J Clin Oncol 23: by American Society of Clinical Oncology INTRODUCTION

VOLUME 23 NUMBER 22 AUGUST 1 2005 JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Neoadjuvant Treatment of Postmenopausal Breast Cancer With Anastrozole, Tamoxifen, or Both in Combination: The Immediate Preoperative Anastrozole, Tamoxifen, or Combined With Tamoxifen (IMPACT) Multicenter Double-Blind Randomized Trial Ian E. Smith, Mitch Dowsett, Stephen R. Ebbs, J. Michael Dixon, Anthony Skene, J.-U. Blohmer, Susan E. Ashley, Stephen Francis, Irene Boeddinghaus, and Geraldine Walsh From the Breast Unit, Royal Marsden Hospital, London and Sutton; Mayday University Hospital, Croydon; Edinburgh Breast Unit, Edinburgh; Royal Bournemouth Hospital, Bournemouth, United Kingdom; Universitaets Klinikum Charite, Berlin, Berlin, Germany; and Macclesfield, United Kingdom. Submitted April 1, 2004; accepted December 14, 2004. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Ian E. Smith, MD, Department of Medicine, Royal Marsden Hospital, Fulham Road, London SW3 6JJ; e-mail: ian.smith@ rmh.nhs.uk. 2005 by American Society of Clinical Oncology 0732-183X/05/2322-5108/$20.00 DOI: 10.1200/JCO.2005.04.005 A B S T R A C T Purpose The Immediate Preoperative Anastrozole, Tamoxifen, or Combined With Tamoxifen (IMPACT) trial was designed to test the hypothesis that the clinical and/or biologic effects of neoadjuvant tamoxifen compared with anastrozole and with the combination of tamoxifen and anastrozole before surgery in postmenopausal women with estrogen receptor (ER) positive, invasive, nonmetastatic breast cancer might predict for outcome in the Arimidex, Tamoxifen Alone or in Combination (ATAC) adjuvant therapy trial. Patients and Methods Postmenopausal women with ER-positive, invasive, nonmetastatic, and operable or locally advanced potentially operable breast cancer were randomly assigned to neoadjuvant tamoxifen (20 mg daily), anastrozole (1 mg daily), or a combination of tamoxifen and anastrozole for 3 months. The tumor objective response (OR) was assessed by both caliper and ultrasound. Comparisons were also made of clinical response with ultrasound response, actual and feasible surgery with feasible surgery at baseline, OR in human epidermal growth factor receptor 2 (HER2) positive cancers, and tolerability. Results There were no significant differences in OR in the intent-to-treat population between patients receiving tamoxifen, anastrozole, or the combination. In patients who were assessed as requiring mastectomy at baseline (n 124), 44% of patients received breast-conserving surgery (BCS) after anastrozole compared with 31% of patients after tamoxifen (P.23); this difference became significant for patients who were deemed feasible for BCS by their surgeon (46% v 22%, respectively; P.03). The OR for patients with HER2-positive cancer (n 34) was 58% for anastrozole compared with 22% for tamoxifen (P.18). All treatments were well tolerated. Conclusion Neoadjuvant anastrozole is as effective and well tolerated as tamoxifen in ER-positive operable breast cancer in postmenopausal women, but the hypothesis that clinical outcome might predict for long-term outcome in adjuvant therapy was not fulfilled. J Clin Oncol 23:5108-5116. 2005 by American Society of Clinical Oncology INTRODUCTION Currently, the main clinical aim of neoadjuvant endocrine therapy before surgery is to downstage estrogen receptor (ER) positive breast cancers so that mastectomy might be avoided or to achieve operability in previously inoperable cancers. 1 In advanced 5108

Neoadjuvant Tamoxifen Anastrozole Combination breast cancer, the third-generation aromatase inhibitors anastrozole and letrozole have been shown to be either superior to tamoxifen 2,3 or at least as good. 4 Small studies have suggested that these agents may also be more effective than tamoxifen as neoadjuvant therapy in older women with large cancers. 5-8 The first randomized trial in this field showed no significant difference between the aromatase inhibitor vorozole (now discontinued from clinical study) and tamoxifen, 9 although this study was small and underpowered. More recently, a larger randomized trial (024) involving 337 postmenopausal women with breast cancers otherwise requiring mastectomy showed that 4 months of preoperative treatment with letrozole was more effective than tamoxifen in terms of both objective response (OR) rate and the number of women achieving sufficient downstaging to receive breast-conserving surgery (BCS). 10 Anastrozole has been assessed as neoadjuvant therapy in two nonrandomized studies of postmenopausal women with locally advanced or large operable breast cancers and was shown to achieve high rates of tumor regression, allowing BCS in some women. 8,11 Against this background, a key aim of the neoadjuvant Immediate Preoperative Anastrozole Tamoxifen or Combined With Tamoxifen (IMPACT) trial was to compare the relative efficacies of anastrozole and tamoxifen in achieving tumor regressions and, thus, allowing BCS in postmenopausal women who would otherwise require mastectomy. However, there was a second important aim underlying the IMPACT trial. Currently, the standard research tool for assessing the efficacy of new therapies in early breast cancer is the phase III randomized adjuvant trial. Such trials are large, expensive, and take years to achieve their outcome. A good example is the adjuvant Arimidex, Tamoxifen Alone or in Combination (ATAC) trial, which has recently shown that adjuvant anastrozole in postmenopausal women with early breast cancer achieves a significantly improved disease-free survival compared with tamoxifen or a combination of the two agents. 12 This trial, started in 1996, involved 9,366 patients and reported its first results 6 years later, with a median follow-up of 33 months. Neoadjuvant therapy trials might offer the eventual possibility of a more rapid and less resource-demanding alternative, at least for some therapies, if short-term surrogate clinical, pathologic, or biologic end points could be identified that might predict for long-term outcome in adjuvant trials. The IMPACT trial was designed to test this hypothesis and is the neoadjuvant equivalent of ATAC; 330 patients were randomly assigned to neoadjuvant anastrozole, tamoxifen, or a combination of anastrozole and tamoxifen, with a 12-week end point of clinical response and a 2-week biologic end point of change in proliferation as assessed by Ki67 staining. The aim was to determine whether or not either of these surrogate end points might predict for longterm outcome in the adjuvant setting. For this reason, postmenopausal patients with smaller breast cancers not necessarily requiring mastectomy were also included, and in this respect, the IMPACT trial differs from the earlier neoadjuvant letrozole trial. 10 Clinical results are reported here, and biologic results will be reported separately. PATIENTS AND METHODS Study Design This was a phase III randomized, double-blind, doubledummy, multicenter trial in which patients were randomly assigned 1:1:1 to receive a daily dose of anastrozole 1 mg and tamoxifen placebo, tamoxifen 20 mg and anastrozole placebo, or a combination of tamoxifen 20 mg and anastrozole 1 mg for 12 weeks before surgery (Fig 1). The trial was designed to be the Fig 1. Trial design. www.jco.org 5109

Smith et al neoadjuvant equivalent of the adjuvant ATAC trial. 12 Initially, it was planned that patients in IMPACT would continue on adjuvant study medication for a total of 5 years, but once the results of the ATAC trial became available, a protocol amendment was made to unblind our patients. Patients in the single arms were advised to continue on current therapy, whereas patients on the combination were given the choice of switching to either of the single agents based on ATAC guidelines. Patients with confirmed invasive histology and ER positivity had baseline clinical (bidimensional using calipers) and breast ultrasound (bidimensional) measurements recorded on entering the trial, along with a core needle biopsy and blood sampling. Two weeks after starting treatment, the clinical bidimensional caliper measurements were repeated, along with further blood sampling, and patients were invited to have a further core biopsy. Six weeks after starting treatment, clinical and ultrasound measurements were repeated. Final clinical and ultrasound measurements were made at 12 weeks, before surgical excision, and the final blood sample was taken. Any patients not having surgery for whatever reason were invited to have a further core biopsy. Patients who had BCS and some patients with involved axillary nodes after mastectomy were administered postoperative radiotherapy based on individual hospital protocols. Patients younger than 70 years who were found to have involved nodes or other pathologic indicators of high-risk disease at surgery were also offered postoperative adjuvant chemotherapy. Patient Inclusion Criteria Eligible patients were postmenopausal women with untreated, core needle biopsy proven, invasive, ER-positive breast cancer that was bidimensionally measurable clinically (caliper) and on ultrasound, operable or potentially operable locally advanced disease, and without evidence of metastatic spread. Women whose tumors were treatable with conservative surgery and women requiring mastectomy were included. The potential for BCS was defined prospectively by surgeons based on standard criteria including size and tumor to breast ratio. Women were defined as being postmenopausal if they were older than 60 years of age; had undergone a bilateral oophorectomy; were younger than 60 years, had a uterus, and had been amenorrheic for at least 12 months; or were younger than 60 years, did not have a uterus, and had follicle-stimulating hormone levels greater than 20 U/L. ER status was assessed locally and subsequently confirmed centrally. In the early phase of the study, centers lacking the facility to measure ER positivity were allowed to enter patients as ER unknown while awaiting ER status from the central laboratory. Women on hormone replacement therapy stopped treatment before study entry. Exclusion Criteria Exclusion criteria included the following: inoperable disease considered to be irreversible with neoadjuvant endocrine therapy; inflammatory breast cancer; any severe coincident medical disease; concurrent use of hormone replacement therapy; any invasive malignancy within the previous 10 years (other than basal cell carcinoma or cervical carcinoma-in-situ); investigational drug treatment within 30 days; and any medical or psychiatric condition making informed consent impossible. Study End Points The primary study objective was to compare clinical tumor OR by bidimensional caliper measurement between anastrozole, tamoxifen, and a combination of anastrozole and tamoxifen after 3 months of treatment. OR was calculated as the percentage of patients with a clinical complete response (CR) or partial response (PR) at 3 months. CR was defined as the clinical disappearance of tumor maintained until the 3-month point, and PR was defined according to WHO criteria 13 as a 50% decrease from baseline in the product of two perpendicular diameters maintained until 3 months. Minor response was defined as a decrease of 25% to 50% in the area of the tumor from baseline, and no change was defined as a decrease of less than 25% or an increase of less than 25% in the area of the tumor from baseline (both of these groups were included in the category of stable disease). Progressive disease was defined as an increase of 25% in the area of the tumor from baseline. The secondary objectives of the study were to compare the following measures between treatments: biologic changes in the tumor, including, in particular, proliferation as assessed by Ki67 staining, after 2 weeks and 3 months of treatment (to be reported separately); the conversion rates to conservative surgery (both actual and deemed by the surgeon) in patients deemed by their surgeon to require mastectomy at baseline before starting treatment; clinical OR in patients deemed to require mastectomy; ultrasound response and ultrasound response compared with clinical response; clinical response in patients whose tumors also overexpressed human epidermal growth factor receptor 2 (HER2); safety and side effects; and changes in circulating estradiol levels, lipids, and markers of bone resorption after 3 months of treatment. These end points were all prospectively defined in the protocol. Biologic results will be reported separately. Protocol Violations The two categories for major protocol violation were ER negativity (three patients) and no evidence of histologically proven invasive breast cancer (five patients, all of whom had positive cytology). Statistical Analysis Previous data have shown that approximately 60% of patients whose tumors were ER positive but not locally advanced achieved a clinical response after 6 months of treatment with neoadjuvant tamoxifen. 14 Because preoperative treatment duration in this trial would be limited to 3 months and it was anticipated that approximately 10% to 15% of patients would have locally advanced tumors, it was estimated that the OR rate for tamoxifen would be 40%. Assuming this, 102 patients per treatment arm were needed to detect an increase in response on anastrozole to 60% with 80% power and a two-sided 5% significance level. For comparative data, 102 patients were also needed in the combination arm. To allow for missing data, 110 patients per arm were recruited. The analysis of all end points was on an intent-to-treat basis, with the exception of safety outcomes, which were based on the treatment received, and biologic markers including response to HER2, which were analyzed on the per-protocol population. A blinded clinical review of all departures from protocol was undertaken to determine whether exclusion from the per-protocol analyses was warranted. All data were recorded on case report forms designed for the study, and the study database was populated by Syne Qua Non (Norfolk, United Kingdom). Data validation was carried out by the clinical study coordinator (G.W.) in association with Syne Qua Non. Statistical analyses were carried out by S.F. at AstraZeneca (Macclesfield, United Kingdom) and overseen by an independent statistician (S.E.A.). Analysis of the overall response rate was performed by means of a logistic regression analysis, with the treatment effect being 5110 JOURNAL OF CLINICAL ONCOLOGY

Neoadjuvant Tamoxifen Anastrozole Combination tested at the two-sided 5% significance level. For the primary treatment comparison (anastrozole v tamoxifen), the odds ratio and its 95% CI were calculated. This was prespecified in the statistical analysis plan, and in line with standard practice, no adjustments for multiple testing were made in any of the other end points in the study. As a robustness check, the analysis was repeated with the prognostic covariates of age ( 65 v 65 years) and clinically involved nodes (yes or no) included in the model. The reduction in surgical requirement, as defined by surgeon preference, was also analyzed by logistic regression, with the treatment effect tested at the two-sided 5% level. Tumor shrinkage was analyzed by using analysis of variance comparing anastrozole versus tamoxifen and combination therapy versus tamoxifen. Differences were tested at the two-sided 5% level. Normality assumptions of the model were examined using normal probability plots and the Kolmogorov-Smirnov statistic. ER, HER2, and Epidermal Growth Factor Receptor Analysis ER status was initially performed locally for entry onto the trial and then reviewed at a central laboratory (Royal Marsden Laboratory, London, United Kingdom) using the 6F11 antibody (Novocastra, Newcastle upon Tyne, United Kingdom), with tumors with more than 1% staining nuclei described as ER-positive tumors. Level of ER expression was assessed by H score, which incorporates both intensity of staining (0 to 3) and percentage of cells stained to provide a score of 0 to 300. HER2 was also assessed in the central laboratory, which is a national reference laboratory for this analysis. Tumors were considered as overexpressing if they scored 3 by immunohistochemistry using the Dako Hercept Test (Dakocytomation, Ely, United Kingdom) or if they showed greater than two-fold amplification of the HER2 gene as assessed by fluorescent in situ hybridization using the Vysis PathVysion kit (Vysis, Downers Grove, IL). Fluorescent in situ hybridization testing was only carried out for tumors that were 2 by immunohistochemistry. Epidermal growth factor receptor (EGFR) was measured using a previously validated immunohistochemical assay. 15 Tolerability Assessments Adverse events (defined as the development of a new medical condition or the deterioration of a pre-existing medical condition) were recorded at 2 weeks, 6 weeks, and 3 months. No prespecified checklists were used. Serious adverse events (defined as fatal or life threatening, requiring hospitalization, causing disability or incapacity, or requiring medical intervention to prevent incapacity) were recorded as they occurred. Ethical Considerations The trial was conducted in accordance with the principles of Good Clinical Practice as specified in the Declaration of Helsinki (1996 revision). The study protocol was approved first by a national multicenter research ethical committee and subsequently by individual local research ethics committees. All patients gave written informed consent before study enrollment. RESULTS Patients Between October 1997 and October 2002, a total of 330 patients (median age, 73 years) from 19 oncology centers across the United Kingdom and Germany were randomly assigned to receive treatment with anastrozole (n 113), tamoxifen (n 108), or the combination (n 109; Fig 1). The median time of follow-up was 13 weeks, and the follow-up period for the final patient ended in January 2003. The groups seemed well balanced with respect to patient characteristics and demographics (Table 1). Tumors were confirmed as ER positive in 98%, 99%, and 96% of patients in the anastrozole, tamoxifen, and combination groups, respectively, and the two patients with ER not recorded were entered early in the study, when centers could randomly assign patients on positive fine-needle aspirate, with pathology failing to confirm invasive cancer. When ER was assayed centrally, five patients were found to have ERnegative cancers (Table 1). OR Rates Clinical OR rates (caliper) for anastrozole, tamoxifen, and the combination were 37%, 36%, and 39%, respectively; there were no significant differences between any of these treatment groups (Table 2). Ultrasound response rates were 24%, 20%, and 28% for anastrozole, tamoxifen, and the combination, respectively; again, none of these differences were significant. Details of response, including CR, PR, stable disease (including minimal response and no change), progression, and not assessable, measured by caliper and ultrasound are listed in Table 3. The results from the adjusted analysis based on a logistic regression model, including treatment, age ( 65 v 65 years), and clinically involved nodes, supported the results of the primary analysis. A subgroup of 124 patients were assessed by the surgeon as requiring mastectomy at baseline (see next section). Characteristic Table 1. Patient Characteristics Anastrozole (n 113) Tamoxifen (n 108) Combination (n 109) Age, years Median 73.2 71.5 73.2 Range 51.8-90.2 49.8-88.4 51.6-85.7 Patients recorded as having 22.1 25.9 18.3 previously received HRT, % Tumor diameter at baseline, cm By caliper Median 3.8 3.8 4.0 Range 1.0-7.2 1.8-10.0 1.5-15.0 By ultrasound Median 2.6 2.3 2.4 Range 1.0-8.5 1.0-11.0 0.9-5.8 Receptor status, % ER positive 98.2 99.1 96.3 ER negative 1.8 0 2.8 ER not recorded 0 0.9 0.9 Abbreviations: HRT, hormone replacement therapy; ER, estrogen receptor. www.jco.org 5111

Smith et al Table 2. Patients Achieving the Primary End Point of Objective Tumor Response at 3 Months Patient Group Anastrozole Tamoxifen Combination Anastrozole v Tamoxifen Combination v Tamoxifen Odds Ratio 95% CI P Odds Ratio 95% CI P All patients Patients, No. 113 108 109 Caliper, % 37 36 39 1.05 0.61 to 1.81.87 1.15 0.67 to 2.00.61 Ultrasound, % 24 20 28 1.23 0.65 to 2.32.53 1.48 0.79 to 2.79.22 Mastectomy at baseline Patients, No. 46 36 42 Caliper, % 39 28 36 1.67 0.65 to 4.28.28 1.44 0.55 to 3.79.46 Ultrasound, % 33 19 29 2.01 0.72 to 5.62.19 1.66 0.57 to 4.80.35 ORs were observed in 39% of patients receiving anastrozole compared with 28% receiving tamoxifen and 36% receiving the combination. These differences were not statistically significant; the odds ratio for anastrozole versus tamoxifen was 1.67 (95% CI, 0.65 to 4.28; P.28), and the odds ratio for combination versus tamoxifen was 1.44 (95% CI, 0.55 to 3.79; P.46; Table 2). Change From Requirement for Mastectomy to BCS Two hundred twenty patients (67%) had baseline pretreatment surgical assessments recorded as requiring either mastectomy or BCS. Of these, 124 patients (56%) were deemed to need a mastectomy, and 96 (44%) were eligible for BCS. In the 124 patients considered to require mastectomy at baseline, 44% treated with anastrozole had BCS compared with 31% receiving tamoxifen (odds ratio, 1.75; 95% CI, 0.70 to 4.38; P.23.); in the combination arm, 24% had BCS (not significant; Table 4). Not all patients deemed by their surgeon to be eligible for BCS accepted this recommendation, and 46%, 22%, and 26% of patients were deemed to have achieved tumor regression sufficient to allow BCS after treatment with anastrozole, tamoxifen, and combination therapy, respectively. The improvement here with anastrozole compared with tamoxifen was statistically significant (odds ratio, 2.94; 95% CI, 1.11 to 7.81; P.03). There was no significant difference between the tamoxifen and combination groups (odds ratio, 1.24; 95% CI, 0.44 to 3.53; P.68; Table 4). Response Related to Level of ER Expression Only four patients had low ER H scores of between 1 and 20. Thus, we analyzed the effect of ER level on response Table 3. Tumor Response by Response Category at 3 Months Response Anastrozole (n 113) % of Patients Tamoxifen (n 108) Combination (n 109) Caliper CR 3 4 3 PR 35 32 37 SD 47 55 47 MR 26 25 19 NC 21 30 28 PD 9 5 5 NA 7 5 9 Ultrasound CR 0 1 0 PR 24 19 28 SD 30 34 28 MR 13 16 15 NC 17 19 14 PD 10 9 7 NA 36 36 37 Abbreviations: CR, complete response; PR, partial response; SD, stable disease; MR, minimal response; NC, no change; PD, progressive disease; NA, not assessable. Table 4. Surgeon s Assessment of Feasible Surgery at 3 Months and Actual Surgery at 3 Months in 124 Patients Assessed As Requiring Mastectomy at Baseline Treatment Group Surgeon s Assessment at 3 Months (%) Actual Surgery Performed at 3 Months (%) Anastrozole (n 46) BCS 46 44 M 41 44 NR 13 2 NP 11 Tamoxifen (n 36) BCS 22 31 M 64 58 NR 14 0 NP 11 Combination (n 42) BCS 26 24 M 69 67 NR 5 2 NP 7 Abbreviations: BCS, breast-conserving surgery; M, mastectomy; NR, not recorded; NP, not performed. Odds ratio 2.94 (95% CI, 1.11 to 7.81; P.03); no other differences significant. 5112 JOURNAL OF CLINICAL ONCOLOGY

Neoadjuvant Tamoxifen Anastrozole Combination rates according to H-score quartiles (Fig 2). We fitted a logistic regression model for OR, with ER at baseline as a covariate. Overall, there were statistically significantly more responders with higher ER levels (P.02). According to treatment, P.76,.38, and.002 for anastrozole, tamoxifen, and the combination, respectively. A nonparametric analysis was also performed, which confirmed the results of the parametric analysis. Response in Tumors Overexpressing HER2 We considered it appropriate here only to assess patients who had actually been treated per protocol because we wished to study the extent to which HER2 amplification or overexpression influenced response to treatment. Two hundred thirty-nine patients were assessable, of whom 34 (14%) were HER2 positive. One of these patients was also EGFR positive (and only two patients in total were EGFR positive). ORs were observed in seven (58%) of 12 patients with anastrozole, two (22%) of nine patients with tamoxifen, and four (31%) of 13 patients with the combination. Using Fisher s exact test, this difference between anastrozole and tamoxifen was not significant, with an odds ratio of 4.90 (95% CI, 0.53 to 63.22; P.18), but because of small numbers, the analysis was underpowered. The difference between the combination and tamoxifen was not significant. Tolerability and Adverse Events All treatments were generally well tolerated. The most common adverse event in all groups was hot flashes. This showed a nonsignificant trend towards a lower incidence with anastrozole (18%) than with tamoxifen (26%) or the combination (28%). The only significant difference was in vaginal discharge, which was not reported in any patients on anastrozole (0%) compared with 6% of patients on tamoxifen and 8% of patients on the combination. Thromboembolic events (eg, deep vein thrombosis and pulmonary embolism) were recorded in the neoadjuvant period and also for 30 days after surgery. No episodes were reported in patients treated with anastrozole alone compared with two events with tamoxifen alone and three events with the combination; four of these five events occurred during the 30 days after surgery. Adverse events occurring in the neoadjuvant treatment period in 5% or more of patients in any treatment group are listed in Table 5. Withdrawal Data All three treatments were well tolerated, with only 2%, 3%, and 2% of patients withdrawing as a result of adverse events in the anastrozole, tamoxifen, and combination arms, respectively (Table 6). DISCUSSION The primary end point of the IMPACT trial was clinical OR. In the main intent-to-treat analysis, no significant difference was seen between the three arms (anastrozole, 37%; tamoxifen, 36%; and the combination, 39%). Therefore, the primary end point did not predict for long-term outcome in the ATAC trial, in which anastrozole achieved a significantly superior disease-free survival compared with tamoxifen or the combination. 12 In a predefined analysis, there was a nonsignificant trend towards more patients requiring mastectomy at baseline actually receiving BCS with anastrozole than with tamoxifen (44% v 31%, respectively; P.23); this difference became significant Fig 2. Objective response rate versus estrogen receptor (ER) H score, by quartiles, for anastrozole, tamoxifen, and the combination. www.jco.org 5113

Smith et al Table 5. Adverse Events That Occurred During the Neoadjuvant Period at a Rate of 5% in Any Arm Adverse Event Anastrozole (n 113) % of Patients Tamoxifen (n 108) Combination (n 109) Hot flashes 17.7 25.9 27.5 Dizziness 8.8 3.7 4.6 Headache 8.8 6.5 6.4 Nausea 8.8 10.2 9.2 Nasopharyngitis 6.2 2.8 0.9 Fatigue 5.3 7.4 7.3 Lethargy 5.3 4.6 2.8 Arthralgia 4.4 0.9 5.5 Vaginal discharge 0 5.6 8.3 Thromboembolism 0 1.9 2.8 NOTE. All mild, moderate, and severe side effects have been combined together. Only significant difference for anastrozole v tamoxifen (P.013). Thromboembolic events for neoadjuvant plus 30-day period. Table 6. AEs, Withdrawals, and Deaths in the Neoadjuvant Phase Event Anastrozole (n 113) % of Patients Tamoxifen (n 108) Combination (n 109) Total AEs 69.0 73.1 78.0 All serious AEs 7.1 3.7 4.6 Drug-related AEs 45.1 48.1 55.0 Total withdrawals 6.2 6.5 8.3 Withdrawals as a result of an AE 1.8 2.8 1.8 Total deaths 0 0 0 Abbreviation: AE, adverse event. for patients deemed by their surgeon to be eligible for BCS after treatment (46% v 22%, respectively; P.03). In the neoadjuvant letrozole 024 trial, 45% of patients initially requiring mastectomy also achieved BCS after letrozole compared with 36% of patients after tamoxifen (P.036). 10 Therefore, the IMPACT trial provides some further supportive data that the third-generation aromatase inhibitors are significantly more effective than tamoxifen in downstaging large breast cancers and reducing the need for mastectomy in postmenopausal women. The comparative response rates in the larger cancers initially requiring mastectomy showed a nonsignificant trend of 39% v 28% in favor of anastrozole compared with tamoxifen, respectively. If this were confirmed in other trials, then it would suggest that serial clinical measurements in smaller cancers during neoadjuvant endocrine therapy might be exposed to larger errors because response may be slow; in addition, follow-up core biopsies for biologic studies after 2 weeks of treatment could further confound accurate measurement because of subsequent hematoma and tissue edema. For the time being, we recommend caution in including patients with small cancers in neoadjuvant endocrine therapy trials with primary clinical end points, although smaller breast cancers could still be appropriate for trials with a biologic end point. The only significant difference in adverse events between the treatment arms during the 3-month neoadjuvant period was in vaginal discharge, which occurred only in patients on tamoxifen either alone or in combination. This reflects similar findings with longer treatment exposure in the much larger adjuvant ATAC trial. 16 Thromboembolic events were monitored not just during the neoadjuvant period but for 30 days afterwards; these events occurred in five patients on tamoxifen alone or in combination but were not seen in any patient receiving anastrozole alone. It should be noted that four of these five episodes were in the postoperative period. Therefore, our data indicate caution in the use of tamoxifen in the immediate postoperative period and might be considered a further advantage for anastrozole. In the preoperative letrozole trial, 15% of patients whose tumors were ER positive also showed overexpression of EGFR and/or HER2. 17 In this group, 15 (88%) of 17 patients treated with preoperative letrozole achieved a clinical response compared with only four (21%) of 19 patients on tamoxifen. Despite the small numbers, this difference was highly significant (P.0004). 17 In the IMPACT trial, 14% of assessed patients overexpressed HER2, including one who also overexpressed EGFR. In this subgroup, seven (58%) of 12 patients treated with anastrozole responded compared with two (22%) of nine patients treated with tamoxifen and four (31%) of 13 patients treated with the combination. The patient population and treatment duration differed between the two studies, but, despite the differences in the IMPACT trial being not significant, the trend in favor of anastrozole reflects the trend observed with letrozole and reinforces the hypothesis that aromatase inhibitors may be more effective than tamoxifen in the treatment of ER-positive early breast cancer that also overexpresses HER2. In the letrozole versus tamoxifen neoadjuvant study, 17 among the few patients with low ER scores, there were no responders to tamoxifen but several responders to the aromatase inhibitor. In the current study, we were unable to examine this directly because even fewer patients had low scores. Overall, there was a significantly lower response for patients with the lowest ER scores, but there was no trend for this to differ between anastrozole and tamoxifen. In conclusion, the IMPACT trial did not fulfill the hypothesis that short-term clinical response in the intent-to-treat population might be used as a surrogate end point to predict for the ATAC trial outcome in the adjuvant setting. Results relating to the second part of our hypothesis, namely that differences in the biologic end point of Ki67 after 2 weeks of treatment might predict for long-term outcome in the adjuvant setting, are being reported separately. 5114 JOURNAL OF CLINICAL ONCOLOGY

Neoadjuvant Tamoxifen Anastrozole Combination Acknowledgment We thank Alison Norton for her wise secretarial and editorial assistance in the preparation of this article, and we also thank the research nurses on all the units for their skill in making this trial run successfully and for the support they provided to the patients under their care. Appendix The following members and institutions of the IMPACT Trialists Group entered patients onto this trial: W.H. Allum, S. Ashley, A. Bradley, I. Boedinghaus, D. Brett, G. Gui, J. Diggins, J. Holborn, A. Ring, N. Sacks, C. Shannon, I. Smith, and G. Walsh, Royal Marsden Hospital, London, United Kingdom (n 88); S. Detre, M. Dowsett, M. Hills, and J. Salter, Royal Marsden Laboratory, London, United Kingdom; S. Ebbs, J. Kember, and C. Chu, Mayday University Hospital, London, United Kingdom (n 65); I. Batty, K. Kazim, and A. Skene, Royal Bournemouth Hospital, Bournemouth, United Kingdom (n 37); J.M. Dixon, J. Murray, and L. Renshaw, Western General Hospital, Edinburgh, United Kingdom (n 43); F. McNeill and K. Rooke, Essex County Hospital, Colchester, United Kingdom (n 7); C. Griffith and J. Bevington, Royal Victoria Infirmary, Newcastle, United Kingdom (n 13); A. Evans and M. Pidgley, Poole General Hospital, Poole, United Kingdom (n 11); J.-U. Blohmer and W. Lichtenegger, Universitätsklinikum Charité, Berlin, Germany (n 11); P. Sauven and K. Rooke, Chelmsford and Essex Centre, Chelmsford, United Kingdom (n 10); C. Holcombe and K. Makinson, Royal Liverpool University Hospital, Liverpool, United Kingdom (n 9); L. Barr, N.J. Bundred, and T. Pritchard, University Hospital of South Manchester, Manchester, United Kingdom (n 8); N. Harbeck, Frauenklinik der TU München, München, Germany (n 6); J. Clarke and J. Mansi, St. George s Hospital, London, United Kingdom (n 6); H. Stehle, Marienhospital, Stuttgart, Germany (n 6); T. Reimer, Universitäts-Frauenklinik, Rostock, Germany (n 5); K. Brunnert, Zentrum für Senologie und Plastische Chirurgie, Osnabrück, Germany (n 2); M. Lansdown and J. Hepper, St. James s University Hospital, Leeds, United Kingdom (n 1); D. Dubois and H. Stansby, Portsmouth Oncology Centre, Portsmouth, United Kingdom (n 1); and Z. Rayter, Bristol Royal Infirmary, Bristol, United Kingdom (n 1). The AstraZeneca Scientific Team included Peter Barker, Stephen Bird, Phil Davies, Jo Diver, Sonia Harris, Karen Langfeld. Authors Disclosures of Potential Conflicts of Interest Although all authors have completed the disclosure declaration, the following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Authors Employment Leadership Consultant Stock Honoraria Research Funds Testimony Other Ian E. Smith Novartis (B) AstraZeneca (A); Novartis (B) Mitch Dowsett AstraZeneca (B); Novartis (B) AstraZeneca (A); Novartis (A); Pfizer (A) J. Michael Dixon AstraZeneca (A); Novartis (A); Pfizer (A) Anthony Skene AstraZeneca (A) AstraZeneca (A); Novartis (A) J.-U. Blohmer Stephen Francis AstraZeneca AstraZeneca (B); Novartis (B) AstraZeneca (C); Novartis (C) AstraZeneca (C); Novartis (C); Pfizer (C) AstraZeneca (A) AstraZeneca (A) Dollar Amount Codes (A) $10,000 (B) $10,000-99,999 (C) $100,000 (N/R) Not Required AstraZeneca (A); Novartis (A) REFERENCES 1. Dixon JM, Anderson TJ, Miller WR: Neoadjuvant endocrine therapy of breast cancer: A surgical perspective. Eur J Cancer 38:2214-2221, 2002 2. Nabholtz JM, Buzdar A, Pollak M, et al: Anastrozole is superior to tamoxifen as first-line therapy for advanced breast carcinoma in postmenopausal women: Results of a North American multicenter randomized trial. J Clin Oncol 18:3758-3767, 2000 3. Mouridsen H, Gershanovich M, Sun Y, et al: Superior efficacy of letrozole versus tamoxifen as first-line therapy for postmenopausal women with advanced breast cancer: Results of a phase III study of the International Letrozole Breast Cancer Group. J Clin Oncol 19:2596-2606, 2001 4. Bonneterre J, Thurlimann B, Robertson JFR, et al: Anastrozole versus tamoxifen as first-line therapy for advanced breast cancer in 668 postmenopausal women: Results of the Tamoxifen or Arimidex Randomized Group Efficacy and Tolerability study. J Clin Oncol 18:3748-3757, 2000 5. Dixon JM, Renshaw L, Bellamy C, et al: The effects of neoadjuvant anastrozole (Arimidex) on tumor volume in postmenopausal women with breast cancer: A randomized, double-blind, single-center study. Clin Cancer Res 6:2229-2235, 2000 6. Miller WR, Dixon JM: Endocrine and clinical endpoints of exemestane as neoadjuvant therapy. Cancer Control 9:9-15, 2002 7. Dixon JM, Jackson J, Renshaw L, et al: Neoadjuvant tamoxifen and aromatase inhibitors: Comparisons and clinical outcomes. J Steroid Biochem Mol Biol 86:295-299, 2003 www.jco.org 5115

Smith et al 8. Milla-Santos A, Milla L, Rallo L, et al: Anastrozole as neoadjuvant therapy for hormonedependent locally advanced breast cancer in postmenopausal patients. Proc Am Soc Clin Oncol 21:40a, 2002 (abstr 156) 9. Harper-Wynne CL, Sacks NPM, Shenton K, et al: Comparison of the systemic and intratumoral effects of tamoxifen and the aromatase inhibitor vorozole in postmenopausal patients with primary breast cancer. J Clin Oncol 20: 1026-1035, 2002 10. Eiermann W, Paepke S, Appfelstaedt J, et al: Preoperative treatment of postmenopausal breast cancer patients with letrozole: A randomized double-blind multicentre study. Ann Oncol 12:1527-1532, 2001 11. Anderson TJ, Dixon JM, Stuart M, et al: Effect of neoadjuvant treatment with anastrozole on tumour histology in postmenopausal women with large inoperable breast cancer. Br J Cancer 87:334-338, 2002 12. Arimidex Tamoxifen Alone or in Combination Trialists Group: Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: First results of the ATAC randomised trial. Lancet 359:2131-2139, 2002 13. WHO: WHO Handbook for Reporting Results of Cancer Treatment. Geneva, Switzerland, WHO, WHO Offset Publication 48, 1979 14. Gaskell DJ, Hawkins RA, de Carteret S, et al: Indications for primary tamoxifen therapy in elderly women with breast cancer. Br J Surg 79:1317-1320, 1992 15. Newby JC, A Hern RT, Leek RD, et al: Immunohistochemical assay for epidermal growth factor receptor on paraffin-embedded sections: Validation against ligand binding assay and clinical relevance to breast cancer. Br J Cancer 71:1237-1242, 1995 16. Arimidex Tamoxifen Alone or in Combination Trialists Group: Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: Results of the ATAC trial efficacy and safety update analyses. Cancer 98:1802-1810, 2003 17. Ellis MJ, Coop A, Singh B, et al: Letrozole is more effective neoadjuvant endocrine therapy than tamoxifen for ErbB1- and/or ErbB2-positive, estrogen receptor-positive primary breast cancer: Evidence from a phase III randomized trial. J Clin Oncol 19:3808-3816, 2001 5116 JOURNAL OF CLINICAL ONCOLOGY