Background: CBF AML, with t(8;21)(q22;q22), inv(16)(p13q22) or t(16;16)(p13;q22) and the associated fusion proteins AML1/ETO or CBFβ/MYH11, has a favourable clinical prognosis although significant numbers of patients still relapse. We examined the prognostic utility of MRD monitoring by RQ-PCR and propose a simple model for prediction of impending haematological relapse.

Methods: Patients with CBF AML had samples collected at diagnosis, after induction and consolidation chemotherapy and at routine regular intervals thereafter. RQ-PCR, using the Applied Biosystems 7700 Sequence Detection System, was performed in triplicate and the final result was calculated by averaging 3 values, expressed relative to PGK levels. Stratified Cox regression was used to assess the impact of predictor variables (diagnostic, post-induction, post-consolidation RQ-PCR levels and presence of a 1 log10 increase in sequential RQ-PCR levels) on leukaemia-free survival (LFS).

Results: Of 46 patients identified with CBF AML, 29 had diagnostic, regular longitudinal samples and clinical follow up allowing further evaluation; 12 AML1-ETO and 17 CBFβ-MYH11. The median age was 39 years (range 7–68) with 52% male. Median follow up was 34 months (range 1–106) and median sample number was 6 (2–15). Twelve relapses occurred at a median of 11 months (range 4–17) from diagnosis. There were significant differences between transcript levels at diagnosis (median 1.9), post-induction (8.96*10−04), post-consolidation (5.01*10−05), in remission (1*10−06) or relapse (0.15) (p=0.01). Diagnostic, post-induction and post-consolidation RQ-PCR levels did not predict outcome. A log10 rise in a remission bone marrow sample correlated with adverse LFS and imminent risk of haematological relapse (HR 8.6). Relapses occurred a median of 60 days (range 45–272) after a log10 rise.

RQ-PCR levelsHazard Ratio (HR) LFS95% CIP-value
Diagnosis 0.22 0.02–2.9 0.25 
Post-induction 1.3 0.8–2.1 0.36 
Post-consolidation 0.8 0.5–1.2 0.27 
1 log10 increase 8.6 1.8–42 0.008 
RQ-PCR levelsHazard Ratio (HR) LFS95% CIP-value
Diagnosis 0.22 0.02–2.9 0.25 
Post-induction 1.3 0.8–2.1 0.36 
Post-consolidation 0.8 0.5–1.2 0.27 
1 log10 increase 8.6 1.8–42 0.008 

Conclusions: A 1 log10 rise in transcript levels was a highly significant predictor of haematological relapse. Transcript levels at early post-treatment time points did not predict long term outcome, cautioning against de-escalation protocols based on these results. Prospective identification of high risk patients will enable clinical trials to address the efficacy of treatment initiated at molecular progression.

Disclosure: No relevant conflicts of interest to declare.

Author notes

*

Corresponding author

Sign in via your Institution