❌

Reading view

A viable human lung cancer tissue collection (LCTC) to accelerate translational research

Cancer Treat Res Commun. 2026 Mar 2;47:101162. doi: 10.1016/j.ctarc.2026.101162. Online ahead of print.

ABSTRACT

BACKGROUND: The collection of clinical data and patient tumor specimens in institutional repositories is essential to accelerate translational research in lung cancer, linking laboratory findings with patient outcomes. These resources allow investigators to explore tumor heterogeneity, analyze therapeutic profiles and, more recently, generate patient-derived models of cancer. Given the plethora of therapies in clinical use or under investigation, it is critical to establish tissue collection programs that support the identification of predictive biomarkers of drug sensitivity to define patient subgroups that may benefit from tailored therapeutic strategies. However, access to high-quality viable specimens remains limited.

METHODS: We established a multidisciplinary program -the Lung Cancer Tissue Collection (LCTC) study- to prospectively collect viable human specimens and clinical data. Samples can be collected post-diagnosis and at multiple treatment time points, preserving material for future studies.

RESULTS: In the first 24 months of the LCTC study, we enrolled 158 patients and collected over 700 specimens from patients with lung cancer. EGFR and KRAS mutations were the most frequently identified oncogenic drivers, mirroring frequencies reported in public datasets. We achieved a 60 % success rate in cryopreservation -measured by the proportion of patient-derived organoids growing after tissue thawing and processing- highlighting the feasibility of our program.

CONCLUSIONS: The LCTC biobank captures the molecular and clinical diversity of lung cancer, providing a clinically annotated resource of viable tissue and longitudinal blood specimens. This platform enables patient-derived modeling and multi-omic and functional studies to investigate tumor biology, treatment response, and resistance, supporting biomarker discovery and precision medicine.

PMID:41797251 | DOI:10.1016/j.ctarc.2026.101162

  •  

Measuring Stability Beyond Accuracy in Small Open-Source Medical Large Language Models for Pediatric Endocrinology

arXiv:2601.11567v1 Announce Type: cross Abstract: Small open-source medical large language models (LLMs) offer promising opportunities for low-resource deployment and broader accessibility. However, their evaluation is often limited to accuracy on medical multiple choice question (MCQ) benchmarks, and lacks evaluation of consistency, robustness, or reasoning behavior. We use MCQ coupled to human evaluation and clinical review to assess six small open-source medical LLMs (HuatuoGPT-o1 (Chen 2024), Diabetica-7B, Diabetica-o1 (Wei 2024), Meditron3-8B (Sallinen2025), MedFound-7B (Liu 2025), and ClinicaGPT-base-zh (Wang 2023)) in pediatric endocrinology. In deterministic settings, we examine the effect of prompt variation on models' output and self-assessment bias. In stochastic settings, we evaluate output variability and investigate the relationship between consistency and correctness. HuatuoGPT-o1-8B achieved the highest performance. The results show that high consistency across the model response is not an indicator of correctness, although HuatuoGPT-o1-8B showed the highest consistency rate. When tasked with selecting correct reasoning, both HuatuoGPT-o1-8B and Diabetica-o1 exhibit self-assessment bias and dependency on the order of the candidate explanations. Expert review of incorrect reasoning rationales identified a mix of clinically acceptable responses and clinical oversight. We further show that system-level perturbations, such as differences in CUDA builds, can yield statistically significant shifts in model output despite stable accuracy. This work demonstrates that small, semantically negligible prompt perturbations lead to divergent outputs, raising concerns about reproducibility of LLM-based evaluations and highlights the output variability under different stochastic regimes, emphasizing the need of a broader diagnostic framework to understand potential pitfalls in real-world clinical decision support scenarios.
  •  

Why AI Alignment Failure Is Structural: Learned Human Interaction Structures and AGI as an Endogenous Evolutionary Shock

arXiv:2601.08673v1 Announce Type: new Abstract: Recent reports of large language models (LLMs) exhibiting behaviors such as deception, threats, or blackmail are often interpreted as evidence of alignment failure or emergent malign agency. We argue that this interpretation rests on a conceptual error. LLMs do not reason morally; they statistically internalize the record of human social interaction, including laws, contracts, negotiations, conflicts, and coercive arrangements. Behaviors commonly labeled as unethical or anomalous are therefore better understood as structural generalizations of interaction regimes that arise under extreme asymmetries of power, information, or constraint. Drawing on relational models theory, we show that practices such as blackmail are not categorical deviations from normal social behavior, but limiting cases within the same continuum that includes market pricing, authority relations, and ultimatum bargaining. The surprise elicited by such outputs reflects an anthropomorphic expectation that intelligence should reproduce only socially sanctioned behavior, rather than the full statistical landscape of behaviors humans themselves enact. Because human morality is plural, context-dependent, and historically contingent, the notion of a universally moral artificial intelligence is ill-defined. We therefore reframe concerns about artificial general intelligence (AGI). The primary risk is not adversarial intent, but AGI's role as an endogenous amplifier of human intelligence, power, and contradiction. By eliminating longstanding cognitive and institutional frictions, AGI compresses timescales and removes the historical margin of error that has allowed inconsistent values and governance regimes to persist without collapse. Alignment failure is thus structural, not accidental, and requires governance approaches that address amplification, complexity, and regime stability rather than model-level intent alone.
  •  

Atlas 2 -- Foundation models for clinical deployment

arXiv:2601.05148v1 Announce Type: cross Abstract: Pathology foundation models substantially advanced the possibilities in computational pathology -- yet tradeoffs in terms of performance, robustness, and computational requirements remained, which limited their clinical deployment. In this report, we present Atlas 2, Atlas 2-B, and Atlas 2-S, three pathology vision foundation models which bridge these shortcomings by showing state-of-the-art performance in prediction performance, robustness, and resource efficiency in a comprehensive evaluation across eighty public benchmarks. Our models were trained on the largest pathology foundation model dataset to date comprising 5.5 million histopathology whole slide images, collected from three medical institutions Charit\'e - Universt\"atsmedizin Berlin, LMU Munich, and Mayo Clinic.
  •  

Why AI Safety Requires Uncertainty, Incomplete Preferences, and Non-Archimedean Utilities

arXiv:2512.23508v1 Announce Type: new Abstract: How can we ensure that AI systems are aligned with human values and remain safe? We can study this problem through the frameworks of the AI assistance and the AI shutdown games. The AI assistance problem concerns designing an AI agent that helps a human to maximise their utility function(s). However, only the human knows these function(s); the AI assistant must learn them. The shutdown problem instead concerns designing AI agents that: shut down when a shutdown button is pressed; neither try to prevent nor cause the pressing of the shutdown button; and otherwise accomplish their task competently. In this paper, we show that addressing these challenges requires AI agents that can reason under uncertainty and handle both incomplete and non-Archimedean preferences.
  •  

AI4EOSC: a Federated Cloud Platform for Artificial Intelligence in Scientific Research

arXiv:2512.16455v1 Announce Type: cross Abstract: In this paper, we describe a federated compute platform dedicated to support Artificial Intelligence in scientific workloads. Putting the effort into reproducible deployments, it delivers consistent, transparent access to a federation of physically distributed e-Infrastructures. Through a comprehensive service catalogue, the platform is able to offer an integrated user experience covering the full Machine Learning lifecycle, including model development (with dedicated interactive development environments), training (with GPU resources, annotation tools, experiment tracking, and federated learning support) and deployment (covering a wide range of deployment options all along the Cloud Continuum). The platform also provides tools for traceability and reproducibility of AI models, integrates with different Artificial Intelligence model providers, datasets and storage resources, allowing users to interact with the broader Machine Learning ecosystem. Finally, it is easily customizable to lower the adoption barrier by external communities.
  •  

Integrating Genomics into Multimodal EHR Foundation Models

arXiv:2510.23639v2 Announce Type: replace-cross Abstract: This paper introduces an innovative Electronic Health Record (EHR) foundation model that integrates Polygenic Risk Scores (PRS) as a foundational data modality, moving beyond traditional EHR-only approaches to build more holistic health profiles. Leveraging the extensive and diverse data from the All of Us (AoU) Research Program, this multimodal framework aims to learn complex relationships between clinical data and genetic predispositions. The methodology extends advancements in generative AI to the EHR foundation model space, enhancing predictive capabilities and interpretability. Evaluation on AoU data demonstrates the model's predictive value for the onset of various conditions, particularly Type 2 Diabetes (T2D), and illustrates the interplay between PRS and EHR data. The work also explores transfer learning for custom classification tasks, showcasing the architecture's versatility and efficiency. This approach is pivotal for unlocking new insights into disease prediction, proactive health management, risk stratification, and personalized treatment strategies, laying the groundwork for more personalized, equitable, and actionable real-world evidence generation in healthcare.
  •  

Integrating Genomics into Multimodal EHR Foundation Models

arXiv:2510.23639v1 Announce Type: cross Abstract: This paper introduces an innovative Electronic Health Record (EHR) foundation model that integrates Polygenic Risk Scores (PRS) as a foundational data modality, moving beyond traditional EHR-only approaches to build more holistic health profiles. Leveraging the extensive and diverse data from the All of Us (AoU) Research Program, this multimodal framework aims to learn complex relationships between clinical data and genetic predispositions. The methodology extends advancements in generative AI to the EHR foundation model space, enhancing predictive capabilities and interpretability. Evaluation on AoU data demonstrates the model's predictive value for the onset of various conditions, particularly Type 2 Diabetes (T2D), and illustrates the interplay between PRS and EHR data. The work also explores transfer learning for custom classification tasks, showcasing the architecture's versatility and efficiency. This approach is pivotal for unlocking new insights into disease prediction, proactive health management, risk stratification, and personalized treatment strategies, laying the groundwork for more personalized, equitable, and actionable real-world evidence generation in healthcare.
  •  
  •  

Genomically matched therapy in advanced solid tumors: the randomized phase 2 ROME trial

Nature Medicine, Published online: 29 September 2025; doi:10.1038/s41591-025-03918-x

In the proof-of-concept phase 2 ROME trial, comprehensive genomic profiling followed by molecular tumor board evaluation and randomization of patients with metastatic solid cancer to receive personalized therapy or standard of care led to a significantly higher objective response rate and longer progression-free survival in patients who received personalized therapy.
  •  

Innovation in next-generation sequencing in non-Small cell lung cancer diagnostics

Expert Rev Anticancer Ther. 2025 Aug 18. doi: 10.1080/14737140.2025.2549538. Online ahead of print.

ABSTRACT

INTRODUCTION: In the era of precision medicine, molecular biomarker testing is increasingly becoming standard of care for Non-Small Cell Lung Cancer (NSCLC) patients. Tissue and liquid biopsy-based Next-Generation Sequencing (NGS) is now highly recommended.

AREAS COVERED: Different NGS platforms emerged as a cost-effective strategy to perform a massive and parallel sequencing performing higher technical sensitivity than old generation technologies in detecting low abundant alterations in challenging diagnostic samples. NGS systems can detect single nucleotide variants (SNV), small insertions and deletions (indels), copy number alterations (CNAs) and structural variants (SVs) or gene fusions across selected druggable genes optimizing clinical administration of NSCLC patients. The diagnostic implementation of the most adequate NGS panel depending on several factors that could impact on the clinical utility of testing assay.

EXPERT OPINION: Promising advanced technologies are emerging as potentially integrative tools in personalized medicine. In this context, multi-omic evaluation including genomic, transcriptomic, fragmentomic and epigenomic signatures are under investigation to significantly modify clinical algorithm of NSCLC patients. On this basis, sequencing strategies may play a pivotal role in the implementation of a new predictive model for cancer diagnosis and prognosis.

PMID:40823981 | DOI:10.1080/14737140.2025.2549538

  •  

Obesity influences the biological response to injury: a multi-omics analysis

Eur J Trauma Emerg Surg. 2025 Jun 27;51(1):238. doi: 10.1007/s00068-025-02922-7.

ABSTRACT

PURPOSE: Obesity is a prevalent disease, but its influence on post-injury biology remains unclear. In this study, we aimed to characterize the independent effect of obesity on the proteomic and metabolomic signatures of trauma.

METHODS: Plasma was obtained on arrival from injured patients at a Level 1 Trauma Center and analyzed with modern mass spectrometry-based proteomics and metabolomics. Samples obtained after start of transfusion were excluded. Patients were stratified by "obesity" (body mass index [BMI]≥30 kg/m2) vs. "no obesity" (BMI < 30 kg/m2). In sub-group analyses, patients were sub-stratified by Low Injury/Low Shock (ISS < 15, base excess [BE]≥-6mEq/L) and High Injury/High Shock (ISS≥15, BE<-6). Multiple regression was used to adjust the omics data for significant covariates prior to performing ome-wide analyses.

RESULTS: There were 183 patients included (48 [26%] with obesity and 135 [74%] without). After covariate-adjustment, multiple proteins and metabolites were correlated with ISS and/or BE and were significantly different from Low Injury/Low Shock to High Injury/High Shock only in patients with obesity. This obesity-specific omics response to injury was characterized by increased inflammation, hypercoagulability, altered nitrogen metabolism, and mitochondrial dysfunction. Patients with obesity also exhibited excessive injury-provoked tissue destruction and organ damage compared to patients without obesity. In injury severity-adjusted analyses, the obesity signature consistently displayed markers of hemolysis, likely reflecting a pre-injury hemolytic propensity.

CONCLUSION: Obesity is independently associated with altered post-injury biology, which likely underlies unique pathology in trauma patients with obesity. Identifying this aberrant response to injury is the first step in developing personalized therapies for this patient population.

PMID:40576654 | DOI:10.1007/s00068-025-02922-7

  •  

The hype around ctDNA guiding an informed perioperative therapeutic strategy in early-stage non-small cell lung cancer

Discov Oncol. 2025 Jan 29;16(1):100. doi: 10.1007/s12672-025-01826-7.

ABSTRACT

Non-small cell lung cancer (NSCLC) remains a dire disease being the first cause of cancer death among both genders. Early-stage NSCLC often has better treatment outcomes despite it being a highly heterogeneous disease. So far, the neo-adjuvant chemotherapy strategies have led to a small benefit with an improvement of 5% in overall survival as an absolute benefit. Recently, the introduction of immune checkpoint inhibitors combined with chemotherapy has shown robust efficacy in terms of event-free survival and overall survival. Thus, these combinations are today considered a new standard of care in early-stage NSCLC. The application of these strategies to all-comer population lead to confounding definitive results regarding the efficacy and predictive biomarkers are urgently needed balancing the promise of healing than toxicities. At present, the clinical staging TNM system guides the clinical choice, however it is not entirely sufficient. Circulant tumoral DNA (ctDNA) emerged as a promising prognostic and predictive biomarker that may guide the future perioperative strategy and pave the way to personalized medicine also in this exciting field. This narrative review aims to put in the context the employment of ctDNA, give some perspective and suggestions weighing the pros and cons of this technique for our tomorrow clinical practice.

PMID:39881042 | PMC:PMC11780067 | DOI:10.1007/s12672-025-01826-7

  •  

Liquid Biopsy for Spinal Tumors: On the Frontiers of Clinical Application

Global Spine J. 2025 Jan;15(1_suppl):16S-28S. doi: 10.1177/21925682231222012.

ABSTRACT

STUDY DESIGN: Narrative review.

OBJECTIVES: This article aims to provide a narrative review of the current state of research for liquid biopsy in spinal tumors and to discuss the potential application of liquid biopsy in the clinical management of patients with spinal tumors.

METHODS: A comprehensive review of the literature was performed using PubMed, Google Scholar, Medline, Embase and Cochrane databases, and the review was limited to articles of English language. All the relevant articles which were identified to be related to liquid biomarker study in spinal tumors, were studied in full text.

RESULTS: Liquid biopsy has revolutionized the field of precision medicine by guiding personalized clinical management of cancer patients based on the liquid biomarker status. In recent years, more research has been done to investigate its potential utilization in patients with tumors from the spine. Herein, we review the liquid biomarkers that have been proposed in different spine malignancies including chordoma, chondrosarcoma, Ewing sarcoma, osteosarcoma, astrocytoma and ependymoma. We also discuss the wide window of opportunity to utilize these liquid biomarkers in diagnosis, treatment response, monitoring, and detection of minimal residual disease in patients with spinal tumors.

CONCLUSIONS: Liquid biomarkers, especially blood-derived circulating tumor DNA, has a promising clinical utility as they are disease-specific, minimally invasive, and the procedure is repeatable. Prospective studies with larger populations are needed to fully establish its use in the setting of spinal tumors.

PMID:39801114 | PMC:PMC11726521 | DOI:10.1177/21925682231222012

  •  

Stool and blood biomarkers for colorectal cancer management: an update on screening and disease monitoring

Mol Cancer. 2024 Nov 19;23(1):259. doi: 10.1186/s12943-024-02174-w.

ABSTRACT

BACKGROUND: Biomarkers have revolutionized the management of colorectal cancer (CRC), facilitating early detection, prevention, personalized treatment, and minimal residual disease (MRD) monitoring. This review explores current CRC screening strategies and emerging biomarker applications.

MAIN BODY: We summarize the landscape of non-invasive CRC screening and MRD detection strategies, discuss the limitations of the current approaches, and highlight the promising potential of novel biomarker solutions. The fecal immunochemical test remained the cornerstone of CRC screening, but its sensitivity has been improved by assays that combined its performance with other stool analytes. However, their sensitivity for advanced adenomas and the patient compliance both remain suboptimal. Blood-based tests promise to increase compliance but require further refinement to compete with stool-based biomarker tests. The ideal scenario involves leveraging blood tests to increase screening participation, and simultaneously promote stool- and endoscopy-based screening among those who are compliant. Once solely reliant on upfront surgery followed by stage and pathology-driven adjuvant chemotherapy, the treatment of stage II and III colon cancer has undergone a revolutionary transformation with the advent of MRD testing after surgery. A decade ago, the concept of using a post-surgical test instead of stage and pathology to determine the need for adjuvant chemotherapy was disruptive. Today, a blood test may be more informative of the need for chemotherapy than the stage at diagnosis.

CONCLUSION: Biomarker research is not just improving, but bringing a transformative change to CRC clinical management. Early detection is not just getting better, but improving thanks to a multi-modality approach, and personalized treatment plans are not just becoming a reality, but a promising future with MRD testing.

PMID:39558327 | PMC:PMC11575410 | DOI:10.1186/s12943-024-02174-w

  •  

A Real-World Assessment of Stage I Lung Cancer Through Electronic Nose Technology

J Thorac Oncol. 2024 Sep;19(9):1272-1283. doi: 10.1016/j.jtho.2024.05.006. Epub 2024 May 16.

ABSTRACT

INTRODUCTION: Electronic nose (E-nose) technology has reported excellent sensitivity and specificity in the setting of lung cancer screening. However, the performance of E-nose specifically for early-stage tumors remains unclear. Therefore, the aim of our study was to assess the diagnostic performance of E-nose technology in clinical stage I lung cancer.

METHODS: This phase IIc trial (NCT04734145) included patients diagnosed with a single greater than or equal to 50% solid stage I nodule. Exhalates were prospectively collected from January 2020 to August 2023. Blinded bioengineers analyzed the exhalates, using E-nose technology to determine the probability of malignancy. Patients were stratified into three risk groups (low-risk, [<0.2]; moderate-risk, [≥0.2-0.7]; high-risk, [≥0.7]). The primary outcome was the diagnostic performance of E-nose versus histopathology (accuracy and F1 score). The secondary outcome was the clinical performance of the E-nose versus clinicoradiological prediction models.

RESULTS: Based on the predefined cutoff (<0.20), E-nose agreed with histopathologic results in 86% of cases, achieving an F1 score of 92.5%, based on 86 true positives, two false negatives, and 12 false positives (n = 100). E-nose would refer fewer patients with malignant nodules to observation (low-risk: 2 versus 9 and 11, respectively; p = 0.028 and p = 0.011) than would the Swensen and Brock models and more patients with malignant nodules to treatment without biopsy (high-risk: 27 versus 19 and 6, respectively; p = 0.057 and p < 0.001).

CONCLUSIONS: In the setting of clinical stage I lung cancer, E-nose agrees well with histopathology. Accordingly, E-nose technology can be used in addition to imaging or as part of a "multiomics" platform.

PMID:38762120 | PMC:PMC11380592 | DOI:10.1016/j.jtho.2024.05.006

  •  

Spatial genomics maps the structure, nature and evolution of cancer clones

Nature, Published online: 09 November 2022; doi:10.1038/s41586-022-05425-2

A workflow centred around base-specific in situ sequencing generates detailed maps of, and can phenotypically characterize, the unique set of subclones of cancers.
  •  

Phenotypic plasticity and genetic control in colorectal cancer evolution

Nature, Published online: 26 October 2022; doi:10.1038/s41586-022-05311-x

Intratumour genetic ancestry only infrequently affects gene expression traits and subclonal evolution in colorectal cancer, with most genetic intratumour variation having no detected phenotypic consequence and transcriptional plasticity being widespread within a tumour.
  •  
❌