National Cancer Research Month: Highlighting UW Radiology Cancer Research

May is National Cancer Research Month, and the Department of Radiology is rich with research projects that have a high degree of innovation and substantial potential impact. Here are some researchers working on cancer-related projects within our department:

Majid Chalian, M.D., “Predicting Treatment Response to Neoadjuvant Radioimmunotherapy (NRIT) in High Grade Soft Tissue Sarcoma (STS) with MRI-Based Radiomic Signature,” RSNA R&E Scholar Grant

Soft tissue sarcomas (STS) constitute a challenging group of malignancies with wide range of biological behavior and rapidly fatal subtypes in children and young adults. Early treatment response prediction is crucial for therapy planning. Current treatment planning is based on clinical and pathologic factors with a one-size-fit-all approach. Cancer patients have varying degrees of responses to therapies, which have been attributed to tumor heterogeneity and immune microenvironment. Response Evaluation Criteria in Solid Tumors (RECIST), as the only widely adopted metric for response assessment in STS, is known to have substantial limitations, especially in molecular-targeted therapies. Advances in imaging technology have introduced novel quantitative imaging biomarkers. Radiomics is a method to extract higher-dimensional imaging biomarkers that are not necessarily captured by visual assessment. Preliminary data support utility of radiomics for immune phenotype assessment of tumors.

Dr. Chalian and colleagues are investigating the role of MRI-based radiomics, alone or in combination with clinical and semantic MRI features, for prediction of pathologic treatment response to neoadjuvant radioimmunotherapy in STS patients. They will test this hypothesis by applying radiomics on a validated retrospective cohort of STS patients with standard of care MRI. Response predictive models will be created based on MRI radiomics, alone or in combination with clinical and semantic MRI features. An independent cohort of STS patients from a running prospective trial will be used to evaluate these models. The expected output of the project will be a multiparametric method that will provide sarcoma clinicians and researchers with new tools to predict treatment response and guide management. It will also serve as preliminary data for an NIH funding proposal for a multicenter registry of STS to validate models and investigate histology-specific radiomics analysis of STS.

Paul Kinahan, Ph.D., FIEEE, FAAPM, FSNMMI, FAIBME, “Plasma and Imaging Biomarkers to Risk Stratify CT Detected Pulmonary Nodules,” National Cancer Institute (NCI)

Small lung nodules are sometime found in patients, either by x-ray computed tomography (CT) screening studies or incidentally, such as from a x-ray collected in a clinic (say looking for a shoulder fracture that also images part of the lung). These indeterminant pulmonary nodules are worrisome and problematic as they have a small chance of being lung cancer, but there is no easy or risk-free way to find out. The two options up until now are to either wait and see if a nodule grows, or perform an invasive lung biopsy.

Making the right choice between benign or cancerous status is a critical part of the choice to treat or observe. Dr. Kinahan’s group is working to pull together all the information that can be collected in a risk-free manner to improve the odds of making the right choice. This includes (1) clinical information such as age and smoking history, (2) plasma samples from a blood draw (sometimes called a ‘liquid biopsy’) that look for specific antibodies, proteins, and other compounds, (3) specific descriptive features from radiologists reviewing CT images, such as shape and location (called ‘semantic’ features), and (4) information from artificial intelligence analysis applied to the CT images (called ‘radiomics’). The information is combined into a statistical model that predicts the likelihood whether an indeterminant pulmonary nodule is cancerous or benign. The ultimate goal is to reduce lung cancer mortality and patient care costs.

To date, the group has spent three years collecting data and building the statistical model, and they are now readying to test the model on independent data shared by other medical centers. The team involves radiologists, pulmonologists, physicists, data scientists, molecular biologists, biostatisticians, and research coordinators at the University of Washington the Fred Hutchison Cancer Center.

Christoph Lee, M.D., M.S., MBA, “Artificial Intelligence for Improved Breast Cancer Screening Accuracy: External Validation, Refinement, and Clinical Translation,” National Cancer Institute (NCI)

In this academic-industry partnership, they are expanding on work that began with the Digital Mammography DREAM Challenge to refine and scale a promising 2D mammography AI algorithm for cancer detection to be applied to 3D mammography (digital breast tomosynthesis). The 3D mammography AI algorithm was evaluated in a pivotal reader study demonstrating its improved accuracy, and the AI algorithm is now under consideration for FDA approval. After FDA approval, they will test the algorithm’s effectiveness in real-world clinical settings.

Janie Lee, M.D., M.Sc., “Risk-based Imaging Strategies to Improve Breast Cancer Surveillance Outcomes,” National Cancer Institute (NCI)

“Risk-based Imaging Strategies to Improve Breast Cancer Surveillance Outcomes” is one of three complementary projects led by Dr. Janie Lee within the research study “Risk-based Breast Cancer Screening and Surveillance in Community Practice” funded by a Program Project grant from the National Cancer Institute. Currently 2.8 million women in the United States with a personal history of breast cancer (PHBC) face years of imaging surveillance. As with breast cancer screening, imaging surveillance after breast cancer treatment aims to detect early second breast cancers, either local recurrence or new cancers in the other breast, with the goal of reducing breast cancer deaths and maintaining quality of life.

Current surveillance guidelines recommend annual mammography for all women with a PHBC regardless of risk. Dr. Lee and colleagues propose that imaging surveillance tailored to a woman’s risk factors, primary breast cancer characteristics, and treatment choices will improve surveillance outcomes. The overall goal of this project is to support a shift from a single strategy of annual mammography for all women to individualized, risk-based strategies, incorporating the full range of available imaging modalities and reliable estimates of a woman’s risk of surveillance outcomes.

Robert Miyaoka, Ph.D., “At home monitoring for 177Lu DOTATATE treatment personalization,” National Cancer Institute (NCI)

Dr. Robert Miyaoka and scientists from the Imaging Research Laboratory in the Department of Radiology are working on technology to enable a patient friendly and economically viable method to personalize peptide receptor radionuclide therapy (PRRT) for neuroendocrine tumor (NET) patients. PRRT using Lu-177 DOTATATE for NET patients is an internal radionuclide treatment that was approved by the US FDA in 2018, as a standardized protocol of four treatments administered at 8-week intervals. While this protocol is safe for a vast majority of patients, most patients can safely receive 5, 6 or even up to 11 treatments. Further when the number of treatments is personalized to what a patient’s body can safely receive versus the standard protocol, overall survival more than doubles (i.e., from 25 to 54 months). Unfortunately, current methods for treatment personalization require 3-4 imaging studies over a 7-day period, ideally for each treatment dose. This methodology is very expensive and very inconvenient for the patient.

The technology that the group is developing will allow the patient to make similar measurements from their homes. They call their device the Personalized Remote Radiation Tracking Portable Organ Dosimetry Device or PRRT PODD. The PRRT PODD is a low cost, easy to use, portable device that consists of 10-15 small radiation detectors that are strategically placed in the PODD according to the patient’s unique anatomy. Used daily, the PRRT PODD provides information about the washout kinetics of the Lu-177 DOTATATE from the patient’s organs at risk. The patient lies down into the PRRT PODD once a day for a 2-minute measurement for up to 2-weeks. After each 2-minute measurement session, the data is securely transmitted to a remote workstation where it is quality control checked and stored for processing. After the 2-week collection period, special software processes the data and reports the dose received by each organ at risk. As currently designed, all components of the PODD are reusable. They will being first patient trials in Summer 2022.

Savannah Partridge, Ph.D., “Validation and Standardization of the Apparent Diffusion Coefficient as a Quantitative Imaging Biomarker in Breast,” National Cancer Institute (NCI)

Diffusion-weighted MRI (DW-MRI) is a powerful functional imaging technique, which indirectly assesses tissue microstructure and can provide complementary information to conventional imaging for breast cancer detection and diagnosis. The apparent diffusion coefficient (ADC) derived from DW-MRI is an imaging marker that has shown potential value to improve ability to distinguish cancer from benign lesions and significantly reduce unnecessary biopsies, as well as to evaluate response to therapy. However, heterogeneity in approaches to measure breast ADC along with susceptibility to a variety of technical factors have limited its clinical utility. This project aims to standardize and validate technical approaches to overcome hurdles in translating ADC as diagnostic biomarker into clinical practice. Image corrections to address known inaccuracies related to gradient nonlinearities, misregistration and geometric distortion will be implemented, and methodology for measuring lesion ADC will be optimized to maximize diagnostic performance and reliability.

Habib Rahbar, M.D., “MRI Characterization of Biological Risk of Ductal Carcinoma in Situ,” National Cancer Institute (NCI)

Ductal carcinoma in situ, or DCIS, is the earliest form of breast cancer we can detect, and it is almost exclusively identified by screening women who have no symptoms of cancer. While we can treat it very effectively with nearly 100% cure rates, there are real concerns that some DCIS is “overtreated” with too much surgery or radiation therapy since it may not ever become lethal if left alone. Because we cannot determine which of these lesions are likely to become invasive cancer and potentially lethal with basic pathology measures, Dr. Rahbar and colleagues are exploring the ability of advanced imaging to assess the biology of DCIS. Their goal is to show value of MRI to both better depict DCIS extent and assess it biology so that it can be integrated into clinical decision-making tools to optimize its treatment.

Share this story: