Imaging the Tumor Microenvironment Beyond Anatomy: Functional, Immune, Stromal, and Metabolic Signatures for Precision Oncology

ABSTRACT

The tumor microenvironment (TME) plays an important role in predicting tumor initiation, progression, immune evasion, therapeutic resistance, relapse, and prognosis. However, cycles of oncologic imaging are mostly anatomy‐based, using tumor size and morphology as surrogate variables of the disease burden. The key limitation of this approach is that critical TME processes frequently change before structural changes and may have a decisive impact on the response to modern therapies, such as immune checkpoint inhibitors, antiangiogenic agents, and metabolism‐targeted interventions. Noninvasive whole‐body evaluation of vascular dysfunction and hypoxia, immune infiltration and activation, stromal remodeling, and metabolic reprogramming is now comprehensible to a host of functional and molecular imaging techniques, and these changes often occur before measurable changes in tumor size. In this review, we integrate emerging ideas and clinically relevant strategies for defining the TME based on imaging of biological signatures as opposed to individual modalities or the use of single‐parameter readouts. We discuss fibroblast activation protein inhibitor‐positron emission tomography (PET) imaging of the stromal phenotype, immune‐PET for immune mapping of the entire body, whole‐body systems‐level kinetics analysis by long‐axial field‐of‐view/total‐body PET, and real‐time metabolic flux evaluation by hyperpolarized ¹³ C‐magnetic resonance imaging. We also discuss how high‐dimensional multiparametric imaging can be transformed into reproducible and predictive biomarkers using radiomics and artificial intelligence. Taken together, the data indicate that TME imaging is an enabling technology for the early detection, risk stratification, response phenotyping, and adaptive treatment selection, reframing oncologic imaging as a biology‐centric decision‐support technology and not a descriptive adjunct.