I noticed that you are looking to hire some data engineers at Datavant. Do you hire people from pharma?
T. May. Yes. Our biggest priority I think even 5 years from now is going to be, how do we get the best data scientists in the world and how do we get the best software engineers in the world? But our strategy is absolutely that we want to have the right combination of tech DNA and healthcare DNA.
Can you tell me what your vision is for Datavant?
T. May. Absolutely. Our 50-year goal is we want to organize all the world’s healthcare data. It’s big and aspirational and we recognize that it’s an impossible challenge but we’re ambitious. For the next 5 years our focus is on organizing data that’s helpful specifically for clinical trial design and interpretation.
The thesis we have is that the data exists today to dramatically change the odds of success for clinical trials, but the biggest challenge is that it’s extremely siloed, it’s extremely unstructured, and it is extremely difficult to use. Our goal for the next 5 years is – Can we organize all that data and ultimately double the odds of success for clinical trials? Something like 8 per cent of clinical trials today succeed. We think the data exists to more than double the odds of success for clinical trials in the future.
How will you use the data in order to reach that goal?
T. May. There’s really three challenges around the data. Accessing the data is a huge challenge. The data is siloed. Some data is held by pharmaceutical companies, and even within a pharmaceutical company data is siloed across different teams. Some data is sitting with pharma companies, some data is sitting with CRO’s, some is sitting with payers, some is sitting within hospitals and EMR’s, some is sitting with regulators, some with academics, etc. And so that’s a huge challenge. How do you get access to all of the disparate data that exists?
The second challenge is – How do you organize and structure that data? A lot of that data is sitting in a PDF form. EMR data could be sitting in handwritten notes. Some of the clinical trial data that’s publicly available might, for example, be in the appendix of a presentation that was given at some conference a few years ago. So, how do you organize and structure that data into a usable way?
The third challenge is, Now that you’ve got access to all the data you want to structure, how do you actually apply it to a given question and apply it to improving the design of a trial? I think those are the three dimensions that we’re focused on. In the early days, the biggest challenges are around solving for data access and data organization.
I understand that Datavant has already compiled data from 85 different datasets comprising over 20 million patient visits. What do you do with the data once you get it organized?
T. May. What we’re primarily focused on is improving the design and interpretation of a clinical trial, so the 3 biggest areas where more data could help you make a better decision. One is inclusion/exclusion criteria. Can you understand what’s succeeding and what’s failing in the real world? Can you understand some of the tradeoffs that you face with inclusion/exclusion criteria to do a better job targeting who you want to be in your trial?
The second one is dosage. You can narrow to some extent the dosage that’s optimal for a patient, but you can’t know it that deeply or that precisely, and so the more data that you have from other trials that have happened in the past, the more you can hone in on what’s the right dosage to include in your own trial.
The third is more aspirational. How can you make trials more adaptive and Bayesian trials? Any form of adaptive trial requires a thesis on how the world works, and you can form a better thesis on how the world works by using real world evidence as a foundation for your hypothesis.
Do you sell the data? Or do you use it for Roivant Science’s companies?
T. May. We see Roivant as a client of ours, but our goal is to build a utility for the whole industry to be able to have better data to improve its trials. And we think that ultimately, if you can change the odds of success, that will lift all boats, and that’s great for the whole industry if we can be successful. Our goal is to service the whole industry as we develop.
There are claims that AI can cut the expenses of drug development by more than half, and reduce the cost of clinical trials by up to 70 percent. Are these outrageous claims?
T. May. It’s always difficult to quantify. AI’s name has a lot of hype around it, and I think it’s always tough to separate the hype from the reality. From our perspective, I think the 5-year goal of – Can we double the odds of success of clinical trials? If we do that, that should at least halve the cost of successful drugs. If you double the odds of success, then your expected value of doing a trial is doubled.
I think, from everything I’ve seen in the industry and from everyone I’ve talked, to, that level of impact is attainable and isn’t far away. I think that the fundamentals of how a clinical trial is run and how the design of a trial works haven’t changed in decades, and there’s been a real explosion of data and of techniques to use the data in the last 5 or 10 years, and I would not be shocked to see a transformational impact of that.
Do you predict any changes in the people that will be working in clinical trials in the future?
T. May. I think that a clinical trial design is in large part science and I don’t believe that, in the next 5-10 years, you start outsourcing clinical trial design to an algorithm. I think that a company like ours, that brings in extra data and extra analytics, will be used to augment the thought process that a team is already going through for how to design trials. Some changes in the skill set are needed, but I see it as being an industry that’s just augmented by better technology versus truly disrupted by better technology.
How were you able to get data from 85 different datasets combined so quickly, when earlier data sharing initiatives have not been as successful?
T. May. There have been a lot of initiatives around data sharing in the past. I think that one of the big challenges around data sharing is that you have to design the incentives so that everyone wants to participate. In the marketing world, nobody shares data out of altruism. That’s not the reason why data is being shared. Data is being shared because it’s beneficial for the companies involved when it’s shared. The challenge in a lot of the initiatives in the pharmaceutical industry is that they’ve been too governed by assuming altruism, where in my mind you have to get the incentives right so that the self-interest of all parties is to share data…I think that’s one of the main challenges in the data sharing portion of the industry today.
—————————————
BIOCOM Medical Device Breakfast: Designing and Implementing Effective Clinical Trials
CONTRIBUTED ARTICLE BY GREG DOMBAL, COO, Halloran Consulting
On October 12, BIOCOM’S Medical Device Committee hosted a breakfast program on designing and implementing effective clinical trials. Jeff Anderson, SVP of Clinical/Regulatory Affairs at REVA Medical and Art Weber, Director of Clinical Affairs, Banyan Biomarkers, presented their experience and lessons learned from successfully managing pivotal trials for their company’s product. Greg Dombal from Halloran Consulting moderated.
Despite having very different products in different therapeutic areas, both study managers were driven by the same principles to a successful outcome.
REVA Medical is a medical device company headquartered in San Diego that has developed a bioresorbable scaffold, as an alternative to metal stents, to treat coronary artery disease. Jeff Anderson presented the trial that he led- REVA’s Fantom® II pivotal clinical trial of their Fantom® scaffold, which was implanted in 240 patients between 2015 and 2016. Banyan Biomarkers, Inc., also headquartered in San Diego, is creating the first blood test to diagnose traumatic brain injury (TBI). Art Weber presented the company’s ALERT-TBI pivotal trial of 2,000 patients, which he managed. If approved, this will be Banyan’s first commercial product.
The environment at both REVA and Banyan is not dissimilar from many other startup medical device and diagnostic companies in SoCal that pose a unique challenge for clinical research directors – limited resources, a non-revenue generating company, no out-of-U.S. staff, just a handful of full-time clinical staff in their San Diego headquarters, and a reliance on CROs and consultants.
Funding sources can add to the challenge. REVA is venture-backed, increasing the pressure to deliver results quickly, and creating a “need for speed” culture. Banyan is funded by grants, contracts, and private investments, including a $26 million contract from the U.S. Department of Defense for the development of a diagnostic test for TBI, and the ALERT-TBI study is high profile within the Defense Department.
How do we see through potential data overload to focus on the critical factors that will make the trial we are designing one that can be run on time, on budget, to deliver data that move a product forward?
Running efficient and effective clinical trials for medical device and diagnostic companies has become increasingly complex. The sheer volume of available genetic and epigenetic data on patients, outcomes data on different procedures and treatments, and the variety of data from personal health trackers can be overwhelming.
There is often the temptation to design more data into our protocols, and to ask more questions, in the belief that this will ensure success. Many times, however, a theoretically well thought out study collapses under the weight of its own complexity- narrowing inclusion criteria so much that patients become elusive; collecting so much data that a site cannot fit trial operations into their busy clinical practice; or the study becomes too costly due to the inclusion of mobile devices.
Define your endpoints carefully; don’t collect data that you don’t need; and design your trial with reimbursement in mind. Both REVA & Banyan spent considerable time planning their clinical program and developing their protocol, seeking input from key opinion leaders, clinical sites, and other stakeholders. This allowed them to simplify their protocol and collect only the data they needed to support a regulatory approval.
Selecting the right partners is key, and bring in help only as needed. Both REVA and Banyan were fanatical about using locally based resources when possible. REVA kept study management and statistics in-house, and hired CROs for study monitoring and regulatory. REVA decided to use multiple regional CROs rather than one “global” CRO for OUS studies because medical practices differ across regions, and “local” CROs can more effectively troubleshoot with their regulatory authorities. Banyan chose to contract with independent consulting CRAs rather than a CRO for study monitoring.
Both clinical programs committed to building and maintaining relationships with clinical staff, vendors, and regulators. Jeff and Art emphasized the importance of relationships with clinical sites and study coordinators; including study coordinators in operational decisions; and clinical site training, making sure that coordinators are included in investigator meetings and are trained on the EDC.
The outcome for both companies was phenomenal success in clinical operations, delivering critical data to advance important medical programs one step closer to market. In early 2017 the Fantom® scaffold received European CE Marking, and Banyan recently completed data collection for ALERT-TBI.