By Paul Nicolaus
February 9, 2022 | During a time when a historic health crisis has continued to garner plenty of attention and investment, a recent global analysis of all clinical trials conducted last year revealed that breast cancer edged out COVID-19 as the most studied disease area.
As 2021 wound down and the New Year kicked off, new developments in the realm of breast cancer detection continued to emerge, from recent research and partnerships newly forged to the launch of new tools and technologies.
New Partnerships Formed
California-based biotech company Freenome has partnered with Siemens Healthineers, a med tech company headquartered in Germany, to collaborate in multi-omics and radiomic breast cancer diagnostics. The partnership will enable the companies to connect imaging and clinical data with molecular data in hopes of identifying new markers and diagnosing early-stage cancer.
Siemens Healthineers previously announced the FDA clearance of two mammography optimazation solutions. The MAMMOVISTA B.smart reading platform quickens the reading workflow for imaging, and the teamplay Mammo Dashboard tool is designed to optimize workflows using the visualization of performance indicators.
Along with Philadelphia-based UE LifeSciences, Siemens Healthineers also recently revealed the launch of the iBreastExam in the United States. The non-invasive, hand-held, FDA-cleared breast health test is administered by health care workers and takes less than 10 minutes to complete, with results available right after.
In early 2022, Agendia and Illumina announced a partnership designed to develop diagnostic tests for oncology testing. Agendia, a company that focuses on precision oncology for breast cancer, will collaborate with the sequencing giant to advance the use of next-generation sequencing (NGS) for decentralized oncology testing.
Together, the two California-based companies intend to develop new tests to improve the care of breast cancer patients using the Illumina MiSeqDx sequencing platform to grow the range of gene panels available for solid tumor analysis.
Agendia anticipates that its MammaPrint NGS test will allow major clinical hospitals worldwide to provide the clinical utility and benefits of MammaPrint—a 70-gene prognostic test that determines risk of recurrence—to patients and doctors in-house.
The company will be providing early access to its Digital MammaPrint platform in Brazil, enabling the country’s doctors and patients to benefit from the genomic insights generated. Digital MammaPrint is powered by the cloud-based Paige Platform. Paige, the company behind the Paige Platform and a startup that spun out of the Memorial Sloan Kettering Cancer Cancer, presented new data on its detection and classification systems at the San Antonio Breast Cancer Symposium in late 2021.
New Data, Research Revealed
At the same event, California-based precision diagnostics company Oncocyte revealed research findings focused on its DetermaIO. The 27-target gene expression test is designed to assess the tumor microenvironment and predict response to immunotherapy.
Notably, the research results demonstrated that the test was, indeed, finding patients who respond only to immune checkpoint inhibitors, Rob Seitz, Oncocyte’s Head of Immune Oncology, told Diagnostics World.
There are a lot of tests that apparently find responders to immune checkpoint inhibitors but in the end are only finding patients who generally do better, regardless of the therapy they are given, he explained. Because there are toxicities and costs associated with immune checkpoint inhibitors, the ability to isolate the patients who will benefit from them has advantages to both the patient and the healthcare system at large.
The IO score maintained its predictive value in PD-L1 negative patients treated with chemotherapy and immunotherapy, revealing the inability of existing biomarkers to identify all candidates for immune therapy treatment.
Furthermore, according to Oncocyte, the data revealed the high-level concordance (92%) between the commercial test compared to the research-based NGS version of the same test, which demonstrates the test’s high reproducibility regardless of the platform utilized.
In a blinded study performed by partners in Milan, the company has shown that DetermaIO delivers the same results measured on data from NGS as the company’s CLIA certified assay, according to Seitz.
“Researchers, both clinical trial groups and pharmaceutical groups, are constantly taking their precious clinical trial samples, and they are basically getting all the gene expression that they can get out of those samples,” he explained. Those tens of thousands of genes are then entered into a computer system. “We can go into that and pull out the 27 genes that we’re interested in and get a DetermaIO score.”
In some scenarios, Oncocyte is asked to generate that data and then obtain a DetermaIO score from the relevant genes. On other occasions, the company works with entities that cannot let clinical trial data out of their facility for patient confidentiality reasons. In those instances, Oncocyte staff members travel to perform on-site analysis.
But you cannot generate 30,000 genes on every patient, he continued, so there’s always a notable concern for those who work with Oncocyte in the latter scenario. If there are good results, will it work with a test that can be used in the clinic?
And the answer, based on this data, was “a resounding yes.” Seitz and colleagues found what he called “incredibly high agreement” between the NGS test data presented in the fall at the European Society for Medical Oncology and the data revealed more recently at the San Antonio Symposium focused on the clinically-approved qPCR test.
“We can use your massive data to generate data,” he said, “and we remove your concerns about whether or not it translates to the clinically-approved, clinically-deployable test.”
The test essentially takes a complex set of factors that clinicians examine—the tumor immune microenvironment, the treatment of immune checkpoint inhibitors, and all of the clinical factors—and simplifies it down to one simple point: Will the patient respond to this drug, yes or no?
Seitz added that this makes it easier to give physicians actionable information without overloading them.
AI Model Predicts Who Will Develop Cancer
Although AI-based risk models have demonstrated a notable advance over the risk models currently used in clinical practice, the responsible use of new AI calls for careful validation across diverse populations, according to an international group of researchers.
This in mind, they recently revealed their efforts to predict whether healthy individuals will develop breast cancer.
Screening mammograms and pathology-confirmed breast cancer outcomes were collected from locations in the United States, Israel, Sweden, Taiwan, and Brazil. A total of over 128,000 mammograms from over 62,000 patients were gathered across seven sites, and a cancer diagnosis followed nearly 4,000 of those patients within five years.
The findings, detailed in the Journal of Clinical Oncology (DOI: 10.1200/JCO.21.01337), revealed that their mammography-based risk model obtained concordance indices of 0.75, 0.75, 0.77, 0.77, 0.81, 0.79, and 0.84 at Massachusetts General Hospital, Novant, Emory, Maccabi-Assuta, Karolinska, Chang Gung Memorial Hospital, and Barretos, respectively.
The researchers concluded that their technology, called Mirai, maintained accuracy across diverse test sets and called this the “broadest validation to date of an AI-based breast cancer model.”
The findings suggest the technology can offer care improvements, they noted, but cautioned that prospective trials are still needed “to confirm the benefit of identifying improved high-risk cohorts and to establish Mirai-based guidelines.”
Ultrasound AI Software for Diagnosis
Women with dense breast tissue often require an alternative to mammography, according to New York City-based Koios Medical, which develops medical software and applies deep machine learning methods to help physicians interpret ultrasound images.
In December 2021, the company received FDA clearance for Koios DS, its AI-based software platform. The system is meant to help doctors accurately diagnose disease and improve speed to treatment while reducing avoidable surgical procedures.
“It’s not as widely known as it should be that more than 4 out of every 10 women have dense or heterogeneously dense breast tissue, which renders a mammography or a mammogram essentially inconclusive,” Koios Medical President & CEO Chad McClennan told Diagnostics World. “It’s standard of care to supplement a mammogram with either an MRI, which is extremely costly, or a much more low-cost ultrasound exam.”
When something suspicious is found on a mammogram, whether the patient has dense tissue or not, a diagnostic ultrasound is performed to determine the proper care pathway, he explained, which typically involves one of three scenarios. It’s either benign; it’s suspicious but not suspicious enough to trigger the decision to go in and sample the tissue; or it calls for a closer look via biopsy.
“What the software is designed to do is analyze a suspicious ultrasound exam and, in doing that analysis, predict the likelihood of malignancy,” McClennan said. Using archives of images tied to pathology results, AI analyzes and classifies suspicious breast lesions and predicts cancer or benign in two seconds or less to help increase early cancer detection and reduce biopsies on benign tissue.
“The purpose of the software is to much more accurately dictate to a physician the recommended path based on the risk of that lesion,” he added, to increase survival and reduce treatment cost.
The FDA clearance process required that the software demonstrate its ability to accurately interpret images from all major manufacturers of ultrasound hardware. It is compatible with major Picture Archiving and Communication System (PACS) workstation viewers, according to Koios Medical, and is integrated into LOGIQ E10—GE Healthcare’s ultrasound scanner.
The new system is built upon ultrasound data from a network of nearly 50 sites worldwide. This data sourced from international research partners improves the accuracy and reliability across different ethnicities and hardware manufacturers.
“We analyze the image that’s captured and created by a GE ultrasound machine, and a Siemens ultrasound machine, and a Philips ultrasound machine, and Canon and Toshiba and Mindray,” he said. Koios has data from over a dozen major manufacturers and across all major ethnicities so that the product can be used “confidently and accurately across a diverse patient population.”
The American Medical Association (AMA) has announced new CPT Category 3 codes for occasions when Koios DS software is used in the ultrasound exam process. The company anticipates that this ability for doctors and health systems to code and bill for their technology will ultimately speed along adoption.
“There’s a shortage of radiologists around the world, and time is precious,” McClennan said. “We are excited because the physician gets back precious time.”
Beyond that, patients experience less anxiety, avoiding callbacks for biopsies when there is no need for them. And risk holders such as employers and insurance companies are poised to benefit from earlier detection, which means higher survival rates and lower treatment costs.
So there are lots of stakeholders who win, he added, by taking advantage of the data the company has been acquiring for over a decade.
Paul Nicolaus is a freelance writer specializing in science, nature, and health. Learn more at www.nicolauswriting.com.