Abstract 2182

Introduction:

A pivotal trial of histamine dihydrochloride (HDC) + interleukin-2 (IL-2) showed a statistically significant benefit for leukemia-free survival (LFS) in a randomized comparison versus standard of care (SOC) for remission maintenance in 320 patients with acute myeloid leukemia (AML) (Brune et al. Blood. 2006;108:88-96). HDC+IL-2 has not been randomized against IL-2 alone for this indication. This raises the question of whether and to what extent the combination of HDC+IL-2 is an improvement over IL-2 alone. In AML remission maintenance studies, both HDC+IL-2 (n=1) and IL-2 alone (n=5) have been randomized against SOC, providing a common reference for comparison. We exploit this commonality. Our objective is to compare HDC+IL-2 versus IL-2 alone on the basis of LFS in AML patients in remission.

Methods:

We take a Bayesian hierarchical modeling approach. This allows for heterogeneity of treatment effects across the 6 trials. We fit two different models, each using Bayesian piecewise exponential modeling. The first model assumes proportional hazards such that treatment effects remain constant throughout the follow-up period. The second model allows for non-proportional hazards such that treatment effects are allowed to vary by year of follow-up. The sufficient statistics for both models are the exposure times and number of events in each yearly interval for each study and treatment arm. From the 5 randomized clinical trials of IL-2 alone versus SOC in patients with AML, we extracted the survival probabilities from the published Kaplan-Meier curves of each trial and estimated the exposure time and number of events in each yearly interval. For the HDC+IL-2 trial, we used individual patient data to calculate these quantities. These analyses consider the first 5 years of follow-up.

Results:

Assuming proportional hazards, the posterior mean 3-year LFS probabilities (%) are 29.5, 33.0, and 39.6 for SOC, IL-2, and HDC+IL-2. The mean 5-year LFS probabilities are (%) 25.8, 29.3, and 35.9 for SOC, IL-2, and HDC+IL-2. Allowing for non-proportional hazards, the posterior mean 3-year LFS probabilities (%) are 30.5, 30.6, and 41.3 and the mean 5-year LFS probabilities (%) are 26.6, 27.8 and 38.0 for SOC, IL-2, and HDC+IL-2. The table shows the mean cumulative months of improvement for each treatment versus the others. By year 5, the cumulative benefit of IL-2 over SOC in LFS is 1.42 months. The cumulative benefit over 5 years for HDC+IL-2 is an estimated 5.65 months over SOC and 4.23 months over IL-2 alone. The cumulative probability that IL-2 is superior to SOC over the 5-year period is 0.636. The cumulative probability that HDC+IL-2 is superior to SOC over the 5-year period is 0.989. The cumulative probability that HDC+IL-2 is superior to IL-2 over the 5-year period is 0.956.

Conclusion:

By both models, the LFS of IL-2 is similar to that of SOC. In contrast, the benefit of HDC+IL-2 over SOC and IL-2 is highly likely, taking into account all available information. This conclusion is robust with respect to the modeling assumptions. The observed LFS benefit of HDC+IL-2 over IL-2 and SOC is clinically relevant and should regenerate interest in immunotherapy as a means of maintaining remission and preventing relapse in AML patients.

Absolute Magnitudes of Effect on LFS, Truncated at Year 5

Through yearMean Cumulative Months Improved
IL-2 over SOCHDC + IL-2 over SOCHDC + IL-2 over IL-2
0.39 0.43 0.04 
0.62 1.26 0.64 
1.19 3.11 1.92 
1.27 4.36 3.09 
1.42 5.65 4.23 
Through yearMean Cumulative Months Improved
IL-2 over SOCHDC + IL-2 over SOCHDC + IL-2 over IL-2
0.39 0.43 0.04 
0.62 1.26 0.64 
1.19 3.11 1.92 
1.27 4.36 3.09 
1.42 5.65 4.23 
Disclosures:

Berry: EpiCept Corporation: Consultancy. Off Label Use: IL-2 monotherapy is not approved for remission maintenance in AML. The European Medicines Agency has approved histamine dihydrochloride as a maintenance therapy for adult patients with AML in first remission concomitantly treated with IL-2. Broglio: EpiCept Corporation: Consultancy. Berry: EpiCept Corporation: Consultancy.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution