Ongoing Projects
1. Developing Artificial Intelligence Imaging Biomarkers for Dynamic Evaluation of Cancer Immunotherapy Response
Project leader Jason Tan, Supervisor Aydin Eresen
Pancreatic ductal adenocarcinoma (PDAC) is projected to become the second deadliest cancer in the U.S. by 2025, with 5-year survival less than 10%. Surgical resection, only curative treatment, is suitable in less than 20% of PDAC patients and provides a 5-year survival of only 20% due to a high recurrence rate (60%). Despite significant advances in chemotherapy, radiation, and targeted therapy, clinical outcomes remain poor. Recently, cancer immunotherapy (CIT) has emerged as a very promising treatment strategy. Programmed death-1 (PD1) and programmed death ligand-1 (PDL1) pathways have been studied intensively. Anti-PD1 inhibits T-cell-suppressive signals conveyed by PD1 ligands expressed on cancer cells, including PDAC. Other CITs garnering strong interest include adaptive immune cell therapy exploiting dendritic cells (DCs), natural killer cells (NKs), or T-cells, have been promising, including our recent studies. DCs are highly effective for activating naïve, and memory T-cell responses, but the efficacy in initial clinical trials for PADC was low. Since different CIT targets different mechanisms, combination therapy has emerged as a promising strategy. For example, dual agents using anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4, ipilimumab) plus anti-PD1(nivolumab) are shown to work well together. Inspired by the success, and in this project, we will test the combination of anti-PD1 and DC vaccine, using three genetically engineered mouse models (GEMMs) that recapitulate heterogeneity in human PDAC subtypes.
In addition to developing effective treatments, another critical research strategy is to find companion assays for selecting candidate patients and monitoring treatment response. DCs prime a tumor-specific cytotoxic T lymphocyte (CTL) response to restore the efficacy of checkpoint blockade, while blockage of PD1/PDL1 pathways upregulates activation of CTLs and reduces the production of regulatory T-cells (Tregs). Therefore, combining them may result in a positive feedback loop yielding superior antitumor immunity. DC vaccine has been found to increase immune cell tumor infiltration, and inhibition of PD1 in DCs can decrease secretion of IL-10 and indirectly inhibit Treg differentiation (Figure 1).
Furthermore, anti-PD1 inhibition of DCs can increase interferon-gamma (IFNγ) and interleukin (IL-12), and the production is further increased in the positive feedback loop. Combination treatment of the anti-PD1 and DC vaccine may drive PDAC immune environment into a pro-inflammatory state that improves patient outcomes. In this study, clinically relevant “liquid biopsy,” including serum biomarker IFNγ, IL-12, and CA19-9; and tumor-derived circulating tumor cells (CTCs), will be measured to develop combined assays for monitoring early response to CITs, including DC vaccine alone, anti-PD1 alone, and DC plus anti-PD1, for predicting overall survival (OS).
Although histological assays can directly measure CIT-induced changes, they are difficult to translate to PDAC patients. Imaging provides a noninvasive and easily translatable method for monitoring response and predicting outcome. However, for CIT, conventional imaging assessment based on tumor size change is not reliable. The immune-RECIST (iRECIST), immune-modified RECIST (imRECIST), and immune-related response criteria (irRC) are proposed to detect novel patterns under CIT. However, there are several weaknesses: 1) they do not consider tumor heterogeneity, 2) they suffer from incorrect assessments due to pseudoprogression and new immune-related patterns of mixed responses, and 3) they cannot evaluate early response or predict progression-free survival (PFS) or OS. To address these clinical gaps, one main goal of this research project is to identify artificial intelligence (AI)-derived novel imaging biomarkers extracted from conventional magnetic resonance imaging (MRI) data for monitoring pathophysiological responses induced by CITs. Lastly, we will translate the findings to evaluate PDAC patients receiving CIT.
2. Novel MRI of Brown Adipose Tissue as Biomarkers for Cancer cachexia
Cachexia is a metabolic disorder and occurs in association with malignant diseases and with multiple chronic non-malignant diseases32,33. So far, treatments for cachexia have been ineffective; therefore, there is an urgent need to identify new measurement and assessment techniques for early determination of the effectiveness of new therapies. Only curing underlying disease can completely reverse cachexia symptoms. Detecting the presence of cachexia early and non-invasively is crucial, as is assessing the response to early intervention. The key characteristic of cachexia is higher resting energy expenditure levels compared to those in healthy individuals; this energy expenditure has been linked to thermogenesis by brown adipose tissue (BAT). Thermogenesis in BAT is regulated by the mitochondrial uncoupling protein 1 (UCP1), which is increased in both mouse models and patients with cachexia, including cancer-associated cachexia36. Thus, there has been heightened interest in quantifying BAT function, and, more specifically, understanding UCP1-mediated BAT thermogenesis in cachexia models as a possible target for novel therapies.
Ghrelin, a natural ligand for the growth hormone (GH)-secretagogue receptor, regulates BAT metabolism by decreasing the expression of UCP1, which may contribute to anti-cachexia by increasing the food intake stimulus. Ghrelin exhibits anti-cachectic actions via both GH-dependent and GH-independent mechanisms. We aim to develop MRI techniques to facilitate a better understanding of the molecular basis of the anti-cachectic effects of ghrelin in cachexia and evaluate therapeutic responses for novel clinical applications.
Currently, 18F-FDG PET/CT is used as a surrogate biomarker to measure BAT activity in vivo. However, PET/CT is inherently sensitive only to active tissue and underestimates the total BAT mass. Moreover, radiation exposure from PET/CT imaging does not allow for its frequent use in longitudinal studies41. For clinical applications, MRI offers several advantages, such as superior soft tissue contrast and spatial resolution. Dixon multi-echo MRI is often used to calculate the fat-water fraction (FWF), used for measuring BAT volume. However, FWF is not sensitive to BAT metabolic activation. We aim to develop Z-Spectrum MRI (ZS-MRI) for simultaneous in vivo measurements of BAT metabolic activation and volume, bridging this technical gap in clinical applications. ZS-MRI acquires a set of MR images after applying a saturation radio frequency (RF) pulse at different frequency offsets. Since tissue metabolites continuously exchange protons with water, ZS-MRI, with its chemical exchange saturation transfer (CEST) contrast, is sensitive to metabolite concentrations, which are expected to increase with BAT activation43. At the same time, protons of water and fat are saturated directly at their resonance offsets, allowing measurements of their relative concentrations in a pixel44. This leads to the calculation of FWF, which can be used to quantify BAT volume. The goals for this proposal are 1) to develop ZS-MRI imaging biomarkers for monitoring the changes of BAT volume and function during initiation and progression of cachexia and 2) to demonstrate that these biomarkers can be used for evaluating in vivo response to early intervention (Figure. 1). This project will use a transgenic KPC mouse model of pancreatic cancer cachexia to address the following specific aims.
3. Develop a novel Z-Spectrum MRI technique for noninvasively longitudinal monitoring of the hypoxia, edema, metabolic stress of visceral adipose tissues
Adipose tissue (AT) hypoxia and metabolic stress are key pathophysiological processes associated with metabolic diseases, such as obesity and type 2 diabetes. Due to invasive sampling methods, human studies of AT metabolic stress and hypoxia have focused mainly upon subcutaneous AT. However, visceral adipose tissue (VAT) exerts more adverse effects on human metabolic diseases than subcutaneous fat.
The 2015–2020 Dietary Guidelines for Americans recommend that Americans consume less than 2,300 mg of sodium each day as part of a healthy eating pattern. However, about 90% of Americans 2 years old or older consume too much sodium. The available evidence indicated a positive relationship between higher sodium intake and subsequent risk of cardiovascular disease, obesity, and type 2 diabetes. The high sodium intake independently associated with an elevated risk of obesity and central obesity, and a positive association between dietary sodium intake and type 2 diabetes in the general U.S. adult population.
The variation in dietary salt intake leads to alterations in glucose metabolism and insulin sensitivity. Moreover, our recent work demonstrated that short-term high-salt interventions reliably result in measurable changes in VAT metabolic stress, hypoxia, and edema by biological techniques. Monocyte activation and tissue infiltration are quantitatively associated with high-salt intake-induced target organ inflammation (edema). These findings provide novel links between dietary salt intake, innate immunity, and end-organ inflammation. The role of monocyte/macrophage infiltration in initiating VAT metabolic stress, hypoxia, and altering insulin sensitivity has been suggested as a critical pathophysiological process in human metabolic diseases, such as obesity and type 2 diabetes.
Critical gaps in clinical knowledge: VAT function is evaluated with invasive biopsy in clinical research settings. PET/CT methods have been at the forefront of noninvasive quantitative functional VAT assessment in clinical settings. However, magnetic resonance imaging (MRI) provides advantages over PET/CT with higher spatial resolution, no radiation exposure, and repeated use for longitudinal studies. Dixon’s MRI has been used for clinical applications to calculate the fat-water fraction (FWF or FF) for measuring VAT volume and fat-water composition, which represents the indirect measurement of VAT edema. However, FWF is not sensitive to VAT hypoxia and metabolism.
To bridge this gap in clinical applications, we will develop multiparametric Z-Spectrum MRI (ZS-MRI) for in vivo measurements of VAT edema, hypoxia, and metabolic stress (EHMS). ZS-MRI acquires a set of Z-spectral MR images after applying a saturation radio frequency (RF) pulse at different frequency offsets. The direct saturation (DS) effect in Z-spectrum allows us to quantify FWF or FF in VAT, an imaging biomarker for VAT volume and edema changes. Most importantly, ZS-MRI, with its blood oxygen level-dependent magnetization transfer (BOLDMT) (Figure 1), is sensitive to tissue oxygenation. Its chemical exchange saturation transfer(CEST) contrast, by measuring tissue metabolites with exchangeable protons, is sensitive to tissue metabolic function.
