Abstract Background: Splicing-derived neoantigens represent a largely unexplored class of immunotherapy targets in diffuse large B-cell lymphoma (DLBCL). While mutation-derived neoantigens have been extensively studied, aberrant splicing can generate tumor-specific peptides absent from normal tissues. The SSNIP pipeline identified tumor-specific splice junctions as a prevalent source of neoantigens in solid cancers but did not include DLBCL. Methods: We analyzed RNA-seq from 775 DLBCL transcriptomes (phs000235), identifying tumor-specific junctions by filtering against 1, 303 GTEx normal lymphoid samples and validating against 17 FACS-sorted tonsillar B-cell populations spanning naive through memory stages. Neopeptides (8-11 mers) were predicted from junction-spanning reading frames, with MHC-I binding and antigen processing scores assessed via MHCflurry 2. 1 across 19 HLA alleles. Immune cell fractions (quanTIseq) and purity-adjusted partial correlations inferred cellular origin. Somatic mutations from 739 samples assessed mutation independence. Samples were classified by DLBCL-IQ immune subtype and LymphGen. Results: From 125, 062 protein-coding junctions, 7, 260 (5. 8%) were tumor-specific; none were detected in normal B-cell samples spanning all differentiation stages. After expression filtering, 3, 494 neojunctions produced 47, 849 predicted MHC binders (IC50 ≤500 nM) and 7, 871 strong binders (IC50 ≤50 nM) across 712 genes. Neojunction-producing and somatically mutated genes were largely non-overlapping (11/805, 1. 4%), and junction burden correlated with splicing factor expression (152/174 significant, led by HNRNPK, rho=0. 50) rather than mutation load (rho=-0. 01), establishing neojunctions as a mutation-independent antigen class. Immune deconvolution identified M1 macrophage infiltration as the dominant cell-type signal (87. 5% of 1, 188 significant associations), suggesting most high-prevalence neojunctions reflect tumor-associated macrophage activity. CD79A, an essential BCR component, showed the strongest correlation with the B-cell fraction (r=0. 36, FDR=1. 3×10-21) and produced 9 predicted strong MHC binders, identifying it as a candidate tumor-intrinsic neoantigen. Immune quadrant stratification revealed quadrant-specific candidates: GCB-Hot (MBOAT7, 71 strong binders, processing score=0. 90) ; ABC-Hot (UBE2F, IC50=9. 9 nM) ; GCB-Cold (SPIDR, 55 strong binders, processing score=0. 92) ; ABC-Cold (TARS1). Hot-enriched genes included LAIR1 (49% prevalence, 39 strong binders, processing score=0. 91) ; GCB-enriched genes included LMO2 and RRAS2. Conclusions: Splicing-derived neoantigens in DLBCL are abundant, mutation-independent, and stratified by immune subtype, cell-of-origin, and inferred cellular origin. CD79A emerges as a priority tumor-intrinsic candidate, while the dominant M1 macrophage signal suggests many candidates may be amenable to myeloid-directed strategies. These findings define a subtype-specific neoantigen landscape for immunotherapeutic investigation in DLBCL. Citation Format: Reid Shaw, Eric Perkey, Justin Kline. Splicing-derived neoantigens in DLBCL reveal subtype- and cell-type-specific predicted neoantigen targets abstract. In: Proceedings of the Fifth AACR International Meeting on Advances in Malignant Lymphoma: From Discovery to Clinical Impact; 2026 Jun 24-27; Philadelphia, PA. Philadelphia (PA): AACR; Blood Cancer Discov 2026;7 (3Suppl): Abstract nr A031.
Shaw et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: