Clinical characterization of acute myeloid leukemia with myelodysplasia-related changes as defined by the 2008 WHO classification system

The classification of acute myeloid leukemia (AML) has evolved from being based on morphologic and cytochemical findings, as included in the French-American-British proposal,^1,2  to systems that incorporate cytogenetic abnormalities.^3,–5  In 2001, the World Health Organization (WHO) classification for tumors of hematopoietic and lymphoid tissues was proposed in an attempt to define more biologically homogeneous entities that have clinical relevance. As it relates to AML, this includes limited cytogenetic findings, presence of morphologic dysplasia, and prior therapy.⁵  Although later studies have validated this system,^6,–8  including the importance of multilineage dysplasia, others have suggested that multilineage dysplasia correlates with unfavorable cytogenetics and has no independent impact on prognosis.^9,10 

In 2008, a revision of the WHO classification has incorporated recently acquired genetic information into an updated classification scheme of AML.¹¹  One of the revisions includes a new “AML with myelodysplasia-related changes” (AML-MRC) group. Patients are assigned to this group for any one of 3 reasons: (1) AML arising from previous myelodysplastic syndrome (MDS) or an MDS/myeloproliferative neoplasm, (2) AML with a specific MDS-related cytogenetic abnormality, and/or (3) AML with multilineage dysplasia.¹¹  The goal of the current study was to clinically characterize this newly defined AML-MRC subgroup as well as to evaluate frequent mutations present in AML, including NPM1, FLT3, and CEBPA.

A total of 100 consecutive AML patients diagnosed at Stanford University Medical Center between 2005 and 2007 with adequate material for mutation analysis were studied. All cases were diagnosed with bone marrow aspirates, blood smears, trephine biopsies, and/or flow cytometry. Clinical parameters, hemogram data, and flow cytometry results at the time of diagnosis were reviewed. Clinical follow-up information was obtained by retrospective review of the electronic charts. Cytogenetic risk group stratification was performed using Southwest Oncology Group criteria.^7,12  This study has been approved by Stanford's Institutional Review Board.

The FLT3-ITD, FLT3-D835, and exon 12 NPM1 insertion mutations were detected by multiplex polymerase chain reaction followed by restriction enzyme detection and capillary electrophoresis.^13,–15  The entire coding region of CEBPA was polymerase chain reaction amplified and sequenced (Snaddon et al,¹⁶  and J.-Y. Ahn, K.S., O.K.W., S.D. Boyd, and D.A.A., J Mol Diag, manuscript in review).

Overall survival (OS), progression-free survival (PFS), and complete response (CR) were defined as previously described.¹⁷  These parameters were compared using Kaplan-Meier methods and log-rank test. Univariate and multivariate Cox proportional hazard models were performed. Quantitative factors were treated as continuous variables inthese regression models. Categorical variables were compared using Fisher exact test.

The cases included 57 men and 43 women with a median age of 56 years (range, 17-81 years). Follow-up and therapy information was available for 90 patients. Most patients received idarubicin and cytarabine as induction therapy (81 of 90, 90%) and high-dose or standard-dose cytarabine for consolidation (75 of 90, 83%). Twelve patients underwent a bone marrow transplantation.

Among the 90 patients with follow-up, the median OS was 373 days (95% confidence interval, 284-503 days) and the median PFS was 254 days (95% confidence interval, 222-349 days). CR was achieved in 60 patients (67%). A univariate analysis showed that advanced age (> 60 years) predicted worse OS (P = .001) and PFS (P = .04). Stratification of patients into cytogenetic risk groups resulted in 9 patients with favorable, 65 with intermediate, and 19 with unfavorable risk status and correlated with significant differences in OS (P = .001), PFS (P = .001), and achievement of CR (P = .001).

Using the 2008 WHO criteria resulted in the distribution of AML subcategories listed in Table 1. The percentage of patients encompassed by the AML-MRC category was 48%, compared with prior reports of AML with multilineage dysplasia comprising 24% to 38%.^6,–8,18  Overall, 26 patients had an NPM1 mutation (16 were FLT3 mutated), 25 had FLT3-ITD alone, 8 had FLT3-D835 alone, and 9 had a CEBPA mutation (3 were FLT3 mutated). The frequency of these mutations is within the range of prior studies.¹⁹ CEBPA mutations, associated with favorable prognosis,²⁰  were significantly absent from AML-MRC (P = .017) with no significant differences in the distribution of other mutations.

Comparison of the clinical outcome of the newly defined group AML-MRC with AML-not otherwise specified (NOS) showed that AML-MRC had significantly worse OS, PFS, and lower CR rate (P = .001; Figure 1). Even after excluding the 14 patients with unfavorable cytogenetics from the AML-MRC group, the remaining AML-MRC patients had worse outcomes compared with all AML-NOS patients (OS, P = .013; PFS, P = .012; CR, P = .008). Among 65 patients with intermediate-risk cytogenetics, the outcome difference between the AML-MRC and AML-NOS remained significant (OS, P = .029; PFS, P = .023), also indicating prognostic significance of multilineage dysplasia. This confirms the previously observed clinical significance of multilineage dysplasia,^6,–8  when strictly defined by the WHO criteria.

A multivariate Cox proportional hazard analysis, performed on the entire group, identified unfavorable cytogenetic risk group, advanced age (> 60 years), FLT3-ITD, and AML-MRC status as significant predictors of worse OS (Table 2). Checking the interaction terms in this Cox model confirmed that AML-MRC predicted poor survival, independent of age or cytogenetic status.

Patients with AML-MRC were significantly older (59 vs 51 years, P = .014) and had higher frequency of unfavorable cytogenetics (14 of 46 vs 3 of 36, P = .014) compared with AML-NOS. The association of multilineage dysplasia with unfavorable cytogenetics has been previously reported^7,9,10 ; however, the difference in age could be attributed to the new definition of AML-MRC as some prior studies have not reported a significant age difference.¹⁰  Patients with AML-MRC presented with lower hematocrit (28% vs 33%, P = .014) and their blasts more frequently expressed CD14 compared with AML-NOS (10 of 46 vs 4 of 36, P = .048), with no other significant differences in antigen expression.

Within the group of 46 patients with AML-MRC, a low platelet count (< 20 000/μL) correlated with worse OS (P = .046) and shorter PFS (P = .029). A wild-type NPM1/mutated FLT3 pattern in AML-MRC resulted in significantly worse PFS (P = .038) compared with other AML-MRC cases. The presence of FLT3-D835 mutation alone in this category also correlated with worse OS (P = .026) compared with wild type FLT3 cases. Although the importance of this mutation has been controversial,¹⁹  Whitman et al recently showed that FLT3-D835 mutation correlates with worse clinical outcome in younger adults with AML.²¹ 

The clinical outcome of patients with a history of MDS was not significantly different from the remaining cases of AML-MRC (OS, P = .249; PFS, P = .265), consistent with prior studies.⁷  The presence of MDS-related cytogenetic abnormalities correlated with a significantly worse OS (P = .002) and PFS (P = .001). Of the 14 patients with MDS-related cytogenetic abnormalities, 7 had morphologic dysplasia. Further analysis showed that 32 patients with multilineage dysplasia in the absence of cytogenetic abnormalities have a better outcome than 7 patients with MDS-related cytogenetic abnormalities but without dysplasia (OS, P = .053; PFS, P = .023). However, the group with dysplasia still had worse outcomes compared with all AML-NOS patients (OS, P = .013; PFS, P = .012; CR, P = .008), suggesting that, whereas the absence of cytogenetic abnormalities in AML-MRC indicates a possible better prognosis, the presence of multilineage dysplasia, as defined by the WHO, retains prognostic significance.

In conclusion, the newly defined WHO category of AML-MRC exhibits a significantly worse clinical outcome compared with AML-NOS and is predictive of worse OS in the multivariate analysis of AML patients, independent of age or cytogenetic risk group. These findings support the clinical, morphologic, and cytogenetic criteria for this 2008 WHO AML category.

Contribution: O.K.W. and D.A.A. designed the research, analyzed results, and wrote the manuscript; K.S. and L.M. performed experiments; M.S. and J.G. collected the clinical data; L.R. performed statistical analysis; and J.L.Z., J.D.M., and J.G. assisted with writing the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Olga K. Weinberg, Stanford University Medical Center, Department of Pathology, 300 Pasteur Drive, Room L235, Stanford, CA 94305; e-mail: okw@stanford.edu.