Supplementary data
Suppl. Table 1: Adverse event listing according to treatment arm
Suppl. Table 2: Cardiac adverse events according to treatment arm
Suppl. Table 3: Infection related adverse events according to treatment arm
Suppl. Table 4: Early death rates in both treatment arms
Supplementary Information on statistical issues
Supplementary Information on statistical issues
Median event free survival times in the treatment arm and the control arm were assumed to be
3 and 4.5 months, respectively (hazard ratio = 0.667). It was assumed that about 1/3 of the
recruited study patients would be lost to follow-up. The form of the (event free) survival curve
as well as the loss curve was assumed to be exponential in both the treatment and control arm. A
single stage design without interim analysis with a fixed sample size was chosen. A total number
of 191 events was necessary to achieve a power of 80% to detect the supposed treatment effect
with a two-sided Logrank test with specified α=0.05. Accrual time and follow-up were planned
to be 24 and 12 months, respectively. Therefore, a total of 216 patients had to be recruited (108
per treatment arm).
The role of the DMC was to monitor the trial for safety issues. Efficacy data was not analyzed
prior to the final analysis, neither by the DMC nor by anyone else.
Definition of EFS:
Time interval from day 1 of study treatment until treatment failure, relapse from CR, relapse
from morphologic leukemia-free state, or death from any cause, whichever occurred first. For a
patient with none of these events before the end of study follow-up, observation of EFS was
censored at the date of his or her last follow-up examination. Patients undergoing allogenic bone
marrow transplantation are censored for EFS at the time of bone marrow transplantation.
Definition of RFS:
Time interval from the day of documentation of CR until relapse or death from any cause.
Patients undergoing allogenic bone marrow transplantation are censored for RFS at the time of
bone marrow transplantation.
Definition of OS:
Time interval from day 1 of study treatment to the day of death. For a patient who is not known
to have died by the end of follow-up, OS will be censored on the date the patient was last known
to be alive. Patients undergoing allogenic bone marrow transplantation are censored for OS at
the time of bone marrow transplantation.
Regarding the primary endpoint (EFS), a total number of 147 events were observed (arm A: 72,
arm B: 75). The resulting power is 69%. This power is large enough to conclude that no relevant
treatment effect exists, taking also into account the estimated hazard ratio that is very close to 1.
Patients were recruited in 25 trial sites, each site recruiting between 1 and 29 patients.
In the whole study population, a total number of 72 TX were documented (ITT population: 70
TX, 30 in arm A, 40 in Arm B). This resulted in censoring regarding EFS in 11 cases in arm A and
in 19 cases in arm B in the ITT population. These cases are mentioned as number of TX in first
CR in the text. In all other cases TX was performed after an event with respect to EFS had
occurred.
Regarding the handling of allogeneic TX, it would be possible not to censor the respective cases
but to use methods for competing events instead. According to the study protocol it was prespecified to censor the cases at the time of TX in the primary statistical analysis.
Of note, there are several way how to account for TX in the analyses setting. Choosing the
appropriate approach it has to be taken into account that TX does not represent an event that
occurs as a result of disease progression or of a therapeutic success or failure. Instead of that,
allogeneic TX rather represents a regular part of the therapeutic strategy: All patients with a
matching stem cell donor in a sufficiently well condition are transplanted. And after TX,
observation of patients is not stopped, but is continued ongoing just as before TX. Any occurring
event with respect to EFS is documented. For this reason the appropriate way of statistical
analysis is a Cox regression, including TX as a time-dependent covariate. A binary covariate is
included in the model that during follow-up may switch from 0 to 1 at the time when a patient is
transplanted. The corresponding hazard ratio reflects the effect of TX and in a multivariable
model provides effects of other factors such as the treatment effect that are adjusted by the TX
effect. The result of a Cox regression of EFS with the time-dependent covariate TX and fixed
covariate study therapy are shown below. It confirms the conclusions regarding the treatment
effect that have been drawn before.
Analysis of Maximum Likelihood Estimates
Parameter
Parameter
DF
Estimate
treatment AZA+Standard
1
-0.09864
cytogenetic high risk
1
1.64801
cytogen intermed. risk
1
1.09895
ECOG
ECOG-Status 1
1
-0.05704
ECOG
ECOG-Status 2
1
-0.08483
ECOG
ECOG-Status 3
1
2.72742
Allogeneic transplantation
1
-0.43146
Standard
Error Chi-Square
0.16852
0.3426
0.72573
5.1566
0.71829
2.3407
0.18684
0.0932
0.27661
0.0940
0.78821
11.9735
0.29771
2.1004
Pr > ChiSq
0.5583
0.0232
0.1260
0.7601
0.7591
0.0005
0.1473
Hazard 95% Hazard Ratio
Ratio Confidence Limits
0.906
0.651
1.261
5.197
1.253
21.551
3.001
0.734
12.265
0.945
0.655
1.362
0.919
0.534
1.580
15.293
3.263
71.685
0.650
0.362
1.164
We investigated the association of ECOG, LDH and cytogenetic risk with the number or reported
Adverse Events per patient to analyze the influence of poor prognosis features for the larger
number of Adverse Events in the Azacitidine plus Chemotherapy cohort.. No consistent
association was noticed, see below.
Number of
patients
Therapy, „As
treated“
ECOG
5-Azacitidine +
Standardchemo
.
Standardchemo
Mean
Std
4
12.25
8.85
ECOG-Status 0
23
14.39
14.95
ECOG-Status 1
59
15.83
12.66
ECOG-Status 2
13
16.54
12.57
All
99
15.44
12.95
5
20.80
14.65
ECOG-Status 0
36
9.86
9.84
ECOG-Status 1
55
15.51
13.15
ECOG-Status 2
12
12.42
5.93
ECOG-Status 3
2
13.00
1.41
110
13.52
11.72
9
17.00
12.53
ECOG-Status 0
59
11.63
12.17
ECOG-Status 1
114
15.68
12.84
ECOG-Status 2
25
14.56
9.98
ECOG-Status 3
2
13.00
1.41
209
14.43
12.32
ECOG
.
All
All
Number of Adverse
Events
ECOG
.
All
Number of
patients
Number of Adverse Events
Mean
Std
Therapy, „As
treated“
Cytogenetic risk
group
5-Azacitidine +
Standardchemo
n.k.
2
18.50
19.09
Good
4
25.50
24.28
Intermediate
56
16.27
13.34
High
37
12.95
10.30
All
99
15.44
12.95
n.k.
3
9.00
7.00
Good
4
27.50
17.37
Intermediate
78
13.81
12.33
High
25
10.92
7.18
110
13.52
11.72
n.k.
5
12.80
11.95
Good
8
26.50
19.57
134
14.84
12.77
62
12.13
9.16
209
14.43
12.32
Standardchemo
Cytogenetic risk
group
All
All
Cytogenetic risk
group
Intermediate
High
All