Biology subjects

Filliol, A.

Publications and source records attributed to Filliol, A..

5 recordsLinked to original sources

bioRxiv2026

Beyond Segmentation Accuracy: Evaluating Biological Fidelity in Spatial Pathology

Multiplex immunofluorescence (MxIF) resolves cellular identity, state, and spatial organization within tumor tissue, enabling higher-order representations of cancer ecosystems including immune infiltration, multicellular neighborhoods, and tissue niches. Yet these analyses depend on a deceptively simple first step: accurately identifying individual cells. Segmentation errors can alter marker quantification, cellular phenotypes, cell-cell relationships, and ultimately the spatial organization inferred from an image. Consequently, improvements in conventional segmentation benchmarks matter biologically only if they propagate to more faithful representations of the tumor microenvironment. We test this relationship directly using an expert-curated benchmark of 31,924 expert-reviewed cells on three slides under a leave-slide-out protocol. We compare a domain-specific self-supervised encoder, adapted to multiplex tissue imaging, with supervised, microscopy-specific, prompted generalist, and general-purpose vision models, and pass every method's predicted cells through an identical downstream pipeline. We evaluate fidelity across four levels: cell segmentation, phenotype assignment, spatial neighborhood structure, and tissue organization, comprising niches and tumor-immune interactions. The resulting hierarchy reveals that these levels do not improve in lockstep. Cellular phenotype assignment is comparatively robust to segmentation differences, whereas missed cells increasingly disrupt neighborhood structure and tumor-immune interactions. As a stringent spatial-interaction test, tumor-CD8+ contact recovery is poorly predicted by conventional segmentation quality (Panoptic Quality: {rho} = 0.18), while errors specifically at tumor-immune interfaces are strongly associated with contact fidelity ({rho} = -0.80). Methods with nearly identical segmentation accuracy can therefore produce substantially different biological representations. These results argue that cell segmentation for spatial pathology should be evaluated not only by boundary and instance accuracy, but also by preservation of the biological relationships that downstream analyses are intended to measure.

bioinformatics↗

bioRxiv2026

Discovering Latent Scientific Concepts through Discrete Representation Learning

Scientific imaging is undergoing a fundamental transformation. Advances in multiplex immunofluorescence, computational pathology, and electron microscopy now enable observation of biological and physical systems at unprecedented spatial resolution. Rather than simply classifying known objects, these data offer the opportunity to discover previously unknown organizational principles. However, scientific image datasets are typically highly curated, information-rich, and limited in size, making reliable representation learning particularly challenging. We show that standard continuous self-supervised adaptation can become unstable in this regime, allowing teacher representations to drift and degrading downstream transfer. We introduce Kodiak, a discrete representation learning architecture that discovers latent scientific concepts through Sinkhorn-balanced codebook assignments. Instead of learning unconstrained continuous embeddings, Kodiak constructs a reusable vocabulary of discrete concepts that provides stable supervision for both masked-patch prediction and cross-view representation learning within a single-stage framework. Across five scientific imaging domains, including multiplex immunofluorescence for pancreatic cancer pathology and scanning electron microscopy, Kodiak consistently improves representation quality under both continued foundation-model adaptation and training from scratch. Beyond improved classification and transfer,the learned concepts are interpretable, reusable, and produce spatially coherent concept maps that support applications such as cell phenotyping, segmentation-free patch phenotyping, spatial niche discovery, tissue organization, and materials characterization. We view Kodiak as the representation-learning engine of our foundation model for spatial biology. More broadly, our results suggest that discrete representation learning provides a principled foundation for discovering reusable latent scientific concepts that can accelerate scientific interpretation across imaging domains.

cancer biology↗

bioRxiv2026

Senescence-directed nanotherapy ameliorates fibrosis and overcomes immune exclusion in cancer

Fibrotic remodeling of tissues and tumors establishes immunosuppressive microenvironments that drive organ dysfunction and, in cancer, limit response to immunotherapy. Senescent-like cells are conserved drivers of fibrosis and therapeutic targets, yet their functional heterogeneity complicates therapeutic intervention. Here, we show that P-selectin is expressed by a subset of senescent-like cells in fibrotic tissues and tumors. Leveraging fucoidan-based nanoparticles that bind P-selectin, we developed senescence-modulating nanoparticles (SMNPs) to selectively target these disease-associated states. SMNPs exerted potent antifibrotic and immunomodulatory effects while improving therapeutic index. Mechanistically, we identified a pathogenic, immunosuppressive macrophage population as a functional target in vivo. In fibrotic tumors, niche remodeling restored immune infiltration and sensitized tumors to immune-checkpoint-based therapies. These findings establish SMNPs as a generalizable strategy to target pathogenic senescent cell subsets across fibrosis and cancer.

cancer biology↗

bioRxiv2025

Oncogenic and tumor-suppressive forces converge on a progenitor-orchestrated niche to shape early tumorigenesis

The benign-to-malignant transition is a defining step in cancer progression. To investigate when and how malignancy initiation occurs and tissue reorganization proceeds, we combine single-cell and spatial transcriptomic profiling in mouse models of pancreatic ductal adenocarcinoma (PDAC) that capture spontaneous p53 loss. Among Kras-mutant cells, we find that oncogenic and tumor-suppressive programs, including those controlled by p53, CDKN2A, and SMAD4, are co-activated in a discrete progenitor-like population, engaging senescence-like responses. Using a framework we develop for spatial analysis, we show that a niche centered on these cells undergoes stepwise remodeling during tumor progression, mirroring invasive PDAC. Transient KRAS inhibition depletes progenitor-like cells and dismantles their niche, delaying malignancy initiation. Conversely, p53 suppression enables progenitor cell expansion, epithelial-mesenchymal reprogramming, and immune-privileged niche formation. These findings position the progenitor-like state at the convergence of cancer-driving mutations, plasticity and tissue remodeling, revealing a critical window for intercepting malignancy.

cancer biology↗

bioRxiv2022

Collagen remodeling dictates pancreatic cancer bioenergetics and outcome through DDR1 activation or degradation

Pancreatic ductal adenocarcinoma (PDAC) is a highly desmoplastic, aggressive cancer that frequently progresses by liver metastasis1. Cancer-associated fibroblasts (CAF), extracellular matrix (ECM), and type I collagen (Col I) support2-5 or restrain PDAC progression and may impede blood supply and nutrient availability6-8. The dichotomous role of the stroma in PDAC, and the mechanisms through which it influences patient survival and enables desmoplastic cancers escape nutrient limitation remain poorly understood. Here we show that matrix metalloprotease (MMP)-cleaved or intact Col I (cCol I and iCol I, respectively) exert opposing effects on PDAC bioenergetics, macropinocytosis (MP), tumor growth and liver metastasis. While cCol I activates DDR1 (discoidin domain receptor-1)-NF-{kappa}B-p62-NRF2 signaling to promote PDAC growth, iCol I triggers DDR1 degradation and restrains PDAC growth. Patients whose tumors are enriched in iCol I and low in DDR1 and NRF2 have improved median survival compared to those enriched in cCol I, DDR1 and NRF2. Inhibition of DDR1-stimulated NF-{kappa}B or mitochondrial biogenesis blocked tumorigenesis in wildtype mice but not in mice expressing MMP-resistant Col I. In summary, the diverse effects of tumor stroma on PDAC growth, metastasis, and patient survival are mediated through the Col I-DDR1-NF-{kappa}B-NRF2-mitochondrial biogenesis pathway, presenting multiple new opportunities for PDAC therapy.

cancer biology↗