Considerations to inform decision-making.
Since the COVID-19 pandemic, the U.S. Food and Drug Administration’s Expanded Access (EA) pathway, also known as compassionate use, has featured in headlines as a mechanism by which unapproved but potentially beneficial treatments were given to patients outside randomized clinical trials (RCTs). One example was the largest-ever EA program, involving convalescent plasma given to COVID-19 patients. Originally planned to accommodate 5,000, this treatment program ultimately enrolled 105,717 patients.
“The size of the convalescent plasma program points to the need to put a cap on future EA programs,” notes Hayley Belli, PhD, MS, an Assistant Professor of Biostatistics in the Department of Population Health at New York University Grossman School of Medicine. “The data collected from the EA program did not provide conclusive answers as to convalescent plasma’s efficacy, suggesting that there may be efficacy in some subpopulations, but not in the overall population.”
Belli suggests that limiting the EA program to a certain enrollment number and requiring other patients to participate in RCTs to get access to an unproven product would result in a definitive answer on whether the therapy works or not. She adds that this would avoid ethical issues relating to the delay in obtaining evidence.
“If a product does not work, we should limit the number of patients exposed to it; if it does work, we should identify in which patients and at what dose and what at time points as quickly as possible for the benefit of patients,” Belli urges. Meanwhile, she says, “Biopharma companies increasingly wish to collect real-world data (RWD) from EA programs to include in regulatory submissions, based on increasing acceptance of these data by U.S. and European regulatory bodies. RCTs are still the ‘gold standard’ for evidence, but sponsors tend to only enroll select populations defined by tight inclusion/exclusion criteria that do not resemble the larger patient population. However, RWD cannot replace well-designed RCTs in evaluating investigational product safety and efficacy, since real-world studies lack the control groups, randomization, and blinding needed to generate robust, generalizable findings.”
Belli continues, “So, EA programs can usefully provide information from a more generalized population to supplement RCT results – including laboratory values, functional assessments, and patient-reported outcomes. Where no RCTs have been carried out, these RWD can support rapid decision-making of the kind that was needed early in the COVID-19 pandemic.”
As Belli explains, EA is intended for treatment of patients rather than for research. As such, it is important to be cautious in interpreting results, since data collection is a secondary activity in EA, rather than a primary one as in research. RWD are susceptible to biases since they are not subject to randomization and perhaps come from a sicker population, possibly after failing to qualify for a clinical trial of the desired therapy, she cautions.
Join Hayley at ACRP 2023 [April 28 – May 1; Dallas, TX], where she—and Dr. Alison Bateman-House—will delve into real-world data that is generated both within and outside EA programs, and how to evaluate real-world data as a potential source of real-world evidence. View complete schedule.
“We need to consider when it is acceptable to collect RWD from EA programs,” says Belli. “There must be a pressing need for these data, for example, to aid in the design of an RCT or supplement results from RCTs. When weighing the evidence from different sources, it is helpful to have a checklist of the population characteristics, who collected the data, and why. This helps in assessing the rigor involved, along with potential sources of bias.”
Belli notes that potential uses for RWD from EA programs include:
- Providing safety information about an investigational product or intervention
- Supporting hypothesis generation
- Informing design and patient stratification for subsequent RCTs
- Helping identify biomarkers
- Providing useful data in cases where RCTs are not feasible, including in some rare diseases
- Supplementing RCTs by enhancing the generalizability data
“Data from EA is by its nature pragmatic and generalizable,” says Belli. “It comes with costs and may ‘muddy the waters’ when trying to assess effectiveness, as the data come from patients, not from controlled trials. Yet these data have value, provided that collection methods ensure high quality and rigor of the data, with attention paid to minimizing bias and sources of error and avoiding missing data. Thoughtful data collection and sharing can help avoid possible misinterpretation.”
Author: Jill Dawson