Recombinant interleukin-2 in patients aged younger than 60 years with acute myeloid leukemia in first complete remission: results from Cancer and Leukemia Group B 19808.

BACKGROUND: Recombinant interleukin-2 (rIL-2) induces cellular cytotoxicity against leukemia blasts. Patients with acute myeloid leukemia (AML) in first complete remission (CR) may harbor minimal residual disease that is susceptible to rIL-2-activated effector cells. METHODS: In the Cancer and Leukemia Group B (CALGB) 19808 study, patients with AML in first CR were randomly assigned after all planned chemotherapy to receive a 90-day course of subcutaneously administered rIL-2 or no further therapy. The primary objective was to compare disease-free survival (DFS) between the 2 treatment arms. A total of 534 patients achieved a CR, 214 of whom were randomized. Six courses of low-dose daily rIL-2 were given for the expansion of cytotoxic effector cells, each followed by 3-day high-dose boluses given to trigger cytotoxicity against minimal residual disease. RESULTS: On the protocol-specified intention-to-treat analysis, the hazards ratio for DFS was 0.75 (95% confidence interval, 0.52-1.09; P = .13); the 5-year DFS rate was 42% in the observation arm and 53% in the rIL-2 treatment arm. The hazards ratio for overall survival (OS) was 0.88 (95% confidence interval, 0.54-1.23; P = .34); the 5-year OS rate was 58% for the observation arm and 63% for the rIL-2 treatment arm. Twenty-five of the 107 patients randomized to treatment with rIL-2 either refused or were unable to initiate therapy and 30 patients did not complete their assigned therapy. However, significant toxicities were not commonly observed. The trial design did not anticipate the difficulties patients would encounter with protocol compliance. CONCLUSIONS: The efficacy of immunotherapy with rIL-2 administered after intensive postremission treatment was not assessed as planned because of unexpected refusals by patients and/or their physicians to comply with protocol-directed therapy. Neither DFS nor OS was found to be significantly improved.

Duke Scholars

Author Stephen L. George Biostatistics & Bioinformatics, Division of Biostatistics