How Real-World Data Will Advance Oncology Clinical Trials
By Ed Miseta, Chief Editor, Clinical Leader
Real-world data (RWD) is coming to fruition in clinical trials. One area where it is expected to have a huge impact on innovation is oncology studies. This topic was discussed by a panel of experts at The Future of Health Data conference hosted by Datavant. The panel included Susan Winkler and Andy von Eschenbach of the Reagan Udall Foundation for the FDA, Jeff Elton of ConcertAI, Todd Golub of the Broad Institute of MIT and Harvard, and Alpa Patel of the American Cancer Society.
The panel notes innovation efforts in oncology are changing. In oncology drug development there are pre-clinical studies, the clinical trials themselves, and post-clinical trial real world clinical practice. In pre-clinical studies, the view has been that you can't predict whether a drug is going to demonstrate effectiveness and tolerability in the clinic. Therefore, companies have tried to move as quickly as they can to get drugs into a clinical trial. That thinking may now be giving way to a more scientific, data-driven approach. Researchers now understand that cancer is a disease of the genome that is genetically heterogeneous. While the pre-clinical cancer models are imperfect, those models also reflect the genetic heterogeneity of the disease and are better at predicting which drugs have a greater likelihood of showing activity in the clinic. Putting more resources into deciding which programs go into the clinic will help reduce the number of failures.
Second, most of the measurements taken in early and late clinical trials are focused on tumor volume, with patient benefit measured by survival. But it's a relatively unsophisticated perspective on what's happening during the clinical trial. Golub notes we now have a whole set of molecular tools that are increasingly becoming more powerful and available. These tools will help researchers determine whether a drug is hitting the intended target. Blood tests will also be increasingly used to make better decisions around clinical trial planning. For example, test results will provide indicators about whether there is reason to have confidence in an early clinical trial.
The final step is RWD. Once a new drug is approved, most patients are treated in community clinics. Usually, the information gathered is lost and researchers do not learn from it. That needs to change, and Golub believes we need to figure out how to capture that information and cycle it back into the discovery process and use it to make better predictions.
Use Data To Answer Questions
The additional information about patients collected outside the clinical trial could answer a lot of questions if it could be linked to the trial data. Von Eschenbach notes in clinical trials there are the left and right ends of the bell-shaped curve where some patients do not respond at all and immediately fail. At the other end are patients who see results for an exceedingly long period of time. The question is, what's the difference between the two?
“RWD gives us the opportunity to see that at scale and be able to use some of these new tools to interrogate those two ends of the spectrum,” he says. “If we can figure out why one patient did exceedingly well and why one didn't, that opens up a whole new horizon for opportunity.”
In the future, as sponsor companies think about how to better design studies, the comparator should be a standard of care and should be where most of the population receives that care. Different sub-populations may be disproportionately affected by a disease state, but those patients may not have been adequately represented in clinical trials. Elton notes RWD can inform researchers of the populations of interest and where a focus needs to be placed.
“RWD is starting to play a critical role in study design,” says Elton. “It is informing endpoints of interest and how a new therapeutic may compare more directly to outcomes of current standard-of-care treatment approaches. Concepts for the registrational phase can be derived from data sources and evidence coming from a wider array of clinical settings.”
“Cancer is not an event, it's a process and it's going to continue to evolve over time,” adds von Eschenbach. “The way in which we manage it is also a process that must evolve over time. There is so much richness we could harvest if we get RWD processes right.”
Use RWD To Improve Trials
RWD creates a great opportunity to improve clinical trials. Patel looks at COVID as an example. The American Cancer Society rapidly deployed a daily symptom tracker for COVID symptoms in one of its cohorts. Although the studies were for cancer patients, changes were made to expand to COVID data collection at a large scale.
“As we think about the longer-term nature of these cohorts, we have extremely well annotated data at the participant level,” says Patel. “We have a select population in clinical trials, so how can you ensure that you're accounting for those similarities or dissimilarities? To do that we need a lot of data. Most of the people in these populations are not being treated at cancer centers. We can get comparison groups and potentially partner to bring these data together. There's a tremendous number of opportunities in that space if we start to think more broadly about these types of cross sector partnerships.”
The panel believes the industry wants to get to a data ecosystem where it no longer must rely on a single method or single source for data collection. Clinical trials should be able to take in similar or disparate types of data, collected in different ways, that will each present their own challenges.
For example, the conventional wisdom is that patient provided data couldn't possibly be accurate enough to be useful. Golub notes that is simply not true, particularly if the goal is to understand if there is a substantial benefit for the patient versus tumor response.
“This is an area where cancer patients and their families are very engaged,” says Golub. “They know what drugs they're taking, and they know what responses they've had. For that reason, we need to be able to draw on patient provided data, clinical trial data, and data that comes in via third-party sources.”
How Much Data?
That raises the question of how much data is needed by researchers. With all the data currently being generated, drug developers will need to parse that data and ensure they are receiving relevant pieces of information. The industry will need to determine if RWD meets the standards expected in clinical trials. With enough data, and the right tools, researchers will be able to determine what data is and isn’t relevant.
“Our data scientists are always talking about the completeness of the data rather than the quantity,” adds Elton. “Some may talk about data like it is one big thing, but in oncology, there is an array of different data types and significant quantities of unstructured data across notes, radiologist reports, next-generation sequencing reports, and a variety of other clinical files. Brought together with the right connections understood, these can give us a more complete, high-fidelity view of patient response, disease progression, and therapeutic benefit. The formalization of how we treat these data, consistent with the FDA’s preliminary guidance document issued today, allows us to integrate RWD-derived Real-world Evidence data and analyses with high reliability and veracity across an array of registrational and post-approval studies. We're getting a lot more sophisticated about how we can achieve this more complete and accurate view of the data informing our questions. In some of these data sets there is a lot more richness and potential clarity than is oftentimes realized.”
For RWD to be used to drive innovation in clinical trials, the right data needs to be gathered in the right manner and shared with the right people. In a clinic, staff will acquire the data, aggregate it into information about the patient, and then act on it. Staff will continue to accumulate knowledge as it evolves. RWD will attempt to do that at scale, which will make it a powerful tool for researchers. The patient experience will help researchers determine their next hypothesis. In the oncology clinical realm, access to RWD can inform study design and may mean determining the right treatment for a patient in less time.
What Needs to Change?
Still, for RWD to have an impact on clinical trials there is a lot that must change. Each panelist was asked if they could fix one thing relating to RWD to advance its use in oncology trials, what would it be?
One panelist would like to see RWD become an integral part of regulatory decision making. RWD needs to be acquired, aggregated, and analyzed. Only then will regulators be able to act on it. Progress has been made in acquiring and aggregating data, but additional progress will need to be made in data analytics.
Another panelist would like to see data aggregation driven by the individual, noting that patients are motivated to share their data if they know it will be used ethically and will maintain their privacy.
Another panelist agreed, noting they would make it possible for any patient to opt to share their data. Many physicians are not familiar with the research community, and patients will not know they own their data and have the right to share it. Therefore, a culture shift will need to occur. Incentives may also need to be created to encourage this flow of data. For example, a physician not getting paid for a test unless they ask the patient if they wish to share their data.
Progress is being made, although not at the pace many would like to see. There is no magic wand that will fix the issues that still exist, but the industry working with healthcare institutions and regulators will hopefully make RWD a powerful tool to be used in the future.