Matthias Löhr

Pancreatic cancer is one of the deadliest malignancies, with limited options for early detection and relapse monitoring. While exosomes—nanoscale extracellular vesicles carrying tumor-derived RNA—show promise as biomarkers, current technologies lack the resolution to explore their diagnostic potential at the subpopulation level. This project introduces NanoSort, a next-generation platform that combines AI-driven microfluidic design with tumor-specific organoid models to isolate and analyze exosomal subpopulations with unprecedented precision. Our recent breakthrough in elasto-inertial microfluidics enables label-free, sheathless focusing of particles as small as 25 nm. We will integrate Large Language Models (LLMs) to guide channel geometry generation, accelerating the development of customizable sorting platforms. We will produce exosomes reflective of disease subtypes using PDAC organoids that recapitulate the tumor microenvironment. These will be size-fractionated and profiled for miRNA cargo. Finally, we will assess their clinical utility in plasma from pancreatic cancer patients, comparing performance to standard markers such as CA19-9. By combining frontier technologies in AI, microfluidics, and organoid biology, NanoSort has the potential to transform liquid biopsy approaches for pancreatic cancer and pave the way for more precise, early-stage diagnostics and risk stratification.

Research problem and specific questionsHealth economic evaluations constitute an important basis for decisions on the reimbursement and use of oncology drugs in Sweden. At the time of decision there is high degree of uncertainty in survival effects from a lifetime perspective, which needs to be predicted from clinical trials with short follow-up time.The project aim is to investigate the accuracy in health economic evaluations in the assessment of oncology drugs and to investigate how these evaluations can be improved to guide decisions on the reimbursement and use of these drugs. The research questions are:how accurate are long-term survival predictions for oncology drugs compared to outcomes in clinical practice?how can relevant information outside the clinical trial and statistical methods be used to improve survival predictions of oncology drugs in health economic evaluations?Data and methodA database with longitudinal data from Swedish registries is used to answer the research questions. Question 1 is answered by studying long-term survival in clinical practice in cancer patients who have used drugs previously assessed by the Dental and Pharmaceuticals benefits agency (TLV) to inform decisions on the reimbursement or use of drugs at hospitals as recommended by the New Therapies Council (NT-Rådet). At the time of the decision the predicted long-term survival effect of the drug will be compared to the long-term survival effect in clinical practice. Question 2 is answered by validating two statistical methods which use external data outside the clinical trial.Societal relev,ance and utilisationThe project contributes to improving health economic evaluations supporting decisions on the reimbursement and use of oncology drugs in Sweden. This benefits TLV, NT-rådet, and the health care system and will increase the opportunity to attain the highest level of health benefits for limited resources in society. The project results will be published in scientific journals and will be presented on seminars at NT-rådet and TLV, which will support the dissemination of the results of the project to a wider community, authorities and healthcare providers.Plan for project realisationThe project is part of a doctoral project and will be conducted in collaboration with TLV and between three research groups at Karolinska Institutet, providing extensive experience in health economic evaluation, biostatistics and cancer survival, and strong clinical experience in oncology.

Advancing personalized approaches in cancer therapy, aiding identification and adaptation of multi-modal treatment strategies for improved outcomes depends on clinical implementation of novel diagnostic technologies. For most cancer types the risk-features used to select individuals for post-operative adjuvant multimodal therapy are suboptimal, where many patients are overtreated and others undertreated. Liquid biopsy has opened a new diagnostic avenue to detect and monitor minimal residual disease (MRD) in individual cancer patients, especially for selecting patients for multi-modal therapies post-operatively. However, despite many circulating tumor DNA (ctDNA) diagnostics being developed there is a lack of standardization, harmonization, and robust data to demonstrate clinical validity. GUIDE.MRD is a consortium of leading academics, technology companies, pharmaceutical companies, and experts in multi-stakeholder engagements. Together, we will tackle the critical questions by developing reference standards for ctDNA diagnostics, clinically validate promising ctDNA diagnostics and develop data to guide the use of multi-modal therapies with a non-invasive diagnostic test. With robust engagement with regulatory authorities, payers and importantly patients themselves, we will develop recommendations and guidelines based on objective data to use ctDNA diagnostics to guide multi-modal therapy selection to improve patient outcomes.

Pancreatic cancer (PDAC) is usually detected at late stages and most patients die within one year after diagnosis. In PANCAID we will therefore develop a blood test for early detection of PDAC. Despite tremendous technological advances in Liquid Biopsy Diagnostics (LBx), this goal is very ambitious because small tumors release only minute amounts of cells or cellular products (e.g., DNA, RNA, protein, metabolites) into the circulation. Thus, tests with a high sensitivity are required but increases in sensitivity are usually achieved on the expenses of reduced specificity which can lead to significant overdiagnosis leading to unnecessary stress for the individuals with a false-positive blood test and high costs for the health system. In PANCAID, we will therefore establish a blood test with high accuracy by analyzing large cohorts of patients with PDAC and its precursor lesions, individuals at risk to develop PDAC and appropriate age-matched control groups (healthy and non-cancer diseases frequent in the targeted population). Ambitious objectives of PANCAID include (1) establishment of a unique resource of blood samples of early PDAC and risk groups (WP1)

(2) Establishment of a breakthrough blood test for early diagnosis of PDAC (WP2)

(3) Identification of the best composite biomarker panel by integrating multimodal features in an AI-assisted computational analysis

(4) Analysis of the socio-economic impact of early PDAC diagnosis (WP4)

and (5) Definition of the ethics parameters relevant to early PDAC detection (WP5). A robust multi-biomarker panel will be determined during the training period (year 1-3) and subsequently validated on bio-banked blood samples (year 4-5). Depending on the outcome of this comprehensive analysis, PANCAID will provide the design of a future prospective study for validation of the developed composite blood test in an international multi-center setting required to introduce LBx into screening programs for high-risk individuals. This action is part of the Cancer Mission cluster of projects on ‘Prevention, including Screening’.

It is well known that early diagnosis dramatically increases survival rates of cancers. The merging of advanced diagnostics with therapeutics (theranostics) is poised to reshape future cancer care. Pancreatic cancer is the 4th leading cause of cancer-related death in the western world with a 5-year survival rate much below 10%. This interdisciplinary programme aims to develop beyond the state-of-the-art theranostic fiber optic technology for pancreatic cancer. Our vision is to take the next step of functionalization of optical fibers, creating a palette of novel components and methods beyond state-of-the-art. On one hand, we take minimal invasiveness of fine-needle aspiration (FNA) based cytology to the next level by developing a complete “in-vivo cell picking platform” capable of selectively retrieving cells of interest for in-vitro single cell based omics. In parallel, based on the highly predictive flow characteristics in microfluidics we will develop systems for in-vivo high dose local chemo-therapy,currently impossible due to toxicity to other cells. Finally, fibers will be developed for advanced imaging and phototherapy applications. We direct our work to pancreatic cancer, but the nature of the platform will not preclude translation to other disease theranostics. Our synergies will strengthen research that is not possible without this interdisciplinary team.