Skip to main content
Premium Trial:

Request an Annual Quote

Sophia Genetics, GE Healthcare Begin DEEP-Lung-IV Data Collection, Technology Integration


CHICAGO – Sophia Genetics and GE Healthcare this week expanded their partnership by agreeing to add GE's Imaging Fabric Core and Imaging Fabric Annotation Template — part of that firm's Edison digital health platform — to an ongoing collaboration.

The companies said at the American Society of Clinical Oncology annual conference here that they would include the technologies in their DEEP-Lung-IV clinical study by using Imaging Fabric to visualize, segment, and annotate lung lesions to accelerate radiomics analytics workflows.

"I think we've got an opportunity to profoundly impact how patients can be served more effectively and outcomes can be delivered more efficiently as well," said Ben Newton, GE Healthcare's UK-based general manager.

DEEP-Lung-IV is a multimodal study of metastatic non-small cell lung cancer that Sophia and GE announced at the Radiological Society of North America conference here in December after hinting at it a few weeks earlier. The firms are partnering with research institutions to develop AI-driven analytics and workflow technologies for oncology, with an emphasis on matching treatments based on genomic profiles and cancer types.

Sophia had an initial public offering last July that grossed $234 million. A private placement from GE Healthcare brought in another $20 million and accompanied a letter of intent to codevelop new AI-driven analytics and workflow technologies to improve the matching of treatments based on genetic and tumor profiles of cancer patients. DEEP-Lung-IV is an outgrowth of that, which has its roots in discussions the companies started in late 2020, Newton said.

Sophia said that DEEP-Lung-IV will leverage the company's machine learning-based analytics for multimodal predictions of immunotherapy response in patients with advanced lung cancer. The company eventually will embed learnings from DEEP-Lung-IV into its CarePath module on its flagship Data Driven Medicine (DDM) platform.

Plans are to follow 4,000 patients at 30 sites worldwide. Sophia said this week that it has recruited 500 patients and signed 19 sites in seven North American and European countries, most recently Roswell Park Comprehensive Cancer Center in Buffalo, New York, which came online in the last two weeks, according to Philippe Menu, Sophia's senior VP and chief medical officer.

The patients will be broken into three cohorts, with 1,500 receiving Merck's Keytruda (pembrolizumab), 1,000 receiving chemotherapy, and 1,500 being treated by both methods. The companies will first test whether they can predict treatment response using clinical, biological, genomic, and imaging data and eventually attempt to predict progression-free survival and overall survival.

"We to make sure not only that [the technology] works, but that we're doing it with the user in mind, the patient in mind, and also the customer in mind," Newton said. "That's only possible when you start to have close relationships with academic medical centers because you want to make sure that it's delivering on their needs."

Initially, the partners are working to integrate data between DDM and GE's Edison, but Menu has previously said that this is the first of three areas of focus. The firms will also be engaging hospitals to bring together imaging and genomics and, later, looking to develop multimodal analytics applications that draw from both technology platforms.

Menu said that Sophia and GE are particularly concentrating on radiogenomics, but noted that the biggest early challenge will be prioritizing activities.

Sophia announced in April that it had analyzed 1 million genomic profiles on DDM. Newton said that GE has a global installed base of about 4 million medical imaging devices, which collectively turn out 2 billion images per year.

"When you bring those genomic insights and those radiomic or imaging insights together, you start to really understand disease in a much more detailed way," Newton said. "Then you can start to make much better decisions around clinical prognosis, also around response to therapy, around course, and prediction of side effects."

The global customer bases of GE and Sophia also improve the chance that the combined technology will be adopted on a wide scale, Menu said. " If you start from having nothing, a foothold into the clinical workflow is incredibly difficult," he said.

However, the firms want to start somewhat small with a clinical study and a series of pilot projects so they can begin to understand the challenges various healthcare organizations in different regions of the world have in combining genomic and imaging data before GE and Sophia can scale up their joint offering.

The analysis has not incorporated any genomics yet because Menu said that the cohort is still too small to draw any meaningful conclusions from molecular data. Early analysis has been completed on about 100 patients based on phenotypic, biological, and CT information, he said.

"The genomics and the imaging suddenly start to become more relevant when you've got both sides of the coin visible," Newton said.

DEEP-Lung-IV, which is expected to continue through February 2024, is currently a retrospective study as GE and Sophia train the machine learning algorithms to make individual-level predictions.

Eventually, Menu wants to switch to prospective prediction to inform clinical decision support. "It would be a clinical decision support system where you can [conduct] stratification of patients or you can enable selecting the right patient for the right therapy," he said.

That transformation could happen in the next six to nine months when the study starts to include prospective analytics and brings in pharma partners.

"There's a lot of deep work around the prototyping and the interconnectivity yet to do, but there's a tremendous pull from customers wanting to have this kind of capability, wanting to exploit it to do clinical trials more efficiently, and to find patients to apply those drugs to," Newton said.