EarlyDx: Detecting and Locating Cancer Early
EarlyDiagnostics, Inc. (“EarlyDx”), a Los Angeles based liquid-biopsy and big data company, is developing a blood-based test, the CancerRadar Test, for not only detecting cancer at early stages but also pinpointing its location.
According to 2019 Cancer Statistics published by CA: A cancer Journal for Clinicians, the 5-year survival is 90.5% for all cancers when diagnosed early (localized stage) but only 36% for cancers diagnosed at all stages. Traditional imaging-based early cancer diagnosis modalities only cover a handful of cancer types and remain mostly unchanged in the past decades. A new approach, called liquid biopsy, analyzing the tumor signal in a patient’s blood, is expected to revolutionize the early diagnosis market.
“Most companies in the liquid biopsy field focus on detecting a few biomarkers flowing freely in the patient’s blood, which may not carry the target features.” said Jasmine Zhou, a professor at the UCLA and co-founder of EarlyDx. Zhou’s lab previously developed the CancerLocator and CancerDetector algorithms to aggregate blood methylation features to increase the detection power. Zhou added, “one needs to integrate more features, such as methylation, copy number variations, fragment lengths, and even microbial abundances in blood to achieve higher sensitivity and specificity for early cancer diagnosis”.
With support from National Cancer Institute Small Business Innovative Research program, EarlyDx has developed an assay, the CancerRadar assay, that can provide multiple genomic and epigenomic features from a simple tube of blood. The company will begin to market this assay as a research use only reagent kit soon.
EarlyDx’s machine learning algorithm integrates all features obtained from their assay for determining the presence and type of cancer. In collaboration with researchers at the University of California, Los Angeles, the company has used CancerRadar to analyze 479 clinical samples from four types of cancer (liver, lung, liver, and gastric) patients and healthy controls. The CancerRadar achieves a superior performance of a sensitivity of 80.9% at the specificity of 99% for detection, and an accuracy of 91.0% for locating the early-stage (I and II) cancers. Dr. Mary Stackpole noted, “the methylation features are most informative, especially for locating cancer, but including other features derived from our data significantly increases the performance.” Stackpole is the VP of Bioinformatics at EarlyDx and the lead developer of the CancerRadar algorithm, who added that she expected that the detection power of CancerRadar will continue to increase with larger samples sizes. The CancerRadar results will be presented in the upcoming American Association of Cancer Research Annual Meeting 2021 in April.
To facilitate a decentralized test, EarlyDx has deployed CancerRadar in a cloud computing platform. Their platform, called EarlyDx-Cloud, currently serves as a hub for collaborative research and clinical trials. “Our platform will collect and analyze genomic and epigenomic data from a patient’s blood over time. This type of data can also provide insight into health status outside of cancer,” said Dr. Xiaohui Ni, the CTO of the company. Ni further stated, “I envision our continuous growth from a reagent kit supplier and a decentralized cancer screening provider, to a big data analytics company. We fortunately have several world-known big data experts in the company’s founding team.”