Tomohiro Aoki

Summary: While end‐of‐treatment (EOT) imaging after front‐line therapy is prognostic in nodal peripheral T‐cell lymphoma (PTCL), its subtype‐specific significance remains unclear. Retrospective review of patients with nodal PTCL included in the Australasian Lymphoma and Related Diseases Registry and the Princess Margaret Cancer Centre lymphoma database between 2011 and 2024. Progression‐free survival (PFS), overall survival (OS), and positive and negative predictive values (PPV/NPV) of an EOT positron emission tomography/computed tomography response were evaluated. We identified 451 patients. Median follow‐up was 43.1‐months (95% confidence interval [CI] 38.8%–48.3%). Patients achieving an EOT imaging complete remission (CR) had a 36‐month PFS of 43.1% (95% CI 34.3%–51.5%). The 36‐month OS was significantly higher in patients achieving CR compared with those who did not achieve a CR (73.5% [95% CI 65.3%–80%] vs. 20.9% [95% CI 12.9%–30.2%]; p < 0.001). EOT imaging CR remained prognostic for OS in all nodal PTCL subtypes ( p < 0.001). At 36‐months, EOT PET demonstrated a PPV of 90% (95% CI 80.5%–96.2%) and a NPV of 43.1% (95% CI 34.3%–51.5%). EOT PET/CT CR is strongly prognostic of OS across nodal PTCL subtypes with a high PPV for predicting death but poor NPV for predicting survival. Outcomes remain poor for patients not achieving a CR, with suboptimal outcomes even in those achieving CR.

TableS1. Patient characteristics of the CHL study training cohorts. Table S2. Gene list for the targeted sequencing panel and HLGen. Table S3 Patient Demographic and molecular alterations. Table S4. HLGen correlation with Age and EBV in Training and Validation cohort. Table S5. Antibodies used for immunohistochemistry staining. Table S6. Differential gene expression analysis followed by gene ontology analysis in L-428 model. Table S7. Differential gene expression analysis followed by gene ontology analysis in U-HO1 model + IL-3. Table S8 Differential gene expression analysis followed by gene ontology analysis in U-HO1 model + IL-5. Table S9 Gene expression changes across all three experimental models. Table S10. Classification methods for HLGen. Table S11 HLGen Prediction in Training cohort. Table S12 HLGen Prediction in Stanford cohort (Alig and colleagues, Nature 2024). Table S13. Antibodies used for immunohistochemistry staining. Table S14. IMC Clustering Strategy. Table S15. Antibodies used for flow cytometry staining. Table S16. Antibodies used for Western blotting.

Figure S1. Co-occurrence of mutation in CHL. Figure S2. STAT6 copy number gain. Figure S3. Frequent mutations according to key clinical features. Figure S4. Copy number alterations according to EB Virus (EBV) status on HRS cells. Figure S5. Multivariable analyses on thymic mutation. Figure S6. Coding mutation count according to key clinical features in CHL. Figure S7. Impact of molecular profile on survival in CHL. Figure S8. STAT6 mutations in CHL. Figure S9. Impact of molecular profile on survival in elderly CHL. Figure S10. MHC status on HRS cells and their molecular correlates. Figure S11. Molecular Subgroups of CHL. Figure S12. Correlation between molecular subtypes and key clinical features. Figure S13. Molecular subtype and progression free survival (PFS) in Hodgkin lymphoma. Figure S14. Overview of the imaging mass cytometry (IMC) analysis and panel. Figure S15. Spatial tumor-microenvironment correlates with molecular subtype in CHL. Figure S16. Membrane map depicting major immune cell populations in each molecular subgroup. Figure S17. Correlation between molecular subtypes and key features. Figure S18. Evidence of CSF2RB mutant allele expression at the protein level in L-1236 KO model. Figure S19. CSF2RB mutations confer cytokine-dependent gain-of-function signaling in the L-428 model. Figure S20. CSF2RB mutations promote activation of the STAT5 signaling pathway. Figure S21. Co-expression of CSF2RB-WT and E788* mutant enhances downstream cytokine responses. Figure S22. Characterization of extrinsic phenotypes of CSF2RB mutations. Figure S23. Tumor-microenvironment characterization with molecular subtype in CHL. Figure S24. Cohort and study design overview for the HLGen development. Figure S25. Oncoplot of the CHL samples from Stanford ctDNA cohort (n = 293), with clinical annotation according to molecular subtype. Figure S26. Impact of feature list reduction on HLGen classifier performance. Figure S27. Graphical summary of molecular subtypes and their clinical and tumor-microenvironment ecosystem correlates in classic Hodgkin lymphoma. Figure S28. GeoMx analyses on CD30+ HRS cells. Figure S29. Segmentation on Imaging Mass Cytometry. Figure S30. Gating Strategy of regulatory T cell population.

TableS1. Patient characteristics of the CHL study training cohorts. Table S2. Gene list for the targeted sequencing panel and HLGen. Table S3 Patient Demographic and molecular alterations. Table S4. HLGen correlation with Age and EBV in Training and Validation cohort. Table S5. Antibodies used for immunohistochemistry staining. Table S6. Differential gene expression analysis followed by gene ontology analysis in L-428 model. Table S7. Differential gene expression analysis followed by gene ontology analysis in U-HO1 model + IL-3. Table S8 Differential gene expression analysis followed by gene ontology analysis in U-HO1 model + IL-5. Table S9 Gene expression changes across all three experimental models. Table S10. Classification methods for HLGen. Table S11 HLGen Prediction in Training cohort. Table S12 HLGen Prediction in Stanford cohort (Alig and colleagues, Nature 2024). Table S13. Antibodies used for immunohistochemistry staining. Table S14. IMC Clustering Strategy. Table S15. Antibodies used for flow cytometry staining. Table S16. Antibodies used for Western blotting.

Figure S1. Co-occurrence of mutation in CHL. Figure S2. STAT6 copy number gain. Figure S3. Frequent mutations according to key clinical features. Figure S4. Copy number alterations according to EB Virus (EBV) status on HRS cells. Figure S5. Multivariable analyses on thymic mutation. Figure S6. Coding mutation count according to key clinical features in CHL. Figure S7. Impact of molecular profile on survival in CHL. Figure S8. STAT6 mutations in CHL. Figure S9. Impact of molecular profile on survival in elderly CHL. Figure S10. MHC status on HRS cells and their molecular correlates. Figure S11. Molecular Subgroups of CHL. Figure S12. Correlation between molecular subtypes and key clinical features. Figure S13. Molecular subtype and progression free survival (PFS) in Hodgkin lymphoma. Figure S14. Overview of the imaging mass cytometry (IMC) analysis and panel. Figure S15. Spatial tumor-microenvironment correlates with molecular subtype in CHL. Figure S16. Membrane map depicting major immune cell populations in each molecular subgroup. Figure S17. Correlation between molecular subtypes and key features. Figure S18. Evidence of CSF2RB mutant allele expression at the protein level in L-1236 KO model. Figure S19. CSF2RB mutations confer cytokine-dependent gain-of-function signaling in the L-428 model. Figure S20. CSF2RB mutations promote activation of the STAT5 signaling pathway. Figure S21. Co-expression of CSF2RB-WT and E788* mutant enhances downstream cytokine responses. Figure S22. Characterization of extrinsic phenotypes of CSF2RB mutations. Figure S23. Tumor-microenvironment characterization with molecular subtype in CHL. Figure S24. Cohort and study design overview for the HLGen development. Figure S25. Oncoplot of the CHL samples from Stanford ctDNA cohort (n = 293), with clinical annotation according to molecular subtype. Figure S26. Impact of feature list reduction on HLGen classifier performance. Figure S27. Graphical summary of molecular subtypes and their clinical and tumor-microenvironment ecosystem correlates in classic Hodgkin lymphoma. Figure S28. GeoMx analyses on CD30+ HRS cells. Figure S29. Segmentation on Imaging Mass Cytometry. Figure S30. Gating Strategy of regulatory T cell population.

Bridging therapy (BT) is frequently used in patients with relapsed/refractory (R/R) large B‐cell lymphoma (LBCL) undergoing chimeric antigen receptor T cell (CAR T) therapy, but the optimal BT approach remains uncertain. We evaluated BT strategies and their associations with response and survival outcomes. Accordingly, we included patients with R/R LBCL evaluated for third‐line or later CD19‐directed CAR T at Princess Margaret Cancer Center from 2017 to 2024 (follow‐up to 01/2025). BT modalities included chemotherapy, polatuzumab‐based chemotherapy, radiotherapy, corticosteroids, and none. We assessed the associations of BT modality with overall response rate (ORR), progression‐free survival (PFS), and overall survival (OS) using logistic regression and Cox models. Among 219 patients, 188 underwent apheresis and 157 (84%) received CAR T. Of those apheresed, 79% received BT (radiotherapy 41%, polatuzumab‐based 26%, third‐line chemotherapy 16%, corticosteroids 17%, while 21% received no BT). ORR was highest with polatuzumab (50%) and radiotherapy (40%) versus chemotherapy (16%) ( p = 0.009). Median ΔSUVmax in 84 patients was higher in polatuzumab and radiotherapy treated patients, compared to chemotherapy ( p < 0.001). Radiotherapy encompassing all active disease had higher ORR (51%) versus non‐comprehensive (12%) ( p = 0.04). Two‐year PFS was highest with radiotherapy (47%) and polatuzumab (37%) BT versus chemotherapy (21%). Adjusted analyses showed improved PFS with radiotherapy (adjusted hazard ratio [aHR] 0.29) and polatuzumab (aHR 0.41). OS was also improved with radiotherapy (HR 0.40). Chemotherapy BT was associated with the highest 2‐year lymphoma‐related mortality (74%). In conclusion, polatuzumab‐based and radiotherapy BT as well as no BT were associated with improved responses and outcomes compared to chemotherapy. Although these results may have been influenced by patient selection, prospective studies integrating these approaches are needed to define optimal, individualized BT strategies.

The tissue architecture of classic Hodgkin lymphoma (cHL) is unique among cancers and is characterized by rare malignant Hodgkin and Reed–Sternberg cells that coevolve with a complex ecosystem of immune cells in the tumor microenvironment (TME). The lack of a comprehensive systems-level interrogation has hindered the description of disease heterogeneity and clinically relevant molecular subtypes. In this study, we employed an integrative, multimodal approach to characterize cHL tumors using malignant cell sequencing, spatial transcriptomics, and imaging mass cytometry. We identified four molecular subtypes (CST, CN913, STB, and CN2P), each characterized by distinct clinical features, mutational patterns, malignant cell gene expression profiles, and spatial architecture involving immune cell populations. Functional modeling of CSF2RB mutations, a characteristic feature of the CST subtype, revealed dysregulated oncogenic signaling and unique TME cross-talk. These findings highlight the significance of multidimensional profiling in elucidating patterns of molecular alterations that drive immune ecosystems and underlie therapeutically exploitable vulnerabilities. Significance: Our systematic and comprehensive genomic and spatially resolved profiling of Hodgkin lymphoma revealed a paradigmatic link between a genetic disease subtype and TME composition and function. Insights into mutation-driven mechanisms of deregulated cytokine signaling will pave the way for therapeutic interventions that interfere with oncogenic signaling and cellular ecosystems in cancer.

Classical Hodgkin lymphoma (CHL) is characterized by a complex tumor microenvironment (TME) that supports disease progression. Although immune cell recruitment by Hodgkin and Reed-Sternberg (HRS) cells is well documented, the role of nonmalignant B cells in relapse remains unclear. Using single-cell RNA sequencing (scRNA-seq) on paired diagnostic and relapsed CHL samples, we identified distinct shifts in B-cell populations, particularly an enrichment of naïve B cells and a reduction of memory B cells in early-relapse CHL compared to late-relapse and newly diagnosed CHL. The enrichment of naïve B cells in early relapse biopsies was confirmed in independent validation cohorts using scRNA-seq and immunohistochemistry. Notably, naïve B cells in early-relapse samples exhibited high expression of LGALS9, an immunosuppressive gene encoding galectin-9, which binds to HAVCR2 (T-cell immunoglobulin and mucin domain–containing protein 3 [TIM-3]) on regulatory T cells (Tregs). Cell-cell interaction analysis revealed the importance of interactions between LGALS9+ naïve B cells and HAVCR2+ Tregs in the early-relapse setting. Spatial analysis by imaging mass cytometry confirmed close proximity of galectin-9–positive naïve B cells with TIM-3+ CD4+ T cells and HRS cells, pointing to their role in shaping an immunosuppressive niche. Our findings highlight a previously unrecognized population of galectin-9–positive naïve B cells with immunoregulatory potential in early-relapse CHL and provide new insights into the spatial and transcriptional architecture of the relapsed TME in CHL.