Distribution and prognostic impact of LymphGen subtypes in Korean diffuse large B-cell lymphoma using clinical NGS panel.

Several molecular subclassifications of diffuse large B-cell lymphoma (DLBCL) with prognostic and therapeutic potential have been proposed; however, validation in diverse populations and real-world clinical settings remains limited. We investigated the genetic subtypes of Korean DLBCL using the LymphGen classifier and an institutionally developed 121-gene targeted next-generation sequencing (NGS) panel. We also assessed quality control metrics, copy number reliability, and benchmarking performance against an external reference data set (Memorial Sloan Kettering Cancer Center-Integrated Mutation Profiling Actionable Cancer Targets). LymphGen subtypes differed by entity. Patients with DLBCL, not otherwise specified (DLBCL-NOS; n = 123) were classified as MCD (25.2%), A53 (14.6%), ST2 (7.3%), BN2 (5.7%), EZB (3.3%), composite EZB/ST2 (1.6%), and unclassified ("other"; 42.3%). Patients with primary DLBCLs of the central nervous system (n = 77) were classified as MCD (70.1%), composite MCD/A53 (2.6%), A53 (3.9%), BN2 (2.6%), ST2 (1.3%), EZB (1.3%), and "other" (18.2%). Compared with Western cohorts, the LymphGen subtype distribution in our Korean DLBCL-NOS cohort differed notably, with MCD being significantly more prevalent. The MCD subtype tended to be associated with non-germinal center B-cell-like DLBCL and DLBCL with MYC/BCL2 double expression (DE). In R-CHOP (rituximab, cyclophosphamide, doxorubicin, Oncovin [vincristine], and prednisone)-treated DLBCL-NOS, patients with MCD, BN2, and A53 subtypes had poorer overall survival than those with EZB and ST2, particularly among patients with concurrent MYC/BCL2 DE. Using a clinically applicable NGS panel, we successfully implemented the LymphGen classification system. This study underscores the distinct genetic landscape of Korean DLBCL and highlights the utility of LymphGen in identifying high-risk subgroups, providing a basis for prognostic stratification and therapeutic optimization.