Using decentralized approaches to unleash the power of real-world evidence in long-term follow-up studies
By Gordon Cummins
There has been no shortage of words written during the past year on the potential benefits of applying decentralized approaches to randomized controlled trials (RCT) — especially in the realms of accelerating patient enrollment, improving retention and diversity, and generally getting investigational compounds to market faster for those who need them most. And with good reason. These approaches continue to show great results.
We also know that decentralized clinical trials are not just short-term novelty tactics for coping with COVID, but they represent a fundamental shift in the way clinical research is conducted that will far outlast the pandemic. In fact, our own research reveals that, in 2022, and for the first time, more sponsors and CROs are running agile, or hybrid, clinical trials (featuring at least some decentralized components) than are planning to run traditional clinical trials.
And that’s great for accelerating and improving RCT research. But what about long-term follow-up (LTFU) studies? Can decentralized and agile approaches also be applied to LTFUs and the collection of real-world evidence (RWE)? What does that look like, who’s interested in pursuing it, and what are the perceived benefits and pitfalls?
In the world of biopharmaceutical research, there are two distinct audiences of sponsors interested in harnessing the power of decentralized-driven LTFUs. The first is the study team, or the clinical operations, who are thinking about how to implement this type of design, extending the main study and transitioning into an LTFU. The second comprises the core RWE and health outcomes economics research (HEOR) teams, whose interest in data sets generally goes beyond that of a standard LTFU protocol.
Why clinical operations execs care
The core value propositions for the first group, the “clin ops” folks, are similar to those making the case for using decentralized approaches in the main clinical studies — namely, that the use of DCT components eases the burden on patients, providers, and sites, which, in turn, improves retention rates and provides better, more reliable data sets.
In some cases, LTFU studies are mandated by the FDA, as is often the case in the oncology space. In fact, any trial that is investigating CAR T-cell therapy is required by FDA to do between a five- and a 15-year LTFU. In other cases, there might be more of an agreement with FDA — or even just an internal rationale at the sponsor — to evaluate such things as long-term safety. Either way, sponsors and regulators alike are trying to figure out what is the most efficient manner of implementing LTFUs.
LTFU studies are not without challenges, of course. We hear from the market that patients, providers, and study sites have each questioned the value of participating in LTFUs, from time to time. The chief reason on the provider/site side is financial. Since these are not interventional or investigational types of studies, then the financial compensation is not as strong as for traditional clinical trials. Let’s say providers are seeing in their practices maybe three to five patients a year, and according to the study design, most of these visits are standard of care, collected maybe once or twice a year, with maybe some additional patient-reported outcomes (PROs) on the patient side. For this typical scenario, the clinician might make $500 a year at the most. So, for sporadic data collection at that level of reimbursement, it can become more of a nuisance to the provider than a value proposition.
Plus, for sites, the sheer duration of some of these studies can be off-putting — it’s not unheard of to work on studies of maybe 30 years, and some of the more apprehensive sites might be wondering if they’ll even be in business in 15 or 20 years.
Major life events
Patients may have similar reservations regarding the duration of LTFUs, especially when it comes to potential scenarios around major life changes. “What happens when I move from New York to California?” they may ask. So. the patient’s perceived burden may become greater. In this example, either the patient will have to pay out of pocket to travel, which means that they’re probably going to discontinue participation in the study, or the sponsor will have to cover the patient’s additional costs — again, for a once- or twice-a-year in-person visit.
Therefore, the classic core value propositions of DCTs hold strong, in terms of eliminating geographic barriers and reducing the burden — and the long-term concerns — of patients, providers, and traditional clinical trial sites.
The patient is, of course, paid a stipend, or reimbursed for the activity that they are requested to do, so the design, endpoints, and frequency become important. If the approach is to look at patients’ medical records, say, every quarter or twice a year, and we’re going to ask them to fill out a questionnaire once or twice a year, they would only get reimbursed for the questionnaire, which might be worth as little as $50 or $100 a year. So, we have to ensure that the study design is going to provide them with value, too.
The diversity issue
It has been well-documented that “regular” randomized clinical trials are not diverse enough and rarely reflect the patient populations for the treatments they are investigating. The rapid adoption of DCTs and agile clinical trials have been shown to improve diversity in these main studies (by three times across the board, according to our own experiences), but because the LTFU is predicated on the composition of the patients in the main study, the challenge on the diversity side is that, purely by the nature of the design, we can’t go out and find a new set of more-diverse patients.
And so, while there is no quick-fire solution to the diversity issue, as clinical trials become more diverse through adopting decentralized approaches, intrinsically, so will LTFUs. In other words, we expect to see in LTFUs a continuation of the improvements regarding the diversity of participants that we are currently experiencing in regular clinical trials with DCT elements.
RWE: Going beyond the study data
On the RWE side, there are a couple of major value propositions in doing LTFU studies that deploy DCT components.
Firstly, this is an opportunity to develop a customized, longitudinal data set — or longitudinal patient population — beyond the confines of the study protocol. Let’s take an example where the LTFU is focused on primary endpoints for safety, and there are also some PROs. Given that we are prospectively engaging with patients, we have a robust level of personally identifiable information, or P.I.I., and one major thing we can do is tokenize that information and bring in additional data outside of the protocol.
So, now we can start to look at longitudinal data, whether it’s medical claims, pharmacy, additional labs, genomics, or other data types that would provide value for different audiences within the sponsor organization — but outside of the study team. Functions can include RWE, HR, market access, payer reimbursement, and any number of these types of stakeholders. Again, this is an opportunity for the sponsor to create customized longitudinal datasets, which represent true value for the RWE team.
If the LTFU is mandated, we can point to the information we have collected for purposes of regulatory requirements. And then later, in the protocol, we can always outline the additional endpoints that we will be exploring, such as healthcare utilization and costs.
As we look at how to execute LTFU studies for maximum success, what we find is one of the best practices is to execute a “warm handoff” or “warm transition,” from the main study to the LTFU. For every patient, the final visit of the main study becomes the first visit of the LTFU, or the first transition into the decentralized model, in other words.
What happens is the patient and the main study investigator have a conversation on that last main study visit, and there’s a smoother transition as a result. Because, with a decentralized approach, if you’re going from the main study into the LTFU, there has to be some proactive contact to engage the patient. Without it, there would be a much lower uptake.
Patient engagement is key
If there is any pushback of taking a decentralized approach to LTFUs, it’s that it might be perceived to lack in-person contact. If that were true, patients would be far less likely to be retained in LTFUs. But it’s important that LTFU designs take into account the crucial need to engage with the patient for the duration of the study. This may include notifications, one-on-one outreaches, and other appropriate tactics, as well as more of a human engagement approach, where clinical research coordinators (CRCs) call patients to, effectively, see how they are doing. Because, again, from the perspective of the patient in a 5 to 15-year study, there needs to be some loopback to the patient in terms of the value they’re providing to the study.
From the sponsor perspective, one perceived challenge is that sites or investigators from the main study might not be interested in, or open to, adopting DCT approaches for the LTFU — having just treated a patient for the duration of main study, they may question having to hand them over to another organization for a virtual LTFU. But the benefits of deploying decentralized models, particularly pertaining to the tearing down of geographic barriers, is overriding, and consequently, we’re seeing that a lot more sites and investigators are interested in having DCT components for LTFUs. It’s worth noting that such decentralized study models should also be able to accommodate individual investigators who want to continue working with specific patients in the LTFU.
Efficiency and agility
Also key, of course, is the fact that the sponsor wants to get the LTFU study up and running and completed in the most efficient manner possible. So, from the sponsor standpoint, it might seem daunting to continually engage with, say, 100 sites over the course of 5 to 15 years. Purely from an efficiency standpoint, then, sponsors would prefer to have 100% of sites transition to virtual models. In reality, this would depend upon a number of factors, such as the condition of the patient population and the endpoints of the study. So, an agile clinical trial design—comprising a mix of traditional and decentralized approaches—would appear to be the way forward, in most cases.
If the investigator or the patient wants to continue with their current treating relationship in the LTFU, then the decentralized model needs to be flexible enough to accommodate it. And, assuming you get it right, you’re also taking a giant step toward listening and adhering to the personal preferences of the patients.
As for the ROI of using DCT approaches to drive LTFUs, an analysis that was done for multiple sponsors, based on how much it would cost to keep sites open, showed a cumulative breakeven point early in the study duration (e.g., year 4), with positive savings for the sponsor from the fourth year.
At the end of the day, LTFU studies represent some of the “easiest” areas for the adoption of decentralization approaches, because all stakeholders—the sponsor, the sites, the patients, the providers—benefit greatly. It is the ultimate win-win. And so, as a result, I see a huge increase in demand ahead.
|Gordon Cummins is vice president, global real world evidence at Science 37.|