Primary Cutaneous Diffuse Large B-cell Lymphoma

Classification and subdivision of primary cutaneous diffuse large B-cell lymphoma (PCDLBCL) are a matter of ongoing debate. In this study we assessed the morphologic, immunophenotypic, and clinical features of 30 cases of PCDLBCL identified during a review of all primary cutaneous B-cell lymphomas in the Scotland and Newcastle Lymphoma Group database. We also determined the number of cases harboring t(14;18) using a polymerase chain reaction and primers to the major breakpoint cluster region. The effect on prognosis of a variety of clinical and pathologic factors was assessed for the group of 30 PCDLBCL and the 5-year disease-specific survival (DSS) of this cohort compared with that of 195 cases of stage I diffuse large B-cell lymphoma arising primarily in lymph nodes, also identified from within the Scotland and Newcastle Lymphoma Group database. Location on the leg was the only independent prognostic factor for determining outcome in PCDLBCL (67% 5-year DSS compared with 100% for the upper body; P = 0.0047). The presence of multiple lesions, involvement of more than one body site, and expression or not of CD10, bcl-2, bcl-6, and CD10 and bcl-6, had no effect on survival. Compared with cases arising above the waist, those on the leg were more often female, were of an older age, and had a significantly higher incidence of bcl-2 expression (P = 0.002) as well as the aforementioned poorer prognosis. They also showed more frequent co-expression of CD10 and bcl-6, supporting a follicle center cell origin for some, but this difference was not statistically significant. Although there was no significant difference in the 5-year DSS between the group of PCDLBCL and the cases of stage I nodal diffuse large B-cell lymphoma (88% 5-year DSS vs. 78%; P = 0.06), the latter were generally treated with more aggressive therapy. Moreover, a significant difference in 5-year DSS was seen when the nodal DLBCLs were compared with PCDLBCLs arising above the waist (78% vs. 100% respectively; P = 0.0135). These results support the current EORTC approach of subdividing PCLBCL on the basis of site to produce prognostically relevant groupings.