Surveilling the Distinctive Vascular and Metabolic Features of Tumor Progression and Response to Therapy

John James Walsh, Yale University Graduate School of Arts and Sciences


Glioblastoma (GBM) is the most common malignant primary brain tumor in adults. Despite maximal treatment with surgical resection, radiotherapy and temozolomide chemotherapy, prognosis is dismal with median survival around 15 months. GBMs are highly infiltrative tumors that invade into surrounding brain tissue, which makes defining the extent of tumor spread difficult and recurrence common. Radiological identification of GBMs with magnetic resonance imaging (MRI), using transvers (T2) or longitudinal (T1) relaxation contrasts, is a mainstay in the initial diagnosis as well as tracking therapeutic response in GBM. However, there is extreme variability in the structural appearance, size, metabolism, and genetic landscape of GBMs, making imaging characteristics highly heterogeneous and hard to define with tumor progression. Although T2-weighted and contrast-enhanced T1-weighted MRI provides anatomical details of the tumor architecture, these methods can be confounded by pseudoprogression and pseudoresponse in the context of therapy.The GBM microenvironment is characterized by immature vasculature and extracellular acidification due to a metabolic shift towards aerobic glycolysis (Warburg effect). The reduced extracellular pH (pHe) has been associated with promoting angiogenesis and invasion as well as creating an immunosuppressive environment. Given the important contribution of vascular changes and extracellular acidosis to shaping the tumor microenvironment, advanced MRI techniques are needed to better characterize the tumor microenvironment to provide more specific readouts of tumor progression and therapeutic response. Biosensor Imaging of Redundant Deviation in Shifts (BIRDS) is a magnetic resonance spectroscopic imaging (MRSI) technique that utilizes the temperature and pH-dependent hyperfine shifts of paramagnetic agents (e.g., TmDOTP5-) for high resolution, three-dimensional, quantitative temperature and pHe mapping. BIRDS has been used to demonstrate the acidic pH in preclinical models of GBM, where the intratumoral space is highly acidified (pH<6.8) in comparison to healthy brain tissue (pH~7.2) and acidic pH spread beyond the anatomically defined tumor core relates to the invasiveness of the tumors. However, a limitation of the BIRDS technique is the necessity of detectable (>1 mM) levels of contrast agent, which are cleared rapidly by the kidney. To obviate need for surgical intervention (e.g., renal ligation) to stop rapid agent clearance, here we demonstrate that pharmacological inhibition of renal clearance of these agents using probenecid to allow for longitudinal imaging of pHe throughout tumor progression and show that acidosis develops early in tumor progression in human-derived GBM tumors (U87 and U251). Since other tomographic pHe mapping methods are non-quantitative and directly altering pHe in a specific tissue is difficult to implement, we looked to assess the BIRDS-based temperature measurements for verification of the quantitative BIRDS readout. A localized cooling system was used to induce hypothermia in sheep brain to levels suggested to be neuroprotective in hypoxic states. Quantitative temperature mapping using BIRDS showed significantly decreased cerebral temperatures with cooling over all defined brain regions and was in agreement with thermocouple measurements. While pHe is a useful metric, tumor vascularity also shapes tumor metabolism and the microenvironment. BIRDS can be combined with other imaging modalities such as dynamic contrast enhanced (DCE) MRI, which allows quantification of vascular parameters (e.g., permeability) through modeling the dynamic uptake of Gd3+-based contrast agents. Multiparametric characterization of the spatiotemporal changes in cellularity, vascularity and acidosis of U87 and U251 tumors throughout progression showed unique patterns that could be used to identify tumor features and differentiate between tumor types. Finally, pHe readouts have potential as a biomarker of therapeutic response. After finding an increase in pHe after treatment with temozolomide in U251 tumors, we used BIRDS longitudinally to demonstrate normalization of pHe in U87 tumors treated with sorafenib, a nonselective tyrosine kinase inhibitor. Both treatments slowed tumor progression and led to increases of pHe which establishes a role for pHe imaging as an early and sensitive marker of evaluating therapeutic response prior to observable changes in the tumor appearance on standard MRI. The potential of BIRDS is vast and not limited to GBM, or cancer in general. Additional work has demonstrated that an acidic pHe is not limited to preclinical tumor models, but is also found in patient-derived xenograft (PDX) models of metastatic melanoma in the brain. BIRDS can also be utilized in evaluating tumors in any organ, as BIRDS has also shown acidic pHe in models of liver cancer. In summary, this work further expands BIRDS into a broadly applicable longitudinal platform for characterization of the tumor microenvironment and may aid in evaluation of many targeted therapeutic strategies.