Innovating Clinical Trials | Amgen

Clinical trials are desperate for innovation. Speed and efficiency need to improve as many patients cannot wait over a decade for new, potentially lifesaving medicines, and trial participants often do not reflect the patient population. Because clinical trials are complex and multidisciplinary, there is not a single, simple solution for accelerating progress.  

In this four-part series, host Rob Lenz, senior vice president of Global Development at Amgen, explores the latest approaches in clinical trial design and execution and highlights real-world examples of how scientists can run trials better and faster to develop potential new medicines that benefit patients. 

Released Episodes

Operational Innovation with Dr. Cynthia Verst, President, Design and Delivery Innovation, R&D Solutions at IQVIA  


Episode 1: Operational Innovation

The Scientist:  Welcome to Innovating Clinical Trials, a special edition podcast series produced by The Scientist's Creative Services Team.

This series is brought to you by Amgen, a pioneer in the science of using living cells to make biologic medicines. They helped invent the processes and tools that built the global biotech industry and have since reached millions of patients suffering from serious illnesses around the world with their medicines.

Until recently, the model used for traditional randomized clinical trials had not changed since it was first introduced in 1948. Now, transformation is underway. Speed and efficiency need to improve as many patients cannot wait over a decade for new, potentially lifesaving medicines, and trial participants need to better reflect the whole patient population. Because clinical trials are complex and multidisciplinary, there is not a single, simple solution. What does innovation in clinical trials look like? In this series, host Rob Lenz, Amgen's Senior Vice President of Global Development and experts leading next-generation clinical trials explore trends and drivers in design and execution to improve trial quality and safety, decrease costs, and improve predictability, reliability, and speed. 

Rob: Our understanding of human biology and disease is progressing at an unprecedented pace, and clinical trial development and execution needs to evolve just as quickly to deliver potentially lifesaving medicines to patients who can't wait. We also need to recruit underrepresented patients into trials, which requires us to think differently about how we identify and recruit patients. In this episode, I talk to Cynthia Verst, president of Design and Delivery Innovation for Research & Development Solutions at IQVIA, a global provider of advanced analytics, technology solutions, and clinical research services. We discuss the drivers for applying innovation in the trial execution space and how the clinical trials model is evolving across the industry thanks to modern innovations.

Over the last decade, there's been tremendous innovation in the clinical trials area, including the use of things like trial simulations to optimize trial design and improve efficiency, and advances in statistics and computing to support such simulations. But there was relatively little new innovation in executing these trials, including identifying potentially successful trial sites with investigators and patient populations appropriate for a trial's needs.

One part of trial execution is monitoring patient vital signs and responses to treatments. Cyndi, will you compare and contrast the way we traditionally monitor our clinical trials with the advancements that are happening now, especially in the area of AI and machine learning, or AI/ML?

Cyndi: Yesteryear, when we were conducting clinical monitoring, we were sending clinical research associates at the investigator site physically and compare the data that the investigator entered into the electronic data capture system, otherwise known as EDC. The CRA, the clinical research associate, would sit with the hard copies or the electronic files that were blinded and do what's called a source data verification. So, what the investigator entered into the EDC was the authentic data in the patient's hard copy binder. It was then the CRA's individual responsibility to look at some of the data trends and to ensure when they were writing this up in their monitoring visit report, if they were seeing issues. We're leaving the impetus on the human to find these signals and trends.

As we're advancing, we're using more AI/ML and more algorithmic predicting capabilities, where we're running in the background in a blinded manner, where patients are de-identified, within the data of the EDC, we're able to identify duplicate subject registration looking at their clinical data, their lab data, etc. Also being able to look at trends over time and identifying safety concerns, as an example with abnormal lab tests, where we're not relying solely on humans.

I'm not suggesting that the human on-site visits will be obviated altogether. It's the human component being augmented with machine learning capabilities that's helping to detect and respond very quickly to ensure two things. We are in the eternal pursuit of ensuring patient safety, first and foremost, and secondly, integrity of clinical data so that we can draw the appropriate efficacy and safety conclusions.

Rob: This is one of the rare instances where you have all three levers being pulled in the same direction—that's improving quality and patient safety, decreasing cost through the automation, and increasing speed as one can interrogate these data literally real time, without the need to send folks out on a plane to an investigative site. One approach that's been getting a lot of attention is decentralized trials. We know that about 70% of patients live more than two hours from a clinical site, making it highly inconvenient or simply not feasible for them to make that commute routinely. Unlike the traditional site-based trials that require participants to regularly visit brick and mortar sites and have their assessments done there on-site, decentralized trials bring the trial to the subject, say at their home. There's a lot that's required to make such a fundamental shift in how these trials are conducted. Share with us what some of those foundational elements and differences are.

Cyndi: Decentralized clinical trials were being developed and deployed prior to the pandemic. If it was not for these DCT capabilities, we would have never conducted and completed the vaccine trials when we did. I liken this to three major reasons why this will continue to be the case of DCT helping us to accelerate clinical trials today and in the future. The first is the digital technology components. That is the backbone behind DCT, from everything we're doing in terms of accessing and recruiting patients to conducting the trial and ensuring that we're collecting the right data at the right time with the right quality. The digital technology component is incredibly important. Being able to use digital footprints, digital journeys with patients, meeting them where they are in the communities is of top importance. Part two, to bring trial components to the patient's home. So, those are services like phlebotomy—going and collecting blood samples—or delivering investigational product to the patient's home or delivering follow up and observational data collection via nurses or other staff at the patient's home. The third pillar is strategic partners like pharmacy networks. Each American is, on average, three to five miles away from a pharmacy. We see this as another huge opportunity in offering decentralized trials and affording much greater participation for diverse populations and underserved patient populations.

Rob: It's incredible just how rapidly the adoption of these decentralized approaches have come into the clinical trial space. I imagine it's going to continue to play an incredibly important role with the end result of dramatically increasing the proportion of patients who can participate in a trial. As we think about moving forward, one of the big questions that the FDA has had is what do we think the durability of this will be?

Cyndi: The sustainability of those pandemic novel trial executional capabilities is really on everyone's mind. I suspect the jury is still out relative to data quality. It is going to be up to us, clinical research organizations, industry associations demonstrating that, there is no negative trend in terms of data quality. When we talk about DCT, we also have to include connected devices, where there is that continuous stream of data, in a way that's reducing burden on patients and sites, ensuring patient safety along the way. Regulators are going to be a very important body that will enable the sustainability of DCT.

Rob: One other barrier that could preclude the use of decentralized approaches, or virtualized clinical trials, is it's probably not equally applicable to all diseases. When there's very little known about the safety of a medicine or disease areas where the patients are particularly sick, or those drugs may have particularly high risks in our patient population, those may be ultimately less amenable to decentralized approaches. But when we think about the vast majority of later phase large trials, it certainly seems like those would be amenable to these types of approaches.

One of the other advantages is to increase the participation in underrepresented populations Are we seeing an actual increase in participation from these underrepresented populations through these approaches, or is it just simply too early to know?

Cyndi: We're heading in the right direction, but I think that we've got more work to do. In certain therapeutic areas, like hepatitis, like HIV, we're seeing some movement in terms of beginning with what I call diversity by design, and that that's something that we adopt here at IQVIA. It's going to take the whole healthcare ecosystem leaning in here to help with the building of trust and in the building of transparency. Designing protocols as well as executional operational strategies, that's going to reduce the burden, enhance, augment, and encourage participation from under-representative, diverse, and inclusive populations. And then it's also how we operationalize. We're going to need an investigator network that is comprised of diverse investigators. Diverse operational clinical research coordinators are more apt to recruit diverse patient populations. So, how do we educate our investigators and ensure that we've got the ability to lean into community grassroots efforts, that we are partnering with our patient advocacy associations, that we are leaning into the industry associations? That's where we'll begin to see the needle moving in terms of ensuring clinical have diverse populations, epidemiologically speaking.

Rob: One potential enabler of decentralized or virtualized trials is the use of wearables or other passive data collection mechanisms. There's definitely been some interesting advances in devices that have received FDA approval for monitoring things like vital signs or heart rhythm movement, or even glucose monitoring. Have these really transformed how we're doing clinical trials? What do you think some of the barriers have been?

Cyndi: We're already making dramatic improvement in terms of reducing the patient-derived and investigator-derived burden of data collection. Just five, seven years ago, we had investigator-derived data being inputted to the tune of about 70% of data coming from clinical trials. And we're now seeing that 70% being reduced down to 35% as a direct consequence of connected devices, wearables, patches, etc. Passive data collection comes with some terrific upsides, not only like DCT and risk-based monitoring, where we're going to be able to increase the data quality, we're going to be able to do this in more real time, affording the ability to track the data and sense and respond if there is patient safety concerns. We've got some time before we see a complete revolution because we'll always need the doctor in the equation.

Rob: We have access to unprecedented amounts of data, with clinical trial data through things like the TransCelerate Placebo Standard of Care Sharing Initiative, which allows companies to share subject level data from study participants who received placebo or standard of care, and real-world data such as claims data or electronic health records data. Where do you see some of the greatest potential and greatest applications of such datasets to clinical trials? How are we using these large data sources to optimize the clinical trial executional parts, particularly feasibility studies, where we need to project if a clinical trial will be successful if it is conducted in a certain geographical region, site, or with certain investigators?

Cyndi: We're shifting the paradigm that we don't conduct feasibility by firstly going to sites and saying, how many patients do you have Dr. Smith that might be eligible for this protocol? We're moving toward obviating that step to improve and accelerate timelines and accelerate productivity.

Now, another related area that we're seeing in terms of innovative trial designs, because of the complexity of cell and gene therapy trials, we have some post-marketing commitments that range from 10 to 15 years of long-term follow up. Conducting those in a traditional clinical trial setting is simply cost prohibitive and unduly burdensome for patients being able to follow patients in a more simplified approach, whereby we're able to use the real-world claims data, pharmacy data, hospital data sets, we're able to sense and respond and follow those patients in a way that's meeting regulatory requirements as well as satisfying long-term safety data collection.

Rob: You mentioned enrollment is one of the one of the areas that we're seeing the utilization of machine learning. Industry metrics tell us that anywhere between 11% up to a third of sites never enroll a single patient. The traditional approach, sending surveys out to potential site participants, is simply very inefficient and terribly poor at predicting who will be a good versus not a good enroller. One of the particular areas of focus for us is using this data along with advanced predictive analytics, including machine learning, to help us identify who will be those performing sites as well as the nonperforming sites. The machine learning algorithms we build allow us to look across thousands of attributes at a site simultaneously. So, things like what their specific patient population is, what the trial complexity is, requirement for genomic screening capabilities, etc. Do you see this movement towards utilizing machine learning specifically in identifying high performing sites as a broader trend or movement in industry?

Cyndi: Absolutely, Rob. Over the last five and a half years, we've been able to prove that through machine learning we're able to connect that individual investigator with attributes such as their current real-time eligible patient population, linking that to past and contemporaneous clinical trial experience, quality of that experience, were they associated with any critical or major audit findings, as well as looking at their clinical trial landscape in their particular setting. We can take all those attributes and running those predictive algorithms, we are able to sort investigators by tier. We have statistically significant data suggesting that the engine is properly predicting the top enrollers.

Rob: What do you see is really transformative opportunities that are not quite ready for primetime today, but maybe they will be, say five years from now?

Cyndi: Using drug discovery AI/ML on the clinical development side is going to move needles for us, being able to be more predictive. For instance, biomarker identification, being able to precisely identify and target patient phenotypes and genotype profiles, patient segmentation and being able to target those populations more expeditiously.

Rob: It's really impressive to see how much innovation is being applied specifically in clinical trial executional aspects. Cindy, it's been great and fun and educational for me. I've really enjoyed the conversation.

Cyndi: Indeed, Rob, it's been a pleasure joining you on this exciting topic.

The Scientist: Thank you for listening to Innovating Clinical Trials, and thanks again to Cynthia Verst, president of Design and Delivery Innovation for Research & Development Solutions at IQVIA. To dive further into this topic, please join Amgen scientists at the Innovating Clinical Trials Q&A webinar discussion on September 28, 2022. Register for the event at the link provided in the episode notes.

A lot of useful information concerning patient health is collected outside of clinical trials. Through advancements in data analysis, these data can be used in different ways. In the next episode of Innovating Clinical Trials, we'll talk to Brian Bradbury, vice president of the Center for Observational Research at Amgen about using real world data to revolutionize clinical trials.

To keep up to date with this podcast, follow The Scientist on Facebook and Twitter, and subscribe to The Scientist's LabTalk wherever you get your podcasts. 


Real World Data in Drug Development  with Brian D. Bradbury, D.Sc., Vice President, Center for Observational Research at Amgen


Episode 2: Real-world Data in Drug Development

The Scientist:  Welcome to Innovating Clinical Trials, a special edition podcast series produced by The Scientist's Creative Services Team.

This series is brought to you by Amgen, a pioneer in the science of using living cells to make biologic medicines. They helped invent the processes and tools that built the global biotech industry and have since reached millions of patients suffering from serious illnesses around the world with their medicines.

Until recently the model used for traditional randomized clinical trials had not changed since it was first introduced in 1948.  Now, transformation is well underway.  Speed and efficiency need to improve as many patients cannot wait over a decade for new, potentially lifesaving medicines, and trial participants often do not reflect the patient population. Because clinical trials are complex and multidisciplinary, there is not a single, simple solution. What does innovation in clinical trials look like?    In this series, host Rob Lenz, Amgen's Senior Vice President, Global Development and experts leading next-generation clinical trials explore trends and drivers in design and execution to improve trial quality and safety, decrease costs and improve predictability, reliability and speed. 

Rob: Randomized clinical trials are the gold standard for evaluating the efficacy and safety of medicines, but they come with many drawbacks including high monetary and time costs, a lack of representation compared to the general public, and ethical limitations. Historically, these trials were the main, and often, only mechanism to understand the effects of a medicine. But more recently real-world data, from sources such as electronic health records, or EHRs, insurance claims and billing activities, disease registries, and wearable devices, is having a greater impact on understanding  a medicines usage and effects. Although this information is collected outside of clinical trials, clinical researchers incorporate it during the earliest phases of clinical development  to gain additional information and speed up the drug development process.  In this episode, I talk to Brian Bradbury, vice president of the Center for Observational Research at Amgen, about the increased utilization of real-world data and its potential to revolutionize every stage of clinical research, from trial design to regulatory requirements to outcomes measurement. 

Hey, Brian, thanks for joining me for this discussion on real-world data. I believe we're in the midst of a transformation. One really exciting approach which is underlying this transformation is real-world data. Brian, highlight the various uses that you see for real data in drug development.

Brian: Real-world data is information on patients' health status and/or healthcare delivery, as is routinely captured in a variety of different sources. And real-world evidence is the evidence on the usage of medicines, or their benefits and risks as derived from analysis of real-world data. That spans a wide expanse in the clinical development and commercialization arena of medicines. And there's a number of different dimensions by which real-world data and real-world evidence are being used today. Certainly, in product development, there's a large focus on understanding target diseases, natural history, understanding where patients may exist. In the regulatory space, using these data to inform on potential clinical programs, on getting medicines approved through an accelerated approval pathway or label expansions. And in the post-marketing safety arena, real-world evidence are also used to support medical affairs research, where we're looking at the effectiveness of medicines when they're administered in clinical practice, the evaluation of things such as policy decisions, and the application of pharmacologic interventions in clinical practice. And then finally, in the arena of value-based payment decisions and payer discussions, we're really interested in understanding the comparative effectiveness and cost effectiveness of medicines to inform on payer needs.

Rob: The focus on real-world data has been compelled by the 21st Century Cures Act, which is designed to accelerate medical product development. This requires the regulatory agencies to create a program evaluating the use of real-world evidence, predominantly for two purposes:  One is to support a new application for an approved drug, then second is to satisfy post-approval study requirements. Can you share where you see the value of real-world evidence specifically from a regulatory enablement perspective?

Brian: 21st Century Cures was a huge step forward for everyone. The idea being that we can use these data to inform on regulatory decisions if those data are fit for purpose, the study design is well thought through, and the analytics makes sense. Some of the places where we see this really taking shape—in rare diseases and in the cancer space—we've seen the use of external control arms to support accelerated approval of medicines, and a number of medicines have gained accelerated approval through that mechanism. Here are situations where you have a very high unmet need, potentially you have a new medicine that is showing very significant clinical effectiveness, and you're able to use an external comparator arm of patients with that target indication to understand what the natural history or the true control experience for patients with that disease looks like and compare that against the new medicine that you're trying to bring forward.

Rob: In clinical trials, participants are divided up into different groups or arms. For example, a study may have an arm where people receive the drug being tested, while others are randomly placed in a placebo or comparator arm within that trial. There is now consideration—particularly in rare diseases and late-line oncology trials—to use external controls rather than within a trial comparator. What are some of the analytic considerations in taking this step -- when you take a placebo-controlled arm within a trial and replace it with an external or synthetic comparator one composed of real-world evidence?

Brian: When we think about a randomized control trial, we think about bringing patients in all with a target indication and we randomly allocate them to either get the new medicine that we're looking to study, or we randomized them to some standard of care with placebo. When we do a single arm study and we want to then compare it to an external population or synthetic control, there's many, many considerations that we have to go through. The first being, are we able to find patients who represent that placebo-like experience in the real-world that we're going to feel comfortable basing a comparison on? And that means, do we have the information on those factors that are going to drive patient prognosis captured in a real-world control arm or external comparator arm? We're always going to be concerned about confounding are the patients materially different between those getting the new medicine versus those in the experimental control population. Having those important confounders, those factors that predict the prognosis of the patient, that's critical whenever we consider these studies. The second piece is if we have that information and we're able to then use statistical adjustment to control for those differences between those patients, are we able to believe that we don't really have residual uncertainty? Do we have the appropriate endpoints in the real-world data that are comparable to the endpoints that are measured in a clinical trial where you have very structured data collection? The types of populations, are they comparable, such that if we then apply principal analytics, we can get an answer that we believe is robust and is going to enable regulators to make a decision about the effectiveness of this medicine?

Rob: Do you think we'll be able to migrate the value of real-world data beyond that rare disease space or in the late-lines of oncology, to disease areas where there are either no available therapies or there are some ethical considerations and concerns around running placebo-controlled trials? Given the progress that's been made, do you think using real-world data is ready for primetime for more common diseases where there isn't an ethical limitation to doing a placebo-controlled trial?

Brian: There's a lot of work underway to test whether in disease areas that are not significant unmet need, a sizable benefit can be evaluated. If the magnitude of the benefit is not so profound, there still will be some residual uncertainty. Under 21st Century Cures, the focus is on label expansions, where you have evidence around a medicine, established clinical trials, and now you may look to see whether you can use real-world evidence to study the use of the medicine and an off-label indication in the real-world and determine whether there's, in fact, benefit consistent with that which was demonstrated in the original studies. There have been some recent examples whereby folks have used real-world data and shown that the medicine's benefit in new populations is consistent with what was seen in the original trials. But there's a belief that, save a few specific kinds of examples, using a randomized experiment still remains the gold standard for bringing this sort of evidence to patients.

Rob: You're referring to the randomized clinical trial duplicate project. This is looking to if principle, non-randomized study approaches using real data can consistently match the results of completed trials. If they can replicate that, then that would give us confidence in using real data approaches in lieu of randomized trials moving forward. There was a publication last year in the journal Circulation and they presented the results from the first 10 studies and show that eight out of the 10, the estimate of the treatment effect was very similar, and six out of the 10 resulted in the same regulatory conclusion. What do you think the biggest insights and takeaways are from that study?

Brian: That team went through a process to identify a number of clinical trials that potentially could be replicated using available real-world evidence or real-world data. They did a fit for purpose evaluation. Do the data coupled with the appropriate analytic methods and analyses inform on a regulatory decision? Six out of the 10, there was regulatory agreement between the clinical trials and the real-world evidence studies that were emulations of the clinical trials. This study highlighted that when we do that fit for purpose evaluation, we have clear understanding of what we're trying to emulate in a real-world evidence study. That should give us confidence as we move forward.

Rob: It seems like certain disease indications would be more amenable to this approach. Diseases where key outcome measures are routinely captured in the claims databases, such as cardiovascular outcome, stroke or myocardial infarction, revascularization procedures or hospitalizations, or achieving a particular lab level that's captured in a structured dataset. It may be more difficult with diseases where those outcomes aren't routinely captured. For a lot of diseases we measure in clinical trials, it would be a qualitative improvement or quantitative improvement or worsening in the symptoms of the disease, something that's captured through a patient reported outcome measure. There's been a fair amount of work trying to extract data from the clinician notes, etc. Share a little bit on what progress is being made there.

Brian: There's many other kinds of tests or assessments of patients that are not either routinely done or the results of those tests are not routinely captured. Think about measures of bone mineral disk density, low density lipoproteins levels and labs are not routinely captured in all the systems that we operate. When we talk about the work ahead of us, can we use the extensive information captured in physician notes and electronic health record systems across this country and beyond and make that information into much more credible structured data. Here's where the use of AI and machine learning and natural language processing, I think, is going to be very promising. In the real-world data space, we see AI-assisted technologies whereby we can use an AI platform that reads physician notes in a HIPAA compliant way and is able to extract and validate markers of disease, tumor staging information on tumor size, information on other markers of disease, validating that against a traditional chart. Once we've done that, we build a pathway to take that unstructured information, make it structured and combine it with all of the other rich data that's captured in EHRs and claim systems to make a more complete picture of the patient. The ability to bring that information forward, make sure that it's trustworthy, and then integrate it into these larger systems is going to really enable us to answer many more questions using real-world evidence.

Rob: There's no doubt that the COVID pandemic catalyzed some pretty significant changes in how we conduct our clinical trials. Things like decentralization of certain activities that instead of being conducted at the study site could happen in the patient's home or in a place that's just more convenient for the patient, direct to patient shipment of study drug, just to name a few. And there's certainly been a lot of attention paid to that in hopes of ensuring the durability of those approaches in a post-COVID era. As you contemplate the COVID pandemic, where do you think real-world data had the biggest impact?

Brian: Two years ago, from a basic epidemiology perspective, we needed to use a lot of the available information from real-world data sources to understand who were the patients that were getting COVID. What was the natural history of this disease? What were the medicines that were being used by physicians across the United States and more broadly around the world? We were able to use the rapidly accumulating data to inform on a lot of questions. This pandemic was moving in an asynchronous way across the United States and across the globe. There was a peak in one part of the world and then it was flattened in another, and then a month or two later the peak was then appearing in different parts of the world. This was very challenging to think about how you could design studies. Where would the patients be that would be enrollable into studies with the changing landscape of the standard of care? When we think about designing clinical studies, we think about a standard of care that we might use as a background, that we would then randomize to therapies, but that standard of care was changing very rapidly, particularly as there were ideas about repurposing of medicines that might prove beneficial. And the healthcare community trying to figure out how to treat these patients and their prognosis. Prognosis can be quite poor in many, many instances. So, there was this huge dynamic piece of trying to design studies in the heart of this, and where I think real-world data ultimately proved quite beneficial is that we can use these large healthcare systems to begin to understand who the patients were that were getting new therapies such as vaccines and understanding in large populations who would have gotten the vaccine or didn't, their relative benefit and the relative safety.

Rob: One of the major limitations of these approaches that I often hear raised, especially as compared to a randomized clinical trial, is that you can only establish an association. There are just so many potential biases that limit the use to establish causation. These biases are exactly what randomization in theory addresses in a randomized trial. Could you share your perspective on that and speak to some of the analytic approaches that are addressing the issues of bias.

Brian: We've done a really good job over the last 30 years of uncovering many of the sources of bias when we use real-world evidence approaches. Early on, real-world evidence has been thought of as a data mining exercise. You have a large data set, run some analytics, you get some answers. There's been a growing recognition of the sources of bias, how they can arise, and what are the things that we can do with appropriate research, design, analytics, and analysis frameworks. When we randomize patients, we believe that we balanced their clinical characteristics. All of the things that may be prognostic in real-world evidence, we don't have the benefit of that random allocation. We have the decision that the physician made about what medicine a patient should get. And so, we ultimately have to figure out how to balance patients in one treatment arm versus another. And that's largely done by things such as propensity scores where we understand the propensity of a patient to get one treatment versus another. But it also requires that we have the data on cardiovascular disease history, blood pressure measurements, or the stage of their disease. If we have that information, we're able to apply and deliver evidence that is going to be much more credible. The other thing that is more common today is quantitative bias analysis, where we are empirically testing the biases that we think exist in our data. By doing that, we get a better understanding of how strongly we can stand behind our estimates.

Rob: One of the other areas that I hear discussed in terms of opportunities for improvement is data standardization. In the clinical trial space, we have data standards, which most clinical researchers adhere to, and this has really been critical for regulatory authorities around the world. What's going on in the real-world evidence space? What will be needed to get us to a point where we'll have regulatory-grade standardized data?

Brian: The success of real-world evidence is going to be based in large part on the quality of the data. Recently, the regulators released a number of guidance documents; they're really focused on data quality. We know that there's lots of different sources of data, and having a better understanding of the provenance of the data, the reliability of the data, the transformations that are made to the data before they come into an analytic dataset that we might use to answer a specific question, the whole effort now on the regulatory side, the process is being much more standardized, there's more transparency about how that's occurring, and the quality standards are being elevated, such that the end game is datasets that are much richer. They're much more trustworthy and we're more confident that we don't have this big missingness in the data that would undermine our ability to make a decision. The last piece is, think about the work that's going to be done around information such as omics and biomarkers. A whole new era is going to be coming forward with precision medicine-like information being integrated into commonly available data systems across the globe. So, I see that there's going to be a real rich fabric of data that's going to provide a longitudinal record, and that's going to require a great deal more standardization.

Rob: Brian, look into your crystal ball. Let's say you and I connect again about five years from now. What are the things that we're going to discuss then that we didn't cover today?

Brian: We're moving into an era where we're going to be able to deliver evidence across multiple jurisdictions, hopefully simultaneously. Can we study the natural history of disease in 10 countries simultaneously? Can we understand the effectiveness of a new medicine in five countries at the same time using principle design methods because we have interoperability of the data systems, we have interoperability of the analytic approaches, we have a common understanding of the strengths of those data, and we have the ability to prosecute analyses in a near real time to deliver evidence to assist decision-making by folks across the healthcare ecosystem? Understanding the safety of a medicine when it's administered to hundreds of thousands of citizens in multiple jurisdictions, that that will be an enormous step forward for us as a global society.

Rob: Brian, I certainly share your enthusiasm for what the future holds with increased utilization of real-world data. There are certainly some areas where some work still remains, but you've highlighted just how transformative real-world data is already today and will continue to be in drug development. Thanks for sharing all your great insights in this exciting area.

Brian: Well, thank you, Rob. I appreciate it. And I share your enthusiasm and I look forward to a return engagement.

Rob: Five years.

The Scientist: Thank you for listening to Innovating Clinical Trials, and thanks again to Brian Bradbury, vice president of the Center for Observational Research at Amgen. To dive further into this topic, please join Amgen scientists at the Innovating Clinical Trials Q&A webinar discussion on September 28, 2022. Register for the event at the link provided in the episode notes.
To provide the best care possible, doctors must prescribe medicines that they know will work for their patients. However, most treatments are tested in clinical trials composed of participants not representative of the general population. In the next episode of Innovating Clinical Trials, we'll talk with Ponda Motsepe-Ditshego, vice president and Global Medical Therapeutic Area head in General Medicine at Amgen about engaging underrepresented populations in clinical trials. To keep up to date with this podcast, follow The Scientist on Facebook and Twitter, and subscribe to The Scientist's LabTalk wherever you get your podcasts. 


The Right Patients  with Dr. Ponda Motsepe-Ditshego, Vice President and Therapeutic Area Head of Global Medical, General Medicine at Amgen 


Episode 3: The Right Patients

The Scientist:  Welcome to Innovating Clinical Trials, a special edition podcast series produced by The Scientist's Creative Services Team.

This series is brought to you by Amgen, a pioneer in the science of using living cells to make biologic medicines. They helped invent the processes and tools that built the global biotech industry and have since reached millions of patients suffering from serious illnesses around the world with their medicines.

Until recently the model used for traditional randomized clinical trials had not changed since it was first introduced in 1948.  Now, transformation is well underway.  Speed and efficiency need to improve as many patients cannot wait over a decade for new, potentially lifesaving medicines, and trial participants often do not reflect the patient population. Because clinical trials are complex and multidisciplinary, there is not a single, simple solution. What does innovation in clinical trials look like? In this series, host Rob Lenz, Amgen's Senior Vice President, Global Development and experts new explore trends and drivers in design and execution to improve trial quality and safety, decrease costs and improve predictability, reliability and speed. 

Rob: The questions scientists face in clinical development are very similar to the questions that doctors face when treating patients. Patients want to know whether their disease is going to go away or worsen and whether they will benefit from a therapy. Doctors generally can only share with the patient what the typical or average effect of a medicine is, but they cannot specifically address the potential benefit or harm in that specific patient. With advances in genetics and other human data, researchers and doctors will one day be able to practice truly precision medicine. This inability to predict how a patient will respond to a medicine is even more challenging in under-represented patients because we often don't have the information because they were not included in the clinical trials. This is due in part to systemic issues that deter people from participating in research, especially those who have been historically excluded due to factors such as race, ethnicity, sex, and age. In this episode, I talk to Dr. Ponda Motsepe-Ditshego, vice president and Global Medical Therapeutic Area head in General Medicine at Amgen, about the recognized differences in disease incidence among racial and ethnic groups and new approaches to increase representation in clinical trials.  The benefits are myriad, for patients and for drug developers. Many of us view the inclusion of under-represented patients in our trials as not only a scientific necessity but an ethical imperative.  
Can you share a little bit of your background as a medical doctor, and how that shaped your work and commitment to improving diversity and representation in clinical trials?

Ponda: Sure, thanks, Rob. I am an MD qualified in South Africa, and I've lived and worked on three continents and in four different countries throughout my time in industry as well as a clinical practitioner. As a native of South Africa and a black woman who's now living in the US, I've really witnessed firsthand how equality and equity hasn't always been present for a lot of people. And disparities in health are a major example of this, especially for chronic diseases. This persists among different communities, which really highlights that health equity is a global issue. At Amgen, I love working as the global vice president of the general medicine therapeutic area where I oversee work done in cardiovascular disease, lupus, obesity, osteoporosis, among others, where we often see racial and ethnic disparities in incidence and prevalence of diseases. But most importantly, I'm also really excited about the work that I'm leading with our team called RISE, which stands for representation in clinical research. This team is dedicated to improving diversity and representation of participants in our clinical trials, especially for those who've been historically excluded from research.

Rob: Since the inception of the modern clinical trial around 70 years ago, the general approach has been to study a drug in a relatively small but very homogeneous population of patients with a particular disease to get a very precise estimate of the treatment effect, the efficacy, as well as the safety in that discrete population, and then make basically a giant leap of faith assumption, which is that the medicine will behave the same in other populations that weren't included in that trial. So, generalizing the results to patients of different age, or genders, or ethnic groups, but we now know that's not always true. I was fortunate to have been trained by Elijah Saunders, who, in addition to being the first black cardiologist to practice in the state of Maryland, was a visionary in identifying differences in treatment responses in black versus non-black individuals. He taught me very early on in medical school that black patients respond differently to the various antihypertensive medicines that were available. Fast forward 30 years to today and we still find ourselves in this reality, where more times than not we're missing critical information on the safety and the efficacy of medicines in these underrepresented populations. In 2018, the Census Bureau populations estimates reported that non-Hispanic white Americans represent about 61% of the US population, but they comprise more than 90% of the population in clinical trials. Ponda, can you give us a bit of an overview of how did we get here? Why do clinical trials continue to have a diversity and representation problem?

Ponda: The first one is that there have been several historic events in many countries, including the US, that have really shaped the relationship communities of color have with medical research, as well as the healthcare system. That's led to, unfortunately, quite a lasting mistrust of both systems and has impacted how they approach the healthcare system. I think institutional and structural racism, both explicit and implicit, is a huge factor. We know that biases among healthcare providers lead them to overestimate sometimes level of distrust of the medical system by people of color, and that leads them to communicate less with their patients from those communities about participating in clinical trials. There are institutional biases that can also lead to study designs and eligibility criteria that disproportionately exclude people of color. So, there's multiple things and as well as lack of trial availability at hospital systems where people of color get their care, there's often lack of community engagement. All of these factors, in my opinion, have led to less participation in clinical trials. This is why it's so critical to address social determinants of health, which includes factors like racism, socioeconomic status, education, and the environment when we address health disparities. We know that social and economic factors drive an estimate of 40% of health outcomes; this is twice as influential as factors related to clinical care.

Rob: I might just add one additional to the list that you highlighted. Oftentimes, we lack the epidemiologic data of diseases to understand if there's differences in those diseases in certain populations. And we certainly very often lack the biological understanding as to whether there are differences in, say, how a black patient versus a Caucasian patient might be expected to respond to therapy. For instance, are there different biological or genetic underpinnings of the disease in different populations? What are some of the things that are happening across the clinical trial ecosystem to address some of these foundational problems?

Ponda: There's a strong acknowledgement that this is a topic that must be addressed with a sense of urgency, and I believe there's a strong desire to work together as industry or the entire healthcare ecosystem to address the health inequity that we're seeing. We're starting to make really great strides. What I'm seeing is industry striving to partner with more clinical trial investigators. I see some initiatives to bring awareness to disparities in health and what we can do about them. I see many peer companies serving on multiple cross-industry collaborations with other sponsor companies that also include patients, providers, payers, legislators, manufacturers, academia, and others, to really find impactful sustainable solutions to address these healthcare disparities. The other thing that we're doing that's really important is supporting STEM education for underserved communities. Championing the work of organizations who are already invested in addressing social determinants of health, including community organizations, minority-serving institutions like historically black colleges and universities. And maybe one would argue that what happened with COVID-19, particularly to communities of color, underrepresented communities, has really sparked the need to, first of all, have these conversations cross-industry and be really clear that this work is mission critical and it's time to move the needle now. We now created a team called RISE, which grew out of grassroots efforts of like-minded Amgen colleagues under the umbrella of an employee resource group called ABEN, which stands for Amgen's Black Employee Network. We took a group of cross-functional experts who had a passion for change and an employee resource group that incubated an idea to become a reality.

Rob: You made it clear that this is going to require a comprehensive approach and will certainly require tremendous effort. With the risk of sounding overly daunting, are there particular diseases one could focus on that are disproportionately impacting minority communities? Where do you think that a particular focus could have a positive impact?

Ponda: Cardiovascular disease is one area that we should focus on. We know that nearly half of all American men and women have some form of heart disease, but black Americans are 32% more likely to die from cardiovascular disease. There's clearly a need to make a difference in this patient population. The other population I would think of is diabetes, which has a very close link to cardiovascular disease. 13% of the general adult population in the US have diabetes. We know that American Indian or Alaska, Native, Asian, Black and Hispanic people are all at much higher risk of developing diabetes than white populations in the United States. Their rate is significantly higher, it's almost 15% And then also autoimmune disease, Rob. More than 7% of Americans suffer from some type of autoimmune disease. If we look at lupus, we know that the prevalence varies from about 20 to 150 cases per 100,000, and there's higher rates in minority populations such as black, Hispanic and Asians. We also know that systemic lupus erythematous disproportionately affects women of childbearing age, with a higher incidence in black versus white females. So, again, racial and ethnic minorities are at higher risk for developing this disease. Asthma is another condition. 8.3% of Americans have asthma, but we know that black and African Americans as well as Hispanics have two times more of a likelihood to suffer from severe asthma, and black Americans are three times more likely to die from asthma. And then if you factor in sex, black women have an even higher rate. When it comes to cancer, among males, incidence and death rates are higher among non-Hispanic blacks than non-Hispanic whites for all cancers combined. Black or African Americans are twice as likely as white Americans to develop multiple myeloma. These conditions that I've just is where we should be focusing on because we need to ensure that the medicines we develop are based on clinical research populations that are proportional to the real world setting, and that we're creating solutions to health inequities with the needs of patients and their communities at the very center of everything that we do.

Rob: We're in an era of unprecedented access to data and advanced analytics. How can we use more comprehensively various human data to inform us in how we develop our therapy, due to things like identifying patients to potentially participate in clinical trials, to those that are most likely to respond to our therapies and derive benefit?

Ponda: We can ensure that we have access to consistent, current patient demographic data via collaborations and partnerships with key organizations working to improve study feasibility and landscaping processes, so that the data is collected and analyzed with racial and ethnic considerations by default. The other thing is, encouraging the diversification and greater inclusiveness of biobanks, which eventually informs a wide range of research with their samples and data. But unfortunately, we know that they have historically been excluded from multiple communities, including racial and ethnic minority groups. So, there's an opportunity for us to continue to increase our diversity in terms of our biobanks and genomic research. The other thing is identifying clinical trials sites that have high potential capacity to enroll patients from our target racial and ethnic minority communities and finding the clinical trial investigators who are like-minded as well. Collecting not just disease descriptors, but genetic data can also help identify new patients. And we're looking at using different types of omics to identify those potentially new targets.

Rob: Yeah, I think there's also a potential opportunity once one collects the phenotypic and the genomic data to potentially identify new targets for disease. By increasing the diversity of the populations that we study, almost certainly there'll be identification of increased genetic diversity underlying the diseases, so another real opportunity for target identification.

Ponda: Absolutely.

Rob: You mentioned the prevalence of cardiovascular disease in African Americans. One of the emerging risk factors in atherosclerotic heart disease is a lipoprotein called Lp(a), which is a relatively well-studied subspecies of LDL cholesterol. And it's an interesting risk factor because one's levels are almost entirely genetically determined. So, your diet, your lifestyle, things like exercise really have little to no impact on levels of Lp(a), and we don't have any therapies today that address this. One thing that's been interesting is that the studies that we do have pretty consistently shown race-based differences. Curious to get your thoughts around what you think the significance of this is as it relates to developing a potential therapeutic.

Ponda: I think it's absolutely significant and absolutely important. The risk factors that contribute to this, from what we're seeing and understanding, disproportionately higher in African Americans and probably even in women. 1.4 billion people globally have elevated Lp(a) levels of more than 50. And we know that elevated Lp(a), as you mentioned, is associated with various disease states, including coronary heart disease, peripheral vascular disease, stroke, and heart failure. There are differences in expression among people of different races. African as well as South Asian individuals generally have a higher level of Lp(a). I think it's really critical that we continue to study Lp(a)-lowering therapies especially in a racially and ethnically representative population of patients to ensure that we're truly collecting a complete profile of safety and efficacy data in this population. It's incumbent on society to understand this predominantly controlled by genetics Lp(a) has a disproportionate higher incidence within African Americans and South Asians.

Rob: It's interesting that African and South Asian individuals have on average higher levels of Lp(a), but we don't understand yet whether or not that confers a greater risk of atherosclerotic cardiovascular disease. It's an important question that can be addressed in part through things like observational studies that can attempt to understand initially the association between Lp(a) and risk in the African populations, and then try to attribute some causal association there as well. These types of dedicated studies within the underrepresented populations in particular are really critical to gaining that foundational understanding of diseases and then linking the underlying biology to diseases in in more diverse population. So, we can be armed with those insights going into our interventional trials, rather than historically what we've done is either going blind or simply don't collect sufficient information within the actual clinical trial to assess this population. Maybe we can get your thoughts on what role the biopharmaceutical industry has here. Do you think that we, as an industry, have an important role?

Ponda: We are part of this ecosystem and we have a very important role to play. There's a couple of things that we are starting to do to address this. Working to identify diverse investigators and collaborating with those who serve diverse patients, that is a big issue. There are opportunities to work across the different stakeholders within the industry. One of the things that I've also seen is a huge opportunity for collaborating with non-profits in communities that can help remove those barriers to participation in clinical trials. There is a long history of working with our communities to address mistrust. At Amgen, we've focused on improving patient enrollment and retention and with minority participation in oncology trials. As an industry, we could look at partnering with influential community leaders to address the issue of regaining trust. Communication and information sharing are really fundamental to driving change. We have to do this with humility and community collaboration. I really believe that it's the key to genuine impact and sustainability.

Rob: The notion of lack of trust seems very foundational. What are some of those intermediate signals that we're making progress there?

Ponda: We have community advisory boards providing a platform where you have members of the community—patients, healthcare professionals, different stakeholders—being able to sit and have a conversation where they're being heard. That's been a very big and humbling experience. I don't think we'll be able to overcome the issue of mistrust overnight, but by having these opportunities to engage, providing a platform where communities can engage with us, having a true plan, shared decision-making on improving diversity and clinical trials, having those opportunities to, as sponsors, to come in with, I want to listen, I want to understand, I want to meet you where you're at, and not have the expectation of you meeting me where I am at, those are opportunities that can show some level of success in the future.

Rob: Clinical trials tend to be very burdensome to patients. One could argue that the onus that it puts on study subjects may disproportionately impact lower income individuals that don't have the means to take off days from work or the flexibility. In those community advisory boards, have you heard some practical input and advice that we can be implementing?

Ponda: There are opportunities to optimize decentralizing clinical trials as part of a solution for communities. I think that communities feel that they want to not just be told, here's a protocol, but they want to be part of providing input into both the protocol or even the design of the study. There are also opportunities to provide protocols that are more friendly from a language perspective, and there are different efforts to continue to simplify the language, making it more palatable for patients to be able to understand so that they want to participate in clinical trials.

Rob: We as an industry need to make sure we're listening and actually taking what we're hearing and implement these changes. I think it's absolutely critical to help expand and drive the participation of more diverse and underrepresented populations in our trials. It simply has to just be easier for them to participate.

One area that we haven't touched on yet is the FDA and the role that they're playing. They've become quite vocal around the need to increase diversity in trials. They've created a diversity in clinical trials initiative that's aimed at some of the things you were mentioning about public education to address some of the barriers preventing diverse groups from participating in clinical trials. Recently they released a draft guidance that recommends sponsors submit what's called a race and ethnicity diversity plan to the agency pretty early on in clinical development. And they provided a framework that's outlined within the in the guidance. I think it's fair to say that's also helping to focus sponsors across the industry on ensuring that robust plans are in place. Have those been helpful in in stimulating some of those efforts across the industry?

Ponda: I would say absolutely, Rob. The FDA's most recent updates and efforts is really providing teeth for all of us to move the needle on this topic. Here at Amgen, we've had the opportunity to help our clinical trial program leads and teams with the diversity plans that are being requested. The regulatory piece provides us with more strength to ensure that this time around, we really, really make a change. I welcome all the efforts that we're hearing from other industry bodies who are working to change the landscape and ensure that we have better representation of underrepresented groups.

Rob: Ponda, it's clear we have our work cut out for us. But this is quite simply something that it needs to be improved and having somebody like you who is not only thoughtful about the approach, but really passionate about it certainly gives me tremendous hope. It's been a real pleasure chatting with you today.

Ponda: Thank you, Rob.

The Scientist: Thank you for listening to Innovating Clinical Trials, and thanks again to Ponda Motsepe-Ditshego, vice president and Global Medical Therapeutic Area head at Amgen. To dive further into this topic, please join Amgen scientists at the Innovating Clinical Trials Q&A webinar discussion on September 28, 2022. Register for the event at the link provided in the episode notes.
For cancer patients, time is of the essence. In the next episode of Innovating Clinical Trials, we'll talk with David Raben, Amgen's vice president of Global Development Oncology, about strategies to speed up clinical trials and make new medicines available sooner. To keep up to date with this podcast, follow The Scientist on Facebook and Twitter, and subscribe to The Scientist's LabTalk wherever you get your podcasts. 

Real-World Clinical Trial Design and Execution in Oncology with David Raben, M.D., Vice President Global Development & Product General Manager, Oncology at Amgen 


Episode 4: Real-World Clinical Trial Design and Execution in Oncology

The Scientist:  Welcome to Innovating Clinical Trials Clinical Trials, a special edition podcast series produced by The Scientist's Creative Services Team.

This series is brought to you by Amgen, a pioneer in the science of using living cells to make biologic medicines. They helped invent the processes and tools that built the global biotech industry and have since reached millions of patients suffering from serious illnesses around the world with their medicines.

Clinical trials are desperate for innovation. Speed and efficiency need to improve as many patients cannot wait over a decade for new, potentially lifesaving medicines, and trial participants often do not reflect the patient population. Because clinical trials are complex and multidisciplinary, there is not a single, simple solution for accelerating progress. In this series, Rob Lenz, senior vice president of Global Development at Amgen, explores the latest approaches in clinical trial design and execution and highlights real-world examples of how scientists can run trials better and faster to develop optimal medicines that benefit patients.

Rob: Cancer is one therapeutic area where patients cannot wait the conventional 10 or 12 years for a new therapy. For these patients, time is of the essence and improved access to faster clinical trials can be the difference between having a new potentially life-saving medicine available and it being too late. In this episode, I talk to David Raben, vice president of Global Development Oncology at Amgen, about the next generation of oncology trial design and execution. 

The last decade witnessed tremendous innovations happening across the clinical trial space, and it's often the case that oncology trials are on the leading edge. Given the imminent lethality of a number of cancers, there's a strong driver to innovate in oncology, and not just in the basic biologic mechanisms underlying cancer, but also innovating how we develop medicines. So, it really seems like a great time to be in oncology drug development. Dave, you're right in the thick of that. You've had a rich and varied career. You're a radiation oncologist, you're a translational scientist, and now you're in oncology drug development. How does your collective experiences impact the way that you think about innovating clinical trials for cancer patients?

Dave: Well, Rob, first of all, thank you so much for having me and thank you for asking about my career because it is different. You don't normally see radiation oncologists coming into the pharmaceutical industries. My career and my curiosity were stimulated as a young man watching my father treat cancer patients. He was the chair of Radiation Oncology at Wake Forest. I have an identical twin brother, Adam Raben, and both of us decided to explore the field of radiation oncology because, honestly, we had never met a doctor like my dad who simply couldn't wait to get to work every day. As we got into medical school, we initially were thinking about being surgeons and ended up wanting to follow in his footsteps. I ended up training at Johns Hopkins Hospital, my twin brother trained at Memorial Sloan Kettering Cancer Center. While at Hopkins, my translational curiosity really blossomed when I started to hear about this epidermal growth factor expression signaling on cancer cells, the emerging science around TP53 mutations. It really intrigued me, the possibilities of using biology to enhance radiation in patients with locally advanced cancers like head and neck cancer. When I finished up residency, I joined the faculty at the University of Alabama at Birmingham in part because they were offering me a chance to delve into the EGFR story where we could look at a drug called Cetuximab, which was an anti-EGFR antibody. We conducted the first in human phase one which led directly to a phase three study, improving local control and survival using Cetuximab with radiation, and that was it for me. I knew, Rob, that I wanted to have a career that included translational research. In 1998, I joined the faculty at the University of Colorado working alongside an amazing mentor, Paul Bunn, an expert in lung cancer and incredible translational oncologist, and we together explored that EGFR pathway with small molecules that led from the bench into the clinic and phase one and two trials.

Rob: With you, your brother, and your father all being in radiation oncology, there's clearly a genetic element underlying your interest in cancer. That's a great segue to start contemplating things like the genetics underlying cancer. One of the greatest advances in oncology drug development and clinical care has been the progress in precision approaches enabled by rapid advances in understanding the underlying biology, including cancer genetics. That then allows us to incorporate those basic understandings into the clinical trials. Enabling things like precision medicine, where patients are selected for targeted therapy based on the somatic mutations that they harbor in their tumor. In your perspective, how much progress has been made in that precision medicine approach in oncology in comparison to five to 10 years ago, in terms of clinical trials?

Dave: When I was training at Hopkins, we didn't really think that we could have a patient-centric or a niche cancer approach to things, but when we think about what we have in the arsenal now, it's outstanding to see where we've come over the last five to 10 years. A majority of the medications that have been approved in the last five years have been precision-based drugs. Some of these drugs are not only hitting the primary target, but they're hitting the resistant areas as well, allowing patients to stay on these drugs longer. We know that monotherapy is not the end all be all, and certainly not in advanced disease. Many of these cancers are going to develop resistant mechanisms. It's an exciting time to utilize precision medicine to amplify immune-modifying drugs, as well as to create long-lasting systemic benefit.

Rob: One of the key enablers of precision medicine and oncology has been the ability to biopsy tumors and assess for the somatic mutations that can direct the therapy. Not all tumors are easily biopsied. The molecular profile the tumor can change over time with therapy. New driver mutations can occur over time, often requiring re-biopsy, and that presents practical challenges and potential risks to the patient. One area of promising research involves the ability to detect circulating tumor DNA in blood, also referred to as ctDNA, which removes the need for invasive tissue biopsies. What promise do you see in the use of ctDNA as it relates to drug development, and what are some of the technical limitations?

Dave: There are so many different ways we can approach using ctDNA, or circulating tumor DNA-based assays. They're under considerable scrutiny now by the FDA. And the FDA brought out that first white paper on how we should be thinking about ctDNA and how we're going to need to get comprehensive data in many patients to be able to say with conviction that this is going to be an important part of how we analyze patient response to therapies and perhaps lead to more accelerated approvals. And utilize this not just in advanced disease, but in earlier stage disease. So, I think the assays are going to continue to improve from sensitivity and specificity because you do get a lot of background noise and the issues around false positives, false negatives can really impact how our clinical trial designs may be effective or not. But I think this is going to be an amazing thing to quickly determine response. What an amazing thing to select which patient is going to benefit from which drugs, to get feedback much earlier in patients with earlier stage disease rather than waiting seven to 10 years for outcomes.

Rob: Another area, Dave, where oncology trials are leading the way is with innovative designs. We're seeing significant increases in the use of things like basket trials, umbrella trials. Maybe you can provide an overview of those and what the differences are, and how you've seen them applied best to date.

Dave: Let's clarify basket trials, which refer to trial designs in which a targeted therapy is evaluated across multiple diseases that have common alterations. These trials are typically not randomized, are typically more exploratory, and a median size of maybe around 200 patients. Many in the past five years investigated only a single agent, so maybe not so efficient, and a majority of these trials are typically done in the US. You'll see some in the UK as well, but they're not as broadly used, certainly not in underserved areas. Umbrella trials, on the other hand, are also generally exploratory. They have had randomized arms. They look to evaluate multiple targeted therapies for a single disease that is typically stratified into different subgroups because that single disease has different parts. The median number of interventions is typically around four or five, and the median sample size is typically a little bit larger than the basket trials, around 350 patients. Platform trials is very, very popular now, continues to explode. It's typically randomized phase three studies. So, there's many more patients and the durations can last a lot longer in terms of therapy. You're talking about 700 or 800 patients here. What I like about platform trials is its flexibility. It can add in more arms and its ability to drop in effective arms. We're seeing a massive uptake in these kinds of trials, and I can see this accelerating now to look at combinatorial approaches of cancer drugs.

So, what's the point of these trial designs? I think it's because we're seeing cancer becoming so much more fragmented; we're discovering new mutations, it's unrealistic to use old school conventional trial designs to interrogate genomic-based cancer mutations. There's the Morpheus trial that that looks at very, very small, randomized registrational trials in specific areas, like patients who have failed immunotherapy, to quickly interrogate in an umbrella fashion. As mentioned earlier, the opportunity to quickly look from a biomarker-based perspective whether patients are responding to the standard of care plus drug A, or drug B, or drug C. But even more interesting to me is the ability to potentially do this at an earlier stage of cancer. A person once told me the best type of cancer to cure is the one you can't see rather than the one you can see. And so molecular profiling and these types of platform studies have moved into earlier stage settings where now we're talking about truly changing the course of a patient's life. We have trials looking prospectively at ctDNA to determine the optimal routes of therapy, and this could provide huge advantages for patients to know quickly if a therapy is actually working before we see that gross failure by imaging. This is for many companies, for academicians, for even community-based centers to test out new drugs rapidly here. We as drug developers need to be thinking more broadly about how we drug develop in parallel going after earlier stage cancers, as well as late-stage cancers, because those patients have different setups, different tumor burdens, different immune microenvironments. So, I personally like the thought of using biomarker-driven basket and umbrella trials to test out novel combinations because there's going to be so many patients who develop resistance to mono therapy-based approaches.

Rob: You clearly articulated that one of the major drivers of these various innovative trial designs is speed, efficiency. There's really no other area where speed is more important than in oncology, where the unmet need is so huge, but we also know that only about 5% of patients with cancer actually participate in a clinical trial. There are numerous barriers, like overly complex protocols, lack of having conveniently-located trial sites. What are some of the other areas that are having an impact in accelerating clinical trials, in broadening that aperture so that more patients can participate?

Dave: Decentralized trials is one way to think about bringing the trial to the patient. It's certainly an area of intense interest within the FDA, within patient advocacy groups, major cancer organizations. Correct me if I'm wrong here, just about everyone in this world lives within about 30 minutes of a Starbucks, yet they have to drive hours and hours to get to a major cancer center. Now there's many decentralized trials that are expanding in the US that offer greater inclusivity. They can help identify rare patients by connecting a local oncologist to the study investigator. They can utilize third party vendors that can facilitate these key connections between mobile nursing research coordinators, get them genomic testing, wearable devices for real time patient monitoring coupled with telemedicine, all to really improve the patient experience. The data actually seems to show that patients who go into clinical trials have better outcomes in general because their monitoring follows so carefully.

Another interesting area is this concept around time toxicity. It's the time spent in coordinating care for the patient, the travels to the clinics, undergoing multiple tests, all for a modest gain in survival in many instances. The amount of time spent receiving cancer care can be really substantial. So, can we use decentralized trials as an example to reduce this burden here? And then, that gets back to the complexity of trials. We've got to find ways to reduce clinical trial complexity—the patient advocacy groups are demanding this. When we look at our inclusion and exclusion criteria, some of it is so outdated. How do we reduce requirements? How do we reduce the need for unnecessary tissue biopsies? How do we reduce the need to visit clinics? How do we improve data collection methods and requirements to facilitate a patient actually wanting to go on to a clinical trial, or in the investigators actually wanting to open that trial? As it goes around to the platform-based approaches that you were mentioning, can we again look at this as a way to do tumor agnostic approaches, utilize ctDNA as an example here, and streamline many of our trial designs, making them much leaner? We have a lot less patients going into these trials because they're much more focused and specific. It's really a coordinated approach with intense dialogue with our FDA partners and our patient advocacy groups to get us more clarified and simplified trial designs that are going to be much more cost effective and impactful for the patient.

Rob: So, Dave, this point of reduced complexity is absolutely critical. Unfortunately, I think all industry trends are heading in the opposite direction. There's a fair amount of data showing that the complexity of our trials is not going down, but it's going up. One of the areas that that gives me enthusiasm is how do we more fully utilize the existing data sources that we have. We have unprecedented access now to various real world data sets to do things like help us understand what the implications are of our inclusion/exclusion criteria is on a population. So, appreciating what the implications are in restricting a large portion of that patient population. We can now use that real world data and make those trade-offs and understand what those are in real time as we're designing the study and putting in place an operationally feasible protocol. And the other area is using those real-world data sources to identify where those patients are, because to your point, the majority of them do not live within 30 minutes of investigative site, right? How can we employ the decentralized trial toolkit to reach out to those patients once they've been identified in an anonymized way and relieve them of the burden of having to come all the way in to a brick and mortar clinical trial site? So, Dave, as you look into your crystal ball, what do you see as important future directions and trends and oncology clinical trials?

Dave: I'm really excited about how we're going into earlier stages of cancer in a parallel drug development approach. We want to detect cancers, treat cancers when they're way earlier because we know that's where the real bang for the buck comes. I'm excited about combinatorial approaches because cancers are not going to be solved by typical monotherapy approaches; they're just too clever. They're resilient, they're rebellious, and they can get around roadblocks. I see us continuing to customize our patients here. So, I'm really excited about ctDNA, to understand from a longitudinal perspective how patients might benefit from our drugs and quickly get away from drugs that are not working. That's a game changer for patients. So, I see all these things evolving over the next five years in a very robust fashion.

Rob: Well, there's no doubt that the pace of biologic insights and oncology has accelerated, and I think the key is going to be for those of us on the clinical trial side to ensure that the clinical trial methodology, the advances, the operational considerations to allow interrogation, precision medicine, combinatorial approaches keep pace with those biologic insights.

Dave: Absolutely, and I just can't wait to see how things are going to evolve. Thank you so much.

Rob: Dave, thanks for joining me today. I really enjoyed the conversation. I'll say I feel even more optimistic about oncology clinical development than I did before our discussion today. So, thank you.

The Scientist: Thank you for listening to this final episode of Innovating Clinical Trials, and thanks again to David Raben, vice president of Global Development Oncology at Amgen. To dive further into this topic, please join Amgen scientists at the Innovating Clinical Trials Q&A webinar discussion on September 28, 2022. Register for the event at the link provided in the episode notes.

To learn about our next podcast series, Innovating Clinical Trials, follow The Scientist on Facebook and Twitter, and subscribe to The Scientist's LabTalk wherever you get your podcasts. 

Subscribe Below