Classifications for all neoplasms have been reviewed and updated in ICD-O-3, but the most extensive revision concerned hematologic malignancies.
Over the past 50 years, many classifications of leukemia and lymphoma have been proposed. Some of these had a major impact on clinical practice while others are now largely forgotten. For most of this period, however, the distinction between lymphoma and leukemia has been regarded as fundamental importance and classifications have tended to evolve separately.
Most lymphoma classifications can be grouped into two major categories. Tumors may be subdivided according to purely morphologic characteristics such as cell size and shape and the pattern of tumor growth within the lymph node or other tissue. In contrast, the Kiel classification and the Lukes and Collins classification were based on the ideas that the cells in a malignant lymphoma have undergone maturational arrest and that tumors could be classified by comparison with the normal stages of lymphocyte differentiation. In the USA, the National Cancer Institute's Working Formulation was an attempt to provide a tool for converting diagnostic data into a common format for comparative purposes. In practice, the Working Formulation became a primary classification based, like the Rappaport classification, mainly on morphologic characteristics.
A grading system was used in most lymphoma classifications to simplify the numerous tumor types into a few categories, primarily for clinical use. It is important to recognize, however, that grades were not strictly comparable between different systems of classification. In the Kiel classification, high and low grade referred to the size of cells in a tumor. Grades used in the Working Formulation were derived from prognostic data collected in the course of the original study that gave rise to the classification; in clinical terms, high grade came to mean an aggressive tumor potentially curable by chemotherapy, while low-grade lymphomas were more indolent but often incurable.
The French-American-British (FAB)(7) system provided a parallel, but distinct, system for the classification of lymphoid and myeloid leukemias and myelodysplasia based on traditionally stained specimens.
In the early 1990s, it was becoming apparent that there were many problems with the existing classification systems for leukemia and lymphoma. The introduction of immunophenotypic and molecular biological techniques had shown that individual categories were, in fact, heterogeneous. It was evident that the use of lymphoma grades as the basis for clinical trials or epidemiological studies was potentially highly misleading. As definitions became clearer, it was increasingly obvious that the distinction between lymphoid leukemias and lymphomas was largely artificial; it reflected patterns of spread in the individual patient rather than basic cellular or clinical differences. The distinction between Hodgkin disease and non-Hodgkin lymphoma was a cornerstone of lymphoma classification. However, various investigations showed that the tumor cells in Hodgkin disease were derived from germinal center B-cells and that Hodgkin disease should therefore be regarded as a distinctive form of B-cell lymphoma rather than as a completely separate group of disorders. Cytogenetic studies revealed the importance of chromosomal translocations with dysregulation of individual genes in the pathogenesis and clinical behavior of several types of leukemia and lymphoma, although achieving a complete understanding of tumor pathogenesis is clearly going to be a lengthy process.
These developments were the basis of the Revised European-American Lymphoma (REAL) classification published in 1994 (6). Although many of the terms used are similar to those used in the Kiel classification, the underlying concepts are different. In the REAL classification definitions of clinico-pathological entities are based on a combination of morphology, immunophenotype, genetic abnormalities, and clinical features. Despite the vast number of possible combinations of these variables, there are in fact relatively few disease entities, and more than 90% of lymphoid malignancies can be classified using this approach. The WHO classification of hematologic malignancies (21,22) is based on the same approach and the section on lymphoproliferative disorders is broadly similar. The approach to subclassification of acute myeloid leukemia (AML) recognizes the central importance of cytogenetic abnormalities and the distinction between "de novo" and myelodysplasia-associated AML.
The WHO classification cannot be regarded as definitive, but it provides a sound basis for future developments. Many of the major categories, such as diffuse large B-cell lymphoma, are clearly heterogeneous in terms of clinical features and response to treatment. In the future these will be further subdivided according to cellular and molecular criteria, but at present there is no consensus as to how this should be done. It is likely that the differences in the hematologic malignancy section of the next edition of ICD-O will be every bit as great as the differences between the Second and Third Editions.
WHO Classification (Part One) shows the WHO classification of Hematopoietic and Lymphoid Neoplasms with ICD-O codes.