Hematopathology / ACUTE MYELOID LEUKEMIA CLASSIFICATION

Most classification systems of acute myeloid leukemia (AML) rely largely on the criteria proposed by the French-American-British (FAB) Cooperative Group. The recently proposed World Health Organization (WHO) classification of neoplastic diseases of the hematopoietic and lymphoid tissues includes a classification of AMLs. The proposed WHO classification of AMLs includes traditional FAB-type categories of disease, as well as additional disease types that correlate with specific cytogenetic findings and AML associated with myelodysplasia. This system includes a large number of disease categories, many of which are of unknown clinical significance, and there seems to be substantial overlap between disease groups in the WHO proposal. Some disease types in the WHO proposal cannot be diagnosed without detailed clinical information, or they are diagnosed only by the cytogenetic findings. In this report, a realistic pathologic classification for AML is proposed that includes disease types that correlate with specific cytogenetic translocations and can be recognized reliably by morphologic evaluation and immunophenotyping and that incorporates the importance of associated myelodysplastic changes. This system would be supported by cytogenetic or molecular genetic studies and could be expanded as new recognizable clinicopathologic entities are described. Pathologic classifications must change continuously to reflect advances in our understanding of disease. The pathologic classification of acute myeloid leukemias (AMLs) used by most physicians today, however, does not provide the most information possible for these diseases. The French-American-British (FAB) Cooperative Group provided clear and useful criteria for the pathologic classification of AML,1,2 and the classification was modified to incorporate new disease types that required ancillary testing not used in the original descriptions.3,4 Some of the FAB categories of AML, such as M3 and M4Eo, correlate with prognostically significant cytogenetic abnormalities; however, the lack of reproducibility and clinical significance of some of the disease groups in the FAB system is now generally acknowledged.5 In addition, the FAB classification does not accommodate the importance of myelodysplasia-associated changes in adult AML. Alternative classification systems have been proposed that incorporate immunophenotyping, cytogenetic, and myelodysplastic changes,6-8 and morphologic and immunophenotypic features are now well described for some AMLs that are highly suggestive of recurring cytogenetic abnormalities.9,10 A list of the provisional World Health Organization (WHO) classification of neoplastic diseases of the hematopoietic and lymphoid tissues has been published that includes a classification for AML ❚Table 1❚.11 In the WHO system, AMLs are subdivided into 4 major categories: AML with recurrent cytogenetic translocations; AML with multilineage dysplasia; AML and myelodysplastic syndromes, therapy-related; and AML not otherwise categorized. The acceptance of cytogenetic findings and associated myelodysplastic changes in AML, which provide essential prognostic information,12,13 are major advances in the classification of these diseases. However, the WHO disease categories are Hematopathology / SPECIAL ARTICLE Am J Clin Pathol 2001;115:552-560 553 © American Society of Clinical Pathologists seemingly overlapping and include disease types that may not be evident by pathologic review alone. Unlike the classification of lymphoid neoplasms by the WHO, which is based largely on the previously reported Revised European-American Classification of Lymphoid Neoplasms,14 criteria for the myeloid categories have not been published. Some categories of the WHO classification, such as “AML with 11q23 (MLL) abnormalities” and the various “AML and myelodysplastic syndromes, therapy-related” types, cannot be classified based on morphologic, cytochemical, or immunophenotypic studies and cannot be offered as timely pathologic diagnoses. In addition, the “AML not otherwise categorized” group includes FAB Cooperative Group–related subtypes that are of questionable clinical relevance and other leukemia types that have not been defined convincingly as clinicopathologic entities. This article proposes a realistic pathologic classification of AMLs that would contain disease types that can be recognized by a combination of morphologic, cytochemical, and immunophenotyping studies. Major subdivisions of de novo AML and myelodysplasia-associated AML are proposed ❚Table 2❚. Because of the well-recognized prognostic significance of molecular genetic changes,12,15-17 all cases of AML in this system should be studied by routine karyotype analysis and, as needed, molecular genetic analysis to supplement the pathologic diagnosis. This classification is proposed as an alternative to the currently proposed WHO system for AMLs. Proposed Classification Similar to the WHO proposal, a diagnosis of acute leukemia would be based on a bone marrow blast cell count of 20% or more in most cases and should be independent of the marrow blast count for cases with changes of acute promyelocytic leukemia, AML with changes suggestive of t(8;21)(q22;q22), or AML with abnormal eosinophils suggestive of inv(16)(p13q22) or t(16;16)(p13;q11). The clinical similarities between patients with 20% or more blast cells and those with acute leukemia diagnosed with the more traditional 30% cutoff have been demonstrated.18 The proposed system is summarized in Table 2.