DECIDER: prospective randomized multicenter phase II trial of low-dose decitabine (DAC) administered alone or in combination with the histone deacetylase inhibitor valproic acid (VPA) and all-trans retinoic acid (ATRA) in patients >60 years with acute myeloid leukemia who are ineligible for induction chemotherapy

Study design and setting

This is a prospective, randomized, observer blind, active control, parallel group,
multicenter, phase II study. The objective of the trial is the investigation of efficacy
and safety of the histone deacetylase inhibitor (VPA) and all-trans retinoic acid (ATRA) as add-on to the epigenetically active drug decitabine (DAC)
in older and unfit acute myeloid leukemia (AML) patients.

The trial has a 2×2 design randomizing the patients to the following four treatment
arms: DAC, DAC?+?VPA, DAC?+?ATRA, DAC?+?VPA?+?ATRA. The effect of VPA will be investigated
by comparing the combined treatment arms (DAC?+?VPA) and (DAC?+?VPA?+?ATRA) as experimental
group versus the combined treatment arms (DAC) and (DAC?+?ATRA) as control group,
and the effect of ATRA will be investigated by comparing the combined treatment arms
(DAC?+?ATRA) and (DAC?+?VPA?+?ATRA) as experimental group versus the combined treatment
arms (DAC) and (DAC?+?VPA) as control group.

Trial sites are located in Aachen, Berlin, Bochum, Braunschweig, Bremen, Düsseldorf
(two sites), Esslingen, Frankfurt, Freiburg, Hagen, Halle, Hamm, Hannover, Jena, Koblenz,
Lahr, Lebach, Leipzig, Lüdenscheid, Marburg, München, Münster, Offenburg, Ravensburg,
Tübingen, Ulm, and Villingen-Schwenningen.

The study and all participating sites were approved by the central ethics committee
(University of Freiburg, 76/10) and the respective local ethics committees. Patient’s
written consent to participate in this clinical trial and translational research program
was obtained prior to any study-specific procedures. The identifying number of the
DECIDER trial in ClinicalTrials.gov registry is NCT00867672 and in German clinical
trials registry DRKS00000733.

Study population

The target population of this trial represents patients older than 60 years with AML
according to WHO (?20 % blasts in the peripheral blood or bone marrow) not qualifying
for, or not consenting to, standard remission-induction chemotherapy or immediate
allografting. Key inclusion criteria are: patient’s written informed consent; 60 years
or older at time of informed consent; patients with primary or secondary AML according
to WHO who are not expected to benefit from standard remission-induction chemotherapy;
30.000 leukocytes/?l and performance status ECOG 0, 1, 2. Key exclusion criteria
are: Acute promyelocytic leukemia (APL), AML of FAB subtype M3; previous remission-induction
chemotherapy for MDS or AML, previous allografting; previous treatment with DAC, 5-azacytidine,
VPA or another histone deacetylase inhibitor, or ATRA; “low-dose” chemotherapy (e.g.,
hydroxyurea, cytarabine (Ara-C), melphalan, clofarabine etc.) within four weeks prior
to DAC treatment, except for cytoreduction of leukocytosis ?30.000/?l with hydroxyurea
or Ara-C as proscribed by the study protocol; treatment with tyrosine kinase inhibitors,
immunomodulating agents or other investigational AML treatment within the last four
weeks or in a time period of drug half-life x 5 (whatever is shorter) before the first
administration of DAC; treatment with cytokines within the previous four weeks. Further
exclusion criteria are related to general medical conditions or contraindications
to administration of investigational products according to Summary of Product Characteristics.

Study treatment and procedures

Study treatment in the four study arms Fig. 1 is as follows: (DAC) intravenous decitabine 20 mg/m² over 1 h, 5 days (total dose 100 mg/m²), repeated every 4 weeks; (DAC?+?VPA) intravenous decitabine 20 mg/m² over 1 h, 5 days (total dose 100 mg/m²), repeated every 4 weeks, and VPA (p.o.) from day 6 of first cycle continuously throughout
all treatment cycles in a dose to achieve a VPA serum level between 50 to 110 mg/l;
(DAC?+?ATRA) intravenous decitabine 20 mg/m² over 1 h, 5 days (total dose 100 mg/m²), repeated every 4 weeks and ATRA (45 mg/m² p.o.) from day 6 to day 28 of each treatment cycle; (DAC?+?VPA?+?ATRA) intravenous
decitabine 20 mg/m² over 1 h, 5 days (total dose 100 mg/m²), repeated every 4 weeks and VPA (p.o.) from day 6 continuously throughout all treatment
cycles in a dose to achieve a VPA serum level between 50 to 110 mg/l and ATRA (45 mg/m² p.o.) from day 6 to day 28 of each treatment cycle. Treatment cycles are repeated
until relapse/progression, hematopoietic stem-cell transplantation, ongoing cytopenia,
other unacceptable toxicity or patient’s death. Response assessment is performed at
the end of the cycles 2, 4 and 6, and thereafter every third cycle. Study visits are
carried out at least once a month during the treatment and every three months after
treatment discontinuation until 12 months after randomization of the last patient.

Fig. 1. Trial flow. *If WBC???30.000/?l: Hydroxyurea (HU) or Ara-C until WBC??30.000/?l.
CR?=?complete remission; PR?=?partial remission; ALE?=?antileukemic effect; SD?=?stable
disease

Study endpoints

The primary endpoint is the objective best overall response, defined as complete remission
(CR) or partial remission (PR). Patients are counted as objective responders if they
achieve a CR or a PR. All other patients are counted as non-responders. Secondary
endpoints for the evaluation of efficacy are the following: (1) the overall best response
defined as CR, PR, or antileukemic effect (ALE). Patients are counted as responders
if they achieve a CR, PR or ALE. All other patients are counted as non-responders.
The determination of CR, PR, and ALE will be performed according to the standard criteria
20], 21], (2) the progression free survival (PFS) time from randomization until relapse/progression
or death. For patients being progression-free and alive at the end of the study, the
PFS time will be censored at the time of the last evaluation of bone marrow, (3) overall
survival (OS) time from randomization until death of the patients. For patients being
alive at the end of the study, the OS time will be censored at the time of the last
visit or follow-up contact, (4) quality of life of the patients using the EORTC QLQ-C30
Quality of Life Questionnaire, (5) number of nights in hospital. Further secondary
endpoint for evaluation of the treatments is safety and toxicity, which are evaluated
by means of adverse events (AEs) occurring from the first administration and until
4 weeks after the last administration of the study treatment, death and laboratory
assessments.

Sample size

Sample size calculation is based on the rate of patients experiencing the primary
endpoint objective best overall response (CR or PR). It is the aim of this phase II
study to come to a decision whether the experimental treatments VPA and ATRA are promising
treatment options being worthy of further investigation in phase III. Therefore, it
is not necessary and desired to come to a definitive and rigorous statistical proof
of the efficacy of VPA and ATRA in this study. Nevertheless, statistical tests on
the effects of VPA, i.e., (DAC?+?VPA) and (DAC?+?VPA?+?ATRA) vs. (DAC) and (DAC?+?ATRA),
and ATRA, i.e., (DAC?+?ATRA) and (DAC?+?VPA?+?ATRA) vs. (DAC) and (DAC?+?VPA), will
be conducted. Both tests will be performed at a one-sided significance level of 10 %.
Based on the results of our single-arm Phase II study of DAC in AML (Study 00331,
NCT00866073) 22], which has recruited 235 patients in Germany, an objective best overall response
rate (ORR) of about 25 % in treatment arm (DAC) is assumed. The study should have
a power of 80 % to detect an increase in the ORR, when the ORR after application of
VPA or ATRA is about 40 %. This corresponds to an odds ratio of 2 between the treatment
arms no VPA vs. VPA and no ATRA vs. ATRA, respectively. It is further assumed that
VPA and ATRA do not show an antagonistic effect. With these assumptions, 176 patients
have to be included in this study. This was calculated with formula (3) in Campbell
et al., 1995 23] for the Chi square-test of the hypothesis that the difference between the rates is
equal to zero, and is adequate also for analysis with logistic regression 24]. To account for the possibility of some protocol violations, 200 patients will be
randomized in total (ratio of 1:1:1:1 to the four treatment arms).

Randomization

Center-stratified block randomization with randomly varying block size is performed
based on computer-generated lists. Block sizes are documented separately from the
study protocol in a document not accessible to investigators. Randomization is performed
centrally in order to guarantee concealment of treatment allocation. Blinding of the
VPA and ATRA intake is impossible since serum levels of VPA will be measured, and
therefore placebo control will not be feasible. However, concealment of randomization
will be guaranteed to minimize selection bias, and reviewers of response will be blinded
with regard to the patient’s treatment arm to minimize assessment bias.

Statistical analysis

The primary efficacy analysis of this clinical trial will be conducted according to
the intention-to-treat principle. This means that the patients will be analyzed in
the treatment arms to which they were randomized, irrespective of whether they refused
or discontinued the treatment or whether other protocol violations are revealed. The
analysis of the effects of the different treatment schedules (VPA vs. no VPA and ATRA
vs. no ATRA) with respect to the primary endpoint objective best overall response
will be performed with a logistic regression model. The treatment effects will be
tested using the Wald test at a one-sided significance level of 10 %, and as estimates
of the effect sizes the odds ratios will be given with 95 %-confidence intervals.
A multiplicative interaction effect between VPA and ATRA will be included in the logistic
regression model and its size will be estimated with 95 %-confidence interval. The
analysis of the treatment effects with respect to the secondary endpoints PFS and
OS will be performed with the Cox model. The analysis of the treatment effects with
respect to the secondary endpoint overall best response (CR, PR and ALE) will be performed
with a logistic regression model. The PFS rates and the OS rates of the different
treatment arms will be estimated by the Kaplan-Meier method, and the ORR of the different
treatment arms will be estimated as relative frequencies. For evaluating QOL, scores
will be calculated according to Fayers et al., 2001 25]. At the different time points during follow-up changes in relation to baseline will
be calculated. With respect to the secondary endpoints QOL and number of nights in
hospital, the treatment groups will be compared descriptively.

Safety is analyzed in the safety population including all randomized patients who
received at least one dose of study medication, and patients are analyzed according
to the received treatment. AEs are coded using Medical Dictionary for Regulatory Activities
(MedDRA). The incidence of AEs is calculated as the number of patients who experienced
at least one AE of a certain category as a percentage of the total number of patients
in the safety population.

Interim analysis

One interim analysis with respect to safety had to be performed three months after
22 patients have been randomized in each treatment arm. The objective of this analysis
is to enable early stopping of one or more of the treatment arms in case of an unacceptable
death rate. A stopping rule was established based on the following considerations:
an analysis of the one arm Phase II study of DAC in AML (Study 00331, NCT00866073)
showed a three-months death rate of about 35 % which is commonly regarded as acceptable.
A three-months death rate of 70 % or higher is regarded as unacceptable and should
lead to an early stop of the treatment arm. The error probability of early stopping
if the three-months death rate is acceptable in reality (35 %) was set as ?5 %, and
the error probability of continuing a treatment arm if the three-months death rate
is unacceptable in reality (70 %) was set as ?5 %.

Central hematopathology

A systematic central hematopathology review of serial bone marrow aspirates, bone
marrow biopsies and matching peripheral blood smears of patients on the study is being
conducted at the Institute of Clinical Pathology, University of Freiburg. The review
is performed by a very experienced hematopathologist who is blinded to the treatment
arm to which each patient is randomized. Written reports (patients are pseudonymized)
are then sent to the local principle investigators at the respective study centers.

The review procedure includes a differential count of peripheral blood (200 cells)
and bone marrow smears (500 cells) as well as histological blast quantification, including
enumeration after immunohistochemistry for CD34 and CD117 on bone marrow biopsies.
It serves to support the study physicians’ treatment decisions, particularly in the
setting of patient relapse/progressive disease. Beyond the quality control for central
response evaluation, and post hoc support of the cytology results of the local centers, the central hematopathology
review process allows for additional, exploratory studies such as those of bone marrow
fibrosis, quantification of dysplasia, enumeration of megakaryocytes, and quantification
of bone marrow cellularity. These measurements, conducted in a highly standardized
fashion, would not be feasible by decentralized diagnostic procedures in a multicenter
trial.

The compliance of the trial sites with the central hematopathology review in sending
samples is 90 %. A repository of bone marrow and blood samples will be very valuable
also for future adjunct studies involving immunohistochemistry, FISH and molecular
techniques.

Central cytogenetics and molecular diagnostics

Cytogenetic and molecular genetic studies are being performed in the central reference
laboratory of the German–Austrian Acute Myeloid Leukemia Study Group (AMLSG), at the
University of Ulm. For conventional cytogenetic analysis, standard techniques are
used and chromosomal abnormalities are described according to the International System
for Cytogenetic Nomenclature 26]. The adherence in sending samples to the central laboratory is 95 % (in DECIDER
sites which are members of the AMLSG) and informative karyotypes are obtained in 90 %
of the patients. In addition, diagnostic samples from all patients are screened for
the recurring gene fusions PML/RARA, CBFB/MYH11, and RUNX1/RUNX1T1 as well as for
the presence of mutations in the genes encoding FLT3 (i.e., the ITD and TKD mutations
at codons D835 and I836) CEBPA and NPM1. In analogy, success rate of molecular genetic
analysis is 95 %; study centers are being informed on molecular screening results
within a time window of 48 h.

Central pharmacy

The central pharmacy of the Medical Center – University of Freiburg has a very long-standing
experience in the preparation and handling of DAC. Senior staff members of the pharmacy
have been involved already in the early European phase II DAC studies since 1996.
Therefore a wealth of knowledge exists in handling of the solubilized solution until
infusion of this notoriously unstable nucleoside. From the very beginning of the study
(before the DAC marketing authorization in the EU) the central pharmacy supported
and consulted study sites on preparation of decitabine. Furthermore, the central pharmacy
has the task of distributing the oral study drug VPA to the different German study
sites in accordance with German laws.

Serial assessment of patient fitness and psychological state

As in the predecessor phase II trial in elderly, non-fit AML patients receiving DAC
(study 00331, NCT00866073), systematic functional assessment of the patients is being
conducted prior to randomization and at several defined time points during and after
the study treatment 27]. It is an additional aim of the study to investigate the prognostic value of geriatric
assessments for elderly AML patients treated non-intensively. This includes determination
of the activities of daily living (ADL) by application of the Barthel index, assessment
of quality of life by the EORTC-C30 questionnaire, application of the Hospital and
Anxiety Depression Scale (HADS), and determination of psychological resilience by
application of the RS-11 questionnaire. In addition, the performance status (Eastern
cooperative oncology group performance status, ECOG) of the patient is captured before
and at defined time points during the treatment. Comorbidities are scored prior to
treatment using the Hematopoietic Cell Transplantation-Comorbidity Index (HCT-CI).
The compliance with this functional assessment is 90 %.

Translational research program

The study provides the opportunity to address not only baseline genetic and epigenetic
characteristics of the patient blasts with relation to the clinical response, but
also mRNA expression and epigenetic changes induced by the treatment. Bone marrow
and peripheral blood cells are procured prior to treatment within the standard diagnostic
workup, and in patients consenting to serial sampling, also at defined, early and
late time points during the treatment. Furthermore serum is being collected and cryopreserved
for future studies on miRNA expression changes with this epigenetic treatment. Serial
anticoagulated blood sample allow for cell sorting for blasts vs. T-cells as well
as isolation of granulocytes (provided patients have sufficient cell numbers at the
different time points).

Quality assurance system

During the clinical trial, quality control is ensured through monitoring, auditing
and supervision by the authorities, if applicable. The Clinical Trials Unit of the
Medical Center – University of Freiburg, Germany, is responsible for project coordination,
statistical planning and analysis, data management, clinical monitoring and pharmacovigilance.
An independent DMC consisting of three hemato-oncologists and one statistician was
established. It is the role of the DMC to monitor and supervise the progress of the
trial (including safety data and adherence to the protocol) and to advise whether
to continue, modify or stop the trial. Composition and responsibilities of the DMC,
structure and content of its meetings, and its relationship to other key study team
members are laid down in a separate DMC charter. The DMC members are continuously
informed about study progress and safety data. The DMC gave recommendations on study
conduct after the review of the extensive report about the interim analysis including
data on patient recruitment, baseline characteristics, eligibility violations, treatment
compliance, compliance with planned visits, completeness of follow-up, and safety
separately for the different treatment arms.

Study status

Since the start of recruitment in December 2011, a total of 189 patients were randomized
in the study until 31.12.2014. The interim analysis was conducted in May 2014 by the
Clinical Trials Unit, and the confidential report was made available only to the coordinating
investigator, medical coordinator, and to the DMC. The interim analysis showed that
in all four treatment arms the critical stopping rule concerning death within three
months after randomization was not reached. Furthermore, no important safety issues
have been observed that would alter the assumed benefit-risk profile. The DMC statement
released in July 2014 recommended continuing the study as planned, as all DMC members
unanimously agreed that at the moment there are no ethical or other concerns on further
conduct of the trial.