Neoadjuvant endocrine therapy has been increasingly employed in clinical practice to improve surgical options for postmenopausal women with bulky hormone receptor-positive breast cancer. Recent studies indicate that tumour response in this setting may predict long-term outcome of patients on adjuvant endocrine therapy, which argues for its broader application in treating hormone receptor-positive disease. From the research perspective, neoadjuvant endocrine therapy provides a unique opportunity for studies of endocrine responsiveness and the development of novel therapeutic agents.

In an early study by Gazet et al (2001), 13 premenopausal women with ER+ breast cancer received neoadjuvant goserelin, a gonadotropin-releasing hormone (GnRH) analogue. At 3 months, seven of the 13 women had an overall response by clinical assessment, suggesting that premenopausal women may also benefit from neoadjuvant endocrine manipulation (Gazet et al, 2001). Torrisi et al (2007) investigated the use of letrozole and a GnRH analogue as primary therapy in premenopausal women with ER+ breast cancer. These patients received a GnRH analogue for a median of 5.2 months and letrozole for a median of 4 months. In the 32 evaluable patients, one achieved a pathological complete response (pCR) and 15 obtained a clinical and imaging partial response (Torrisi et al, 2007). These studies suggest that neoadjuvant endocrine therapy is effective in selected premenopausal women with ER+ breast cancer and further study in this patient population is needed.

Oestrogen receptor positivity remains the single most important criterion for eligibility for neoadjuvant endocrine therapy. Patients with an ER Allred score of 6 and above are the most likely to respond (Ellis et al, 2001). The data on PgR status does not support its use as a selection criterion in the context of ER positivity, and PgR+ ER− tumours are too uncommon to make a definitive statement with regard to their management (Ellis et al, 2001). HER2 status has been investigated in both the IMPACT and the P024 studies. Although the presence of HER2 amplification does not preclude a meaningful response to an AI in the neoadjuvant setting (Ellis et al, 2001; Smith et al, 2005), HER2 positivity is associated with a lower suppression of Ki67, a marker of cell proliferation, in response to either tamoxifen or letrozole (Ellis et al, 2006), suggesting treatment resistance to endocrine therapy alone. In practice, most patients with HER2+ tumours receive trastuzumab in combination with a chemotherapy regimen because of the high pCR rate. Although the combination of trastuzumab and an AI has shown promising results in patients with metastatic disease (Marcom et al, 2007; Kaufman et al, 2009), it has not been adequately tested in the neoadjuvant setting.

Neoadjuvant chemotherapy studies have firmly established that response, and particularly pCR, is a strong predictor of survival in breast cancer patients. However, in the subset of patients with ER+ HER2− disease treated with chemotherapy, pCR has limited value, as these patients often do well despite the absence of pCR because of effective adjuvant endocrine therapy (Buchholz et al, 2001). Pathological CR is very uncommon in response to neoadjuvant endocrine therapy. In three relatively large studies with letrozole, the pCR rate was no more than 1% (Eiermann et al, 2001; Baselga et al, 2009; Olson et al, 2009). Although residual cancer burden after neoadjuvant chemotherapy has been shown to predict distant relapse-free survival (Symmans et al, 2007), the application of this measure to patients who underwent neoadjuvant endocrine therapy remains to be evaluated. To identify alternative post-treatment factors that predict breast cancer survival after neoadjuvant endocrine therapy, Dowsett et al (2007) examined Ki67 expression before and after 2 weeks of endocrine therapy. Patients with higher Ki67 expression after 2 weeks of endocrine therapy had a significantly lower recurrence-free survival (Dowsett et al, 2007). In a multivariable analysis conducted on the P024 trial (Ellis et al, 2008a), four factors were determined to have independent prognostic value for relapse and death after relapse. These included pathological tumour size (T1/2 vs T3/4), pathological node status (positive or negative) and two biomarkers in the surgical resection specimen, the natural logarithm of the Ki67 value and the ER status of the tumour. A prognostic score, the preoperative endocrine prognostic index (PEPI), was developed, which weighs each of these factors according to their associated hazard ratios (Table 2). The PEPI was then validated in an independent dataset from the IMPACT trial (Ellis et al, 2008a). No relapses were recorded in either trial in patients with T1, N0 tumours with a PEPI score of 0 (residual tumour with Ki67 index of 2.7% – natural logarithm of 1 – or less with maintained ER expression) or in the rare patient with a pCR. These patients are not likely to benefit from adjuvant chemotherapy as endocrine therapy alone appears to adequately control their disease. These results supported an amendment to the Z1031 trial (Cohort B), in which chemotherapy was not recommended to patients in the pathological stage 1/0 PEPI 0 category. The acceptability of this advice will be assessed to see if this approach can be made a protocol mandate in the next phase of clinical trial development. In addition, patients with a high Ki67 proliferation index in a 2- to 4-week biopsy (>10%) are triaged to neoadjuvant chemotherapy or immediate surgery, as these tumours are exhibiting primary endocrine therapy resistance (Dowsett et al, 2007). To determine the chemotherapy responsiveness of this group, the rate of pCR with neoadjuvant chemotherapy in the high on-treatment Ki67 group will be determined (Figure 1). If this approach to tailoring neoadjuvant endocrine therapy is feasible, a larger, more definitive study will be considered.

Discovery genomics is a complex process that often requires fresh-frozen material, an adequate sample size, patient consent to genetic studies and access to sophisticated analysis techniques. To date, the number of publications on genomic profiling in the context of neoadjuvant endocrine therapy remains few and the sample sizes are small. For example, Miller et al (2009) reported on a series of 52 patients, 37 were considered letrozole sensitive and 15 resistant. A predictive model was developed that used all three types of gene expression variable: baseline, on-treatment and change in treatment, to differentiate between the two groups. This paper illustrates the variety of bioinformatics approaches that can be taken when before and on-treatment paired samples are available for analysis (Miller et al, 2009). Other investigations are ongoing to assess the predictive value of established molecular signatures, such as the Netherlands Cancer Institute NKI 70 gene (van de Vijver et al, 2002), 21 gene recurrence score (Paik et al, 2004) or the PAM50 (Parker et al, 2009), performed on tumour samples taken after the initiation of endocrine therapy (Ellis et al, 2008b). The Z1031 trial has recently completed accrual of 375 patients, despite mandatory fresh tumour tissue acquisition, through the provision of tissue acquisition kits and the active collaboration of surgical investigators. Z1031 patients are also consented for massive parallel DNA sequencing to define complete cancer genomes (Ding et al, 2010). Very detailed information on the molecular characteristics of endocrine therapy-resistant and -sensitive tumours is therefore a near-term prospect.

The superior efficacy of adjuvant aromatase inhibition vs tamoxifen in reducing the risk of recurrence was mirrored by advantages for aromatase inhibition as neoadjuvant therapy. This was most consistently seen with letrozole, with both the clinically based (Eiermann et al, 2001) and Ki67-based data (Ellis et al, 2003) correctly predicting that letrozole would be the more effective adjuvant treatment (Thurlimann et al, 2005) several years in advance of the actual result. Anastrozole was also shown to be a superior antiproliferative agent in comparison with tamoxifen in the neoadjuvant setting, and the failure of the anastrozole/tamoxifen combination arm to improve outcomes in the ATAC trial (Howell et al, 2005) was mirrored by the inferior antiproliferative response seen in the combination arm of the IMPACT trial in comparison with anastrozole monotherapy (Dowsett et al, 2005). These datasets provide strong arguments for the value of neoadjuvant endocrine therapy trials as a ‘proving ground' to test the efficacy of new endocrine therapy approaches for breast cancer. Phase 2 randomised studies of endocrine therapy and signal-transduction inhibitor combinations have been recently reported for gefitinib and anastrozole (Polychronis et al, 2005; Smith et al, 2007), and the rapamycin analogue, RAD001, with letrozole (Baselga et al, 2009). The combination of anastrozole and gefitinib was not found to be superior to anastrozole alone. The combination of RAD001 and letrozole showed improved antiproliferative response and clinical response by palpation at the expense of more adverse events. Perhaps, the most convincing evidence for progress with a novel endocrine therapy combination would be an enhanced rate of pCR as evidence for the activation of a cell death process that would more effectively delete ER+ breast cancer cells before the acquisition of resistance to endocrine therapy (Crowder et al, 2009) or an increased rate of PEPI score 0 following neoadjuvant therapy.

Neoadjuvant AI therapy is a clinically valuable, logical and increasingly accepted approach to neoadjuvant systemic therapy for postmenopausal women with ER+ HER2− breast cancer. With an array of new agents and predictive biomarkers to investigate, the time for further randomised trials to compare neoadjuvant endocrine treatment with conventional chemotherapy in a broad spectrum of patients with ER+ breast cancer has probably passed – particularly if the trial under consideration is not prospectively testing a predictive biomarker for endocrine responsiveness. For future investigations, we should also focus on the fact that for ER+ endocrine therapy-resistant disease, conventional chemotherapy may not be adequate treatment. We should therefore use the neoadjuvant setting to address two key issues: the development of more effective mechanism-based treatments for endocrine therapy-resistant disease; and the identification of ER+ breast cancers that can be managed without chemotherapy.