inhibit breast cancer cells even in the presence of the much higher than normal serum levels of estrogen typically found in premenopausal patients taking the drug.

Thus, with more than 30 years of follow-up from many studies involving thousands of patients, it is certain that if tamoxifen is given for 5 years to patients selected on the basis of ER status, it is effective in both younger and older women. It is also important to emphasize that the differences in outcome between tamoxifen and no tamoxifen observed after 5 years of follow-up grow even larger during the next 5 years, indicating that the benefits of tamoxifen are very durable over time (12).

No biological reason exists for women with axillary nodenegative ER-positive breast cancer to benefit differently from adjuvant tamoxifen therapy than those with positive nodes, and many trials of adjuvant tamoxifen included both nodenegative and node-positive patients. Fewer recurrences and deaths overall in the node-negative subset make it more difficult to show significant reductions with tamoxifen, but strong trends were evident early on in the larger studies (14, 19). NSABP trial B-14 is by far the largest of the initial
trials of adjuvant tamoxifen (2,644 patients), and it focused on patients with histologically negative axillary nodes (19). Both patients younger than 50 years of age (820 patients) and older patients (1,824 patients) were eligible, and all patients had ER-positive disease. Patients were randomly assigned to receive placebo or tamoxifen for 5 years, and those who received tamoxifen were reassigned at 5 years to stop therapy or to continue for 5 additional years.

Women who received tamoxifen had a significantly higher probability of being free of local and distant recurrences (78% vs. 65%, p < .0001) and survival (71% vs. 65%, p = .0008) than with placebo. Tamoxifen-treated patients also had fewer ipsilateral breast, local-regional, and distant recurrences than placebo-treated patients, and they had a substantial reduction (approximately 50%) in contralateral breast cancer. These benefits persist beyond 15 years of follow-up (19).

The most recent meta-analysis also suggests that the benefit of adjuvant tamoxifen is similar for node-negative and node-positive patients (12). The reductions in the odds of recurrence were 40%, 36%, and 44% for patients with negative nodes, one to three positive nodes, or four or more positive nodes, respectively. The estimated 10-year recurrence rate for node-negative patients was 19% with tamoxifen in the absence of chemotherapy and 35% without tamoxifen, while the recurrence rate for node-positive patients was 42% with and 57% without tamoxifen. Thus, the cumulative data confirm that tamoxifen improves recurrence and survival in both node-negative patients, who have a substantially lower baseline risk of recurrence and death, and in node-positive patients.

Many of the early tamoxifen adjuvant trials included patients with ER-negative or ER-poor as well as ER-positive and ER-unknown tumors. These studies helped assess the potential benefits of tamoxifen in both subsets. The results are difficult to interpret, however, because of varying definitions of ER-positive and ER-negative, concerns about assay quality, and because only a fraction of the patients in some of these early studies had ER assays performed.

Earlier meta-analyses found a small but statistically significant survival benefit in women with ER-poor tumors treated with adjuvant tamoxifen, but the most recent meta-analyses with longer follow-up and a larger sample size do not (12). Patients with ER-poor tumors, a subset that includes tumors with undetectable or borderline-positive ER (4 to 9 fmol/mg protein by ligand-binding assay), showed no reduction in the annual odds of recurrence or death. For unclear reasons patients with ER-negative first tumors who are treated with adjuvant tamoxifen have no reduction in contralateral breast cancer either. Women with tumors known to be definitely positive for ER had a 39% reduction in the annual odds of recurrence and a 38% reduction in contralateral breast cancer when treated with tamoxifen for 5 years. Interestingly even patients with tumors expressing only low levels of ER (10 to 19 fmol/mg protein) had a significant reduction in recurrence from tamoxifen (Table 43-3). Why tumors with only a small fraction of cells expressing ER by either ligand binding or IHC benefit from adjuvant endocrine therapy remains a mystery. It is possible that other cells in the tumor have low ER levels undetectable by standard assay but still sufficient to affect gene transcription. Alternatively, cells not expressing ER at the time of the biopsy might express it later due to cell cycle variation in expression or other factors. Higher ER levels were associated with a greater risk reduction. The ligand binding assay is no longer used for clinical ER measurements but very low ER levels detected by IHC have also been associated with tamoxifen benefit (20).

The College of American Pathologists suggests that a cutoff of greater than 1% positively staining cells be used to define ER-positivity by IHC (21). This committee also provided technical guidelines in an attempt to standardize the entire procedure to reduce the high error rate observed in many studies. This procedure should be followed closely to obtain quality results for patient care. The role of PR in predicting benefit from tamoxifen adjuvant therapy or in predicting relative benefit of tamoxifen compared to aromatase inhibitors is more problematic. The assay itself has changed over time from ligand binding to IHC and has not been as well standardized as that for ER. Also PR, as an estrogen-regulated gene product, is affected by the level of estrogen circulating in blood and therefore would be expected to be higher in premenopausal women and postmenopausal women on hormone replacement therapy at the time of breast cancer diagnosis. PR might be negative in women who have been postmenopausal for some time and who are not taking estrogen replacement therapy even if the ER signaling pathway is intact. In the meta-analysis, among women with ER-positive tumors, the efficacy of tamoxifen was independent of the concentration of PR (18). Studies of patients with metastatic disease have shown consistently that patients with ER-negative, PR-positive tumors, although uncommon, benefit from tamoxifen and other endocrine therapies and that those with ER-positive, PR-negative tumors respond less well than those positive for both receptors. There is level-one evidence from a prospective clinical trial in patients with metastatic breast cancer treated with tamoxifen demonstrating the independent predictive role of PR expression (22). And, there are retrospective data from a very large patient cohort with receptor assays done in a single reference laboratory suggesting that PR does predict benefit from tamoxifen adjuvant therapy (23). But this has not been confirmed in the meta-analysis or in the ATAC or BIG 1-98 adjuvant trials (18, 24, 25). Nor has PR level correlated with selective benefit of either tamoxifen or an aromatase inhibitor in these comparative trials.

Therefore at the present time PR should not be used as a predictive marker for adjuvant endocrine therapy response, although its loss does correlate with a more aggressive luminal B subtype of breast cancer. This issue is discussed in detail in Chapter 26.

It is clear, however, that when the assays are done properly, patients with ER- and PR-negative tumors do not benefit from tamoxifen adjuvant therapy or any endocrine therapy that blocks the ER pathway. Nor do they benefit with a reduction in contralateral breast cancer (see above). On the other hand, patients even with a low but detectable level of ER in their tumors do benefit from tamoxifen and other forms of endocrine therapy, which should be an important component to their treatment.

Because tamoxifen has predominantly antiproliferative effects it was hypothesized early in its use that it might need to be given indefinitely in the adjuvant setting. However, early trials did not show an advantage for continuing tamoxifen beyond 5 years and in fact, two trials showed slightly worse outcome (19, 26). This lack of benefit for more prolonged tamoxifen in these two trials persisted at 14 and 15 years of follow-up, respectively. These trials were small and not definitive leading to more recent trials with larger patient accrual. The aTTom trial in the United Kingdom randomized 7,000 patients, many of unknown ER status, to stop or continue tamoxifen to 10 years and recently reported long-term results. Its results a significant benefit for extending tamoxifen beyond 5 years (27). In late 2012 the ATLAS Trial (Adjuvant Tamoxifen: Longer Against Shorter) reported 15-year data (28). This trial randomized 12,894 women between 1996 and 2005 to stop tamoxifen at 5 years or to continue to 10 years; 6,846 of these patients had known ER-positive tumors. Among these patients, continuing tamoxifen to 10 years significantly reduced recurrence (p = .002), breast cancer deaths (p = .01), and overall mortality (p = .01). The relative risk (RR) for continuing tamoxifen was less impressive during years 5 to 9 than after 10 years (0.9 vs. 0.75, respectively) perhaps due to the known carryover effect of just 5 years of tamoxifen during years 5 to 9 (Table 43-4). A reduction in breast cancer mortality for extended tamoxifen was only seen after 10 years (RR = 0.97 vs. 0.71). The cumulative risk for recurrence during years 5 to 14 for extended tamoxifen was 21% compared to 25% for patients stopping at 5 years. The risk of breast cancer death during years 5 to 14 was 12% for continued tamoxifen compared to 15% for the control group.

Mortality without breast cancer recurrence overall was not affected by continuing tamoxifen for 10 years. Known tamoxifen side effects were higher in women assigned to 10 years of tamoxifen but these were counterbalanced by favorable effects. Relative risks for pulmonary embolus (1.87) and endometrial cancer (1.74) were higher but mortality was minimally affected. There was no increase in stroke and there was a significant reduction in ischemic heart disease (0.76, p = .02) for continuing tamoxifen. These data offer a new option for ER-positive patients being treated with adjuvant tamoxifen, and appropriately selected ER-positive patients should be considered for 10 rather than 5 years of treatment. Low-risk patients for whom the extra benefit of continuing tamoxifen beyond 5 years would be minimal might stop at 5 years while higher-risk patients should be considered for extended treatment. This strategy would have the greatest appeal for premenopausal patients who are still treated predominantly with tamoxifen. However, given that fewer than 10% of the ATLAS study population was premenopausal raises a small concern about young patients, although there is no legitimate reason why they would not also benefit from more prolonged therapy.

Adjuvant therapy is more problematic in elderly patients because of comorbidities more common in these patients that can cause death before breast cancer recurrence (29). The meta-analysis demonstrates a significant mortality reduction in patients older than 70 years treated with adjuvant tamoxifen (Table 43-2) (12, 18). Furthermore, some individual trials have specifically targeted this population. The Eastern Cooperative Oncology Group (ECOG) study randomized 181 patients 65 years of age or older to tamoxifen or placebo for 2 years (30). The drug was well tolerated, and significant reductions in recurrence and borderline significant reductions in mortality were observed. Tamoxifen also reduced the incidence of contralateral breast cancers. Surprisingly, most of the older patients who died in this study (61%) succumbed to breast cancer, although, as anticipated, a significant number of them (22%) died of competing illnesses not related to cancer. Nonadherence to the prescribed dose and schedule of tamoxifen, which is relatively high overall, is even higher in elderly patients (31).

The ATAC and TEAM trials comparing tamoxifen with an AI evaluated comorbidities and age on death without recurrence (32, 33). In the TEAM trial evaluation of disease-specific mortality, as a proportion of all-cause mortality, showed that 78% of deaths in patients less than 65 years of age were due to breast cancer, while in patients between 65 and 74 years or those 75 or greater, 56% and 36% of deaths, respectively, were due to breast cancer (32). Interestingly breast cancer recurrence and disease-specific mortality increased with age but other-cause mortality increased even more dramatically with age with a sevenfold increase in patients 75 years or greater. The increase in disease-specific mortality with age is not explained but could be due to inclusion in the trial of elderly patients with more aggressive disease. Similar findings were observed with longer follow-up in the ATAC trial (33).

Several small, randomized trials have also evaluated the use of tamoxifen as sole treatment without surgery for operable primary breast cancer in elderly patients (34, 35). In one study with 20-year follow-up there was no difference in time to distant metastases or overall survival between tamoxifen alone and mastectomy alone as initial treatment despite the fact that the patients were not selected by ER (35). These findings were confirmed in a large Cochrane review of 7 trials
of 1,446 elderly women unselected by ER (36). This analysis found no difference in overall survival between surgery alone without or with tamoxifen, or tamoxifen alone as initial therapy. Hormonal therapy as a single modality with tamoxifen or an aromatase inhibitor might be reserved for elderly patients not suitable for surgery for temporary disease control. Because these patients have a relatively high risk of thromboembolic complications, an AI may be preferable in debilitated elderly patients in need of treatment, although AIs have not been tested in this setting.

Although a rare situation today, the question whether patients could still benefit from adjuvant hormonal therapy, even if initiated years after their primary treatment, remains important for a small subset of patients. One study addressed this question and found that patients who had ER- or PR-positive tumors, and whose adjuvant therapy with tamoxifen was started 2 years or more after initial diagnosis, had improved DFS and OS, even when the delay in starting adjuvant tamoxifen was more than 5 years (37). Thus, those patients whose tumors are receptor positive and who, for whatever reason, were not started on adjuvant hormonal therapy at the time of diagnosis may still benefit from delayed treatment. It is likely that a similar benefit would be also seen with AIs, although no studies have addressed this topic. However, the remaining risk of recurrence in such patients should be considered. Patients with a low risk at the time of diagnosis may have an extremely low risk of recurrence at the time of initiating endocrine therapy late, not justifying the side effects of therapy if several years have elapsed since diagnosis.

Side effects and deaths from other causes from tamoxifen have been addressed in meta-analyses of 5 years of tamoxifen treatment and in studies comparing tamoxifen and an aromatase inhibitor. The 2005 meta-analysis showed a reduction in all-cause and in breast cancer mortality with tamoxifen (18). Non-breast cancer mortality was similar for tamoxifen and no tamoxifen. There was a borderline statistically significant increased risk of stroke and a similar reduction in deaths from heart disease, primarily myocardial infarction, with tamoxifen. Long-term follow-up of the Cancer Research UK “Over 50s” trial comparing five with 2 years of tamoxifen showed a striking reduction in cardiovascular disease in women aged 50 to 59 that lessened as patients became older (38). Similar data were observed in a Swedish trial (39). Data from the placebo-controlled NSABP tamoxifen prevention trial (Breast Cancer Prevention Trial [BCPT]) provides additional information regarding cardiovascular mortality (40). At 4 years of follow-up, a shorter time frame compared to the other cancer trials, this study found no differences between tamoxifen and placebo in fatal myocardial infarction, nonfatal myocardial infarction, unstable angina, or severe angina. One possible explanation for differences in cardiovascular mortality might be the variable use of lipid lowering agents for hypercholesterolemia, a hypothesis that can be tested with available data.

Tamoxifen has estrogen agonist properties in bone. In postmenopausal women, long-term tamoxifen treatment increases the bone density of the axial skeleton and stabilizes the bone density of the peripheral skeleton (8). In premenopausal women, however, tamoxifen may decrease bone mineral density by antagonizing the more potent activity of endogenous estrogen (41). Although evaluating osteoporotic fracture rates in patients with a diagnosis of breast cancer is problematic, prevention trials of tamoxifen do show a significant reduction in fractures with 5 years of treatment (42). There is a marked reduction in fractures in trials comparing tamoxifen with an aromatase inhibitor (43).

Individual clinical trials in patients with invasive breast cancer, as well as the updated meta-analysis, indicate a nearly 50% reduction in the risk of contralateral breast cancer after approximately 5 years of tamoxifen treatment (18). As described above, there is no reduction in contralateral breast cancer with tamoxifen if the original tumor is ER-negative. Continuing tamoxifen to 10 years compared to 5 years further reduces the risk of contralateral breast cancer, RR 0.88 (0.77-1.00, p = .05).

The most frequently reported side effects in patients taking tamoxifen are menopausal symptoms (44). At least 50% to 60% of these women report some hot flashes, but 40% to 50% of placebo-treated patients report similar episodes. Tamoxifen may cause hot flashes more commonly in premenopausal women than in older women. Approximately 20% of patients report severe hot flashes while taking tamoxifen compared with 3% of patients on placebo. Vaginal discharge and irregular menses are also slightly more common in patients taking tamoxifen than in those receiving placebo. In one study, however, general quality of life scores were similar for tamoxifen and placebo (44). Headaches were reported less frequently with tamoxifen. The incidence of nausea, arthralgias, insomnia, restlessness, depression, and fatigue was similar with tamoxifen and placebo in this study. In the BIG 1-98 trial hot flushes and night sweats were more common with tamoxifen than with letrozole but vaginal dryness was more common with the aromatase inhibitor (Table 43-5) (25, 45).

Depression is not increased in randomized, placebo-controlled trials, but this may reflect underreporting of symptoms that may be brought out by more careful and detailed questioning. A nonrandomized, single-institution study suggests that symptoms of depression can be identified in up to 10% of patients taking tamoxifen (46). Symptoms
are occasionally severe and may require antidepressant medication or even discontinuation of tamoxifen. Failure to identify depression as a side effect in placebo-controlled trials suggests, however, that discontinuation of estrogen replacement therapy or stress from the recent diagnosis of a potentially fatal disease may be causally more important than tamoxifen itself.

The detailed analysis of depression in women on the BCPT adds additional insight on this topic (47), although the women in a prevention trial may not accurately represent breast cancer patients who are diagnosed with a lifethreatening disease and have more reason for depression. Nevertheless, no difference was seen in depression by treatment assignment (tamoxifen vs. placebo) in this study, suggesting that tamoxifen itself does not increase the risk of, or exacerbate, existing depression in women.