Oncohost opens trial sites for study evaluating its AI proteomics profiling technology
Original source here
Oncohost Ltd. has opened eight U.K. trial sites in the study assessing the ability of its artificial intelligence (AI)-driven proteomics profiling technology to single out which cancer patients will respond to treatment with immune checkpoint inhibitors.
The sites will carry out proteomic analyses of blood samples from patients with late-stage melanoma or non-small-cell lung cancer (NSCLC), to predict their likely response to immunotherapy.
The data will add heft to Oncohost’s Prophet system, which combines high throughput protein analysis with machine learning to understand the dynamics of how a patient’s immune system – as opposed to the tumor per se – responds to a given cancer immunotherapy.
Immunotherapy has achieved “excellent results” in certain situations for a number of cancers, inducing remission, improving quality of life and extending survival, noted David Farrugia, consultant medical oncologist at Cheltenham General Hospital, U.K. But he said, “Success with immunotherapy is not guaranteed in every patient.”
Farrugia is principal investigator in the U.K. arm of the Prophetic study, which also is recruiting patients at 10 centers in the U.S. and 14 in Israel. The study is analyzing blood samples to look for changes in protein expression before and four weeks after the first round of treatment, to build a profile of each individual’s response.
The percentage of cancer patients eligible to receive drugs targeting the PD-1, PD-L1 and CTLA-4 checkpoints that are employed by tumors to avoid immune system defenses, has risen to around 50% since the first checkpoint inhibitor, Merck and Co. Inc.’s Keytruda (pembrolizumab) was approved by FDA in 2014.
However, a retrospective study published in JAMA in 2018 showed only around 13% of eligible patients responded to these drugs. Other data indicate there is an objective response rate of 40-45% to Keytruda in first line treatment of melanoma, and 20% in second line treatment of NSCLC.
The level of resistance to immunotherapies has become “the most pressing issue in oncology today,” said Ofer Sharon, CEO of Binyamina, Israel-based Oncohost. “Immunotherapy is not a one size fits all solution,” he said.
Led by the clinical development programs of pharmaceutical companies, currently a 75-year-old woman and a 45-year-old man with NSCLC will be treated according to the same protocol. “Whenever I speak to oncologists they get it immediately,” Sharon said. “They are flying in the dark currently,” he told BioWorld.
The low response rate to checkpoint inhibitors has focused attention on the identification and validation of biomarkers that will predict which patients are likely to respond to immunotherapy.
Tumor mutation burden is one measure that has emerged as being significantly correlated with positive responses. Based on this, Foundation Medicines Inc.’s Foundationone test of mutation burden was approved by FDA in June 2020 as a companion diagnostic for Keytruda, and subsequently for other cancer therapies.
Last year also saw the first FDA approval of next generation liquid biopsy sequencing of tumor DNA for mutation profiling, when Guardant Health Inc.’s Guardant360 CDx was approved.
Other tumor features, including microsatellite instability, neoantigen levels and tumor phenotypes such as levels of PD-1 expression, also have been studied as potential predictive biomarkers.
Oncohost is coming at the problem from a different direction: rather than looking at the changing landscape within a tumor, Prophet is designed to understand how the body’s host immune response protects a tumor from immunotherapy drugs.
No other diagnostic on the market, or in development, can provide insight into the complex interplay between patient, tumor and therapies, Sharon said. “We’re looking downstream, at the [proteomics] response to therapy. Our approach is true personalized medicine,” he said.
In research published in the Annals of Oncology in September 2020, the scientists behind Oncohost describe applying machine learning algorithms to data from 52 patients with stage IV NSCLC receiving anti-PD-1 therapy, in a search of a proteomics signature that distinguishes responders from non-responders.
Plasma samples were taken at baseline and early on in treatment. From these, the researchers found a 3-protein signature they say accurately identifies which patients will benefit. The signature was validated in a separate cohort of 82 patients.
The value of the proteomics assessment goes beyond identifying patients who will benefit from a particular immunotherapy drug, to highlighting specific inflammatory and metastatic biological pathways that are activated in non-responders.
These insights could be used to identify potential combination therapies that could overcome resistance, Sharon said. “In lung cancer, there are not that many approved options [currently]. But if you look in the pipeline there are hundreds of combinations in trials,” he said.
The question for physicians will be how to choose the right second line treatment from an array of possibilities. “We can give direction; we can answer the question,” said Sharon.
The proteomics analysis is based on the research of Oncohost co-founder and chief scientific advisor Yuval Shaked, into how the host immune response to cancer therapies can promote relapse and metastases after initial treatment.
The trial version of Prophet scans more than 1,000 proteins to identify patterns that are predictive of outcome. The 1.0 version of the test is due to get its formal commercial launch later this year. Sharon said partnering with the U.K. National Health Service will enhance the capabilities of the system and Prophet will evolve as more patient data is added, with updates released at different time points.
Oncohost plans to add a further three European trial sites, two in Denmark and one in Germany. The research program also will be expanded to ovarian, head and neck and urogenital cancers.
Oncohost expects to recruit up to 100 patients to the Prophetic trial at the eight U.K. sites. The trial is being financed with an $8 million series B round that Oncohost closed in January.