Institut Curie, Freenome to Leverage Machine Learning for cfDNA Analysis

March 29, 2018
Institut Curie, Freenome to Leverage Machine Learning for cfDNA Analysis
Freenome’s artificial intelligence (AI) genomics platform, designed to process the full range of cell-free (cf) biomarkers in the blood, will be evaluated by Institut Curie as a novel tool to predict patient response to immuno-oncology therapies. [© adimas/Fotolia]

Freenome’s artificial intelligence (AI) genomics platform will be evaluated by Institut Curie as a novel tool to predict patient response to immuno-oncology therapies, the partners said today, through a collaboration whose value was not disclosed.

Freenome’s Adaptive Genomics Engine platform will be evaluated by observing changes in circulating cell-free biomarkers, the company and the institute said.

The platform is designed to process the full range of cell-free (cf) biomarkers in the blood, looking beyond tumor DNA. By assembling a library of cell-free disease signatures, Freenome aims to develop a variety of non-invasive blood tests for the early detection of cancer and early prediction of response to various oncology therapies.

In the initial phase of the collaboration, Freenome plans to analyze samples from the institute’s Analysis of Circulating Tumor Markers in the Blood (ALCINA) Trial (NCT02866149). An estimated 620 participants are to be enrolled in ALCINA, an umbrella trial designed to assess different circulating biomarkers and their correlation with clinical and pathological characteristics pertaining to response to programmed death-1 (PD-1) inhibitors, both alone and in combination with other therapies.

Freenome’s tests would be designed to identify patients who are more likely to respond to PD-1 inhibitors, the company and the institute said. They cited past studies showing that approximately 80% of patients with advanced non-small-cell lung cancer do not respond to PD-1 inhibitors.

“Our approach incorporates a variety of biological signals, such as genomic, proteomic and epigenetic changes, providing new insights into possible mechanisms of resistance and guiding treatment selection for patients,” Blandine Merino, VP of business development at Freenome, said in a statement. “Our machine learning scientists and molecular biologists are evaluating the cell-free genome—given 60% to 80% of cfDNA comes from immune cells—and other analytes. These provide a more complete picture of the dynamic interaction between the tumor and its environment.”

Freenome and Institute Curie reason that analysis of the data they collect could lead to new targets in precision oncology and improved therapeutic decision-making that overcomes limitations associated with current approaches, such as programmed death-ligand 1 (PD-L1) expression testing, and assessing tumor mutational burden.

According to Institut Curie, the collaboration is one of its first with a company specializing in both cell free biomarkers and applying machine learning to large data sets.

The partnership follows the Institut Curie identifying innovation around liquid biopsy and big data as a major focus of medical-scientific research in its 2015-2020 MC21 strategic plan.

The plan calls in part for a comprehensive effort to support, finance, and host startup projects for spinout from the Institut, as well as the promotion of collaborations between Curie researchers and “innovative” companies. Curie has launched partnerships in recent years with pharma giants Roche (specifically its Riche Institute), Bristol-Myers Squibb, and Servier.

Institut Curie has also hired a data director, and has committed to positioning itself as a leader in personalized medicine in oncology.

“The technologies developed by Freenome will fully benefit our patients and its access within the Institute will speed up future collaborations for developing and validating predictive tests,” Amaury Martin, head of Institut Curie’s Technology Transfer and Industrial Partnerships Office, and director of the Institut Curie’s Carnot Institute (Curie Cancer), said in a statement.

Added Prof. François-Clement Bidard, M.D., Ph.D., a principal investigator at Institut Curie: “Freenome is developing a novel approach which could revolutionize the way we analyze cell-free biomarkers for patients with cancer treated with immunotherapy. This approach complements a pipeline of innovative research projects that is ongoing at Institut Curie Circulating Tumor Biomarkers Laboratory.”

Launched in 2014, Freenome is a liquid biopsy diagnostic developer focused on facilitating early detection of cancer by combining genomics research with computational expertise.

Last year, Freenome raised $65 million in Series A financing—which along with a $5.5 million seed funding in 2016 earned the company a number-six ranking by ClinicalOMICs among “10 Liquid Biopsy Companies in the Money.” Andreessen Horowitz led the Series A round, joining with two other investors in Freenome’s seed financing, Data Collective (DCVC), and Founders Fund, as well as new investors GV (Google Ventures), Polaris Partners, Innovation Endeavors, Asset Management Ventures, Charles River Ventures and Spectrum 28.

Institut Curie said the collaboration with Freenome follows on some two decades of study into detection of tumor cells, followed by study of circulating tumor DNA (ctDNA) fragments. In the late 1990s, the institute launched the first research program focused on disseminated tumor cell (DTC) detection in the bone marrow of patients with early breast cancer. About 1,000 patients were studied in the program, which generated a landmark 2005 study establishing bone marrow DTC as a level-of-evidence 1 prognostic factor in early breast cancer.

More recently, the institute said, its Institut Curie’s Circulating Tumor Biomarkers Laboratory has developed numerous innovative ctDNA techniques, such as droplet digital PCR and NGS-based techniques.