This is the first clinical study to evaluate the benefits of combining an HDAC inhibitor with an anti-oestrogen in patients with advanced ER-positive breast cancer. The rationale for this combination stems from extensive preclinical data suggesting epigenetic modulation and post-translational modification of the ER by HDAC inhibitors enhances the anti-tumour effects of tamoxifen (Yang et al, 2001
; Kawai et al, 2003
; Alao et al, 2004
; Jang et al, 2004
; Saji et al, 2005
; Kawai and Arinze, 2006
; Sharma et al, 2006
; Fiskus et al, 2007
; Zhou et al, 2007
; Bicaku et al, 2008
). We found that treatment of cultured breast cancer cells with vorinostat leads to downregulation and reversal of tamoxifen-induced stabilisation of the ER (Bicaku et al, 2008
). The anti-tumour activity of tamoxifen is primarily anti-proliferative. In the presence of an HDAC inhibitor, however, we find that tamoxifen induces apoptosis rather than growth arrest. Further studies suggest that this interaction is mediated through inhibition of HDAC2 (Munster et al, 2007
; Bicaku et al, 2008
; Marchion et al, 2009
). Select depletion of HDAC2 by small interfering RNAs mimics the effects of an HDAC inhibitor on the ER and its downstream signalling (Bicaku et al, 2008
). This led us to pursue a clinical trial evaluating the addition of an HDAC inhibitor to tamoxifen for the treatment of women with ER-positive breast cancer who had progressed on prior hormonal therapy.
Vorinostat was the first HDAC inhibitor approved by the FDA for the treatment of cancer, specifically cutaneous T-cell lymphoma. In addition to the effects in patients with cutaneous T-cell lymphoma, HDAC inhibitors appear to be active in Hodgkin's lymphoma and other haematological malignancies (Duvic et al, 2007
). In contrast, the anti-tumour effects of vorinostat in solid tumour malignancies have been less evident. Vorinostat has been tested as a single agent in patients with metastatic breast cancer. Although disease stabilisation was observed in 30% of the patients, no clinical responses were achieved (Luu et al, 2008
). Similarly, despite reported efficacy as single agents in several preclinical models, clinical benefit with HDAC inhibitors in several other solid tumour malignancies has been modest, limited mostly to disease stabilisation. Preclinical data from our laboratory and others suggest that HDAC inhibitors have the ability to re-sensitise tamoxifen-resistant cells to hormone therapy, and has been hypothesised to prevent the emergence of hormone therapy resistance (Yang et al, 2001
; Zhou et al, 2007
; Thomas et al, 2011
This study's findings suggest that the addition of the HDAC inhibitor vorinostat to tamoxifen results in durable responses (8 out of 43, 19%) and prolonged disease stabilisation (9 out of 43, 21%) in patients who had progressed on at least one prior aromatase inhibitor. Further, more than half of the patients had previously progressed on adjuvant tamoxifen. A significant proportion of the patients also received chemotherapy. The expected response rate in this patient population, based on the EFECT trial, is 7.6% for fulvestrant and 6.7% for exemestane (Chia et al, 2008
). Two more contemporary trials suggested that while stable disease is achievable, objective, and durable responses are rare. The first trial reported disease stabilisation, but no objective responses using estradiol in this setting (Ellis et al, 2009
). The second trial comparing tamoxifen plus placebo vs
tamoxifen plus EGFR inhibitor gefitinib demonstrated that tamoxifen and placebo treatment resulted in a 15% objective response rate in patients with newly diagnosed metastatic breast cancer or recurrent disease after adjuvant tamoxifen. However, in the stratum best comparable to the patient population in this study, patients recurring on or not responding to prior aromatase therapy, no objective responses were observed (Osborne et al, 2011
). While in this study, the observed anti-tumour activity of 19% confirmed partial responses by RECIST criteria in this heavily pre-treated patient population is therefore very encouraging. shows that responses were seen in patients after progression on two or three prior endocrine therapies and chemotherapy.
Furthermore, the correlative studies accompanying this clinical trial suggest that enrichment of responsive patients may be feasible. The pharmacological effects of vorinostat were evaluated in PBMCs. A statistically significant increase in histone H4 acetylation compared with baseline was observed in only 21 out of 36 (58%) patients (). This suggests that 15 out of 36 patients (42% of the evaluated patients) did not reach vorinostat plasma levels that were high enough to induce a change in histone acetylation, or did not express the appropriate histone target. Several studies with vorinostat have shown that histone acetylation occurs at lower concentrations than those required for the modulation of other targets. Our data suggest that day 8 histone acetylation is a strong predictor of response. We have shown previously that vorinostat-induced histone acetylation in PBMCs is comparable to histone acetylation in tumour cells (Munster et al, 2009a
). The mean increase in histone H4 acetylation for all patients in this study was 20% (95% CI: 14–26%) over baseline when treated with 400
mg of vorinostat. This may appear lower than has been previously reported when histone acetylation was measured by western blot analysis (Kelly et al, 2003
). The effects, however, were comparable to our findings from other studies where histone acetylation was measured by immunofluorescence. We found that the observed changes in histone acetylation induced by 400
mg oral vorinostat given daily in this study were comparable to the results observed in a dose escalation phase I trial evaluating vorinostat in combination with the anthracycline doxorubicin, conducted by our group (Munster et al, 2009a
). We reported a 25% increase in mean histone acetylation (95% CI: 3–47) on day 3 in patients receiving 400
mg of vorinostat, with mean plasma levels of 114
n (95% CI: 37–191). Patients treated with higher doses of the HDAC inhibitor resulted in a 20% increase in median H4 histone acetylation (95% CI: 29–70) at 600
mg of vorinostat, and 84% (95% CI: 54–114) for 800
mg. Corresponding mean vorinostat levels were 256
n (95% CI: 133–378) and 760
n (95% CI: 598–924), respectively (Munster et al, 2009a
To confirm that certain patients did not show a change in acetylation, we measured histone modification changes alternatively by western blot analysis. Although the range of histone acetylation measured by western blot analysis was numerically greater, a significant change in acetyl-H4 was only observed in 57% of the patients, comparable to the findings of the immunofluorescence analysis.
Furthermore, as shown in previous studies, baseline expression of HDAC2 is positively correlated with a change in histone H4 acetylation (), suggesting HDAC2 is a potential biomarker and important pharmacological target of vorinostat.
Responders were also more likely to exhibit elevated histone acetylation following treatment and increased baseline HDAC2 expression (; ). An increase in histone acetylation was observed in 13 of 14 evaluable responders (). Thus, by identifying hyperacetylators at the initiation of treatment, patients most likely to benefit from this treatment would be enriched. Furthermore, determining initial acetylation response may provide direction for vorinostat dose modification.
Vorinostat and tamoxifen treatment was well tolerated by many of the patients and long-term exposure of up to 2 years was feasible. A subgroup of patients required dose modifications, however, due to grade 3 and 4 toxicities. In addition to myelosuppression, the predominant toxicities were fatigue and anorexia. These toxicities have been well described for vorinostat, and are less likely due to an interaction between vorinostat and tamoxifen. In the fourth week of each cycle (tamoxifen alone treatment), patients reported temporary alleviation from nausea, fatigue, anorexia, and myelosuppression. Notably, dose modification did not appear to compromise response, as 8 of the 17 patients with clinical benefit were dose adjusted, and maintained their response. Observed toxicities did not correlate with changes in histone acetylation (), suggesting they may be the result of off target effects.
In support of our preclinical findings, baseline HDAC2 expression in patients' PBMCs correlated with the degree of histone acetylation. Thus, the assessment of baseline HDAC2 expression may predict a patient's molecular response. Patients with low HDAC2 expression, or those who do not show a change in acetylation, could then be removed from study or receive higher doses of the HDAC inhibitor if feasible.
In summary, this trial suggests that the addition of vorinostat to tamoxifen in patients with hormone receptor-positive breast cancer results in tumour regression or prolonged disease stabilisation in 40% of the patients who had progressed on prior hormonal therapy and chemotherapy. Although the current study may be limited by its sample size, the results are greater than those of contemporary studies testing endocrine therapy in second- and third-line therapy with response rates reported in <10% of the patients, or no responses when measured by RECIST criteria (Chia et al, 2008
). However, a randomised trial is required to determine the effects of the combination over the potential efficacy of tamoxifen alone. Pharmacodynamic assessment of vorinostat-induced histone H4 acetylation and HDAC2 expression at baseline was strong predictors of biological activity and clinical benefit. This suggests that the absence of histone acetylation could be used as an early negative predictor for patients who are not likely to benefit. These patients could then be removed from study, or be treated with a higher dose. The observed anti-tumour efficacy warrants further testing of HDAC2 inhibitors and hormonal therapy, yet the development of rapid acetylation bioassays and assessment of baseline HDAC2 expression in tumours in future studies may provide a feasible method to enrich for patients more likely to benefit.