Diffuse large B-cell lymphoma (DLBCL) exhibits marked heterogeneity, complicating treatment and prognosis. The role of plasma cell-like phenotypes in DLBCL remains underexplored. Using spatially-resolved transcriptomics, we profiled four distinct plasma cell-like phenotypes in DLBCL based on CD20, HLA-DRA, and PRDM1 markers: CD20(+)HLA-DRA(+)PRDM1(-), CD20(+)HLA-DRA(+)PRDM1(+), CD20(+)HLA-DRA(-)PRDM1(-), and CD20(+)HLA-DRA(-)PRDM1(+). The CD20(+)HLA-DRA(+)PRDM1(-) phenotype (non-plasma cell-like phenotype) was associated with better prognosis whereas the other three phenotypes (plasma cell-like phenotypes) correlated with worse prognosis. Patients with 30% plasma cell-like phenotype cells were classified as DLBCL plasma cell-like phenotype predominant (DLBCLPCPP), and those with ≤30% as DLBCL plasma cell-like phenotype deficient (DLBCLPCPD). Plasma cell-like phenotype cells correlated with reduced proliferation, impaired immune function, and enhanced tumor microenvironment remodeling. DLBCLPCPP demonstrated a lower proportion of patients with Ki-67 ≥85% (29.6% vs. 53.9%, p=0.0198), similar responses to first-line treatment (92.6% vs. 92.6%, p=1.00), but a higher rate of disease progression within 12 months (25.9% vs. 3.7%, p=0.0238) and an increased incidence of B2M alterations (22.2% vs. 3.7%, p=0.0509) compared to DLBCLPCPD. Transcriptomic analyses revealed closer proximity between plasma cell-like phenotype cells and plasmablastic lymphoma. Leveraging plasma cell-like phenotype-associated genes, we constructed a random forest model to predict response to bortezomib. High plasma cell-like phenotype signatures were linked to poorer progression-free survival (PFS) but greater benefit from R-CHOP plus bortezomib, while low-signature patients achieved better PFS with R-CHOP alone. These findings characterize the transcriptomic features of distinct plasma cell-like phenotypes in DLBCL, providing new insights into its prognostic relevance and potential for individualized treatment strategies.
Xie et al. (Thu,) studied this question.