NEW YORK – Israeli startup OncoHost is hoping to validate and launch its proteomics-based Prophet platform next year as a tool for predicting immunotherapy benefit using serum samples researchers are collecting from late-stage lung cancer and melanoma patients.
Binyamina, Israel-based OncoHost has developed the Prophet platform based on the idea that host response — or the characteristics that shape the body's, not just the tumor's, reaction to cancer therapy — can provide valuable information about whether a patient might respond to treatment.
"If you look at the existing products on the market today or in development [for predicting immunotherapy response], you will see that most of the companies are focusing on the interaction between the therapy and the tumor," OncoHost CEO Ofer Sharon said. "And by doing that, we are missing a big chunk of the story, which is … the human being who is actually suffering from the disease."
The Prophet test that OncoHost is currently validating in its observational PROPHETIC trial involves proteomic analysis of serum blood samples from patients in the US, Europe, and Israel. Specifically, OncoHost conducts proteomic profiling of plasma samples collected from melanoma and lung cancer patients both at baseline and after treatment. The platform scans over 7,000 proteins in each sample, and then, using high-throughput machine learning-based algorithms, it identifies patterns in these samples to predict outcomes as well as identify potential drug targets and strategies for overcoming therapeutic resistance.
Among the thousands of proteins that OncoHost screens for in patients' blood samples are certain cytokines, chemokines, growth factors, and enzymes that can promote tumor growth or therapy resistance.
"The tumor is kind of a hybrid entity," Sharon explained. "It's foreign to the body in the sense that these are cells that lost their way, but it's also a part of the body." As a result, when anti-cancer therapy is introduced, Sharon explained that the body might see it as "an insult" against itself and respond against that perceived attack by helping the tumor survive through the creation of new blood vessels, by inducing metastases, promoting tumor cell proliferation, and firing up metabolic and other processes. These mechanisms cancer cells use to hide from therapy can all be traced back to the proteins that the Prophet platform is designed to analyze.
OncoHost has designed Prophet so that it can gauge whether patients are responding to immunotherapy as early as their first dose. This could help patients and oncologists determine whether to continue treatment, switch to a different regimen, or attempt to hinder resistance mechanisms.
As Sharon explained, OncoHost tests responders and non-responders to a certain treatment to identify differences in expressed proteins, then uses Prophet's machine-learning algorithms to understand the role of these proteins in patients' cancers, and from these insights identifies targetable markers. For many of these proteins, there might be clinical trials underway evaluating targeted treatment approaches or approved drugs. The specific potentially targetable proteins that OncoHost has identified in completed validation studies are proprietary, Sharon said.
Validation, product launch
OncoHost is still validating Prophet for broad commercial use, which the company is expecting next year. As of now, Sharon said OncoHost has tested the platform on just over 700 patients in a research context and will test it in as many as 2,000 more patients in the PROPHETIC trial.
The firm has been reporting a steady stream of confirmatory data on Prophet at recent medical conferences. During the 2020 American Society of Clinical Oncology annual meeting, for instance, OncoHost presented data from 34 advanced melanoma patients treated with just a PD-1 inhibitor or anti-PD-1, anti-CTLA4 combination therapy.
After obtaining plasma samples at baseline and at two-to-four weeks post-treatment, OncoHost performed proteomic profiling using ELISA-based protein arrays, then applied a linear model to identify a 10-protein signature that could distinguish between responders and non-responders with a sensitivity of 0.94 and a specificity of 0.79. OncoHost's investigators also performed a pathway enrichment analysis, which revealed different treatment resistance mechanisms, mostly pertaining to immunosuppression and inflammation.
While the ongoing validation study is focused on characterizing non-small cell lung cancer, small-cell lung cancer, and melanoma patients' responses to immunotherapy and chemotherapy, OncoHost plans to enroll patients with other tumor types, such as head and neck squamous cell and urogenital cancers.
Company origins, directions
OncoHost launched in 2017 as a spinoff of the Israel Institute of Technology, and Prophet was the brainchild of Yuval Shaked, the head of the institute's Integrated Cancer Center.
"What [Shaked] was investigating over the last 15 years was the effect of the patient's own body on the interaction between the cancer therapy and the tumor," Sharon said. In 2019, Shaked put out a comprehensive paper in Nature Reviews Cancer, in which he laid out the ways that host response could affect tumor growth and therapy resistance, potential strategies to mitigate host response-related treatment resistance, and the tools and trials needed to turn this biological knowledge into a clinically useful predictive test.
"The majority of studies describing host response mechanisms are currently preclinical, and they are sometimes limited to a specific tumor model or treatment modality," Shaked wrote in the paper, encouraging researchers to focus on "measuring host responses in clinical studies in order to increase … understanding of the 'silent' adverse effects (that is, host responses) that have never been measured before."
Shaked — who is OncoHost's CSO and cofounder — did acknowledge that one potential drawback to the host response approach to predicting treatment response is that it requires at least one sample to be taken after therapy begins. This means that a patient who might not respond to a given therapy would need to begin that therapy before the test could characterize their response to it.
"The prediction of therapy outcome before any treatment would require massive data collection and applied machine learning to correlate proteomic and cellular data with outcomes to a specific therapy, and, as such, is currently unfeasible," Shaked wrote in the Nature Reviews Cancer paper. "Thus, the hope is that the knowledge gained from studying therapy-induced effects can be translated to the clinic, specifically for predicting responsiveness to therapy and developing improved personalized combinatorial drug approaches."
This is exactly what OncoHost hopes to do. The company's plan is to first launch Prophet as a tool for mitigating treatment resistance early in the course of treatment and then gather enough data so eventually the blood test can predict whether a patient will respond to treatment before they receive it.
The five-year-old company, which has a presence in multiple continents, is rapidly expanding so it can gather the data to fuel its ambitions for Prophet. The firm closed an $8 million Series B funding round early this year and used it partly to open a US-based affiliate. OncoHost has recently registered the US affiliate and is hiring its first employees, according to Sharon. The company expects to secure a physical location in the US in late 2021 or early 2022.
Establishing US operations will be an important milestone for OncoHost, Sharon explained, because the firm wants to commercially launch Prophet first in the US market followed by Europe. In the UK, OncoHost has a collaboration with the National Health Service through which it is running its PROPHETIC trial at eight NHS sites.
The fact that Prophet relies on serum blood samples, in Sharon's view, is a major advantage over predictive tests that rely on tumor tissue and an attractive feature for oncologists treating late-stage cancer patients who cannot provide tissue samples. Initially, the blood samples will be analyzed at OncoHost's central labs, which it expects to establish in the US, Israel, and Europe upon Prophet's launch. In the future, it might be possible to expand sample processing to additional sites, he said.
Sharon hopes that OncoHost's research and Prophet platform will move cancer care toward a future where patients are no longer treated based only on whether their tumors are likely to respond to treatment but based on their body's overall ability to respond to a given drug. "This is where precision medicine should go," he said.