BLOODPAC offered a multi speaker seminar last fall on the topic, New Frontiers in Therapy Selection: Beyond DNA Mutations. Find the online resources now.
Here's the home page: link.
Find the two-hour YouTube webinar here: https://www.youtube.com/watch?v=yuYdhbdVcpU
As you scroll the home page, you'll also reach the 37 page white paper.
###
AI CORNER
###
Chat GPT 5.4 reads and summarizes.
###
Overview
The BLOODPAC webinar and accompanying 37-page white paper together outline a shift in precision oncology from single-modality DNA testing toward integrated, multimodal approaches. The webinar presents exploratory perspectives from industry leaders on combining genomics, transcriptomics, epigenomics, and digital pathology, while the white paper distills these into a structured synthesis. Both emphasize that future therapy selection will depend on biologically richer, longitudinal, and clinically validated models that better capture tumor complexity and treatment response.
###
These are two related but different items. The two-hour YouTube webinar is the live BLOODPAC seminar held in November 2025, with speakers, discussion, and a panel. The 37-page white paper is the later written synthesis of that seminar: a more polished, organized document that turns the talks into concise presentation summaries, key takeaways, and a structured Q&A section. In other words, the webinar is the event; the white paper is the curated write-up of the event.
The webinar opens with Lauren Leiman and Asaf Zviran setting a strategic frame for the field. Their starting point is that current precision oncology is still too dependent on narrow companion-diagnostic logic: many patients have no usable biomarker, many who do have a matched biomarker still fail therapy, and many development programs collapse for lack of efficacy. The proposed answer is not merely “more sequencing,” but integration: genomic, transcriptomic, epigenetic, digital pathology, and functional assay data should be combined to capture tumor biology more fully and make therapy selection more widely useful. The tone of the webinar is exploratory and future-facing. The speakers are not claiming that the field has already solved multimodal precision medicine; rather, they are arguing that the next leap forward lies in turning multiple imperfect data streams into a more coherent biological model of response.
In the webinar’s first scientific presentation, Tuvik Beker of Pangea Biomed presents ENLIGHT as a biology-driven response prediction platform intended to go beyond mutation matching. His central claim is that actionable mutations help only a minority of patients, and that conventional biomarkers are too siloed to capture the actual biology of drug response. ENLIGHT therefore uses gene interaction networks, grounded in concepts such as synthetic lethality and synthetic rescue, to generate a standardized “matching score” across drugs, cancers, and data types. In the live webinar, Beker spends more time than a written summary naturally can on the practical ambition of this idea: one platform spanning RNA-seq, digital pathology, and eventually liquid biopsy. He describes ENLIGHT-NGS as already validated across many patients and trials, then moves to ENLIGHT-DP, which attempts to infer gene-expression-relevant information directly from ordinary H&E slides. The most memorable part of the live talk is the sense of translational aspiration: he gives patient examples, discusses rapid turnaround from archival tissue, and frames liquid biopsy as the next step toward repeated, noninvasive treatment matching over time.
The second webinar presentation, by Alexander Savchenko of Novartis, is more operational and more pragmatic. If Beker’s talk is about prediction engines, Savchenko’s is about what digital pathology can already do in serious clinical-development settings. He emphasizes throughput, reproducibility, automation, and the ability to quantify features that traditional microscopy handles poorly or inconsistently. His case studies illustrate three levels of value: first, algorithms that reconstruct tumor masks and other morphologic features without extra staining; second, immune-phenotyping tools that measure CD8 spatial relationships and classify tumors as inflamed, excluded, or desert; and third, harmonization of PD-L1 scoring across assays and sites. In the webinar, this comes across less as abstract AI enthusiasm and more as a clinical operations argument: digital pathology becomes attractive when it is robust enough for large trials, fast enough for workflow, and validated enough to satisfy regulatory expectations.
The webinar then shifts, through Darya Chudova of Guardant, into the liquid-biopsy version of the same overall philosophy. Her key point is that genomics alone is not enough: mutations reveal the tumor’s hardware, while epigenomics can reveal something more like the software or functional state. She argues that combined genomic and epigenomic analysis from the same blood sample can generate a richer tumor phenotype, especially when tissue is limited or repeated biopsies are impractical. In the materials, this includes tumor-origin classification, subtype inference, surrogate detection of alterations such as fusions or deletions, and longitudinal monitoring of tumor evolution under treatment pressure. What stands out in the webinar is the scale argument: Guardant is not just presenting a concept but saying that very large real-world multimodal datasets can now support discovery, validation, and increasingly sophisticated response models. The live discussion also introduces a note of caution: richer models are exciting, but the field must be disciplined about validation and not confuse early positive signals with clinically durable truth.
The final major scientific talk in the webinar, by Timothy Taxter of Tempus, gives the seminar its most clinically concrete case study: endometrial cancer. Here the multimodal theme is not presented as a general platform dream but as a disease-specific implementation problem. Taxter describes how endometrial cancer has evolved from older clinicopathologic risk frameworks toward newer molecular subtyping, including POLE-mutated, p53-abnormal, MSI-high, and NSMP groups. But he also makes clear that these newer categories do not fully solve treatment selection, especially for patients in the murkier middle-risk or indeterminate groups. Tempus’s answer is to combine pathology, clinical data, and molecular tools, including an algorithm licensed from Leiden that predicts distant recurrence and may help identify which patients benefit from chemotherapy. In the webinar, this talk feels like a bridge between research and productization: it shows how multi-omic and digital-pathology tools might move from interesting signals to assays that actually alter management in a defined cancer population.
The panel discussion in the webinar broadens the conversation from individual company showcases to field-level barriers. The panelists discuss what it takes for integrated multimodal tests to reach the clinic: data integration, validation rigor, regulatory expectations, operational realities, and the need to balance innovation with patient safety. A recurring theme is that the science is moving faster than the clinical and regulatory framework. The panel does not sound naïve about this. Instead, it suggests that BLOODPAC’s role is to help define common language, validation approaches, and working-group priorities so the field does not fragment into incompatible one-off claims. The live seminar closes not with a triumphant scientific conclusion but with a practical agenda: proposed deliverables include advancing clinical pathways for multimodal therapy selection, regulatory dialogue, extension of BLOODPAC’s framework beyond blood-only assays, and cross-platform validation strategies.
The white paper, by contrast, is more distilled and more disciplined. It strips away much of the conversational texture of the webinar and reorganizes the content into a formal document with an introduction, speaker summaries, key takeaways, and a Q&A section. Its strongest function is synthesis. Rather than reproducing the full rhythm of the event, it makes the central argument legible: the future of therapy selection lies in multimodal precision medicine beyond DNA mutations alone. Each speaker section is sharpened into a thesis. Beker’s section becomes a statement about biology-informed response prediction across RNA-seq, digital pathology, and future liquid biopsy; Savchenko’s becomes a statement about digital pathology as a scalable, reproducible biomarker engine; Chudova’s becomes a statement about genomic-plus-epigenomic liquid biopsy enabling genotype and phenotype assessment over time; and Taxter’s becomes a statement about clinically implementing multi-omic algorithms in a specific disease setting. The white paper therefore reads less like an event recap and more like a field-positioning document.
As a written product, the white paper is especially useful because it clarifies the common thread running through otherwise different technologies. These speakers are not all solving the same problem in the same way. One is building a mechanistic response model; another is operationalizing digital pathology at trial scale; another is expanding blood assays into phenotype and longitudinal tracking; another is embedding multi-omic tools into endometrial-cancer care. But the white paper pulls them together around one shared message: DNA mutation testing alone is too narrow a basis for next-generation therapy selection. The field now wants models that are richer biologically, more adaptive over time, and better able to handle tumor heterogeneity in space and time.
The white paper also makes the discussion feel more mature by emphasizing takeaways rather than enthusiasm. In the Beker section, for example, the written version highlights not just promise but the conceptual logic of combining spatial context from pathology, molecular detail from sequencing, and temporal dynamics from liquid biopsy. In the Savchenko section it emphasizes reproducibility, site-to-site harmonization, and the ability to extract more information from existing slides. In the Chudova section it spotlights large annotated datasets, methylation-based inference of tumor characteristics, and improved response stratification. In the Taxter section it highlights how clinical classification systems in endometrial cancer are evolving and why extra multi-omic tools are needed precisely where current molecular groupings still leave uncertainty. The white paper therefore functions almost like a consensus-oriented executive summary for scientists, developers, and translational leaders.
Taken together, the webinar and the white paper tell the same story at two different levels. The webinar is richer in voice, ambition, anecdote, and discussion. It lets you hear how the participants think, where they are most excited, and where they are cautious. The white paper is cleaner, tighter, and more strategic. It is what you would hand to someone who wants the field summarized without watching two hours of presentations. If one had to reduce the shared conclusion to a single sentence, it would be this: BLOODPAC is arguing that the next phase of precision oncology will come from integrating multiple biological and imaging modalities into clinically usable, validated decision tools, rather than continuing to rely mainly on isolated DNA mutations as the master key to therapy selection.
If you want, I can next turn this into a tighter 700–900 word polished memo in your preferred essay style.
