There has been one multinational double-blind randomized trial (PO24) comparing preoperative letrozole 2.5 mg with tamoxifen for 4 months prior to surgery (9). This involved 337 postmenopausal women with estrogen receptor (ER) or progesterone receptor (PgR)-positive cancers, defined by at least 10% nuclear staining. All patients would have otherwise required mastectomy at entry to the trial or were considered inoperable (14%). The overall clinical response rate, the primary endpoint, was significantly higher for letrozole than for tamoxifen (55% vs. 36%; p = .001). The median time to response was 66 days for letrozole and 70 days for tamoxifen. Progressive disease during treatment was seen with 12% of patients treated with letrozole and 17% with tamoxifen. Letrozole was also more effective than tamoxifen when the response rate was determined by ultrasound (35% vs. 25%; p = .0042) and by mammography (34% vs. 16%; p = .001).

The main secondary endpoint of the trial was breast conservation, and significantly more breast-conserving surgery was achieved with letrozole than with tamoxifen (45% vs. 35%; p = .022). Pathological complete remission (pCR) in the primary breast lesion was seen in only 2 patients treated with letrozole and 3 with tamoxifen. Only 2 of these 5 patients with pCRs had no involved nodes at surgery.

In a further analysis of the same study, ER and PgR expression were re-assessed in a central laboratory, and 12% of patients were found to have tumors that were both ER and PgR-negative (10). In patients whose tumors were confirmed ER or PgR-positive, the response rate to letrozole was 60% compared with 41% for tamoxifen (p = .004), and 48% versus 36%, respectively, underwent successful breastconserving surgery (p = .036).

In this analysis ER and PgR were quantified using the Allred scoring system in which an intensity score (range, 1-3) is added to a frequency score (range, 1-5) (11). Letrozole response rates were numerically superior to tamoxifen for all ER Allred scores from 3 to 8; furthermore, responses to letrozole were seen in all Allred scores between 3 and 8, whereas responses were only seen in tamoxifen for scores between 6 and 8. Based on this, the authors suggested that letrozole might be more effective than tamoxifen in patients whose tumors show relatively low ER expression, but it is important to note that the numbers were small in each of these Allred groupings and no definite conclusions should be drawn.

There have been 2 multinational double-blind trials comparing preoperative anastrozole 1 mg daily with tamoxifen 20 mg daily for 12 weeks prior to surgery in postmenopausal women with ER-positive breast cancer.

The IMPACT (IMmediate Preoperative Anastrozole, tamoxifen or Combined with Tamoxifen) trial compared anastrozole with tamoxifen or with both in combination given for only 12 weeks (in contrast to 16 weeks for PO24 above) (12); the trial was designed to be the preoperative equivalent of the adjuvant ATAC trial described elsewhere. The main clinical aim was to compare the efficacies of these treatments in terms of response and in downstaging to avoid mastectomy. An important further aim, however, was to determine whether short-term surrogate endpoints of response could be identified to predict for long-term outcome in the adjuvant ATAC trial; these included clinical changes after 12 weeks or biological changes in proliferation assessed by the proliferation marker Ki-67 after 2 and 12 weeks. For this reason, postmenopausal patients with smaller breast cancers not necessarily requiring mastectomy were also included, in contrast to the preoperative letrozole trial described above. This trial involved 330 patients with confirmed invasive histology and ER positivity on core needle biopsy. Median age was 73 years, median tumor size was 4 cm for each of the three groups, and tumors were confirmed in a central reference laboratory as ER-positive in 98% of cases. Objective clinical response rates by caliper measurement for anastrozole, tamoxifen, and the combination were 37%, 36%, and 39%, respectively, on an intent-to-treat basis, and none of these differences was significant. Ultrasound response rates were 24%, 20%, and 28%, respectively; again, none of these differences was significant. Progressive disease during treatment occurred in only 9%, 5%, and 5% of patients respectively.

A subgroup of 124 patients was assessed by the surgeon as requiring mastectomy at baseline. In these, 46%, 22%, and 26% were deemed to have achieved tumor regression sufficient to allow breast-conserving surgery after treatment with anastrozole, tamoxifen, and combination therapy, respectively. The improvement with anastrozole compared with tamoxifen was statistically significant with an odds ratio (OR) of 2.94 (p = .03).

There was no significant difference between the tamoxifen and combination groups.

An important secondary endpoint in the IMPACT trial was the reduction in proliferation as measured by Ki-67 staining after preoperative treatment. This was significantly reduced by all three treatments after 2 and 12 weeks, anastrozole by 76% and 82%, tamoxifen by 60% and 64%, and the combination by 65% and 64% (13). Change after 2 weeks correlated with the change after 12 weeks. The decrease with anastrozole was significantly greater than that with tamoxifen, as assessed by geometric mean ratios of the changes in Ki-67 after 2 weeks’ treatment (p = .04) and again after 12 weeks’ (p = .001), but there were no significant differences between tamoxifen and the combination. The changes in Ki-67 after only 2 weeks of treatment therefore predicted for long-term differences in relapse-free survival in the adjuvant ATAC trial, suggesting that change in proliferation as measured by Ki-67 might be a short-term surrogate for predicting differences in long-term outcome between different endocrine therapies in adjuvant trials.

This possibility is reinforced in an earlier and much smaller preoperative trial comparing vorozole with tamoxifen, which found a similar but nonsignificant trend in favor of an aromatase inhibitor, with mean drops in Ki-67 of 58% and 43% for vorozole and tamoxifen, respectively, after 2 weeks’ treatment (20).

A statistically significant correlation was found between Ki-67 reduction after 2 weeks and response in the IMPACT trial, but this was not seen with Ki-67 reduction at 12 weeks: A weakly significant relationship was seen between the percentage of tumor shrinkage and change in Ki-67 (13).

In the second preoperative anastrozole trial, PROACT (PReOperative Anastrozole Compared with Tamoxifen), also multicenter and double-blind, 451 postmenopausal women with operable or locally advanced but potentially operable
(T2-4b) hormone receptor-positive breast cancers were randomized to anastrozole 1 mg or tamoxifen 20 mg for 12 weeks prior to surgery (14). As in the IMPACT trial, patients with small breast cancers appropriate for breast-conserving surgery were eligible for entry. In contrast to other trials, concomitant chemotherapy was also allowed and was given to 29% of patients on anastrozole and 32% on tamoxifen. Mean age was 67 in both groups, and mean ultrasound tumor diameter was 3.6 cm. Overall ultrasound response, the primary endpoint, was 40% for anastrozole and 35% for tamoxifen (p = .29). Clinical response by caliper measurement was 50% and 46%, respectively (p = .37). Chemotherapy was clearly confounding here, and in the 314 patients treated with endocrine therapy alone without chemotherapy, ultrasound and clinical response rates for anastrozole and tamoxifen, respectively, were 36% versus 27% (p = .07) and 50% versus 40% (p = .08).

In the 262 patients treated with endocrine therapy alone without chemotherapy who would have required mastectomy or had locally advanced disease at baseline, surgical improvement (inoperable to mastectomy or mastectomy to breast-conserving surgery) was deemed feasible in 47% after anastrozole compared with 38% after tamoxifen (p = .15) and actually occurred in 43% versus 31% (p = .04).

Preoperative exemestane has also been shown to be active in achieving clinical responses and downstaging to avoid mastectomy (16), but so far only one small randomized trial, comparing preoperative exemestane with tamoxifen, has been reported. Seventy-three postmenopausal women with hormone receptor-positive status were randomized to receive exemestane 25 mg or tamoxifen 20 mg daily for 3 months before surgery (17). Clinical objective response rates were reported as 89% for exemestane compared with 57% for tamoxifen (p = .05), including complete clinical remission rates of 14% and 11%, respectively (ns). Ultrasound response rates were 70% and 41% (ns), and breast conservation rates 39% versus 11%, respectively (p = .05). Two pathological complete remissions were found with exemestane and one with tamoxifen.

The authors reported without details that responses were more likely with higher levels of estrogen-receptor expression.

In the PO24 trial, preoperative letrozole was markedly superior to tamoxifen in the small subgroup of 36 patients with HER2-positive cancers (88% vs. 21%; p = .004) (10), and a similar marked numerical difference was seen for anastrozole over tamoxifen for the 34 patients with centrally confirmed HER2-positive cancer in the IMPACT trial (58% vs. 22%; p = .09) (10, 12). These differences were not, however, supported by subsequent results from the equivalent large adjuvant trials BIG 1-98 and ATAC (see Chapter 43), neither of which showed an increased benefit for AIs over tamoxifen compared with cancers not overexpressing HER2. These results therefore emphasize the potential pitfalls of extrapolating clinical preoperative data as a surrogate marker for long-term outcomes. A possible explanation for this discrepancy emerged with molecular marker studies from the IMPACT trial, described later in this chapter.

There are some data suggesting that a high body mass index (BMI) may be associated with inferior outcome after adjuvant anastozole. In a recent Japanese study involving 109 patients treated with preoperative exemestane, a low BMI was paradoxically associated with a lower objective response rate (ORRs were 21.7% in low BMI, 56.0% in intermediate BMI, and 60.6% in high BMI, respectively; p = .01) (18). In a multivariate analysis, a low BMI was an independent negative predictor of clinical response. The authors do not postulate an explanation for these paradoxical results.

Although these trials were relatively small compared with large adjuvant trials, the balance of evidence from them shows that the AIs are more effective clinically than tamoxifen in achieving objective responses and in downstaging to avoid mastectomy or to convert inoperable to operable cancers. AIs should therefore be considered the first-line preoperative endocrine therapy of choice for postmenopausal patients with ER-positive breast cancers.

A French trial, UNICANCER CARMINA 02, randomized 116 postmenopausal women with T2-T4, N0-N3, ER-positive, HER2-negative breast cancers to preoperative anastrozole 1 mg daily or fulvestrant 500 mg on Day 1, 15, and 29 and then 4 weekly, for 4 to 6 months. Clinical response rates at 4 months were 54% for anastrozole and 44% for fulvestrant and at 6 months 53% and 35% (a smaller number were treated to 6 months), respectively. These differences were not significant (19).

There are good experimental data to show that cross-talk between the ER and the phosphoinositide-3-kinase (PI3K)/Akt/mammalian target of rapamycin (mTOR) pathways is a mechanism of resistance to endocrine therapy. Preclinical models have shown that blockade of both pathways can enhance antitumor activity (21). A phase 2 randomized trial of 4 months of letrozole in combination with everolimus (RAD001), an oral mTOR inhibitor, in a dose of 10 mg/day, versus letrozole alone was carried out in 270 postmenopausal women with operable ER-positive breast cancer. Response rate by clinical palpation, the primary endpoint, was significantly higher at the preplanned, one-sided 0.1 level in the everolimus group than with letrozole alone (68.1% vs. 59.1%; p = .062). An antiproliferative response, as defined by a reduction in Ki-67 expression to natural logarithm of percentage positive Ki-67 of less than 1 at day 15, occurred in 52 (57%) of 91 patients in the everolimus group compared with 25 (30%) of 82 patients in the placebo group (p = .01) (22). Marked downregulation of phospho-S6, a downstream intermediary of mTOR, occurred only in the everolimus group.

These positive results, particularly with respect to differences in Ki-67 suppression, prompted a subsequent phase 3 trial (BOLERO 1), which confirmed the clinical superiority of the everolimus combination; this trial is described in detail elsewhere in this book.