The value of molecular stratification for CEBPA DM and NPM1 MUT FLT3 WT genotypes in older patients with acute myeloid leukaemia

Summary Older adult patients (≥60 years) with acute myeloid leukaemia (AML) are generally considered to be poor‐risk and there is limited information available regarding risk stratification based on molecular characterization in this age group, particularly for the double‐mutant CEBPA (CEBPA DM) genotype. To investigate whether a molecular favourable‐risk genotype can be identified, we investigated CEBPA, NPM1 and FLT3 status and prognostic impact in a cohort of 301 patients aged 60 years or more with intermediate‐risk cytogenetics, all treated intensively. Overall survival (OS) at 1 year was highest in the 12 patients (4%) that were CEBPA DM compared to the 76 (28%) with a mutant NPM1 and wild‐type FLT3 (NPM1 MUT FLT3 WT) genotype or all other patients (75%, 54%, 33% respectively), with median survival 15·2, 13·6 and 6·6 months, although the benefit was short‐term (OS at 3 years 17%, 29%, 12% respectively). Combination of the CEBPA DM and NPM1 MUT FLT3 WT genotype patients defined a molecular group with favourable prognosis (P < 0·0001 in multivariate analysis), with 57% of patients alive at 1 year compared to 33% for all other patients. Knowledge of genotype in older cytogenetically intermediate‐risk patients might influence therapy decisions.

In recent decades there have been considerable improvements in the long-term outlook for younger adult patients with acute myeloid leukaemia (AML) (Burnett et al, 2011). Current therapy is risk-adapted, based predominantly on cytogenetics and molecular characterization (Dohner et al, 2010;Dohner & Gaidzik, 2011;Ofran & Rowe, 2013), and consolidation of first remission by allogeneic transplantation is not usually considered in patients with either good-risk cytogenetics or a favourable mutation profile, defined as either mutant for nucleophosmin 1 and lacking a fms-like tyrosine kinase 3 internal tandem duplication (NPM1 MUT FLT3 WT ) or double mutant for CCAAT/enhancer binding protein-a (CEBPA DM ) (Cornelissen et al, 2012;O'Donnell et al, 2012). These two mutational categories are almost totally mutually exclusive (Green et al, 2010). In our study of younger patients, the presence of an NPM1 MUT FLT3 WT genotype was associated with a higher complete remission (CR) rate and a lower relapse rate, both contributing to improved survival (Gale et al, 2008). The presence of a CEBPA DM genotype was associated with a non-significantly higher CR rate and a significantly lower relapse rate, leading to improved long-term survival (Green et al, 2010).
However, the median age of AML at diagnosis approximates 70 years (Derolf et al, 2009), so that the majority of patients are considered to be elderly (≥60 years), and the improvements seen in the prognosis of younger patients have not been matched by improvements in this older age group (Derolf et al, 2009;Burnett et al, 2011;Thein et al, 2013).
There is limited information concerning risk stratification in the older compared to younger patients, partly because all older patients have been considered as poor-risk. The reasons for risk stratification in the older age group, however, are different from those in younger patients. In the older patient fit enough to receive intensive therapy, there is a growing consensus that more intensive therapy, similar to that used in younger patients, results in prolonged survival (Derolf et al, 2009;Oran & Weisdorf, 2012), and the quality of life is probably no worse than in those receiving best supportive care or non-intensive therapy (Alibhai et al, 2015). This does not imply that living with AML is not extremely difficult for older patients, and some informed patients might choose not to receive life-extending therapy. One of the factors to be considered in making this decision is how long patients are likely to live if they elect to receive intensive therapy, and prognostic stratification in the elderly is clearly relevant to this issue.
There is some data suggesting a better outcome, at least in the short-to medium-term, in those older patients with intermediate-risk (IR) cytogenetics and an NPM1 MUT or NPM1 MUT FLT3 WT genotype, although this largely manifests as increased duration of survival rather than cure (Buchner et al, 2009;Becker et al, 2010;Lazenby et al, 2014), and may be limited to those ≤65 years of age (Ostronoff et al, 2015). There is very little data on the impact of CEBPA DM specifically in the older age group. We have therefore determined the impact on survival of the NPM1, FLT3 and CEBPA mutation status in a cohort of 301 patients aged 60 years or more with IR cytogenetics who received intensive therapy. We first examined the impact of the presence of a CEBPA DM genotype and then considered the outcome of the combined NPM1 MUT FLT3 WT and CEBPA DM subgroup (considered as the favourable mutational profile in younger patients) compared to those with any other genotype.

Patients and mutation analysis
Genomic DNA was available from diagnostic samples of 301 (45%) of the 662 patients aged ≥60 years with IR cytogenetics and entered on the UK Medical Research Council (MRC) AML11 trial between 1990 and 1998. Median age was 67 years (range, 60-85). Compared to the 361 patients with IR cytogenetics that were not included in the study, there was no difference in age, sex or type of leukaemia (de novo/secondary), CR rate or overall survival (OS), but patients studied were more likely to have a higher present-ing white blood cell count (WBC) ( Table SI). Ethical approval for the trial was obtained from participating institution's ethics review committees and patients gave informed consent. FLT3, NPM1 and CEBPA screening were performed as previously described (Gale et al, 2008;Green et al, 2010).

Therapy, clinical endpoints and statistical methods
Details of the trial protocol have been published elsewhere (Goldstone et al, 2001). CR was defined as a normocellular bone marrow (BM) containing <5% blasts and showing evidence of normal maturation of other marrow elements. Persistence of myelodysplastic features did not preclude the diagnosis of CR. OS was the time from trial entry to death. For patients achieving CR, relapse-free survival (RFS) was the time from the date of first CR to an event (death in first CR or relapse) and cumulative incidence of relapse (CIR) was the incidence of relapse after CR, with death in CR as a competing risk.
Mantel-Haenszel and chi-squared tests were used to test for differences in demographic and clinical data by genotype. Kaplan-Meier curves were constructed for survival data and compared by means of the log-rank test. Surviving patients were censored on 9 August, 2010, with follow-up complete for 98% of patients. Median follow-up for survival was 16Á1 years (range, 13Á7-19Á5 years). Multivariate Cox models were used to analyse CIR and OS, adjusting for age, secondary leukaemia, WBC, performance status and molecular genotype. Models were fitted using forward selection, with variables added to the model if they had a P value, derived using the deviance statistic, of <0Á05. Odds ratios (OR) or hazard ratios (HR) and 95% confidence intervals (CI) are quoted for endpoints. In all cases a ratio of <1 indicates benefit. All P values are two-tailed.

Patient characteristics according to CEBPA genotype
Details of the cohort studied are shown in Table I. Overall, 28 patients (9%) were CEBPA MUT , 16 (57%; 5% of total cohort) had a single mutation (CEBPA SM ) and 12 (43%; 4%) were CEBPA DM (Fig 1, Table SII). All CEBPA DM patients had mutations that would be predicted to lead to complete loss of normal C/EBP-a activity, with N-terminal mutations leading to production of the p30 isoform or frameshift or nonsense mutations leading to a truncated protein and/or Cterminal mutations disrupting the DNA binding or leucine zipper domains. There was no significant difference between CEBPA DM , CEBPA SM and CEBPA WT patients in age, sex, type of leukaemia, WBC and incidence of either FLT3 ITD or NPM1 MUT , although it should be noted that no patient with a CEBPA DM had the NPM1 MUT FLT3 WT genotype (P = 0Á02) ( Table I).

Response to therapy and outcome of patients with CEBPA mutations
There was no evidence of a benefit in CEBPA SM patients, where response to therapy and outcome were either the same or worse than CEBPA WT patients (Table II). CEBPA DM patients had a higher CR rate than CEBPA WT patients (75% vs. 59%). Although a relatively large difference, this was not significant in multivariate analysis adjusting for baseline characteristics (OR = 0Á33, CI = 0Á08-1Á38; P = 0Á12), which is not unexpected as the number of such cases is small  (n = 12) (Table II). CIR was lower in the CEBPA DM patients compared to the CEBPA WT patients, being 44% vs. 55%, respectively, at 1 year and 67% vs. 73% at 3 years (Fig 2A), but again this did not achieve statistical significance (P = 0Á4 for CEBPA DM versus all others). Short-term survival was improved for CEBPA DM patients (median 471 d for CEBPA DM and 248 d for CEBPA WT ), although the benefit was lost by 3 years when the OS was the same (17% vs. 18%) (Fig 2B). In multivariate analysis, there was a trend for a better OS in the CEBPA DM patients when compared to other patients (HR = 0Á57, CI = 0Á31-1Á08; P = 0Á08).

Comparison of response to therapy and outcome in patients with CEBPA DM , NPM1 MUT FLT3 WT and Other genotypes
Although the CEBPA DM genotype appeared to be associated with only a limited benefit in outcome, the above analysis is likely to be influenced by the presence of good-risk NPM1 MUT FLT3 WT patients in the non-CEBPA DM group, which would attenuate any difference between CEBPA DM patients and those with a poor-risk genotype. Overall, 83 95% CI, 95% confidence intervals; CIR, cumulative incidence of relapse; CR, complete remission; CRi, complete remission with incomplete haematological recovery; DM, double mutant; HR, hazard ratio; MUT, mutant; OR, odds ratio; OS, overall survival; RFS, relapse-free survival; WT, wild-type. *Adjusted for age, secondary leukaemia, white blood cell count and performance status. patients (28%) had an NPM1 MUT FLT3 WT genotype and it was mutually exclusive with a CEBPA DM genotype. We therefore divided the patients into three groups, NPM1 MUT FLT3 WT patients, CEBPA DM patients, and all other patients (i.e. those with an NPM1 WT or NPM1 MUT FLT3 ITD genotype), hereafter called 'Other' genotypes. The CR rate in patients with an NPM1 MUT FLT3 WT genotype was 71% compared to 75% in the CEBPA DM patients and 54% in those with 'Other' genotypes (Table III). NPM1 MUT FLT3 WT or CEBPA DM patients had a lower CIR at 1 year than the remaining patients (44%, 56% and 62% respectively). However, this difference was less apparent by 3 years (66%, 67% and 78% respectively) and, although it was statistically significant in univariate analysis (P = 0Á01 for the 3-way comparison), it did not retain significance after adjustment for other factors (P = 0Á3) (Fig 3A). Survival at 1 year was highest in the CEBPA DM group, but this then fell towards the level of the patients in the 'Other' genotype category and showed no difference by 2 years (Fig 3B). Even so, multivariate analysis showed that the OS was significantly better in the CEBPA DM patients than in the 'Other' genotypic group (HR = 0Á52, CI = 0Á28-0Á97, P = 0Á04). Similarly, OS was significantly better in the NPM1 MUT FLT3 WT group compared to the 'Other' genotype group (P < 0Á0001 for univariate analysis; P = 0Á002 for multivariate analysis). Median survival was 13Á6, 15Á2 and 6Á6 months, respectively, in the NPM1 MUT FLT3 WT , CEBPA DM and 'Other' groups.

Outcome in a Molecular
As outcome was broadly comparable for the CEBPA DM and NPM1 MUT FLT3 WT patients, they were combined into a favourable-risk group, as in younger patients, together comprising 32% of the patients in this study. OS was very significantly better in this combined group than the 'Other' genotypes group, 57% vs. 33% at 1 year, and 27% vs. 12% at 3 years (HR = 0Á50, CI = 0Á38-0Á65, P < 0Á0001 in multivariate analysis) (Fig 3C). Median survival was 14Á3 months in the favourable-risk group compared to only 6Á6 months in the remainder. There was no evidence that survival in the favourable-risk group differed according to age: OS at 3 years was 30% for the 56 patients (59%) aged >65 years compared to 23% for those aged <65 years (HR = 0Á92, CI = 0Á59-1Á44; P = 0Á7) (Fig 3D), median survival 14Á3 and 14Á0 months, respectively.

Discussion
In younger adult AML patients the presence of a CEBPA DM genotype is associated with better response to treatment and improved long-term outcome (Green et al, 2010;Taskesen  et al, 2011). It is sometimes assumed that this is also true in older patients (Ossenkoppele & Lowenberg, 2015), but this has never been formally demonstrated, with such results only presented within a much wider age range of patients (Renneville et al, 2009;Wouters et al, 2009;Dufour et al, 2010;Fasan et al, 2014). The study presented here is the first to report on the impact of a CEBPA DM genotype specifically in patients ≥60 years of age. The incidence of 4% CEBPA DM in the present cohort was similar to the 5% incidence reported in our study of younger patients with IR cytogenetics (Green et al, 2010), and is consistent with other studies where age and double/single mutant status have been given (Dufour et al, 2010;Marcucci et al, 2012). A CEBPA DM genotype was associated with improved short-term survival that was significant in multivariate analysis when compared to patients without either a CEBPA DM or NPM1 MUT FLT3 WT genotype. It is clear that this improvement is relatively short-lived and does not equate to a high cure rate.
Older patients with an NPM1 MUT FLT3 WT genotype have already been shown to have an improved 1-year OS (Buchner et al, 2009;Lazenby et al, 2014;Ostronoff et al, 2015), and this was confirmed in the present cohort. The combined favourable-risk genotypic group reported here, of either a CEBPA DM or an NPM1 MUT FLT3 WT genotype, accounted for 32% of all the patients investigated, although it must be acknowledged that this cohort was restricted to patients deemed fit enough to receive intensive chemotherapy. Even in this favourable-risk genotypic group only 57% of the patients were alive at 1 year and the corollary of identifying a favourable group is that, by default, an unfavourable-risk group is also identified. In this cohort of patients, of those without a favourable-risk genotype, nearly 50% had died by 6 months and only 12% were alive at 3 years.
Although the cohort presented here was treated two decades ago, there has been very little progress in the intervening years in improving outcome in this age group and the findings are still likely to apply. This disease-related information needs to be integrated with other patient-related information, including co-morbidities, but for some patients it may influence the decision of whether or not to receive intensive therapy. Therefore consideration should be given to offering molecular screening as part of the diagnostic workup for older patients with IR cytogenetics. Nevertheless, even for the more favourable-risk group, there remains an undoubted need to develop novel therapeutic strategies for older patients with AML.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table SI. Characteristics of UK AML11 trial patients aged ≥60 years and with IR cytogenetics that were excluded and included in the molecular investigation. Table SII. Details of the mutations identified in the CEB-PA MUT cases.