This article has been updated to correctly state that more than Colox tests have been ordered for more than 5000 patients.
NEW YORK – Novigenix is seeking pharmaceutical industry partners to test its newly unveiled artificial intelligence-driven LITOSeek liquid biopsy platform in drug trials with an eye toward eventually developing it as a tool to guide treatment for cancer patients.
The Lausanne, Switzerland-based company has developed a platform that uses AI to analyze immune-related gene-expression levels via RNA in whole blood for early evidence of disease or response to therapy. It currently markets the Colox blood test in Switzerland for detecting early-stage colorectal cancer for people aged 45 and over. The test detects colorectal cancer with 78 percent accuracy and adenomatous polyps with 52 percent accuracy based on analysis of 29 biomarkers that capture how the immune system is responding to malignant cells.
Now with LITOSeek, Novigenix is offering AI-powered transcriptomic analysis as a way for drugmakers to profile the immune system's response to cancer treatments they're developing. Novigenix CEO Brian Hashemi compared the company's approach to accessing the "operating system" of the immune response using RNA analytics. At the same time, he noted that the AI platform overcomes some limitations in the space such as the computational burden of RNA analysis and the noise associated with the data.
Stefan Scherer, CEO of 3T Biosciences and chair of Novigenix's clinical advisory board, noted that it would not be possible to conduct whole-transcriptomic analysis using RNA without AI. "In order to master the complexity [of RNA], you need sophisticated methods," he said.
LITOSeek is built on a database comprising clinical data from about 3,500 patients and 5,000 RNA-seq profiles. That includes healthy and cancer patient populations and samples drawn at multiple time points from some of the patients. LITOSeek is also informed by a library of clinical studies, academic collaborations, and biopharma partner programs and continually monitors public databases for relevant information.
The analytic process begins with the collection and stabilization of the patient's blood or tumor biopsy sample. Novigenix then carries out whole-transcriptome sequencing followed by harmonizing and de-noising the readout using multiple algorithms. Next, another set of machine learning algorithms selects biomarkers and trains a predictive model.
Novigenix is pursuing multiple applications for its models, including early cancer detection, precision oncology therapy optimization, and drug target discovery. The Colox test, which it markets with Unilabs, has given the firm confidence in its AI platform. To date, more than 400 physicians have ordered Colox for over 5000 patients.
According to Hashemi, comparable liquid biopsy tests that use circulating tumor DNA (ctDNA) have a sensitivity in the range of 10 percent to 15 percent for detecting advanced adenomas, because a certain quantity of tumor load is required for the signal to be detectable in blood. In contrast, "we are essentially leveraging the sensitivity of the immune system of the patient and its reaction to the onset of cancer to pick up those signals earlier, and that gives us about a threefold improvement over ctDNA technology."
In January, Novigenix reported its first results from an updated version of the Colox test developed using LITOseek. In that study, researchers used LITOseek to generate transcriptomic profiles from patients with colorectal cancer, advanced adenoma, non-advanced adenoma, other types of cancer, and controls without colorectal cancer. They identified gene-expression signatures using machine learning methods to rank the patients and create a gene classifier, which had 71 percent sensitivity and 94 percent specificity for detection of advanced adenoma. Novigenix is currently validating the classifier in an independent patient group.
The company has also developed an immune signature associated with immunotherapy response in metastatic bladder cancer and tested it in a mouse model. The results of that work, Hashemi said, suggest that a LITOseek-based test could be used to identify patients with bladder cancer most likely to respond to immunotherapy.
"Checkpoint inhibitors typically work in 30 percent of patients," increasing to approximately 50 percent to 60 percent with the addition of newer antibody-drug conjugates, Hashemi said. "The question becomes, 'Why are the remaining patients resistant, and what is the mechanism of resistance of those patients?'"
Novigenix has now completed a validation study in an independent cohort of about 89 patients with bladder cancer and stratified them based on the immune profiles provided by LITOseek. The analysis has revealed activation or suppression of certain immune pathways in some patients resistant to immunotherapy. The results from this study, which Novigenix conducted with collaborators at Radboud University Medical Center, are being prepared for publication. Hashemi said the test can provide insights that inform the development of novel precision therapies and, in the future, may be useful for guiding an individual patient's treatment decisions.
The company has also used LITOseek to identify sets of biomarkers predictive of response to immunotherapy in metastatic melanoma and urothelial cancer.
Other companies are also taking advantage of increasing computer power and newer AI technologies to develop tests based on whole-transcriptomic analysis. Caris Life Sciences' Assure Liquid Biopsy assay uses AI to measure cell-free DNA, cell-free RNA, white blood cell genomic DNA, and messenger RNA in a single assay and assesses tumor-derived nucleic acids plus incidental germline mutation and clonal hematopoiesis of indeterminate potential (CHIP).
"If there's something in the cell that's going wrong, there's something in the molecular system that's going wrong," Caris President David Spetzler said. "If we measure everything, we'll find the patterns that are indicative of that disease." The company has submitted a paper for publication on the clinical utility of the Assure assay for minimal residual disease detection in solid tumors.
However, a significant difference between Novigenix and companies like Caris developing AI-powered whole-genome and whole-transcriptome tests for cancer is that Novigenix is measuring changes in gene expression of the immune system, whereas the other tests in that space are usually looking at signals derived directly from the tumor.
"It's prime time for AI and RNA," Hashemi said. "We've managed to combine the two to bring unprecedented insight into patient populations and response to therapy."
Novigenix has already initiated clinical trial partnerships with biopharmaceutical companies. In 2019, it inked a deal with RadioMedix to develop a precision diagnostic test that measures response to alpha and beta peptide receptor radionuclide therapy in patients with neuroendocrine tumors. In addition, Hashemi said Novigenix is working with other large US public and private organizations that it hasn't disclosed publicly yet, but the firm expects to report results from several large trials from those collaborations this year.