Minimal Residual Disease Monitoring by RQ-PCR in Core Binding Factor Positive AML Allows Risk-Stratification and Predicts Relapse: Results of the UK MRC AML-15 Trial.

The clinical value of serial Minimal Residual Disease (MRD) monitoring in core binding factor (CBF) positive patients was prospectively assessed in the AML-15 Trial which opened in July 2002. The trial compared 3 induction regimens (DA ^V/_S ADE ^V/_S FLAG Ida), followed by randomisation in consolidation (courses 3 and 4) to either MACE or 2 doses of Ara-C (3g/m² or 1.5g/m²) and to stop or have a 5^th course (Ara-C 1.5g/m²). Patients were also randomised to receive Gemtuzumab Ozogamicin (3mg/m²) at induction and/or consolidation. Over 2500 patients have so far been recruited, with 271 CBF patients (155 t(8;21), 116 inv (16)). Complete remission (CR) and relapse rates (RR) at 4 years were 95% and 19% respectively. CBF transcripts (AML1-ETO for t(8;21), CBFB-MYH11 for inv(16)) from bone marrow (BM) and peripheral blood (PB) were measured by real-time quantitative PCR (RQ-PCR) on the 7900 HT ABI machine, at presentation, after each course of chemotherapy and 3 monthly during remission for 2 years. CBF copies were normalised to ABL gene and expressed per 10⁵ABL. The sensitivity of the RQ-PCR assay was 10⁻⁵. Data were analysed in 47 relapsed patients and in 92 patients who were in remission for >1 year. In 66 patients, where the reduction of initial CBF transcript level in BM, following induction chemotherapy, was measured, only 1 of 32 patients with >3 log reduction at remission whereas 20/34 patients with <3 log reduction have relapsed, giving relapse rates of 3% and 61% respectively, (2p<0.00001). With respect to BM post induction transcript levels, in the t(8;21) group (n=50), patients with <500 AML1-ETO copies had a 18% RR compared to 62% for patients with >500 copies (2p=0.003) and in the inv (16) patients (n=38), the RR were 8% and 58% respectively for CBFB-MYH11 copies lower or higher than 100 (2p=0.004). After consolidation and during remission, BM and PB transcript levels were also highly predictive of relapse risk. In t(8;21) patients, all 7 with BM AML1-ETO level >500 copies but only 3/45 patients with <500 copies relapsed (RR 100% ^V/_S 9%, 2p<0.0001). Moreover all 12 patients with PB level >50 copies and only 2/52 patients negative for or with <50 AML1-ETO copies relapsed (RR 100% ^V/_S 4%, 2p <0.00001). In inv (16) patients, 13/13 with >100 CBFB-MYH11 copies in BM and 3/25 patients with <100 copies relapsed (RR 100% ^V/_S 9%, 2p<0.00001). In PB, any positive level resulted in relapse in 18/18 patients compared to 2/29 (RR 7%) with a negative MRD for CBFB-MYH11 (2p<0.00001). The interval between molecular and clinical relapse was ≥3 months and there was a significant correlation between BM and PB MRD levels post induction and at first positivity after consolidation (r>0.40, p<0.05). We conclude that MRD monitoring in CBF AML allows risk stratification based on treatment response, and can predict relapse, thus opening the way to risk-directed or pre-emptive therapy. We propose that MRD monitoring by RQ-PCR should be an integral part of the management of CBF positive AML.