May 18, 2020
Precision Oncology News | May 04, 2020 | Caroline Hopkins | NEW YORK –
Germany-based Indivumed is enlisting the help of cancer centers and research institutions around the world to help expand its omics database and make it a key resource for unraveling the complexities of cancer biology and advancing new precision oncology drugs.
Indivumed last month launched the Oncology Alliance for Individualized Medicine (Onco AI-Med), which is a partnership between leading cancer clinics and research institutions selected based on their research activities and expertise in molecularly informed cancer care. Currently, the collaboration includes over 30 partners from 11 countries, though Indivumed plans to add new sites to the alliance continuously.
As part of the collaboration, the partner sites have access to Indivumed's multiomics database, IndivuType, which includes a biobank of cancer patients' samples deeply annotated with genomic, transcriptomic, proteomic, immuno-phenotypic, and digital histopathology data, combined with clinical information. The multiomics data are integrated using advanced artificial intelligence technology, with bioinformatic tools set in place to perform deep and detailed data analysis.
According to John Marshall of Georgetown University, who chairs the Onco AI-Med advisory board, Indivumed's repository currently contains hundreds of thousands of samples, including tissue, plasma, and urine specimens. The company, which was founded in 2002, has spent nearly two decades building out the repository through a number of joint biobanking collaborations, first within Germany and the US and then in other countries.
Indivumed is now working to develop specific prognostic and predictive algorithms using artificial intelligence tools, an undertaking for which it has partnered with other companies that specialize in genomics technology. In a deal announced two weeks ago, for instance, Indivumed said it would work with Menlo Park, California-based Personalis and use its genomics characterization technologies to further enhance its multiomics database with "genome and transcriptome-level data from thousands of samples."
As evidenced through new collaborations with companies like Personalis, Indivumed is continuously building out the data and algorithmic capabilities within IndivuType. As the company continues to collect different types of molecular data from across the global alliance and new biomarker signatures come to light, Marshall said the expectation is that new drug targets will be identified but also that oncologists will be able to use the database to inform patient care.
"The transition to patient care will happen as we develop algorithms," said Marshall, who shared that development for several of these algorithms is already well underway. "Once these algorithms are validated and approved, they can be used in clinical decision making. Until then, IndivuType is to be considered a research tool."
"We're working very hard to develop the technology with our AI partners in order to image and digitalize [these] molecular signatures so that this can actually be used day-in and day-out for clinicians," said Marshall. The clinical decision making aided by the algorithms, Marshall said, would start with the multiomics analysis of a patient's sample. These algorithms would then further analyze the relevant multiomics data and provide a result, most likely in the form of a score or report, which then could be used by the treating physician to guide therapy decisions.
Carlos Sampaio, founder of Clinica AMO, a company that operates oncology clinics in Brazil and a member of Onco AI-Med, said that the multiomics nature of the IndivuType database was one of the key reasons that his firm agreed to join the effort when Marshall first reached out about a year ago.
"The tools that we use nowadays to assess prognostic factors and [therapeutic] targets only look at the surface of complex biology," said Sampaio. "The platform that this alliance is trying to pull together is looking at … multiomics and is going to answer a whole bunch of questions we've been trying to answer for decades now."
Sampaio, who treats women with breast cancer, said he anticipates that the information gleaned from IndivuType will be particularly valuable for patients who continue to progress after receiving multiple lines of treatment. "Those patients may eventually be able to receive treatments that would not seem obvious with the currently available prognostic factors or target-driven lab reports," he said.
He offered the hypothetical example of a patient with triple-negative breast cancer for whom immunohistochemistry testing does not suggest benefit from immunotherapy. Through the technology and algorithms available with IndivuType, he said, "we may find that a target is up-regulated and that it might be important for tumor growth." In this way, he envisions that the algorithms available within IndivuType could allow patients to receive treatments that they may have missed out on using traditional predictive tests.
Implementing an educational framework
Marshall and Sampaio both emphasized that the information — in the form of a score or report — that partnered oncologists will receive when they run samples and clinical data through IndivuType's forthcoming algorithms, will fulfill its purpose of advancing precision oncology only if oncologists have the working knowledge needed to best apply them to research and patient care.
While access to deeply granular molecular biology information on patients' cancers is indeed one of the primary benefits of being a part of the Onco AI-Med alliance, Sampaio explained that this data is only good if oncologists can use it. However, he acknowledged that advances in molecular analysis in cancer care over the last five years has yielded data faster than many physicians can understand and apply it.
Enabling oncologists to parse the complex data and actually leverage it for research and patient care is something that both Marshall and Sampaio cited as a key goal of the Onco AI-Med alliance. To address this potential knowledge gap, Indivumed, in collaboration with Georgetown University and Oxford University, is working to develop and implement an educational program available to researchers and clinicians within partner centers.
The program, called Precision Oncology Web Education Resource, or "POWER," is funded through philanthropy that Marshall said he put together for the purpose of addressing the widening knowledge gap when it comes to precision medicine.
"We feel it is critical as we further expand the technology and the data that's being collected, that we're going to need to backfill information and education for the practicing physicians," said Marshall. "A very important role for our group is to develop a global training program for how to take these complex data and apply them as quickly as we can to patient care."
Standardizing collection, quality
When asked what makes this global alliance unique compared with other collaborations in precision oncology, both Marshall and Sampaio cited Indivumed's emphasis on the standardization and quality of tissue samples in IndivuType's repository. As Onco AI-Med's partners collect samples, Marshall said they will need to adhere to the standard operating procedures that Indivumed has used since its inception.
This will include standardized limits on the time that can elapse from when tissue or blood is collected from a patient and when it is preserved for analysis in the biorepository. Mean timing for this has been set at less than 10 minutes across the samples included in the IndivuType database. Collected blood is immediately cooled to between 4 and 8 degrees Celsius and processed in under four hours. Onco AI-Med members submitting samples must meet these standards in order to ensure that the detailed omics analysis can happen down the line.
"The quality of tissue collection in every other group I know has not been so well established and standardized as Indivumed has done," said Sampaio. "There's major strength in the standardization of tissue collection. Because if you don't do it properly, you can run the best tests available on Earth, but the answer you're going to get is not going to reflect the reality of tumor biology."
Marshall echoed the importance of this standardization. "There's that classic line, 'garbage in, garbage out,'" he said. "What gets tested has to be of the highest quality, and we need to be able to do that across all tumor types."
"If you take a breast cancer tumor sample and leave it out for let's say 30 minutes, the chemical reactions that will happen before you freeze that tissue will certainly affect the results of your lab work down the road," said Sampaio. "You can up regulate some proteins, you can shut down completely some short-waving RNA material, and you can destroy the targets just because the handling of the sample has not been properly done."
Although Indivumed's stringent sample collection standards are not what's typically expected within other biorepositories, such standards are much needed if the goal is to advance the type of deep multiomics analysis that Onco-AI-Med members hope to have access to. In his more than three decades in research, Sampaio has noted that each institution has its own standards for collecting and storing samples, but across the board, he said, "it's not strict."
Next steps, forthcoming publications
Marshall said that Indivumed has already analyzed a significant number of cases in four initial tumor types — gastrointestinal cancers, breast cancers, lung cancers, and kidney cancers —and research using those datasets is ongoing.
As the research proceeds and IndivuType's technology and algorithms mature, Marshall said that Indivumed will pursue drug development in partnership with pharma and biotech companies. One example he gave was Indivumed's collaboration with Evotec focused on colorectal and lung cancer drug development programs.
Under the terms of the colorectal cancer agreement, Evotec has full access to IndivuType's multi-omics data. Evotec has already applied its PanHunter bioinformatics analysis platform and drug discovery technologies to the database and identified three novel drug targets for the treatment of colorectal cancer. Following the success of that alliance, Evotec and Indivumed announced in January that they had entered a new research collaboration to discover and develop first-in-class therapeutics for the treatment of non-small cell lung cancer.
"Indivumed has its own capabilities for the early phases of drug development, including biomarker discovery and validation," said Marshall, "but typically the later stages would be covered by the pharma partner."
In the coming months, Marshall said that Indivumed plans to put out a position paper on multiomics analysis and its role in advancing precision oncology. The company expects as well that it will soon be able to publish data from the Onco AI-Med alliance that will articulate its value.
"While we're excited about what today holds, what this [alliance] is really about is what tomorrow holds," said Marshall. "I'm a clinician who takes care of patients with bad cancers, and I am hoping that this collaboration will generate information at a level that will transform my patients' care not tomorrow but in a year or two years from now, so that not only do we have better drugs and better targets, but that we also have better efficiency with what we are doing."