ALK+ LBCL and PBL share a similar plasmablastic transcriptional identity but show divergent signaling dependencies and therapeutic vulnerabilities

Abstract

Abstract Introduction: Plasmablastic lymphoma (PBL) and ALK-positive large B-cell lymphoma (ALK+ LBCL) are rare, aggressive lymphomas characterized by poor clinical outcomes and resistance to B-cell–directed therapies. Although historically classified as variants of diffuse large B-cell lymphoma (DLBCL), their biological classification and therapeutic vulnerabilities remain poorly understood. Methods: We performed integrated transcriptomic and mutational profiling of 86 patient lymphoma samples (PBL, n=56; ALK+ LBCL, n=19). Patient-derived xenograft (PDX) and PDX-derived primary culture (XDC) models were established to investigate functional dependencies and therapeutic vulnerabilities. Transcriptomic and mutational profiling were performed on PDX and XDC models to assess fidelity of models to patient samples of origin. Functional studies included CRISPR-Cas9–mediated STAT3 knockout, pharmacologic ALK inhibition, STAT3 degradation, and immune-based cytotoxicity assays using BCMA CAR T cells and monoclonal antibodies targeting CD38 and SLAMF7. Results: Transcriptomic analysis revealed that PBL and ALK+ LBCL share a plasmablastic gene expression program. Unsupervised clustering as well as a gene expression signature validated in an independent subset of samples revealed that ALK+ LBCL and PBL clustered together and fall intermediate between DLBCL and plasma cell myeloma (PCM). Mutational profiling showed that PBL harbors recurrent STAT3 gain-of-function mutations and other oncogenic alterations such as NRAS and MYC, whereas ALK+ LBCL contains relatively fewer recurrent mutations, but universally harbors ALK fusions. We created in vivo patient-derived xenograft (PDX) and PDX-derived primary culture (XDC) models from patient samples of ALK-LBCL (n=2) and PBL (n=3). These models maintained high genomic, transcriptomic, and phenotypic fidelity to primary tumors. Unsupervised clustering of patient, PDX and XDC samples showed clustering primarily by lymphoma type (ALK+LBCL vs. PBL) and secondarily by patient of origin (ALK2 vs ALK3 and PBL4 vs PBL5, etc). Functional interrogation of patient-derived models demonstrated that ALK+ LBCL is dependent on ALK fusion-mediated STAT3 signaling. Pharmacological inhibition of ALK activity most significantly suppressed STAT3 phosphorylation and MYC-associated transcriptional programs. STAT3 deletion (through both CRISPR knockout and targeted degradation) abolished ALK+LBCL cell viability in vitro and significantly reduced growth in vivo. In contrast, PBL models were insensitive to STAT3 degraders and CRSIPR knock-out, including models harboring STAT3-activating mutations, indicating distinct pathway dependencies between the two diseases. Given their shared plasmablastic phenotype, we assessed the expression of plasma cell associated antigens in ALK+ LBCL and PBL, with frequent expression of BCMA, CD38, and SLAMF7 in PBL. Functional assays in patient-derived models demonstrated that PBL is highly susceptible to plasma cell–directed immunotherapies. BCMA CAR T cells induced potent cytotoxicity in PBL models, while elotuzumab (SLAMF7) and daratumumab (CD38) elicited strong NK cell-mediated antibody-dependent cellular cytotoxicity in PBL. These immunotherapeutic approaches represent potentially transformative treatment strategies for PBL, which currently lacks effective therapeutic options. Conclusions: While PBL and ALK+ LBCL share a similar transcriptionally plasmablastic phenotype intermediate between DLBCL and PCM, they exhibit fundamentally different signaling dependencies. ALK+ LBCL is addicted to ALK-mediated STAT3 signaling and is highly sensitive to ALK-mediated STAT3 degradation, whereas PBL is not dependent on STAT3 activation (even when harboring STAT3 activating mutations), and instead demonstrates marked susceptibility to plasma cell-directed immunotherapies. These findings refine the biological classification of these lymphomas, identify new therapeutic targets, and support a shift away from DLBCL-based treatment paradigms toward more precise approaches tailored to the unique biology of plasmablastic malignancies.