- © 2005 by American Society of Clinical Oncology
Biomarker Changes During Neoadjuvant Anastrozole, Tamoxifen, or the Combination: Influence of Hormonal Status and HER-2 in Breast Cancer—A Study from the IMPACT Trialists
- Mitch Dowsett,
- Steve R. Ebbs,
- J. Michael Dixon,
- Anthony Skene,
- Clive Griffith,
- Irene Boeddinghaus,
- Janine Salter,
- Simone Detre,
- Margaret Hills,
- Susan Ashley,
- Stephen Francis,
- Geraldine Walsh and
- Ian E. Smith
- From the Academic Department of Biochemistry and the Breast Unit, Royal Marsden Hospital, London; Mayday University Hospital, Croydon, Surrey; Edinburgh Breast Unit, Edinburgh; Royal Bournemouth Hospital, Bournemouth, Dorset; Royal Victoria Infirmary, Newcastle Upon Tyne, Tyne and Wear; and AstraZeneca, Alderley Park, Macclesfield, Cheshire, United Kingdom
- Address reprints requests to Mitch Dowsett, MD, Academic Department of Biochemistry, Royal Marsden Hospital, London SW3 6JJ, United Kingdom; e-mail: mitch.dowsett{at}icr.ac.uk.
Abstract
Purpose To investigate the relationships between biomarker changes in breast cancer during neoadjuvant (preoperative) endocrine therapy.
Patients and Methods The IMPACT trial compared the preoperative use of tamoxifen with anastrozole alone or in combination in postmenopausal women (n = 330) with primary breast cancer. Biomarkers were measured in tumor biopsy specimens taken at baseline, and after 2 and 12 weeks of treatment.
Results A decrease in the proliferation marker Ki67 occurred in the majority of patients: 52 (93%) of 56, 46 (85%) of 54, and 37 (84%) of 44 patients in the anastrozole, tamoxifen, and combination groups, respectively. There was a significantly greater suppression of Ki67 in the anastrozole-treated group than in the tamoxifen- or combination-treated groups, which is parallel to the greater efficacy seen for anastrozole over these two treatments in the Arimidex, Tamoxifen, Alone or in Combination adjuvant trial. A positive relationship was noted between estrogen-receptor level and Ki67 suppression in all patients. Ki67 was reduced to a greater extent in progesterone receptor-positive tumors compared with progesterone receptor-negative tumors. HER-2-negative tumors tended to show a greater reduction in Ki67 compared with HER-2-positive tumors, but the difference was only significant in the tamoxifen group after 2 weeks, and in the anastrozole group after 12 weeks.
Conclusion These results confirm the value of Ki67 as a molecular marker, and provide information regarding the relationships between treatment-induced changes in Ki67 and other important biomarkers. Studies such as this should help integrate agents targeted at growth factor signaling with endocrine agents in breast cancer.
INTRODUCTION
Neoadjuvant trials provide a unique opportunity to integrate the molecular determinants of response and resistance with the clinical response of primary breast cancer to medical therapy. Although there have been many studies attempting to do this with cytotoxic therapy, including recent studies based on comprehensive expression array profiling, the greater understanding of mechanisms of response to endocrine therapy has made neoadjuvant endocrine studies of particular biologic interest.
It is now clear that the third-generation aromatase inhibitors have greater efficacy than tamoxifen in both early and late hormone receptor–positive breast cancer.1 Correlative science studies in the neoadjuvant scenario provide an opportunity to better understand the molecular determinants of this differential efficacy, and to assess the possibility that, despite the overall greater efficacy of the aromatase inhibitors, there may be identifiable subgroups that derive as much efficacy from tamoxifen.
In our publication2 of the first randomized presurgical trial of an aromatase inhibitor (vorozole in this case) versus tamoxifen, we found that there was a greater suppression of the nuclear proliferation marker, Ki67, with the aromatase inhibitor at both 2 and 12 weeks into therapy. This greater antiproliferative effect was matched by a greater suppression of progesterone receptor (PgR). The high estrogen dependence of PgR expression suggests that vorozole achieved a greater reduction in estrogen signaling and that the antiestrogen effect of tamoxifen was impeded by its partial agonist activity. Letrozole has also been reported to produce a greater suppression of Ki67 compared with tamoxifen, and in this case the difference was found to be particularly marked in patients with human epidermal growth factor receptor 2 (HER-2) -positive disease,3 a finding that was parallel to the greater clinical efficacy of letrozole in this subgroup in the same trial.4
More recently, we have reported the first clinical5 and biologic6 results from the IMPACT trial, a double-blind, randomized, neoadjuvant comparison of anastrozole with tamoxifen alone or combined in postmenopausal patients with steroid receptor–positive primary breast cancer. This trial was identical in its drug comparisons to the Arimidex, Tamoxifen, Alone or in Combination (ATAC) adjuvant trial, which reported that anastrozole led to improved disease-free survival compared with tamoxifen, while tamoxifen and the combination had similar efficacy.7 In the IMPACT trial, there was no overall difference in response rates between the three treatment arms, but anastrozole showed a greater response rate than tamoxifen among those women who were deemed unsuitable for breast conserving surgery.5 As such, this result was similar to the greater response rate to letrozole than tamoxifen in an earlier neoadjuvant trial that was confined to patients requiring mastectomy.8
In contrast to the overall lack of parallelism between the clinical response data from the IMPACT trial and the improvement in disease-free survival in ATAC, the reduction in Ki67 levels in IMPACT patients at both 2 and 12 weeks was parallel to the ATAC outcome, that is, Ki67 suppression was greater with anastrozole than tamoxifen and was similar between tamoxifen and the combination.6 This suggests that early changes in Ki67 in the neoadjuvant setting might be predictive of long-term outcome in the adjuvant setting.
Apoptosis levels were also significantly reduced with anastrozole at 2 and 12 weeks. At 12 weeks this change was significantly different from that in tamoxifen-treated patients who showed no significant change in apoptosis. It is possible that this difference between the two agents might have at least, in part, balanced out the greater reduction in proliferation with the aromatase inhibitor. However, the reduction in the growth index (Ki67 [%]/apoptosis [%]) from baseline to 3 months, which is a first approximation to integrating these two components as an index of growth, was also suppressed to a significantly greater degree with anastrozole than tamoxifen. The lack of increase in apoptosis with anastrozole or tamoxifen (or the combination) is contrary to expectation from studies in xenografts,9,10 but confirms the data seen with smaller numbers of patients in the earlier comparison of vorozole with tamoxifen.2 These data strongly suggest that estrogen is an important proliferative, but not survival, signal for hormone-dependent breast cancer cells.
We describe in this report other biologic results from the IMPACT trial, including data on the influence of estrogen receptor (ER), PgR, and HER-2 on the antiproliferative effects of, and the clinical response to, the agents. Experimental details of the clinical and biologic investigations with IMPACT have been described in detail elsewhere,5 but aspects that are important for the interpretation of the data are given here.
PATIENTS AND METHODS
Study Design
The design of the clinical aspects of this trial are described in detail elsewhere.5 In summary, this was a randomized, double-blind, double-dummy, multicenter trial in which patients with primary breast cancer were randomized 1:1:1 to receive a daily dose of anastrozole (1 mg) plus tamoxifen placebo or tamoxifen (20 mg) plus anastrozole placebo or anastrozole (1 mg) plus tamoxifen (20 mg) for 12 weeks before surgery. Eligible patients were postmenopausal women with previously untreated, core-needle biopsy proven, invasive, ER-positive breast cancer. Tumors were ≥ 2 cm in maximum diameter, operable or locally advanced but potentially operable (after medical downstaging) and without evidence of metastatic spread. Any women receiving hormone replacement therapy stopped this before the trial. Patients had to have ceased such therapy at least 4 weeks before the start of treatment to be assessable for the biomarker end points.
The primary clinical objective of the trial was to compare the differences between the treatments on objective tumor response. Clinical measurements of tumor size (bidimensional by calipers) were made at baseline and before surgery at 12 weeks. Objective clinical response was calculated based on WHO criteria11 as follows: complete response (CR), the clinical disappearance of tumor maintained until the 12-week point; partial response (PR), ≥ 50% decrease from baseline in the product of two perpendicular diameters again maintained until 12 weeks; minor response (MR), decrease of 25% to 50% in the two-dimensional product from baseline; stable disease (SD), a decrease of < 50% or an increase of < 25% in two-dimensional product; progressive disease (PD), increase of ≥ 25% in the two-dimensional product. Written informed consent was obtained from all patients before study entry and a research ethics committee at all study sites approved the protocol.
Core-cut biopsies were taken before starting therapy and at 2 weeks (nonobligatory). Patients not proceeding to surgery for whatever reason were invited to have a further core biopsy at 12 weeks. Core biopsies and the excision biopsy were fixed in 10% neutral buffered formalin for 24 to 28 hours before processing and embedding at local pathology centers in paraffin-wax blocks. These blocks (or unstained sections derived from a small proportion of cases) were sent to the central laboratory (Academic Department of Biochemistry, Royal Marsden Hospital, London, United Kingdom).
Unless otherwise specified, primary antibodies and avidin-biotin complex were purchased from DakoCytomation (Cambridge, UK). Sections of 3 to 4 μm were cut and dried onto charged microscope slides. Sections were dewaxed in xylene, taken to water, and endogenous peroxidase was blocked. Antigen retrieval was carried out by microwaving the sections at 750 W in preheated citrate buffer at pH 6.0 for 10 minutes, after which they were cooled in the buffer to room temperature and normal rabbit serum was applied as a blocking antibody in phosphate-buffered saline (PBS) before the primary antibody. ER and PgR were stained at room temperature for 2 hours using the following antibody dilutions: 1/40 of 6F11 and 1/100 of 312, respectively (both from Novocastra, Newcastle upon Tyne, United Kingdom). The sections were washed and incubated for 45 minutes in 1/200 biotinylated anti-mouse immunoglobulins. Following a further wash, avidin-biotin complex was applied for 30 minutes. After washing, the peroxidase activity was developed to a brown stain by means of 0.05% 3,3′-diaminobenzidine enhanced with 0.07% imidazole and hydrogen peroxide. ER and PgR were scored as percentage of cells positive. The cutoff for ER positivity was > 1% cells positive.
HER-2 was measured immunohistochemically using the HercepTest (DakoCytomation) and by fluorescent in situ hybridization (Vysis Pathvysion, Downers Grove, IL) according to manufacturers instructions. HER-2 was considered positive if immunohistochemical staining was scored 3+ or 2+ if the fluorescence in situ hybridization analysis indicated an amplification ratio of > 2.0.
Statistical Analysis
The study was powered for clinical response rate comparisons, assuming an objective response rate (ie, CR + PR) to tamoxifen of 40%. One hundred two patients per treatment arm were required to detect an increase in response rate with anastrozole to 60% with 80% power and a two-sided 5% significance level. For comparative data, 102 patients were also required for the comparison between tamoxifen and the combination arm. To allow for missing data, 110 patients per arm were recruited.
For changes in Ki67, powering was determined on the basis of previously published estimates of a reduction of 47% after a median of 21 days of treatment with tamoxifen.12 Data from 50 patients in each of the study arms would enable the detection of a further reduction with anastrozole or the combination to 67% with 80% power at a 5% level of significance.
Descriptive statistics for Ki67, apoptosis, and growth index (Ki67 [%]:apoptosis [%]) were expressed as geometric means because of the approximate lognormal distribution of the data. ER and PgR, as measured by the H-score, did not follow approximate normal or lognormal distributions, and so medians were used as descriptive statistics for these markers. Values at 2 and 12 weeks were expressed as a proportion of the baseline, and analyses of variance were conducted for within and between treatment comparisons. For the between treatment comparisons, the nonparametric Kruskal-Wallis test was used to confirm the results from the parametric analysis of variance (ANOVA) test. The nonparametric sign test was preferred to test the significance of percent change in ER and PgR within treatment groups.
The percent change in Ki67 and in growth index (both from baseline to 2 weeks and baseline to 12 weeks) was compared between HER-2-positive and -negative patients, and PgR-positive and -negative patients, using summary statistics (geometric mean) and ANOVA tests. Kruskal-Wallis tests were used to confirm results from the parametric tests.
Baseline ER and PgR, measured on a continuous scale, were correlated with the percentage of change in Ki67 and growth index (both from baseline to 2 weeks, and baseline to 12 weeks) using Spearman's rank correlation coefficient.
Logistic regression was conducted to determine whether percentage of change in the molecular parameters predicted for clinical response. These analyses have been reported previously6 and will not be discussed further in this article.
In common with the ATAC trial and the IMPACT clinical analyses, statistical comparisons were made only for anastrozole with tamoxifen, and for the combination with tamoxifen.
RESULTS
Biopsies were available for 292 patients who were treated with anastrozole (n = 98), tamoxifen (n = 98), or the combination of anastrozole and tamoxifen (n = 96). Major violations or deviations from the trial protocol occurred in 12, 10, and 11 of these patients, leaving 86, 88, and 85 patients per protocol, respectively.
We have previously reported that the geometric mean suppression of Ki67 was significantly greater at both 2 and 12 weeks with anastrozole (76.0% and 81.6%, respectively) than with tamoxifen (59.5% and 61.9%, respectively). Mean suppression with the combination (63.9% and 61.1%, respectively) was similar to that with tamoxifen. After 2 weeks, there was a numerical decrease in Ki67 in all but four (7%) of 56, eight (15%) of 54, and seven (16%) of 44 patients in the anastrozole, tamoxifen, and combination groups, respectively (Fig 1). The number of patients not showing suppression with tamoxifen or the combination of tamoxifen and anastrozole was greater than that observed for anastrozole. Thus, the lesser degree of Ki67 suppression with these treatments than with anastrozole alone appears to depend, in part, on a greater number of refractory patients. It is important to note that for an individual tumor to show a statistically significant increase in Ki67 the change would need to be by at least 50%.13 The apparent increases in Ki67 were generally of more modest levels.
Individual changes in Ki67 according to treatment after 2 weeks. The bold lines indicate the few patients that failed to show a decrease in Ki67 levels. (A) Anastrozole; (B) tamoxifen; (C) combination.
The individual changes in Ki67 for all patients treated with anastrozole are shown in Figure 2A. A subset of six patients selected by visual inspection of the data as showing relatively high pretreatment Ki67 and a modest reduction in Ki67 levels at 2 weeks are shown in Figure 2B. While two of these six patients showed a partial clinical response, they may form a group of patients with tumors that show a relative resistance to anastrozole, and these may form the target of alternative or additional treatment strategies. A greater understanding of the factors in the pretreatment phenotype that associate with, and/or may be determinants of, the variable suppression would enhance the development of new treatment strategies. We have therefore assessed the level of suppression of Ki67 according to expression of ER, PgR, and HER-2.
Individual changes in Ki67 at 2 weeks and 12 weeks for (A) patients treated with anastrozole and (B) a selection of patients who show modest changes on treatment.
The degree of suppression of Ki67 at 2 and 12 weeks according to quartile of ER expression is shown in Figures 3A and B, respectively. When analyzed by Spearman's rank analysis, there was a positive significant correlation between ER level and degree of Ki67 suppression for all groups of patients at both 2 and 12 weeks, except for anastrozole after 2 weeks, although there was also a trend apparent at that time point. Percentage suppression of Ki67 is greater with anastrozole than with tamoxifen at all quartiles of ER, but the differential was greatest for the lowest quartile.
Geometric mean suppression of Ki67 for the different treatments according to estrogen receptor (ER) quartile after 2 and 12 weeks of treatment. (A) Anastrozole; (B) tamoxifen; (C) combination; (D) all.
The suppression percentage of Ki67 according to PgR or HER-2 positivity is shown in Figures 4 and 5, respectively. Except for the combination group at 2 weeks, Ki67 was reduced to a statistically significant greater degree in PgR-positive than PgR-negative tumors in the overall group and in the individual groups at 2 and 12 weeks. The differential between tamoxifen and anastrozole appears to be greater for the PgR-positive group. At 2 weeks, there was a numerically greater reduction in Ki67 in HER-2-negative tumors compared with HER-2-positive tumors in the overall population, but the difference only reached significance in the tamoxifen-treated group. At 12 weeks, there was a significantly greater reduction in Ki67 in HER-2-negative tumors than in HER-2-positive tumors in the overall population and in the anastrozole group. There were, however, no HER-2-positive tumors with high Ki67 levels at baseline for the anastrozole group, which increases the uncertainty in the value of these comparisons (Fig 6).
Geometric mean change in Ki67 for the different treatments according to progesterone receptor (PgR) status after 2 weeks and 12 weeks of treatment. (A) Anastrozole; (B) tamoxifen; (C) combination; (D) all.
Geometric mean change in Ki67 for the different treatments according to HER-2 status after 2 and 12 weeks of treatment. (A) Anastrozole; (B) tamoxifen; (C) combination; (D) all.
Individual changes in Ki67 according to treatment after 2 and 12 weeks in tumors that were positive for HER-2. The dashed lines connect values in the same patient when a 2-week sample was unavailable. (A) Anastrozole; (B) tamoxifen; (C) combination.
In the anastrozole group, there were no significant changes in the median level of ER after 2 or 12 weeks (+3% and −3%, respectively). Similarly, after 2 weeks, there were no significant changes for tamoxifen or the combination (−4% and −3%, respectively), but at 12 weeks, both showed a significant fall in ER levels (−13% and −14%, respectively; P < .05 for both). While these falls in ER were relatively modest, it appears that the post-treatment levels of ER were more homogenous than pretreatment levels, with many fewer H-score values > 225 and a marked increase in values between 175 and 225 with both tamoxifen and the combination (Figs 7 and 8). This pattern was not apparent with anastrozole.
Individual changes in estrogen receptor according to treatment at 2 and 12 weeks. (A) Anastrozole; (B) tamoxifen; (C) combination.
Frequency histograms of estrogen receptor (ER) values at baseline, 2 weeks, and 12 weeks according to treatment showing the loss of high values of ER during treatment with tamoxifen and the combination. (A) Anastrozole; (B) tamoxifen; (C) combination.
There were major differences between the treatments in their effects on PgR. In the anastrozole-treated patients, PgR levels fell by 41% at 2 weeks and 82% at 12 weeks (median, P ≤ .05 for both), while in the tamoxifen-treated group, there was an increase by 48% at 2 weeks (P ≤ .05) with a return close to baseline (+3%) at 12 weeks. At both 2 and 12 weeks the combination-treated group showed no significant change from baseline (+8% and +2%, respectively). Thus the effects were intermediate between anastrozole and tamoxifen at 2 weeks, but segregated with the tamoxifen effect at 12 weeks. At both 2 and 12 weeks the effects were highly significantly different between tamoxifen and anastrozole (P < .0001). The changes for individual patients on anastrozole and tamoxifen at 2 weeks are shown in Figure 9. There was a significant relationship between the reductions in Ki67 and PgR levels for the anastrozole group (P = .003) and a significant relationship between the decrease in Ki67 and the increase in PgR for the tamoxifen group (P = .027) at this time.
Individual changes in progesterone receptor after 2 weeks in the (A) anastrozole and (B) tamoxifen groups.
DISCUSSION
The finding that mean changes in Ki67 and the growth index in the IMPACT trial were parallel to the differences in disease-free survival in the ATAC adjuvant trial7,14 is an important indication of the value of these markers in assessing the relative efficacy of hormonal agents. There was not a significant relationship between the decrease in Ki67 and clinical response but this may be due to clinical response being an imprecise end point and/or that clinical regression, as opposed to reduced tumor growth, may require a particularly marked reduction in cell proliferation accompanied by persistent cell death. We have additionally found that while there were trends to greater suppression of Ki67 at both 2 weeks and 12 weeks in those patients downstaged to being suitable for breast-conserving surgery compared with those still requiring mastectomy, in neither case was the difference statistically significant (data not shown).
At present, the data do not support the use of short-term changes in Ki67 for treatment decisions in individual patients. However, they do support the use of this clinical model for the evaluation of new agents before the conduct of large-scale adjuvant trials. For example, had the results of this study been available before the conduct of the ATAC trial, the close similarity of all data between the tamoxifen and combination arm is likely to have led to the exclusion of the combination arm (which was eventually dropped after the first analysis of ATAC7). Furthermore, had tamoxifen and anastrozole been developed clinically at the same time, these data would have allowed the prioritization of anastrozole over tamoxifen, the former now being known to be the more efficacious.
The results presented here show further detail of the effects of the treatments on biologically defined subgroups that are considered to be of importance to response to antihormonal treatment. In a biologic study of neoadjuvant therapy with letrozole, on-treatment biopsies were collected only at the end of the 4 months of medical therapy3 such that those data cannot be directly compared with the results presented here, although it is likely that the changes with the two aromatase inhibitors are likely to be similar.
A remarkable feature of the changes in Ki67, particularly in the anastrozole group, was that there were very few tumors that did not show a numerical decrease in score. This suggests that almost all ER-positive breast carcinomas show some dependence on estrogen for proliferation. The degree of suppression of Ki67 did, however, vary markedly between tumors, and this indicates that the degree of estrogenic dependence is also highly variable between tumors. Assessment of pretreatment phenotype, and changes in that phenotype with therapy alongside the changes in Ki67, may help establish the mechanisms that contribute to this variable response and lead to strategies that may overcome resistance.
Throughout the biologic studies in the IMPACT trial, results with the combination arm were similar to tamoxifen. This was also seen in clinical efficacy and tolerability end points in both the IMPACT and ATAC trials.5,7,14 The explanation for this is probably due to tamoxifen saturating the ER15 such that estrogen withdrawal with anastrozole has no perceptible effect on ER signaling when given in combination with tamoxifen.
The greater reduction of Ki67 overall in tumors with high ER is consistent with the known greater effectiveness of antihormonal therapy in such tumors. This feature was also apparent in the clinical response data in the IMPACT trial.5 The greater suppression of Ki67 with anastrozole compared with tamoxifen is particularly apparent in tumors with low ER, but numbers in these subgroups makes this comparison uncertain.
It has long been considered that PgR positivity is an indicator of a greater likelihood of response to antihormonal agents.16 The mechanism for this has generally been thought to be that PgR positivity represents an intact estrogen-signaling pathway.17 More recently, it has been found that growth factor stimulation can lead to downregulation of PgR in breast cancer cells18 and this may provide a complicating factor in this interpretation of the interaction. Consistent with this, pretreatment PgR positivity in the IMPACT trial was associated with a greater degree of Ki67 expression in the overall group and in each of the subgroups. The overall greater suppression of Ki67 by anastrozole compared with tamoxifen or the combination was evident in each of the PgR subgroups, with an indication that the differential effect was somewhat greater in PgR-positive tumors. This is not consistent with the recent observation in the ATAC trial that anastrozole showed greater benefit relative to tamoxifen in PgR-negative compared with PgR-positive patients.19 The explanation for this is not clear at this time.
We have previously reported that HER-2 positivity is associated with a reduced suppression of Ki67 in an overview of different hormonal therapies.20 A minority of patients from the IMPACT study was included in that analysis before unblinding. The 12-week therapy data presented here are consistent with that report, although the data at 2 weeks do not show a significant difference between HER-2-positive and -negative disease. In the letrozole trial referred to above, there was a substantially greater suppression of Ki67 by the aromatase inhibitor compared with tamoxifen after 4 months of treatment in the HER-2 and/or HER-1-positive group of patients, a result that is consistent with the markedly greater clinical response to letrozole in that subgroup. In the present study, the difference between the treatments in the pretreatment values of Ki67 made unbiased comparison between them difficult, and reveals the necessity for adequate powering for the valid comparison of subgroups. It is interesting to note, however, that the mean suppression of Ki67 by anastrozole after 2 weeks was greater than that with tamoxifen (71% v 25%), but by 12 weeks the suppression was similar, with an indication that there was an increase from 2 to 12 weeks in the anastrozole arm. At a clinical level in the IMPACT trial, there were more HER-2-positive responders in the anastrozole arm than in the tamoxifen arm (seven of 12 v two of nine, respectively; P = .09).5 It is possible that despite this, the increase in Ki67 is indicative of an early escape from control by the aromatase inhibitor. Response rate in the neoadjuvant setting will not be informative in this regard since clinical change is expected to follow biologic changes. Evaluation of time to relapse may eventually be helpful but will have little statistical power in this small subgroup. We will also evaluate the possibility of early escape of control by the HER-2-positive patients in ongoing studies that combine a type 1 growth factor receptor tyrosine kinase inhibitor with anastrozole in the neoadjuvant setting.
The changes in ER expression produced by anastrozole were not significant at either 2 or 12 weeks. In a randomized trial of vorozole versus tamoxifen,2 there was a decrease in ER values with both drugs, and this was also found in a study comparing letrozole and tamoxifen.3 It is notable that pretreatment ER H-scores were lower in the vorozole study compared with the current study, and this may have contributed to the difference, possibly because of a different sensitivity of the ER assay between the two studies. Comparability is more difficult with the letrozole study since that employed the Allred scoring system. The increase in the homogeneity of scores after treatment with tamoxifen and the combination has not been reported before. It appears to result from what might be described as a “funnel” effect, in which most high values decrease and some low values increase. The consistency of the observation at 2 and 12 weeks in both the tamoxifen- and combination-treated groups supports its significance. It is possible that this might relate to the mixed agonist activity of tamoxifen, with differing effects according to ER levels. For example, high ER expression may be indicative of a higher estrogenic environment in which the antagonist activity of tamoxifen would be more apparent. This is consistent with the greater Ki67 suppression in high ER-positive tumors.
The marked increase in PgR levels with tamoxifen after 2 weeks contrasts greatly with the highly effective suppression of PgR in most patients receiving anastrozole. These results are very similar to those that we previously published on vorozole and tamoxifen. In the present study, we additionally found that in the anastrozole arm the decrease in PgR was significantly correlated with the decrease in Ki67, suggesting that the dependence of the PgR on estrogen in a particular tumor relates to the effectiveness of estrogen deprivation. It was remarkable, however, to find that the increase in PgR seen with tamoxifen after 2 weeks was related to the decrease in Ki67. Thus, although it seems reasonable to consider the increase in PgR as indicative of an agonist effect on that gene, it clearly does not determine a poor antiproliferative effect in these same tumors. Rather, the increase appears to be indicative of the degree of biologic responsiveness of the tumor. This phenomenon of the early increase in PgR being related to greater responsiveness of a breast carcinoma was previously noted by Howell et al,21 with the responsiveness in that case being according to clinical parameters.
It was also notable that although a small number of patients showed a numerical increase in Ki67 levels when treated with tamoxifen or the combination, there were very few in whom this was a meaningful increase. Thus, the data provide little evidence for tamoxifen operating overall as a full estrogen agonist in these tumors.
In summary, these investigations confirm the attractiveness of the neoadjuvant model for bioclinical investigations, a key scenario for translational research. The relationship between the changes in Ki67 in the IMPACT trial and the disease-free survival outcome from the ATAC study provides support for the value of Ki67 as a marker of the biologic efficacy of treatment. Ki67 is likely to become increasingly used as an end point in neoadjuvant and advanced disease studies. It is also likely to gather a prominent role as an end point in the further exploration of molecular determinants of response and resistance to endocrine and other targeted therapies of breast cancer, and possibly other malignancies.
Appendix
The IMPACT Trialists Group
W.H. Allum, S. Ashley, A. Bradley, I. Boeddinghaus, D. Brett, G. Gui, J. Diggins, J. Holborn, A. Ring, N. Sacks, C. Shannon, I. Smith, G. Walsh, Royal Marsden Hospital, London, UK; S. Detre, M. Dowsett, M. Hills, J. Salter, Royal Marsden Laboratory, London, UK; S. Ebbs, J. Kember, C. Chu, Mayday University Hospital, Croydon, UK; I. Batty, K. Kazim, A. Skene, Royal Bournemouth Hospital, Bournemouth, UK; J.M. Dixon, J. Murray, L. Renshaw, Western General Hospital, Edinburgh, UK; F. McNeill, K. Rooke, Essex County Hospital, Essex, UK; C. Griffith, J. Bevington, Royal Victoria Infirmary, Newcastle, UK; A. Evans, M. Pidgley, Poole General Hospital, Poole, UK; J.-U. Blohmer, W. Lichtenegger, Universitätsklinikum Charité, Berlin, Germany; P. Sauven, K, Rooke, Chelmsford & Essex Centre, Chelmsford, UK; C. Holcombe, K. Makinson, Royal Liverpool University Hospital, Liverpool, UK; L. Barr, N.J. Bundred, T. Pritchard, University Hospital of South Manchester, Manchester, UK; N. Harbeck, Frauenklinik der TU München, München, Germany; J. Clarke, J. Mansi, St. George's Hospital, London, UK; H. Stehle, Marienhospital, Stuttgart, Germany; T. Reimer, Universitäts-Frauenklinik, Rostock, Germany, K. Brunnert, Zentrum für Senologie und Plastische Chirurgie, Osnabrück, Germany; M. Lansdown, J. Hepper, St. James's University Hospital, Leeds, UK; D. Dubois, H. Stansby, Portsmouth Oncology Centre, Portsmouth, UK; and Z. Rayter, Bristol Royal Infirmary, Bristol, UK.
AZ Scientific Team
Peter Barker, Stephen Bird, Phil Davies, Jo Diver, Sonia Harris, Karen Langfeld.
Authors' Disclosures of Potential Conflicts of Interest
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. Employment: Stephen Francis, AstraZeneca. Consultant/Advisory Role: Mitch Dowsett, AstraZeneca; Ian E. Smith, AstraZeneca. Honoraria: Mitch Dowsett, AstraZeneca; Ian E. Smith, AstraZeneca; J. Michael Dixon, AstraZeneca. Research Funding: Mitch Dowsett, AstraZeneca; Ian E. Smith, AstraZeneca. For a detailed description of these categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.