Real-world data can help expedite drug approvals for serious diseases with few options

My grandmother was diagnosed with dementia in the early 2000s. We started taking it seriously when she began letting strangers into her Brooklyn brownstone. From there, her behavior quickly deteriorated. She would try to catch her excrement when using the bathroom because she believed it would clog the plumbing. She stashed bottles of whiskey under her bed to cope with insomnia.

The drugs her doctors prescribed were the best medications available at the time. Still, they did not slow my grandmother’s loss of memory or salvage her dignity. They also came with notable side effects. Today, people with dementia still don’t have great treatment options.

I reflect on these painful memories because, as experts consider the evidence required to approve new therapies for serious diseases like Alzheimer’s, it’s important to appreciate the experience of individuals and their families as they deal with diseases for which there is no treatment to slow the condition, let alone cure it. When you look down a path of painful decline to an inevitable end, your calculus becomes “try anything that works, even things that may work” because the alternative is very stark. Some hope is better than none.

This is why the Food and Drug Administration needs to combine the use of accelerated approval pathways with a requirement that drug companies connect real-world data from patients who participated in the registration trials used for FDA review. This approach enables patients to get access to new therapies quickly while offering the fastest path to validating whether they will yield tangible benefits to patients.

Accelerated drug approvals leave open questions about many drugs

The 21st Century Cures Act, signed into law in December 2016, was passed in part to accelerate getting new therapies to patients. It required the FDA to create pathways and tools to help speed drug development.

Under the Drug Development Tool Qualification Programs, companies can leverage clinical outcome assessments, such as a validated patient survey, and qualify biomarkers as surrogate endpoints. Surrogate endpoints are markers of drug effectiveness such as a blood test results in place of clinical outcomes such as death, stroke, or infection. Changes in these markers can be observed faster than waiting to assess traditional clinical outcomes.

FDA has published a table of surrogate endpoints it has accepted in diseases such as leukemia, lymphoma, solid tumors, Fabry disease, Duchenne muscular dystrophy, and others. Alzheimer’s drug trials have used measurement of beta amyloid plaques, the bundles of protein that accumulate between neurons and disrupt brain cell function. Beta amyloid has been measured as a proxy for memory loss in early-stage Alzheimer’s, though whether this relationship has been clearly established is controversial.

The FDA also created several types of accelerated approval pathways to speed new drug approvals in areas like oncology, rare disease, and conditions with dire unmet medical needs. One of these programs, the Accelerated Approval Program, allows for the use of surrogate endpoints to speed drug development and FDA review.

Diseases like my grandmother’s dementia can take years to manifest and progress. Waiting a decade for proof from clinical trials of a definitive impact on memory loss would doom millions of people to suffer with advancing dementia with no hope for treatment. In addition, the cost of requiring decades-long studies before seeking FDA approval would deter drug companies from investing in these kinds of therapies.

Difficult choices in the face of incomplete evidence

Approving drugs based on promising changes to surrogate markers raises two important questions:

• Will these drugs actually deliver clinically meaningful benefits, such as extending patients’ lives or improving their quality of life?
• Are these drugs cost effective?

Some have questioned if it is ethical to approve drugs and devices when there is no clear evidence that the hope they offer translates into tangible improvements. There is also the valid concern that these therapies could cost the health care system billions — even trillions — of dollars without any evidence that their cost has a benefit.

FDA can take several approaches when considering whether to approve a drug under accelerated approval pathways.

One option is to issue a conditional approval based on surrogate endpoints and impose additional study requirements, such as a roll-over study — a continuation of clinic visits and data collection from the patients in the trial. Another is to approve a therapy and require a new Phase 4 study to validate its effectiveness and safety. These studies can take years to plan, execute, and measure, and come with extraordinary expenses as patients must be actively followed for years.

The FDA must ensure that sponsors conduct the required studies and act on the results if they are negative. A recent NPR article looked at 10 years’ worth of drugs approved under the accelerated review pathway and found that 42% of the required follow-up studies had not been started more than one year after the drug’s approval.

Another option is to wait for validation of improvements in clinical outcomes in trials before approving new interventions. This stringent approach would be inconsistent with FDA’s stated intent to quickly make new therapies available to patients with serious diseases and critical unmet needs.

A third option is to approve drugs conditionally and require linkage to longitudinal real-world data on the patients who participated in the registration trial. One example would be to license and link datasets of all deaths in the U.S. Linking mortality data to a cancer trial in which biomarkers suggest that a new drug slowed cancer progression (progression-free survival) would validate whether the therapy actually extended the length of patients’ lives (overall survival). Another example would be linking insurance claims to a Covid-19 vaccine trial. The claims would show whether patients received a later diagnosis of Covid, which enables measurement of how long vaccine protection lasts.

Such linkages can be achieved with technologies that de-identify a patient’s personal information (like name and address) and replace it with an anonymous identifier. These identifiers can be used to match the trial patient to their real-world data while protecting their privacy.

There are many benefits to the linkage of real-world data to registration trials including:

• making a drug available as soon as the registration trial has been reviewed to the FDA’s satisfaction and allowing patients in the trials to continue on the therapy
• saving time and money when compared with designing, recruiting, enrolling, and running a full-scale Phase 4 study or continuing patients in a roll-over study
• providing evidence faster about the intervention’s impact on clinical outcomes
• reducing the burden on the FDA of following up with sponsors on the status of their Phase 4 studies, as this would be part of the original study design
• creating the foundation for cost-effectiveness analysis in tandem with the clinical evidence by linking paid insurance claims to patients’ trial data for the period before, during, and after the trial ends
• collecting supplemental data that might further enable stratification of patients who respond well versus those who don’t and information that may explain adverse events

Augmenting and extending trials with real-world data

Linking real-world data to registration trials as a standard for validating drug effectiveness could be beneficial in many situations. It could help determine if cancer drugs approved using surrogate endpoints like complete pathological response or progression-free survival actually increase overall survival. It could also validate if drugs leveraging imaging and blood biomarkers truly slow cognitive decline in Alzheimer’s disease and other dementias.

How the FDA can help

To accelerate adoption of linking real-world data to registration trials, companies will need help from the FDA in several areas to:

• Support evolved study designs for registration of Phase 2 or Phase 3 trials to accommodate following a cohort of patients using linked real-world data
• Set standards for trusted approaches for de-identifying and matching patients’ trial data to real-world data, both during trials and after they conclude
• Advance the use of different types of real-world data depending on the availability of data for each patient, its quality, and completeness
• Create standards for patient-consented medical record retrieval, a capability that allows patients to have their entire medical record linked to their trial data, which has just a fraction of patients’ health information
• Define “trusted” partners for data via an FDA program for data sources that meet specific data quality standards

Taking the road less traveled

Linking real-world data to trial data won’t be a panacea for answering questions about drug effectiveness and safety across the board. There is likely no single method to guarantee that the benefits of drugs and devices approved today using surrogate endpoints will bear out over time in the real world. Although testing real-world data linkages and creating new standards for post-approval evidence generation would take time — and likely some trial and error — it could help bridge gaps between clinical trials and real-world results. Some therapeutic areas, drugs and devices, like vaccines, oncology drugs, diagnostic blood tests and neurological drugs and digital therapeutics, may make excellent candidates for real-world data linkage, while others less so.

The ubiquity and accessibility of real-world data are at a point where it is possible to evaluate their utility for drugs and devices developed in many diseases. In the not-too-distant future, the traditional development process for Phase 3 and Phase 4 trials could be transformed with registration trials that are routinely connected to real-world data with a conditional approval at the conclusion of Phase 3. Participants would be able to continue taking a therapy deemed effective by surrogate outcomes as their real-world data are linked to measure their impact on their health long term.

Defining novel approaches for evidence generation is a path almost everyone can endorse taking because it can help settle open questions that regulators, researchers, health care professionals, patients, and families will wrestle with until insights about long-term clinical outcomes become available. It may generate evidence more efficiently on effectiveness, safety, and economic value, transforming the speed and depth of insight on every new intervention.

Linking real-world data to registration trials may make all the difference in validating the hope that a new therapy will change the trajectory of diseases like my grandmother’s Alzheimer’s disease and so many others that today have no therapies that can cure or slow their progress.

Elenee Argentinis, an attorney has worked for more than 20 years in biopharma, health care, and health tech organizations, currently works for Datavant, a health data connectivity company, expanding enterprise partnerships.