Reasons For Survival Improvement In Core Binding Factor AML: A 25 Year Analysis Of The UK MRC/NCRI AML Trials

Outcomes in core binding factor AML have improved over the past 25 years from a 5 year survival of 59% in 1988 to 84% post 2008. Patients were originally treated with 2 courses of DA followed by consolidation with MACE (amsacrine/etoposide) and MidAC (mitoxantrone/Ara-C). In 2002, the AML15 trials introduced an induction randomisation including FLAG-Ida, a 1:1 randomisation to add gemtuzumab ozogamicin (GO) to induction, and a 1:1 randomisation to high dose Ara-C consolidation (2 courses). For younger patients with non-CBF AML survival has improved partially because of a reduction in induction deaths and deaths in CR attributable to improved supportive care. We previously reported that GO significantly improved OS in AML [Burnett AK et al J Clin. Oncol. 2011,29: 369-377], and have recently confirmed that high dose Ara-C was superior to MACE/MidAC. Here we investigate which factors have contributed to the improvements in survival, by performing a prognostic factor analysis including date of diagnosis, and the introduction of GO and high-dose ara-C as covariates.

A total of 896 patients with cytogenetically confirmed core binding factor leukaemia from the MRC AML10, 12, 15 trials and the NCRI AML17 trial were included in the analysis. Date of diagnosis ranged from 1988 to 2012. Patients were aged between 16 and 60 years of age. A prognostic model was built using Cox regression with forward selection, at an entry probability of p<0.05, with candidate variables year of diagnosis, age, white blood cell count (WBC, both as WBC and log(WBC)), performance status, sex, secondary disease, type of abnormality (t(8;21) or inv(16)), and the use of FLAG-Ida and/or GO in induction, and high dose Ara-C in consolidation. The use of high dose Ara-C was tested only when considering outcomes post remission, to avoid issues with early deaths. Patient follow-up is complete to 1^st January 2013, with median follow-up for survival in this cohort of 8.24 years.

In multivariate analysis of survival, the following variables were significant, in order of entry to the model: use of GO in induction (HR 0.40 (0.26-0.61) p<.0001); performance status (HR 1.20 (1.07-1.34) p=0.001); age (HR per decade 1.18 (1.07-1.30) p=0.001); log WBC (HR per unit increase 1.38 (1.12-1.70) p=0.002). In particular, year of diagnosis was not significant in multivariate analysis (p=0.2) indicating that improvements over time are due to the introduction of GO, rather than a steady improvement in outcome over many years. The improvement due to GO was consistent across both types of CBF AML (test for interaction between GO and type of CBF leukaemia p=0.6). In analysis of relapse free survival log WBC (HR 1.86 (1.53-2.25) p<.0001), FLAG-Ida (HR 0.38 (0.24-0.61) p<.0001) and high dose Ara-C (HR 0.76 (0.60-0.96) p=0.02) were the only significant prognostic factors; however, for survival from remission, GO was the most significant factor (HR 0.50 (0.31-0.80) p<.0001). The improvements were seen in both cytogenetic subgroups. Although overall survival has improved patients are not immune from relapse, but the introduction of GO and to some extent FLAG-Ida and Ara-C has reduced this risk and improved survival from relapse. Since 2009, all patients' treatment has included GO in induction and high dose Ara-C consolidation (2 courses) which in 176 patients overall has yielded 5-year survival of 89%.

Core binding factor leukaemia is now highly curable with the most important contribution being the introduction of GO in induction followed by the adoption of high dose Ara-C (2 courses) in consolidation. When combined we observe an OS of 89%

Off Label Use: Gemtuzumab Ozogamicin for AML.