The majority of relapses in acute myeloid leukemia (AML) patients occur in the first or second year following complete remission. In routine, AML patients are followed during five years because few relapses can occur after three or five years. These late or very late relapses remain poorly described, particularly at the molecular level, with only few consistent series in the literature. (Medeiros B et al., Leuk Lymphoma 2007; Verma D et al., Leuk Lymphoma 2010; Watts J et al., Leuk Res 2014).

We retrospectively studied all AML relapses occurring after complete remission (CR) obtained with one or two induction cycles between 2000 and 2012 in Toulouse University Hospital, France. Our analyses focused on late relapses (LR, >3 years from CR) and very late relapses (VLR, >5 years from CR) in comparison to early relapses (ER, ≤3 years from CR).

Between 2000 and 2012, out of 636 CR patients, 346 had morphological relapses (54.4%). The median time to relapse was 0.9 years (range, 0.1-11.9 years; interquartile range [IQR], 0.5-1.5 years). There were 198 relapses during the first year (57.2%), 82 during the second year (23.7%), 24 during the third year (6.9%) whereas 42 relapses occurred after 3 years (12.1%) and 16 after 5 years (4.6%). Characteristics at diagnosis, i.e., age, AML status, WBC count, karyotype, FLT3-ITD mutation, CEBPA mutation and induction regimen did not differ between ER and LR or VLR. However, NPM1 mutations were more frequent in LR (NPM1m at diagnosis in relapses >3 years: 46% vs. 28% in relapses <3 years, P=.0532), and in VLR (NPM1m at diagnosis in relapses >5 years: 67% vs. 27% % in relapses <5 years, P=.0070). Allogeneic stem cell transplantation (alloSCT) was more frequently performed in the LR group (24% vs. 14%, P=.0369) and in VLR group (31% vs. 14%, P=.0748). Second CR (CR2) rate and median overall survival from relapse date (OS2) were better in LR and VLR than in ER (CR2ER: 26%, CR2LR: 43%, CR2VLR: 50%; P=.0154; OS2ER: 4.6 months, OS2LR: 10.8 months, OS2VLR: 11.6 months; P=.0024).

Among the 142 CR1 patients with NPM1m, 67 relapsed (47.2%). In patients with NPM1m, relapses occurred during the first year in 39 (58.2% of NPM1m relapses), during the second year in 14 (20.9%) and during the third year in 2 (3%) whereas 12 relapses occurred after 3 years (17.9%), 8 occurred after 5 years (11.9%) and 3 after 8 years (4.5%). In NPM1-wild type patients, LR and VLR were significantly less frequent (<3 years: 91.9%; >3 years: 8.1%; >5 years: 2.5%; >8 years: 0.6%; P=.0317, .0037 and .0783 respectively). NPM1m relapses represented one half of LR (48%) and two thirds of VLR (67%). Among them, genotype was NPM1m/FLT3-wild type in most patients (75% in LR and 88% in VLR patients). In LR and VLR, NPM1 mutational status had no impact on CR2 and OS2: CR2LR/NPM1m: 42% vs. CR2LR/NPM1-WT: 38% (P=.8702); CR2VLR/NPM1m: 50%vs. CR2LR/NPM1-WT: 50% (P=1.0000); OS2LR/NPM1m: 7.4 months vs. OS2LR/NPM1-WT: 19.4 months (P=.2019); OS2VLR/NPM1m: 7.8 months vs. OS2VLR/NPM1-WT: 29.8 months (P=.0917).

Our data show that LR and VLR are not infrequent in AML patients with NPM1 mutations. Although this finding needs to be validated in updated multicentric cohorts with a very long follow-up, it strongly suggests that AML patients with NPM1 mutations should benefit from a prolonged follow-up beyond 5 years from CR.

Disclosures

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution