Using data of the prospective AMLCG 1999 trial, we performed a matched pairs analysis to evaluate outcome in patients (pts) with AML according to postremission allo-SCT or randomly allocated autologous SCT or maintenance chemotherapy. As of 07/2007, 2547 pts accrued to the trial were evaluable. 135 pt pairs in CR1 were identified who matched for the following criteria:

  • AML type (de novo AML, s-AML, t-AML, MDS);

  • cytogenetic risk group [unfavorable (UNF-CG), intermediate (INT-CG), and favorable with the exclusion of t(15;17)];

  • age (± 5 years); and

  • time in CR1 to account for the time to transplant in allo-SCT pts.

30 pt pairs had an UNF-CG, 97 pairs INT-CG, and 8 pairs had favorable cytogenetics. Median pt age was 45 years (range: 16–63). In the control cohort, 86 pts were assigned to cyclic maintenance chemotherapy and 49 to autologous SCT (see

J Clin Oncol
). In the allo-SCT cohort, 87 pts had an HLA identical sibling donor (SIB) and 48 a matched volunteer unrelated donor (VUD). Median follow-up was 1349 days (range:70–2820). Projected 5-year relapse-free survival (RFS) was 57% in the allo-SCT group and 37% in controls (p<.001, log-rank test). Overall survival (OS) was 58% and 41% (p=.115), respectively. Median RFS was significantly improved by allo-SCT in pts with UNF-CG (25 vs. 7 months; p=.019) or INT-CG (not reached after 7 years vs. 25 months; p=.005). Median OS was 32 months in allo-SCT pts with UNF-CG versus 14 months in matched controls (p=.183) and not reached after 7 years in INT-CG pts with allo-SCT versus 71 months in matched controls (p=.257). Of note, 40 pts in the control cohort received an allo-SCT (16 from SIBs, 24 from VUDs) beyond CR1 (9/30 with UNF-CG; 30/97 with INT-CG). Median OS of 40 matched pts receiving an allo-SCT in CR1 was not reached after 7 years, while it was only 31 months in paired pts with allo-SCT beyond CR1 (p<.05). In a subset of pts with INT-CG (35 allo-SCT pts and 35 paired controls), information on NPM1 and FLT3 (ITD) mutation status was available. In these pts, we performed a preliminary analysis comparing AML pts with mutated NPM1 (NPM1-Mt) and wild-type FLT3 (FLT3-Wt) to the remainder of the respective cohorts. With conventional PRT, NPM1-Mt/FLT3-Wt pts (18) had a projected OS at 6 years of 69% versus 41% in the remaining 17 pts (p=.112). Interestingly, NPM1-Mt/FLT3-Wt pts receiving an allo-SCT in CR1 (11) had a projected 6-year OS of 91% versus 45% in 24 pts with other NPM1/FLT3 phenotype combinations (p=.024). We conclude that allo-SCT to date is the most potent PRT for AML with UNF-CG and INT-CG. Its impact on OS is difficult to assess as about a third of pts initially treated with conventional PRT undergo allo-SCT beyond CR1 with significantly worse survival rates. AML pts with NPM1-Mt/FLT3-Wt phenotype appear to constitute a good risk group with conventional PRT, but at the same time seem to particularly benefit from allo-SCT in CR1.

Author notes

Disclosure: No relevant conflicts of interest to declare.