Comparison of the working formulation of non-Hodgkin's lymphoma with the Rappaport, Kiel, and Lukes & Collins classifications. Translational value and prognostic significance based on review of 658 patients treated at a single institution

Six hundred fifty-eight cases of previously untreated non-Hodgkin's lymphoma seen between 1970 and 1979 at the Medical Department, the Finsen Institute, were the basis for a comparative study of the prognostic value of the Rappaport, Kiel, and Lukes & Collins classifications and the new translation system, the Working Formulation of Non-Hodgkin's Lymphoma. Each histopathologic system proved equally effective in separating patients into subgroups with a spectrum of prognoses ranging from a median survival of less than 1 year to greater than 7 years. The established classifications were compared with the Working Formulation in order to evaluate its translational value. The Working Formulation was more similar to the Rappaport and the Lukes & Collins systems than to the Kiel system, since 82%, 89%, and 75% of the cases, respectively, were translatable following the guidelines outlined in the National Cancer Institute (NCI)-sponsored study. Similarities among the four systems were demonstrated in lymphomas with follicular growth pattern, and in diffuse lymphomas composed of small mature appearing lymphocytes or small cleaved lymphocytes. Incongruity among the systems was more marked in lymphomas composed of large lymphoid cells or in lymphomas of mixed cellular composition. A comparison was performed for each classification against the Working Formulation. All such subdivided subsets were tested for prognostic heterogeneity and the following conclusions were reached: the diffuse poorly differentiated lymphocytic category of Rappaport was separated into two subgroups (malignant lymphoma ML] small cleaved cell and ML lymphoblastic) with different prognoses (P = 0.01); the diffuse "histiocytic" lymphomas were prognostically homogeneous, since none of the newer systems were able to identify subpopulations with significantly different prognoses; the subtypes of the Kiel classification were prognostically homogeneous; the only weakness of the Lukes & Collins classification was the undefined cell subtype, encompassing two populations with different prognoses; and (5) the importance of follicular growth pattern was confirmed for small cleaved cell and mixed cell cytology, whereas large cell cytology implied a poor prognosis regardless of pattern. By the use of the Cox regression model it could be demonstrated that the Working Formulation can substitute any of the established classifications in terms of prognostic value.