From comparative oncology to AI-enabled precision medicine: translational biomodels for triple-negative breast cancer.
Authors
Affiliations (4)
Affiliations (4)
- Molecular and Cellular Oncology Research Group, Cancer Biotechnology Laboratory, Technological Development Center, Federal University of Pelotas, Pelotas, Brazil.
- Department of Pathology, School of Veterinary Medicine and Animal Science, University of São Paulo - USP, São Paulo, Brazil.
- INCT T-Bio2: Translational Biodiscovery and Biomodels, CNPq, Fortaleza, Brazil.
- Department of Physiology and Pharmacology, Faculty of Medicine, Federal University of Ceará, Fortaleza, Brazil.
Abstract
Triple-negative breast cancer (TNBC), defined by the absence of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 expression or amplification, comprises a biologically heterogeneous group of tumors with aggressive clinical behavior and limited biomarker-guided treatment options. Although chemotherapy, immune-checkpoint inhibition, poly(ADP-ribose) polymerase inhibitors, and antibody-drug conjugates have expanded the therapeutic landscape, durable benefit remains constrained by genomic instability, homologous recombination deficiency, phenotypic plasticity, immune-stromal interactions, and treatment-driven evolution. This Perspective critically examines how complementary translational biomodels can be organized into a fit-for-purpose framework for TNBC precision oncology. Spontaneous canine mammary tumors provide naturally evolving disease in immunocompetent hosts, whereas patient- and species-derived organoids enable scalable functional perturbation and drug-response profiling. Patient-derived xenografts preserve clinically relevant tumor heterogeneity and treatment-selected states, while genetically engineered mouse models support mechanistic interrogation of defined oncogenic events <i>in vivo</i>. Large-animal platforms, including the Oncopig Cancer Model, may add anatomical, procedural, and longitudinal realism; however, their application to TNBC remains emergent and requires disease-specific validation. We further propose a patient-model-algorithm-patient reverse-translational loop in which histopathology, radiology, molecular profiles, and functional response data are integrated through computational pathology, radiomics, multi-omics factor models, and supervised multimodal learning. Robust implementation will require patient-level data partitioning, external validation, calibration, explainability, and explicit control of batch effects, domain shift, and interspecies bias. Rather than prioritizing a single experimental system, TNBC precision medicine should rely on coordinated evidence across models with distinct and complementary strengths and limitations. This integrated strategy may improve biomarker prioritization, therapeutic hypothesis testing, and selection of clinically actionable interventions.