- Roland B. Walter1,2⇓,
- Bruno C. Medeiros3,
- Bayard L. Powell4,
- Charles A. Schiffer5,
- Frederick R. Appelbaum1,6 and
- Elihu H. Estey1,2
- 1Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA
- 2Department of Medicine, Division of Hematology, University of Washington, Seattle, WA
- 3Stanford University, Stanford, CA
- 4Comprehensive Cancer Center of Wake Forest University, Winston-Salem, NC
- 5Barbara Ann Karmanos Cancer Center, Wayne State University, Detroit, MI
- 6Department of Medicine, Division of Medical Oncology, University of Washington, Seattle, WA, USA
- Correspondence: Roland B. Walter, MD PhD, Clinical Research Division, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, D2-190, Seattle, WA 98109-1024, USA. Phone: international +1.206.6673599. Fax: international +1.2066676519. E-mail:
Histone deacetylase inhibitors such as vorinostat enhance gemtuzumab ozogamicin efficacy in vitro. We, therefore, investigated vorinostat+gemtuzumab ozogamicin for adults aged 60 years and over with untreated acute myeloid leukemia. We stratified patients into 2 groups (group 1: patients aged ≥70 years and performance status 2–3; group 2: aged 60–69 years with performance status 0–3 or aged ≥70 years and performance status 0–1). Responses were monitored separately in group 2 patients with normal or favorable cytogenetics (group 2A) and other cytogenetics (group 2B). Among 31 patients, 6 (19.4%) achieved complete remission, and one (3.2%) achieved complete remission with incomplete platelet recovery; these patients had a higher median overall survival than non-responders (553 vs. 131 days, P=0.0026). Response rates were: group 1, one of 10 (10.0%); group 2A, 6 of 13 (46.2%); and group 2B, none of 8 (0%). These data indicate that vorinostat+gemtuzumab ozogamicin has activity that is mostly confined to patients with normal karyotype disease. ClinicalTrial.gov: NCT00673153.
Age remains an important risk factor for poor outcome in acute myeloid leukemia (AML).1–3 Therefore, novel strategies are needed for older adults, especially for those with poor performance status (PS) who are at high risk of life-threatening or fatal toxicities from intensive therapy. This need for new therapies is reflected by the general recommendation that all older patients with AML should receive investigational treatments whenever possible; this call is supported by current guidelines, e.g. by the National Comprehensive Cancer Network (NCCN).4 One approach uses monoclonal antibodies, most notably with gemtuzumab ozogamicin (GO), an immunoconjugate between an anti-CD33 antibody (hP67.6) and a toxic calicheamicin-γ1 derivative.5 However, when used alone for adults with untreated or relapsed AML, rates of complete remission (CR) and CR with incomplete platelet recovery (CRp) have typically not exceeded 25–35%.6–8 Research has, therefore, focused on developing means to improve GO monotherapy.
hP67.6 facilitates cellular uptake of the calicheamicin-γ1 moiety which, after cleavage from the antibody, causes DNA damage, apoptosis, and cell death.5 This putative mechanism suggests that drugs that augment the DNA damage response and lower the apoptotic threshold could improve GO efficacy. Histone deacetylase inhibitors (HDACis) are of interest in this regard. There is increasing evidence to suggest that HDACis, such as vorinostat, function synergistically or additively with conventional anti-AML chemotherapeutics, partly due to lowering the apoptotic threshold of tumor cells.9 Furthermore, recent in vitro studies indicate that HDACis strongly increase DNA intercalation of the calicheamicin-γ1 derivative and enhance DNA degradation and GO efficacy.10 These findings prompted a phase II study of vorinostat as chemosensitizer with GO in patients aged 60 years and over with untreated AML.
Design and Methods
Patients were eligible if they had untreated primary or secondary AML (other than acute promyelocytic leukemia) according to the 2008 World Health Organization classification.11 Other inclusion criteria included: an Eastern Cooperative Oncology Group (ECOG) PS of 0–3; a white blood cell (WBC) count <10×109/L; total bilirubin ≤2.5 × Upper Limit of Normal (ULN) unless elevation was due to hepatic infiltration by AML, Gilbert’s syndrome, or hemolysis; SGOT/SPGT ≤1.5 × ULN unless elevation was due to hepatic infiltration by AML; serum creatinine ≤1.5 × ULN; and left ventricular ejection fraction of 40% or over. Exclusion criteria were: diagnosis of another malignancy (unless disease-free for >6 months); prior AML-like systemic therapy or use of GO or HDACi; central nervous system AML; positive HIV test; or uncontrolled systemic infection. Cytogenetic risk-group assignment was carried out according to the Southwest Oncology Group (SWOG)/ECOG criteria.12 Treatment responses were defined according to standard criteria.3,13 The study was approved by the institutional review board of all participating institutions (Fred Hutchinson Cancer Research Center/University of Washington/Seattle Cancer Care Alliance, Stanford University, Wake Forest University, and Barbara Ann Karmanos Cancer Center; trial registered at ClinicalTrials.gov as NCT00673153), and patients gave their informed consent in accordance with the Declaration of Helsinki.
Patients received vorinostat 400 mg orally once daily on Days 1–9 and GO 3 mg/m2 on Day 8; hydroxyurea was given to reduce the WBC to less than 10×109/L before beginning vorinostat. Therapy was repeated for residual disease on Day 14; the protocol was amended after 8 enrolled patients to allow a third induction course before response assessment. Patients who achieved either CR or CRp were eligible for one cycle of consolidation treatment with vorinostat and GO at the same doses. Patients could then proceed with maintenance therapy with vorinostat only (400 mg once daily on Days 1–14 every four weeks for up to 4 cycles) if CR/CRp was maintained. Patients entering a CR/CRp could also receive more intensive consolidation therapy including hematopoietic cell transplantation (HCT).
Our main objective was to test whether preliminary efficacy data suggested that further clinical testing of vorinostat and GO in a randomized fashion with ‘standard’ chemotherapy (e.g. ‘3+7’) was warranted. Given the influence of age and PS on treatment outcome,2 patients were stratified into two groups: group 1, patients aged ≥70 years with ECOG PS 2–3; and group 2, aged 60–69 years with PS 0–3 or aged ≥70 years and PS 0–1. For group 1, the primary outcome of interest was 30-day survival, using a historical rate of 50% for similar patients undergoing standard induction therapy.2 For group 2, we focused on the CR/CRp rate. Given the influence of cytogenetics on remission rates,3 the statistical design monitored CR/CRp rate separately in group 2 patients with normal or favorable cytogenetics (group 2A) and other cytogenetics (group 2B) while allowing results from group 2A to affect stopping in group 2B and vice versa. CR/CRp rates of interest were based on those found in the literature. In group 2A, stopping would occur if the probability was over 97% that the true CR/CRp rate was below 45%, while in group 2B, the criterion probability was 98% and the reference CR/CRp rate was less than 30%. These cut offs were chosen to allow a maximum 10% probability of early stopping if the true CR/CRp rates were at least 45 or 30% in groups 2A or 2B, respectively. A planned interim analysis was performed after enrollment of 15 patients into group 2. Follow-up data are current as of August 16, 2011.
Results and Discussion
Between September 2008 and March 2010, 31 adults with a median age of 74 years were enrolled (Table 1). Based on their age, PS, and cytogenetic risk, 10 patients were assigned to group 1 (including one patient with favorable cytogenetics), 13 to group 2A, and 8 to group 2B.
All patients received study therapy; 30 patients completed at least one course of induction therapy, 29 received at least 2 courses, and 9 received 3 courses. Six of the 31 patients (19.4%, exact 95% confidence interval [CI], 7.5–37.5%) achieved CR, and one patient (3.2%) achieved CRp, for a CR/CRp rate of 22.6% (9.6–41.1%; Table 2). The 3 patients with known NPM1-positive/FLT3-ITD-negative normal karyotype AML achieved CR/CRp; this observation is consistent with a previous report suggesting that patients with NPM1-positive/FLT3-ITD-negative normal karyotype AML may benefit from GO-containing therapy.14 All patients who achieved CR/CRp received additional consolidation chemotherapy, including vorinostat alone (n=1), non-myeloablative allogeneic HCT (n=3), and/or cytarabine-based chemotherapy (n=4). Three patients relapsed after 82, 291 and 294 days; 4 patients are in ongoing remission after 455, 496, 956 and 988 days. Median overall survival for patients achieving CR/CRp was significantly longer than for those who failed therapy but lived at least 30 days after treatment initiation (553 [range 291–1035]) vs. 131 [34–998] days, log rank P=0.0026; Figure 1). Three patients (9.7%) died within 30 days after initiating study therapy: 2 patients in group 1 died from progressive disease and one patient in group 2B died from sepsis. Two of 3 patients died before initial treatment assessment (‘death from indeterminate cause’). Treatment was otherwise well tolerated with grade 3–4 bleeding (n=4), dyspnea (n=3) and acute renal failure (n=2) as the most common toxicities besides cytopenias and neutropenic fever and/or infections (Online Supplementary Table S1).
Importantly, our data indicate that the likelihood of achieving CR/CRp depended on age/PS and cytogenetics. Specifically, CR/CRp rates were one of 10 (10.0%) in group 1, 6 of 13 (46.2%) in group 2A (2 patients negative for minimal residual disease by flow cytometry), and none of 8 (0%) in group 2B. This low CR/CRp rate led to discontinuation of accrual in group 2B after a planned first interim analysis. The stopping rule was not met in group 2A, but accrual was stopped after the availability of GO in the US was curtailed.15 Two of 10 (exact 95% CI, 2.5–55.6%) group 1 patients died within the first 30 days of progressive disease, a finding that would not have led to termination of accrual were it not for the general unavailability of GO.
As this was a single-arm clinical study, the relative contribution(s) of vorinostat and GO to the clinical efficacy is unknown. The currently available clinical data with vorinostat suggest that its activity as single agent is relatively limited.16 On the other hand, GO monotherapy has shown efficacy in older adults with newly diagnosed AML in several phase II studies, with overall response rates ranging from 8–27%,17–19 although post hoc stratified analyses in one study indicated better response rates for patients aged 61–75 years than those aged over 75.18 Relative to our study, however, higher doses of GO were used in the previous studies and data were not stratified prospectively based on cytogenetic risk, limiting the ability to directly compare trial results. Only a better-controlled trial would ultimately be able to determine whether vorinostat indeed augments GO efficacy, as pre-clinical data suggest.
A particular characteristic of our trial is the statistical design in group 2 which monitored drug efficacy in cohorts with different cytogenetic risk separately, thereby addressing the patient heterogeneity that poses a major challenge in phase II drug development.20 At the time of trial development, there was little information on the relationship between cytogenetic/molecular abnormalities and likelihood of response to either vorinostat or GO; therefore, our trial design was loosely based on risk stratification schemes that evolved from experience with ‘standard’ AML therapeutics. However, recent results from the Medical Research Council (MRC) AML15 trial on 1,113 predominantly younger adults with newly diagnosed AML, randomized to receive a single dose of GO (3 mg/m2) during induction chemotherapy, indicate that the benefit of addition of GO is most prominent for patients with favorable-risk disease, with some benefit extending to subgroups of patients with intermediate-risk disease but no benefit for those with poor-risk disease.21 These data suggest that the response to GO may be similar to that of conventional AML chemotherapy, providing some basis for the chosen cytogenetic risk stratification in our study. Although direct comparability is limited, our study agrees with data from the MRC AML15, indicating only very limited efficacy of GO in worse-prognosis patients.
While it is well recognized that therapeutic response, likelihood of disease relapse, and chance of cure differs widely between individual patients,3 most phase II trials combine all patients in one group, thus ignoring heterogeneity. Consequently, a new drug might be called ‘promising’ (‘not promising’) purely because of a disproportionate number of favorable (‘unfavorable’) patients given the drug. The small number of patients treated in many phase II trials increases the probability that this situation will arise. Accounting for patient heterogeneity should increase the likelihood of drawing correct conclusions in phase II trials. The simplest method to account for heterogeneity is to conduct distinct trials in various prognostic groups. Although this is preferable to averaging and considering such patients as one group, separate trials increase sample size and study duration. Another method computes observed/expected success ratios with the expected rates accounting for heterogeneity. However, these methods do not formally allow results from one subgroup to adaptively influence trial conduct (e.g. stopping or continuing) in another group. This is particularly problematic when treatment-subgroup interactions exist, i.e. when a treatment has different effects in different prognostic groups. Wathan et al. have proposed a hierarchical Bayesian design to address this problem.22 As with separate trials, stopping rules are subgroup specific. However, in contrast to separate trials, the design examines accumulating data to see whether a given treatment might have similar effects in different prognostic groups and allows data from two groups to be combined to the extent that such ‘borrowing of strength’ is justified by these data. Although the design is computationally complex, advances in computing algorithms and in computing power will facilitate its use. Our trial shows that this approach can identify subgroups of patients that benefit from drug treatment while preventing unnecessary toxicity in others that are unlikely to respond to therapy.
Together, our study indicates that vorinostat and GO has encouraging anti-AML activity with limited toxicity in a well defined subset of older adults with untreated AML, i.e. those with normal karyotype AML (perhaps in particular those with NPM1-positive/FLT3-ITD-negative AML) while it is largely ineffective in others. These data would support further clinical testing of vorinostat and GO, ideally in a randomized design with established therapies (e.g. ‘3+7’) in the subset of older adults with better-prognosis disease, i.e. favorable-risk and normal karyotype AML.
Funding: supported by funding from Merck and Co., Inc (IISP #33067 to RBW) and grants from the Wayne D. Kuni and Joan E. Kuni Foundation (#2008-08 to RBW) and the National Cancer Institute/National Institutes of Health (P30-CA015704-35S6 to RBW).
The online version of this article has a Supplementary Appendix.
Presented in part at the 15th Congress of the European Hematology Association, June 10-13, 2010, Barcelona, Spain (abstract #734).
Authorship and Disclosures
The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
- Received September 21, 2011.
- Revision received November 1, 2011.
- Accepted November 24, 2011.
- Copyright© Ferrata Storti Foundation