Supplement AML 2003 Study

1. results in the Primary efficacy population

The treatment effectin the primary efficacy analysis population is reported here. This population was defined post hoc in the statistical analysis plan of the AML 2003 study. The reason for defining this population was that patients with favorable karyotypes, who were not considered candidates for transplantation, and patients with standard risk stratificationwho had anHLA-identical sibling donor were equally considered for allogeneic HCT in first complete remission in both treatment arms and therefore could not contribute to any difference between the two alternative treatment arms.Further, patients with refractory AML who went off study after first induction chemotherapy and those who died during first induction chemotherapy were excludedfrom the primary efficacy population. The reason for the exclusion of these patients was thattheir clinical course could never be affected by the randomization between two different transplantation strategies which became effective earliest after remission assessment following the first induction chemotherapy. The primary efficacy population thus defines an optimal population to study the effect of the two transplantation strategies.

1.1. Patient characteristics in the primary efficacy population

The full analysis set consisted of 1,179 patients. Yet, the primary efficacy population contained only 713 patients after exclusion of 142 patients with t(8;21) or inv(16)/t(16;16) AML who were generally not considered for transplantation, 143 patients with standard risk stratification and HLA-identical sibling donors who were scheduled for allogeneic HCT in both arms, and 186 patients who died during IT1 or had refractory AML and went off study after IT and therefore did not contribute to the comparison of the two transplant strategies.The characteristics of the patients in the primary efficacy population are shown in Table S1.

In the primary efficacy population, 212 patients (30%) received allogeneic HCT in first remission. In addition 205 patients (29%) were transplanted after primary induction failure or relapse, whereas 286patients (40%) have notreceived an allogeneic HCT.Overall, 63% of patients in the experimental arm and 56% in the control arm received allogeneic HCTduring the observation period.

1.2. Intent-to-Treat Analysis in the primary efficacy population

In the primary efficacy population, OS was 49% (95% CI, 43% to 55%) and 47% (95% CI, 41% to 53%), and event-free survival (EFS) was 34% (95% CI, 28% to 39%) and 26% (95% CI, 21% to 31%) respectively. In the primary efficacy population, the hazard ratio of the treatment effect (experimental versus control) was 0.92 (95% CI, 0.75 to 1.14; p=.46) for OS and 0.85 (95% CI, 0.71 to 1.02; p= .08) for EFS. Non-relapse mortality and the cumulative incidence of relapse did not differ statistically significant.

1.3. Interpretation

Although the primary efficacy analysis allows for more precise estimates of the treatment effect, overall no statistically significant differences in outcome between the more aggressive and the more conservative transplantation strategy can be described.

Table S1. Patient Characteristics in the primary efficacy population (N=713)

Controls
(N=344) / Experimental
(N=369)
Age (years), median [min, max] / 48 [18, 60] / 49 [17, 60]
Women, n (%) / 197 (57.3) / 168 (45.5)
Type of AML, n (%)
de novo AML
therapy related myeloid neoplasm / 19 (5.5) / 13 (3.5)
RAEB-2 / 16 (4.7) / 12 (3.3)
AML with preceding MDS / 33 (9.6) / 34 (9.2)
AML, not further specified / 6 (1.7) / 13 (3.5)
ECOG performance status, n (%)
ECOG 0 / 71 (20.6) / 87 (23.6)
ECOG 1 / 195 (56.7) / 203 (55.0)
ECOG 2 / 55 (16.0) / 56 (15.2)
ECOG 3-4 / 10 (2.9) / 13 (3.5)
ECOG missing / 13 (3.8) / 10 (2.7)
bone marrow blasts in %, median [min, max] / 60.0 [0.5, 97.5] / 67.5 [8, 100]
value missing, n (%) / 16 (4.7) / 21 (5.7)
white blood count (x109 per liter), median [min, max] / 14.4 [0.6, 410.0] / 16.2 [0.2, 395.4]
lactate dehydrogenase in U/l, median [min, max] / 462 [108, 7369] / 485 [93, 5565]
value missing, n (%) / 6 (1.7) / 7 (1.9)
cytogenetic risk group according to AML2003 protocol, n (%)
low- risk / 0 (0.0) / 0 (0.0)
intermediate-risk / 263 (76.5) / 268 (72.6)
high-risk / 81 (23.5) / 101 (27.4)
risk stratification according to AML2003 protocol, n (%)
low-risk / 0 (0.0) / 0 (0.0)
intermediate-risk / 158 (45.9) / 162 (43.9)
high-risk / 186 (54.1) / 207 (56.1)
FLT3 ratio >0.8, n (% of total) / 28 (8.1) / 31 (8.4)
MAC/MAMAC/MAC Consolidation, n (%) / 180 (52.3) / 191 (51.8)

Abbreviations (Table S1): N, number; AML, acute myeloid leukemia; RAEB, refractory anemia with excess blasts; MDS, myelodysplastic syndrome; ECOG, Eastern Cooperative Oncology Group; Gpt/L, Giga particles per liter; U, units; L, Liter; FLT3, Fms-like tyrosine kinase 3; MAC, consolidation chemotherapy with mitoxantrone 10 mg/m2 days 4 – 6 and cytarabine 2 x 1 g/m2 days 1 – 6; MAMAC, consolidation chemotherapy with amsacrine 100 mg/m2 days 1 – 5 and cytarabine 2 x 1 g/m2 days 1 – 5;

Table S2. Effect estimates In the primary Efficacy population

Intent-to-treat Analyses
Overall Survival / Event-free Survival
Populations / N / HR* (95%-CI) / p / HR* (95%-CI) / p
Primary Efficacy Population / 713 / 0.92 (0.75-1.14) / .5 / 0.85 (0.71-1.02) / .08
Intermediate-risk stratification / 320 / 0.99 (0.69-1.41) / .9 / 0.91 (0.69-1.21) / .5
Poor-risk stratification / 393 / 0.87 (0.67-1.13) / .3 / 0.79 (0.63-0.99) / .04

Abbreviations (Table S2): * HR, hazard ratios <1 are in favor of the experimental arm; # hazard ratios <1 are in favor of transplantation; N, number of patients who contributed data; CI, confidence-interval; p, p-value of the Wald test; HCT, hematopoietic cell transplantation;

1

Supplement AML 2003 Study

Figure S1:

The two upper panels show the effect of experimental versus control treatment on overall survival (upper left panel) and event-free survival (upper right panel) in the primary efficacy population.The two lower panels show overall survival in intermediate-risk (lower left panel) and high-risk patients (lower right panel). The hazard ratios and p-values are derived from multivariate Cox regression model.

1