Immunophenotypical Sequential MRD Assessment in AML for Prediction of Relapse and Definition of Putative Future Intervention Time Points.

Abstract

Immunophenotypical assessment of minimal residual disease (MRD) has been shown by us (Feller et al, Leukemia 18:1380, 2004) and others (San Miguel et al., Blood 98: 1746, 2001, Venditti et al., Blood 96: 3948, 2000) to be predictive for clinical outcome in AML both in patients treated with standard chemotherapy (SD) and high dose chemotherapy followed by autologous (AutoTQ) or allogeneic (AlloTQ) transplantation. Directly after consolidation therapy using a cut-off of 0.11% in the high MRD group the risk of relapse was 7.2 fold than in the low MRD group (Feller et al., 2004). Using that cut-off level, in the present study another important issue, i.e. sequential MRD assessments after end of treatment were done to i) prove consistancy of low MRD% in patients that remain in CR, ii) prove the ability to predict relapse using increase of MRD% and iii) assess the minimally required inter-sampling period to make MRD assessment clinically useful. To enable such 124 samples of 33 patients (<60 years) were included: 16 after SD, 8 after AutoTQ and 9 after AlloTQ. All time points refer to post-treatment periods, time point zero reflecting end of therapy.Of 16 patients studied after SD, 15 had MRD <0.11%. Of these, 12 are in CR with a median follow-up of 30.5 (range 9–46) months. In these 12 patients MRD remained consistently <0.11 %, (mean of 3 sequential MRD assessments per patient). Of these 15 patients 3 relapsed after 9, 9 and 26 months, but in these cases the last time point of MRD assessment (still <0.11%) unfortunately was at least 6 months before relapse. One patient who started with MRD>0.11 relapsed within 2 months after that MRD assessment.Of 8 patients studied after AutoTQ, 6 had MRD <0.11%. Of these, 4 are in CR with a follow-up of 6, 15, 40 and 72 months. Again, MRD remained consistently <0.11% (mean of 3 assessments per patient). Of these 6 patients 2 relapsed after 13 and 43 months, with relapse predictable using 3 months assessment intervals. Two patients had >0.11% MRD and relapsed after 5 and 11 months and within 3 months after the last MRD assessment, with no previous sign of forthcoming increase. Of 9 patients studied after AlloTQ, 2 had MRD <0.11%; thet are still in remission after 47 and 53 months. Again, MRD remained consistently <0.11 % (6 and 7 MRD assessments in these 2 patients).Of these 9 patients 2 relapsed too fast to have the opportunity to assess MRD. In addition, 5/9 had MRD >0.11%, of whom 2/5 relapsed after 6 and 7 months which was 4 and 2 months resp., after their last MRD assessment, with no previous sign of forthcoming increase. A remarkable phenomenon was observed in the other 3/5 patients: a large drop in MRD% down to detection level (0.01%) occurred 3–10 months after end of therapy. 2 of these 3 patients are still in CR (at 11 and 33 months), 1/3 relapsed after 26 months, 5 months after last MRD assessment with no previous sign of increase. The unique decrease of MRD% may be speculated to result from graft versus leukemia effects.These results show: 1) consistently low MRD% in sequential MRD assessments during persisting CR; 2) increases of MRD% which consistently predict relapse; 3) a minimally required inter-sampling period of 3 months for patients with MRD <0.11% and probably <3 months for patients with MRD>0.11%; 4) a strong decrease of MRD in part of the allogeneically transplanted patients.Results show the feasibility of sequential post-treatment MRD assessment that ultimately will contribute to design of post-treatment strategies in AML.