Predictive Value of Minimal Residual Disease (MRD) Monitoring by RQ-PCR in WT1 Positive Patients Entered in the UK MRC AML-15 Trial

Relapse remains the main cause of treatment failure in Acute Myeloid Leukaemia (AML). Some 65% of AML patients lack a specific fusion transcript which can serve as a molecular target for detecting residual disease. However, in over 75% of AML cases, the Wilm’s Tumor (WT1) gene is over-expressed and can therefore be used for MRD monitoring by RQ-PCR. The clinical value of serial MRD monitoring in WT1 positive patients was prospectively assessed in the AML-15 Trial which opened in 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 + MidAC 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. Adults in the standard and poor risk groups were scheduled to receive an allograft if a matched donor was available. An “in house” WT1 assay used primers/probes covering exons 7 and 8 in the 3′ region of the WT1 gene. WT1 transcripts from BM and PB were measured by real-time quantitative PCR (RQ-PCR) on the 7900 HT ABI machine, at diagnosis, after each course of chemotherapy, stem cell transplant, and during remission for up to 3 years. WT1 copies were normalised to ABL gene and expressed per 10⁵ABL copies. WT1 was over-expressed in >80% of AML cases screened, when compared to normal BM copies (median 544, range 31–1173), PB (median 10, r 0–58), PBSC (median 71, r 0–483). In this study, 700 patients (excluding Core Binding Factor positive AML) with diagnostic BM WT1 levels >10⁴ copies have been recruited and data on 416 are presented. Follow up is complete to 1/1/08 with 244 relapses after a median follow-up of 40 months. At complete remission (CR) following induction, a greater log reduction in WT1 transcripts was associated with a significantly reduced relapse risk even after adjustment for age, WBC, cytogenetics, performance status and de Novo/Secondary AML (hazard ratio (HR) 0.57 (0.46–0.71) per log reduction p<0.0001). The 5yr relapse rates (RR) for patients with a log reduction of <2, 2–3, and 3 or more were 77%, 50% and 26% respectively (p <0.00001). Furthermore, presence or absence of a 2 log reduction adds to the MRC risk index (developed to identify patients who might benefit from transplant or be candidates for experimental therapies, Burnett et al. ASH 2006, Abstract 18) for survival from CR (HR 0.59 (0.37–0.91), p=0.02). With respect to post induction WT1 transcripts, there were naturally three groups for both BM and PB levels for relapse risk: <1000, 1000–4999, 5000+ copies for BM (5 yr RR 45%, 74%, 82%, p<0.0001 ) and <1000, 1000–2999 and 3000+ copies for PB (5yr RR 53%, 82%, 91%, p<0.0001). After consolidation/SCT, BM and PB transcript levels were also highly predictive of relapse: BM <5000 copies RR 37%, BM 5000+ copies RR 84% (p<0.0001); PB <1000 copies RR 32%, PB 1000+ copies RR 90% (p<0.0001). Looking at all data together, in a multivariate analysis adjusted for MRC risk group, the most important prognostic variables for relapse, relapse free survival and overall survival were post induction log reduction and post chemotherapy PB transcript level >500 copies. In all cases PB transcript levels were more prognostic than levels from BM. We conclude that MRD monitoring in WT1 positive AML patients allows risk stratification based on treatment response, and in this group of patients, MRD monitoring adds to already known risk factors. Whether patients identified as being at high risk of relapse benefit from further treatment, including allogeneic stem cell transplantation, remains to be determined in future studies.