iiinova — where your digital self keeps evolving

Funded by SFB1530, @dfg.de, Deutsche Krebshilfe and others. Proud to work within @goetheuni.bsky.social , @dkfz.bsky.social & NCI. Thank you to all caregivers & patients. ❤️

010

Huge thanks to co-first authors Arber Qoku & Bjoern Haeupl, and mentors Lou Staudt, Florian Buettner & Thomas Oellerich. 🙏

100

🎁 In short: Integrating proteomics, scRNA/ATAC-seq & spatial profiling of the TME revealed common oncogenic themes of high-risk DLBCL tumors & proteomic framework for diagnostic and therapeutic approaches.

100

📌 Spatially, PG4's TME shows depletion & exhaustion of CD8+ T cells — pointing to a potential immune escape mechanism. 🔬

100

📌 At the molecular level, PG4 shows convergence of subsets from distinct genetic subtypes through shared high-risk biology. 📌 Key PG4 features: distinct mutational enrichment (such as BTG1), very high MYC activity & upregulated gene-regulatory activity of transcription factors TCF3/4.

100

📌 Each PG is driven by tumor-cell-intrinsic characteristics or differs in tumor microenvironment (TME) composition. 📌 One PG (PG4) was linked to poor patient outcomes, independent of known high-risk factors: cell-of-origin, prognosis scores, or genetics.

100

We studied two large cohorts of newly diagnosed #DLBCL patients using genomic sequencing, gene expression analysis & proteome profiling. 📌 Using ML-based multimodal integration, we identified molecular programs spanning data layers → 7 distinct DLBCL proteogenotypes (PGs). 🔬

100

👏 Big thank you to all co-authors (especially co-first authors Boya Wang and George Wright), reviewers/editors, colleagues, and institutions involved, especially at the NCI. Additional thanks to funding support from @dfg.de, Deutsche Krebshilfe and SFB 1530!

🚀 There is more to come building on these exciting findings and this data-rich resource.

100

-Expanding the epigenetic analysis and modeling GRNs present in normal and malignant B cells highlighted that transcriptional states of DLBCL genetic subtypes vary along three principal differentation. axes – GC B cell, memory B cell, and PC.

100

-Investigating the functional consequences of genetic alterations among the gen. subtypes revealed REL amp as a mechanism to block terminal memory B-cell diff. -By integrating single-cell RNA and ATAC from tonsillar B cells, we learned gene-regulatory networks (GRN) defining normal B-cell states.

-Inference of DNA CNV revealed that each DLBCL tumor comprised up to 5 distinct malignant subclones. -Analyzing the gene expression profiles of these subclones resulted in signature themes describing B-cell diff., cell proliferation, and cell growth that are distinctive across the subclones.

-DLBCL genetic subtypes differ strikingly regarding their tumor microenvironment. -Gene expression in malignant B cells yielded predicted signatures for each DLBCL genetic subtype and highlighted their phenotypic diversity.

100

The genetic and gene expression subtypes of DLBCL have been defined using bulk tumor profiling. To study their biology in depth, we performed single-cell RNA and ATAC sequencing on 103 tumor biopsies. This revealed multiple fascinating findings: