Precision medicine starts well before the first trial participant is treated, with precise research, discovery, and development. While there are many pieces to this puzzle, predictive biomarker databases powered by AI can be much more than a repository of information. They can seed new lines of inquiry and novel biomarker-drug combinations that apply to diverse tumor types.
With the explosion of sequencing data and computational power, a range of different biomarker databases are now on the market. Some leverage sequencing data and/or EMR patient data, others tap into the ever-growing pool of published research in this space. The key for biopharma companies looking to access this knowledge is understanding where the platform came from and what its core focus is.
If it’s an academic database, or perhaps one developed by an early-stage commercial research company, it likely specializes in biochemistry and biological characterization of genetic alterations or molecular activity of therapeutic compounds. For example, whether a mutation predisposes someone to cancer, or whether a new compound is toxic to humans. Similarly, an EMR-derived healthcare database will be most helpful for clinicians looking to optimize outcomes for specific patients, providing an individualized medical assessment based on their genomic data.
LARVOL’s database, VERI, occupies a specific functional space for biopharma companies, providing focused information on predictive biomarkers for oncology drugs, including modern targeted therapies. VERI's core focus is helping users understand whether a given biomarker influences response to a drug. Specifically, does it predict sensitivity or resistance to that therapy of interest? This information is valuable for assessing genomic alterations in everything from cell lines to FDA-approved therapies. Mutations of unknown significance are omitted; it’s a focused knowledge framework with actionable results.
This information has many potential applications in today’s drug development landscape. One example is in the design of clinical trials. With an easy search of the VERI database, biopharma companies can learn which biomarkers may influence treatment response, based on insights ranging from news sources to social media to scientific conference abstracts. If a biomarker confers resistance to the therapy then the individuals expressing the gene, alteration, or fusion can be excluded from the trial. If a biomarker increases sensitivity to a drug, that patient population may be the focus. This is the foundation of the VERI database – but it’s just the beginning.
For biopharma companies, commercial success is not just about drug approvals. To stay competitive, they need up-to-date information on trends and activity in the field. LARVOL built this service into VERI, creating a one-of-a-kind biomarker database and competitive intelligence engine.
For background research, it employs text-mining technology that makes news and articles readily searchable. VERI also provides weekly alerts, curated by a global team of experts who are surveilling over 150 tumor types, more than 1,000 drugs, countless different genes, 250+ companion and complimentary diagnostics, and 3,000+ biomarkers.
Recognizing the critical overlaps, particularly around companion diagnostics, VERI also provides insights into over 200 clinical diagnostic tests and their associated companies. That includes clinical news and business updates, such as mergers and acquisitions that affect the broader landscape. Once again, this is curated by a team of analysts who make the information accessible and decisive.
What’s on the horizon? Predictive biomarkers in oncology can include a variety of factors ranging from activating oncogenic mutations and deletion of tumor suppressors, to complex changes in gene expression signatures and alterations in serum protein levels, all of which predict response to therapy. It’s a lot of information to manage. VERI is designed to make sense of it all, prioritizing key associations and ranking the level of scientific evidence supporting each result. The next step is to encourage novel approaches for analyzing and visualizing biomarkers and therapies.
LARVOL’s initial goal with VERI was to provide oncology companies with the full landscape of predictive biomarkers at their fingertips – the scientific findings that matter most and the context in which they were found. The VERI team continues to deliver on this goal, adding more features and functions. But with the core foundation in place, a new question surfaced: How can this knowledge base help users approach the unknown? What studies haven’t been conducted (or published) that may show valuable connections between known biomarkers and different therapeutics? Scientists are just scratching the surface when it comes to all the possible connections that can be made, across tumor types and compounds.
To answer the questions that weren’t being asked; to find the connections that weren’t being made requires new approaches like data visualization and artificial intelligence.
Alongside user experience advances in the platform, such as the inclusion of intuitive heatmaps, LARVOL is systematically reviewing what drugs might make sense to explore in novel cancer settings. This is achieved through the parallel use of AI/deep learning and a rules-based machine learning algorithm that analyzes the drugs and the tumor types in the knowledge base, along with the biomarkers and biomarker genes.
VERI is designed to organize and prioritize information, displaying different levels of insight based on the user’s current need. At the highest level, the platform features a heatmap that tracks evidence for the different biomarker associations, providing a snapshot of key findings that can then be investigated at a more granular level. Click here for a free VERI Trial.
LARVOL is a SaaS company, but its real mission is to improve human health. The team is driven by the possibility of facilitating discoveries that impact cancer care – discoveries that otherwise may have been missed. To that end, they're also developing 3D modeling capabilities that allow users to line up the same biomarker-drug combinations in a series of different cancers. This will allow them to see, on a multi-cancer basis, commonalities across drugs, biomarkers, and different tumors.
This tissue-agnostic approach mirrors the industry’s recent shift towards precision oncology, breaking down the theoretical barriers science and medicine have constructed between different tissue types. In mid-2017, the FDA endorsed this new approach by approving a drug, Keytruda (pembrolizumab), for for 29 indications including patients with solid tumors that express so-called mismatch repair genes. In other words, it’s the biomarker that matters, not the tissue of origin.
Scientists and clinicians know that a therapy that works for a person with lung cancer could also prove beneficial for patients with other types of solid tumors. The challenge is implementing this thinking more broadly. Approaching cancers based on their tissue of origin has been at the core of medicine for decades and it often dictates how doctors specialize. With VERI, researchers and clinicians can more readily connect the dots, using artificial intelligence to spark new ideas and advanced data visualization that allows users to easily compare biomarker and therapy information across cancer types.
For more on AI and deep learning as a tool for precision oncology, check out this podcast with LARVOL’s Sabrina Bellisario, Head of Operations, VERI, and Dr. Mark Gramling, Director, Oncology.