b'Catalysts of ProgressPathologist Payal Kapur, M.D., discussing digital pathology approaches with computer scientists Satwik Rajaram, Ph.D., Guanghua Xiao, Ph.D., and Yang Xie, Ph.D. Kidney Cancer Explorer, accessible throughKidney Cancer Explorer is a unique a central web portal, links multiple types of Today, kidney cancer diagnosis is still made by pathologists who look at tissue under a microscope.information, both medical and research (suchand visionary tool to support the as genomics). By providing links to samples, it Researchers at UT Southwestern have developed a novel classification of renal cancer based on geneenables further research by UT Southwestern next generation of research on mutations (see page 40). Instead of judging a book by its cover, you can actually read whats inside,investigators. kidney cancer.says Dr. Kapur, lead pathologist of the Kidney Cancer Program.to evaluate the process whereby tumorsrenal cell carcinoma (ccRCC). However,behind the tapestry and provided novelArtificial Intelligencebecome resistant to drugs; to develop newmuch like colors, which have infiniteinsights into how these patterns evolve radiology tests for kidney cancer (see pageshades, no two ccRCCs are identical underover time. Her studies provide an inno-28); to understand the mutations that causethe microscope. The complexity of tryingvative framework to understand kidneyA 21st-Century Medical of surgery, radiation, and drug treatments),patients they care for. kidney cancer (see page 40); to study howto make sense of subtle morphologicalcancer. (Cai et al., EBioMedicine, 2019)Intelligence Platform longitudinal metrics (such as weight, bloodKCE is a unique and visionary tool to kidney cancer invades and metastasizesdifferences likely accounts for the lack ofOver the last five years, a team led bylaboratory tests), and research data support the next generation of research on (see page 54); and to understand whatprogress. How does one make sense of aDr. Kapur and including Venkat Malladi andincluding next-generation sequencingkidney cancer, says Dr. Brugarolas. nutrients kidney cancers use (see pagetapestry with millions of cells? This dauntingAlana Christie has developed a medical(which is available for more than 1,500 sam-29). (Sivanand et al., Sci Transl Med, 2012;task was tackled by Dr. Payal Kapur. Sheintelligence platform that includes data fromples) and our tumor bank.Building the Future with Artificial Pea-Llopis et al., Nat Genet, 2012; Pava- first defined three different parameters (orover 3,000 patients linked to genomics andKCE is accessible through a pass- Intelligence and Machine Learning Jimnez et al., Nat Protoc, 2014; Wolff etaxes) to classify tumors: cells, architecture,sample availability. This platform, referredword-protected web-based interface. ByOne application of AI (artificial intelligence) al., Oncotarget, 2015; Chen et al., Nature,and borders. She then compiled all of theto simply as KCE (Kidney Cancer Explorer),automatically running preset queries on thebeing explored in the Kidney Cancer 2016; Wang et al., Cancer Discov, 2018;variants. On the whole, she arrived at 33provides a centralized access point for allelectronic health record, KCE self-updatesProgram is digital pathology. AI is ana-Courtney et al., Cell Metab, 2018)descriptors. She and her team then cata- kidney cancer research at UT Southwestern.and stays permanently current.lyzing tumor samples to understand how logued over 500 tumors according to the 33The resource was developed with theKCE enables a wide variety of analyticcancers develop and how they relate to Morphologic Evolutionary Trajectoriesvariables. By studying how these descrip- Lyda Hill Department of Bioinformatics,functions across the different types of infor- their environment. Several scientists at Underpinning Tumor Growth tors relate to each other within a tumor andled by Gaudenz Danuser, Ph.D., and themation: clinical, pathological, and genomic.UT Southwestern are working on this Kidney cancer is classified into differenthow they impact tumor aggressiveness andFoundation for novel classification of ccRCC byQuantitative Biomedical Research Center,Investigators can ask questions aboutproblem, including Satwik Rajaram, Ph.D., types, including most commonly clear cellpatient prognosis, she unraveled the logicDr. Kapur based on 33 descriptors across three axes. led by Yang Xie, Ph.D. Funding was pro- how genomics or pathological featuresGuanghua Xiao, Ph.D., and Yang Xie, vided by the SPORE and CPRIT (Cancerimpact outcomes or treatment respon- Ph.D., together with pathologist Dr. Payal Prevention and Research Institute of Texas).siveness. The platform is fertile ground forKapur. The platform could revolutionize TUMOR DIGITAL PATHOLOGY MERGED KCE is a complex informatics systemartificial intelligence and machine-learninghow tumors are defined. With a virtually that integrates many types of information:approaches. Investigator capabilities are fur- infinite range of tissue properties that demographic and other general patientther expanded through access to genomicmight be captured, AI is set to improve characteristics (such as age of diagnosis,datasets (such as whole exome sequencinghow tumors are classified. Such capability gender, height, and weight), comprehensiveand RNA sequencing). is particularly relevant to tumor behavior pathological information (such as tumorKCE can also be used by healthcaredriven by interactions with its environment, A computer algorithm recognizesstage, histology, and grade), patterns ofproviders, who are able to determine whata feature that only recently is beginning to different cell types in a tumor. Red Blood Cell SupportCell tumor spread, treatment information (typeresearch data or samples are available forbe explored (see page 49).Tumor Cell68 69'