Adjuvant Sequencing of Tamoxifen and Anastrozole Is Superior to Tamoxifen Alone in Postmenopausal Women with Low Proliferating Breast Cancer

Assessment of tumor proliferation plays a central role in the stratification of breast cancer into biologically relevant, distinct subtypes. One of the most feasible and direct ways to measure tumor proliferation is the immunohistochemical determination of Ki67 expression. Ki67, a proliferation marker, is present solely in cycling cells. In recent years, this marker has been drawn into the spotlight of interest as a prognostic factor and predictor of therapy response in early breast cancer (1–9). Several studies have shown that in hormone receptor-positive, early breast cancer, Ki67 serves as a prognostic factor in both pre- and postmenopausal women. In addition, high expression of Ki67 predicts response to chemotherapy in the neoadjuvant setting, and dynamic changes in Ki67 expression have also been found to correlate with response to endocrine therapy with aromatase inhibitors (5, 10–12).

Ki67 is a crucial factor that distinguishes luminal A from luminal B breast cancers with a simple methodology. Luminal A cancers, which bear a better prognosis, possess a lower fraction of Ki67-expressing tumor cells, in contrast to the luminal B subtype with a higher fraction of cycling cells, which indicated worse prognosis (8). However, the optimal therapeutic approaches in the adjuvant treatment of luminal B breast cancer remain to be determined. One central question is whether these patients benefit from endocrine treatment with aromatase inhibitors over tamoxifen (TAM).

It has been reported that high Ki67 labeling index might be predictive of benefit from letrozole over TAM in postmenopausal patients with hormone receptor–positive breast cancer (7). In this study, we investigated the clinical relevance of immunohistochemically determined Ki67 expression in terms of recurrence-free survival (RFS) and overall survival (OS) in postmenopausal women with estrogen receptor (ER)-positive breast cancer and limited risk who were enrolled in the Austrian Breast and Colorectal Cancer Study Group (ABCSG)-8 trial and were prospectively randomized to 5 years of adjuvant TAM or 3 years of anastrozole (ANA) after 2 years of TAM, in the absence of adjuvant chemotherapy. We hypothesized that sequencing of TAM and ANA may have differential impact on RFS and OS in patients with low and high proliferating breast cancers, respectively.

The present investigation is part of the ABCSG translational research program (abcsg.research). All patients included in this study had participated in the large phase III trial ABCSG-8. The study design, inclusion criteria, and the main results of the trial have been reported elsewhere (13). The objective of ABCSG-8 was to determine whether the sequence strategy of TAM followed by ANA improves the outcome of postmenopausal women with hormone receptor–positive, early-stage breast cancer. Briefly, a total of 3,714 postmenopausal breast cancer patients had been randomly assigned to receive either TAM for 5 years or TAM for 2 years followed by ANA for 3 years. Patients included in ABCSG-8 did not receive neoadjuvant or adjuvant chemotherapy as well as trastuzumab. The primary endpoint was RFS defined as time from randomization to the earliest occurrence of local or distant recurrence or death from any cause.

The sequence strategy of 2 years of TAM followed by 3 years of ANA leads to moderate outcome benefits and lends further support to the inclusion of ANA during the first 5 years of endocrine therapy.

A block containing representative formalin-fixed, paraffin-embedded tumor tissue was available from 1,587 of 3,714 (43%) patients enrolled into ABCSG-8 and sent to the central lab of abcsg.research at the Medical University of Vienna. A flowchart illustrating patient and sample selection is shown in Fig. 1. In total, 19 centers contributed tumor samples (see appendix). All tumor specimens were obtained at the time of surgery before adjuvant therapy. Paraffin blocks were stored at room temperature and were identifiable only by an identification number assigned to each patient at randomization. Approval was obtained from the local Institutional Review Boards. From each tumor block sections were cut at 4 μm. One section was stained by hematoxylin and eosin to confirm the presence of invasive carcinoma histologically and further sections were used for immunohistochemical analyses as described previously.

Immunohistochemistry was done and evaluated centrally in the abcsg.research central laboratory at the Medical University of Vienna by using a standardized protocol: Briefly, tissue sections were deparaffinized and rehydrated. After heating for 10 minutes in 10 mmol/L citrate buffer (pH 6.0) in a pressure cooker for epitope retrieval, the tissue sections were incubated for 30 minutes at room temperature with a mouse monoclonal antibody specific for Ki67 (clone MIB1, Dako, dilution 1:100). Antibody binding was detected by means of the UltraVision LP detection system according to the manufacturer's recommendations (Lab Vision Corporation). Color development was done by 3,3′-diaminobenzidine and counterstaining by hematoxylin. HER2 status was centrally assessed by immunohistochemistry. In case of HER2 2+ scores, FISH analyses were used to determine amplification.

Ki67 and HER2 immunostaining was independently assessed by experienced breast pathologists (Z.B.-H., M.R.), who were unaware of the clinical data of the patients. For Ki67, invasive tumor cells in 20 representative HPF (400× magnification) were visually evaluated and only nuclear staining was scored as positive. The results were documented as the percentage of Ki67-stained nuclei regardless of staining intensity. Comparisons of Ki67 expression with clinical variables including survival of the patients were done with Ki67 expression as a continuous variable and as a dichotomized variable classified as low (≤10%) and high (>10%). This cut-off of 10% was previously used in several series published by others (1, 14).

The primary endpoint of this analysis was RFS defined as time from randomization to the first occurrence of local, regional, or distant recurrence or death from any cause, as in the ABCSG-8 trial. OS defined as time from randomization to death from any cause was a secondary endpoint.

To be consistent with the clinical trial data presentation and to maintain consistency, we censored all patients at the time when they were allowed to enter ABCSG-16 (Secondary Adjuvant Long Term Study With Arimidex; SALSA, NCT00295620) or at 5 years. In addition, patients who selectively crossed over to the sequential treatment arm were censored at the time of cross-over (15).

Baseline characteristics of patients with or without tumor blocks were compared using the χ² test. Baseline data according to dichotomized Ki67 status (≤10%, >10%) were compared in univariate analyses using the χ² test and in a multiple logistic model. Survival rates were estimated with the use of the Kaplan–Meier method. The prognostic value of breast cancer subtypes was studied using univariate and multiple Cox models. All multiple regression models (i.e., logistic and Cox) mentioned below were adjusted for age, tumor size, lymph node status, tumor grade, ER expression, progesterone receptor (PgR) expression, HER2, and adjuvant endocrine treatment. All reported P values are results of 2-sided tests. A P value equal to or less than 5% was considered statistically significant. All statistical analyses were done using of SPSS software, version 15.0 (SPSS Inc.).

We centrally assessed immunostaining of Ki67 on whole-tumor tissue sections of 1,587 patients. As shown in Table 1, these 1,587 patients had similar prerandomization characteristics as the clinical trial. Of the 1,587 tumors, 394 (23%) had a high (>10%) Ki67 expression. The median numeric value of Ki67 expression was 10% in all 1,587 patients and also 10% in both treatment arms of the clinical trial.

The association of Ki67 with clinical variables is shown in Table 1. High Ki67 was associated with larger tumors, higher tumor grade, lower PgR expression levels, and HER2 positivity in univariate analyses. In addition, we compared baseline data according to Ki67 with a multivariate logistic model. Ki67 positivity was studied simultaneously according to age, tumor size, nodal status, tumor grade, ER expression, PgR expression, HER2 expression, and adjuvant endocrine treatment. In these analyses, high Ki67 remained associated with tumor size (OR = 1.35, 95% CI: 1.06–1.72; P = 0.01), tumor grade (OR = 1.32, 95% CI: 1.12–1.54; P = 0.001), PgR expression (OR = 0.84, 95% CI: 0.75–0.94; P = 0.002), and HER2 expression (OR = 2.75, 95% CI: 1.80–4.20; P < 0.0001).

As in the ABCSG-8 trial, all patients were censored at 60 months follow-up irrespective of eventual extended adjuvant treatment. Therefore, the median follow-up in this analysis is 60 months. Within the first 5 years of follow-up, 157 of 1,587 (9.9%) patients had local, regional, or distant recurrences, 98 of 1,587 (6.2%) patients had died. The 5-year RFS and OS rates were 89.5% and 93.5% in this study population, respectively. RFS and OS were significantly shorter for patients with tumors with high Ki67 (HR for recurrence = 2.28, 95% CI: 1.66–3.13, P < 0.0001; HR for death = 2.15, 95% CI: 1.44–3.22, P = 0.0002), confirming the prognostic value of Ki67 in this cohort (Fig. 2A and B). Five-year RFS and OS estimates were 92% and 95% for low versus 83% and 89% for high Ki67 labeling index, respectively. In addition to Ki67, age, tumor size, nodal status, and PgR expression were also significantly associated with RFS in the univariate analyses (Table 2). Age, tumor size, and nodal status were associated with OS as well. The Cox models were adjusted for these variables plus tumor grade, ER expression, and HER2 expression. In these multivariate analyses, Ki67 was still significantly associated with RFS (adjusted HR for recurrence = 1.90, 95% CI: 1.37–2.64, P < 0.0001) and OS (adjusted HR for death = 1.78, 95% CI: 1.18–2.70, P = 0.006) of the patients (Table 2). Similar results were obtained when Ki67 was included as continuous variable into the Cox model (adjusted unit HR for recurrence = 1.03, 95% CI: 1.02–1.04, P < 0.0001; adjusted unit HR for death = 1.02, 95% CI: 1.01–1.04, P = 0.007). Thus, Ki67 is an independent poor prognostic factor.

The interaction of Ki67 and treatment benefit was evaluated in a Cox model adjusted for age, tumor size, nodal status, tumor grade, ER expression, PgR expression, and HER2 in the total patient population and in women with medium or high ER expressing breast cancer. In the subgroup analysis of patients with medium or high ER tumors, ER was not included into the Cox model. In all 1,587 patients, the test for interaction between Ki67 and treatment on RFS was marginally significant (P = 0.049), but it was not significant for OS (P = 0.79). These data suggest that adjuvant sequencing of TAM and ANA leads to a significantly prolonged RFS compared with TAM alone in postmenopausal women with low Ki67 (adjusted HR for recurrence = 0.61, 95% CI: 0.40–0.92, P = 0.02) but not in high Ki67 disease (adjusted HR for recurrence = 1.18, 95% CI: 0.72–1.93, P = 0.52). This treatment-by-marker effect was even more pronounced in the 1,438 women with medium or high ER breast cancer (test for interaction p = 0.03). TAM followed by ANA was superior to TAM alone in postmenopausal women with medium or high ER tumors and low Ki67 (adjusted HR for recurrence = 0.53, 95% CI: 0.34–0.83, P = 0.005) but not in high Ki67 disease (adjusted HR for recurrence = 1.18, 95% CI: 0.66–1.89, P = 0.68; Fig. 3). RFS at 5 years was 89% (TAM) and 94% (TAM"ANA) for low and 83% (TAM and TAM"ANA) for high Ki67 labeling index, respectively (Fig. 3).

In this study, we confirmed that high Ki67 is associated with both shorter RFS and OS in early-stage breast cancer. Thus, high Ki67 is an independent poor prognostic factor in women with breast cancer who received endocrine therapy which has been previously described (5, 7, 9). Breast carcinomas with high Ki67 expression are most likely luminal B type cancers and may have, in addition to estrogen-driven proliferation, estrogen-independent pathways activated. This might be one of the reasons why some of these tumors are intrinsically resistant to endocrine therapy (8). Moreover, we observed a significant treatment-by-marker effect. Therefore, Ki67 is also a predictive marker in ABCSG-8 patients, suggesting that adjuvant sequencing of TAM and ANA may be superior to TAM alone, particularly in postmenopausal women with medium or high ER, low proliferating breast cancer.

The BIG 1–98 findings suggested that Ki67 may have predictive value among postmenopausal women with endocrine-responsive early breast cancer (7). Comparison of letrozole and TAM by Ki67 revealed that letrozole may be particularly beneficial at higher levels of Ki67, which is in contrast to our results.

There may be several reasons behind this apparent difference. The positive interaction between high proliferation and letrozole in BIG 1–98 was described in a monotherapy setting (letrozole for 5 years vs. TAM for 5 years), whereas ABCSG-8 tested the interaction in a sequence of ANA after 2 years of TAM. Furthermore, the benefit of ANA is carried over the 60 months of endocrine treatment. A recent update of the ATAC trial clearly indicates that the reductions in risk are sustained well beyond the 5 years of treatment (16). Our analysis does not account for recurrences after 5 years and merely represents the early outcome of these women.

A comparison of characteristics of patients from ABCSG-8 with BIG 1–98 for whom Ki67 was available showed that 72% of ABCSG-8 patients were node negative and 72% had tumors below 2 cm. In the BIG 1–98 subpopulation, a rate of 65% node-negative patients and 65% of tumors below 2 cm was observed. Furthermore, ABCSG-8 was the only large endocrine treatment trial that excluded G3 tumors from the trial population, whereas in BIG 1–98, 16% of patients had G3 tumors. Another important difference may be that ABCSG-8 did not allow chemotherapy, whereas in BIG 1-98, 26% of patients received preoperative or adjuvant chemotherapy. Although the Cox proportional–hazards regression analysis in BIG 1–98 was stratified according to chemotherapy use, this could be of importance because it has been shown that high Ki67 is associated with a better outcome of chemotherapy, thus suggesting that patients with highly proliferating tumors may be undertreated with the adjuvant regimens used in ABCSG-8 (10–12).

The median value of 10% for Ki67 in ABCSG-8 is comparable with the median of 11% in BIG 1–98 and similar to the 10% cut off used in other studies (1, 14). In both trials, ABCSG-8 and BIG 1–98, a correlation between high Ki67 and adverse prognostic factors such as larger tumor size, positive nodal status, higher tumor grade, and HER2 positivity was observed.

The prognostic and predictive value of Ki67 was also assessed in 2 randomized trials of the International Breast Cancer Study Group (IBCSG Trials VIII and IX) that compared adjuvant chemoendocrine therapy with endocrine therapy alone in patients with node-negative breast cancer (9). In postmenopausal women, high Ki67 was associated with other factors that predict poor prognosis such as larger tumor size, higher tumor grade, and the absence of ER and PgR expression and HER2 overexpression. Among patients with endocrine-responsive tumors, high Ki67 was associated with worse disease-free survival, but Ki67 did not predict the relative efficacy of chemoendocrine therapy compared with endocrine therapy alone. Thus, Ki67 was an independent prognostic factor but was not predictive of better response to adjuvant chemotherapy in these studies.

The present findings suggest that adjuvant sequencing of TAM and ANA is particularly beneficial in lowest risk patients with a low Ki67 labeling index. If these results are confirmed by other translational studies from adjuvant endocrine or neoendocrine trials, this would suggest that estrogen depletion via aromatase inhibition is a more effective treatment in breast cancers with high dependence on estrogen metabolites. This is also supported by recent findings of the TAM and Exemestane Adjuvant Multinational (TEAM) pathology substudy, which showed a trend toward benefit in favor of exemestane compared with TAM in patients with breast tumors expressing high ER levels (17). In patients with higher proliferation (associated with larger tumor size, higher tumor grade, and lower ER expression), the type of endocrine treatment may be less relevant.

However, the optimal adjuvant therapy in this patient population with higher risk remains currently unknown and has to be determined in well-designed prospective clinical trials. Thus if our results are confirmed, Ki67 may be another useful biomarker to identify postmenopausal women with endocrine-responsive early breast cancer with low risk. In addition, this population may have particular benefit from adjuvant sequencing of TAM and ANA. Finally, an individual patient level pooling project adding data from all aromatase inhibitor translational studies may have sufficient power and diversity of subjects to build and validate a really persuasive predictive model.

The following investigators also participated in this study: T. Bachleitner-Hofmann, S. Taucher, U. Pluschnig, R. Bartsch, G. Steger, C. Zielinski, M. Seifert, C. Dadak, R. Obwegeser, E. Kubista, E. Asseryanis, R. Möslinger-Gehmayr, E. Hanzal, C. Sam (Medical University of Vienna, Vienna); P. Mayer, C. Menzel, B. Mlineritsch, M. Moik, C. Rass, R. Reitsamer, G. Russ (Salzburg Hospital, Salzburg); K. Haider, W. Kwasny, E. Markis, W. Stiglbauer (Wiener Neustadt Hospital, Wiener Neustadt); R. Grill, M. Medl, A. Nader, M. Stierer (Hanusch Hospital, Vienna); S. Redtenbacher, W. Ulrich (Hietzing Hospital, Vienna); G. Reiner (Weinviertel Hospital, Mistelbach); F. Hofbauer, M. Lang (Oberpullendorf Hospital, Oberpullendorf); W. Horvath (Güssing Hospital, Güssing); P. Kier, A. Reiner-Concin, K. Renner (Sozialmedizinisches Zentrum Ost Hospital, Vienna); P. Balcke, G. Salem, R. Sedivy (St. Pölten Hospital, St. Pölten); E. Nessler, W. Neunteufel (Dornbirn Hospital, Dornbirn); A. Haid, F. Offner (Academic Teaching Hospital Feldkirch, Feldkirch); F. Kugler, G. Michlmayer, R. Schildberger, W. Sega (Sisters of Mercy Hospital, Linz); F. Fridrik, R. Greul, G. Hochreiner, G. Syré, C. Tausch, G. Wahl (Linz Hospital, Linz); A. Chott, H. Hartleb, H. Ludwig, H. Salzer (Wilhelminenspital, Vienna); R. Kocher, G. Leitner, H. Rabl, F. Stangl, R. Stering (Leoben Hospital, Leoben); E. Forsthuber, K. Holzmüller, J. Omann, H. Rogatsch, A. Urbania (Klagenfurt Hospital, Klagenfurt); H. Feichtinger, M. Rottmann, J. Schüller (Rudolfstiftung Hospital, Vienna); O. Braun, F. Burger (Waldviertel Hospital, Horn); G. Bayer-Tiefenbach, K. Mach (Oberwart Hospital, Oberwart); B. Gebhard, W. Rausch (Vöcklabruck Hospital, Vöcklabruck); W. Öhlinger, R. Punzengruber (Amstetten Hospital, Amstetten); W. Döller, E. Melbinger (Wolfsberg Hospital, Wolfsberg); H. Brustmann, P. Riss (Mödling Hospital, Mödling).

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.