At Real-World Evidence Digital Week (March 2021), Dr Boxiong Tang, Head of Health Economics and Outcomes Research (HEOR), U.S., Europe & New Markets Medical Affairs at BeiGene, presented a session on the role of Real-world Evidence in clinical development and product registration.
The session featured a Q&A at the end, with numerous questions from the audience, which we have gathered together here. You can watch the full session on-demand, as well as nine other sessions, here.
What is the key reason that you would say that there are more success stories in recent years? More flexibility from the FDA or other agencies regarding how they rate real-world evidence? Better use of methodology overruled by researchers? Better quality of real-world evidence available currently than before?
Dr Tang: My view is that there are more success stories thanks to all of those reasons. First of all, with many years of experience in with real-world evidence, more and more, good quality of the data has been accumulated. If you talk to the many data providers, they have been collecting good quality data for 10 or 20 years. It's not built up within the last one or two years. Also, if you go beyond that, the quality of the data is not only contributed by the health care sectors, but also by the technology developments.
Secondly, I think there’s been a big push from the regulatory authorities. In the US, I think the FDA's realized they wanted to speed up and to fill the gap of the unmet medical need. They also wanted to revolutionize clinical development in the real-world - compared to a critical trial, in the earliest stage of, we call it the test, the clinical trial is more unscientific. But for maybe 20 years of the life cycles after the approval, there would be the real-world, so real-world data in that piece would be better than the clinical trial. Three or five years ago, even for us, we'd never think about using real-world data to support the regulatory approval because it would not be accepted. But now, the environment is different.
Could you explain the difference between external versus synthetic control arm?
Dr Tang: To be honest, for me, I don't see any difference. I use the external control because it's very easy to understand. If you look at the synthetic, that means the data is not collected or found in this clinical trial, but it's from another, and the external control arm, by definition, is the same. The data for both are not from the current clinical trial, but from either another clinical trial or it could be real-world data analysis.
So for me, those two terms are used interchangeably. The only difference if you want to think a little bit more, is that the synthetic control arm could also use other methodologies such as mathematically matching with published data. It's not necessary an individual patient, one to one match. Synthetic means all the different way you can calculate based on the model simulations, based on the propensity score.
What are the challenges of using real-world data analysis in clinical development and regulatory support?
Dr Tang: First of all, real-world data is still relatively new, relatively young, so even internally within the clinical development team, we don't have a lot of expertise or experience. So I'm not only a presenter today, but I’m also learning.
Resource, should probably not be a big issue as real-world data to support the clinical trial will not only save the time, but also save your budget, your money, so I don't think that resource is a big challenge.
However, internal and external acceptance can be a challenge. Initiative are not only new to the researchers, to the industry, but also new to the regulatory agencies. The FDA or EMA develop documents and engage with many academic experts and professionals to improve. I think we have to work together on this - for example, you'll never find a perfect match because real-world endpoints are still different compared to clinical endpoints. So I think the FDA has to review case by case and sometimes, they invite the advisers to determine whether a real-world endpoint can be acceptable. You'll never find a perfect match, but I think it's whether the endpoint is going to be acceptable to compare with the clinical endpoint.
Do you think of EMR data as external control arm or synthetic control arm?
Dr Tang: It is both synthetic and external control arm as both can come from the electronic medical records, but also, they can come from the previous conductor of the clinical trial. So that's the difference, they're not from the EMR. That is the third category. It's a history control. That kind of control arm can come from published information, which means the control arm can come from multiple data sources, not only from the EMR.