In a Friends of Cancer Research (Friends) meeting devoted to real-world evidence, members of the cancer community generally agreed that its use has an increasingly important role to play in gathering the data necessary to test, evaluate, and bring new therapeutic agents to market. This Friends meeting, which was titled “The Future Use of Real-World Evidence,” was held in July in Washington, DC.
Definition and Problems
Real-world evidence is derived from data about the uses and potential benefits and risks of a therapeutic agent outside a traditional clinical trial. These data are being studied to better understand how they can relate to more traditional clinical endpoints and the long-term benefits of an agent.
Clinical trials do not reflect the breadth of data that can be collected in a more diverse real-world setting. Real-world evidence can fill that gap, although parameters and conditions of use have not been well defined. Clinical endpoints are difficult to obtain and evaluate from real-world evidence alone.
It is necessary to standardize and validate real-world evidence data to establish a usable framework, so data collection can be put into operation. In other words, how do data generated from real-world experience correlate with progression-free survival, overall survival, and other key indicators of disease burden, and how do they compare with prior clinical trials?
These issues will inform ongoing regulatory discussions on the use of real-world evidence and should ultimately result in a standard approach for data collection as well as information about the long-term value of a therapeutic agent.
New Clinical Endpoints
Time to treatment discontinuation may be a potential endpoint in clinical trials using real-world evidence, said Gideon Blumenthal, MD, Acting Deputy Director, U.S. Food and Drug Administration (FDA) Office of Hematology and Oncology Products.
There is an apparent strong correlation between time to treatment discontinuation and progression-free survival, and for some treatment regimens the correlation is even stronger.— Gideon Blumenthal, MD
Tweet this quote
“Progression-free survival is an important efficacy endpoint in oncology, but increasingly, trials allow treatment beyond objective radiographic progression for patients who are benefiting, especially with immuno-oncology and targeted therapy. Thus, time to treatment discontinuation has been proposed as a potential real-world evidence endpoint.”
Dr. Blumenthal said there are two key questions about time to treatment discontinuation as an endpoint. First, how well does it correlate with progression-free survival in a variety of treatment types, and second, how many outliers are there?
He explained: “Late time-to-treatment-discontinuation patients stayed on therapy for 3 months or more after disease progression, and early time-to-treatment-discontinuation patients stopped treatment 3 months or more before disease progression.”
He collected and analyzed data from 18 randomized controlled trials in non–small cell lung cancer begun between 2007 and 2014. For all 8,947 patients, there were 6,745 progression-free survival events and 8,055 time-to-treatment-discontinuation events, with a correlation of 0.87 (7.98% of patients had early time to treatment discontinuation and 5.97% had late time-to-treatment discontinuation).
“There is an apparent strong correlation between time to treatment discontinuation and progression-free survival, and for some treatment regimens the correlation is even stronger.” There are limitations, however, he added. “This was a retrospective, post hoc analysis with differential censoring between progression-free survival and time to treatment discontinuation. There were differences among trials regarding treatment beyond disease progression, thus creating a potential bias to stay on targeted therapy and immunotherapy as long as possible and to come off chemotherapy early. Moreover, time to treatment discontinuation is not a good endpoint if discontinuation was scheduled at the beginning of a trial.”
The next steps, according to Dr. Blumenthal, involve looking at time to treatment discontinuation in other tumor types and therapeutic classes, encouraging prospective data collection in real-world trials, and looking at time to next treatment.
Evaluating Treatment Effect
Pallavi Mishra-Kalyani, PhD, Team Leader, Division of Biometrics V, Office of Biostatistics, FDA Center for Drug Evaluation and Research, wrote a presentation about potential endpoints for evaluating treatment effect when using real-world data. Her talk was given by Rajeshwari Sridhara, PhD.
Pallavi Mishra-Kalyani, PhD
Rajeshwari Sridhara, PhD
Selection of an appropriate endpoint is a critical step in the design of any study, wrote Dr. Mishra-Kalyani. “Ideally, endpoints provide a precise measurement of differences in benefit or efficacy and are clinically meaningful. However, many randomized controlled trial endpoints using real-world data are imprecisely or inconsistently measured and are thus not clinically meaningful.”
She said that endpoint selection for real-world data depends on many factors, including temporality and the structure of data collection. But it is difficult to replicate real-world data endpoints within a randomized controlled trial and to understand their relationship with progression-free and overall survival. She did note, however, that time to treatment failure and time to treatment discontinuation can and should be directly compared with progression-free and overall survival. (In the past, time to treatment failure was not a preferred regulatory endpoint because of the inability to distinguish between efficacy and toxicity.)
“Because time to treatment failure and time to treatment discontinuation are rarely studied as endpoints in randomized controlled trials, little is known about their clinical value in estimating efficacy. But there are strong associations between time to treatment failure/time to treatment discontinuation and progression-free survival, especially in targeted therapies,” wrote Dr. Mishra--Kalyani. Therefore, the following items are worthy of future study:
DISCLOSURE: Drs. Blumenthal, Mishra-Kalyani, and Sridhara reported no conflicts of interest.