NEW YORK – While many oncologists recognize the potential of real-world data to advance precision oncology research and inform patient care, the lack of a uniform infrastructure for collecting these data prospectively has, in many cases, precluded their use in more traditional research settings.
A feasibility trial recently launched by Roche subsidiaries Genentech, Flatiron Health, and Foundation Medicine, called the Prospective Clinico-Genomic (PCG) Study, aims to address this real-world data challenge. The PCG study will prospectively enroll 1,000 patients with metastatic non-small cell lung cancer or extensive-stage small cell lung cancer treated with standard-of-care across 20 sites within the Flatiron Health Network. Ultimately, the companies want to assess whether technology-enabled trial enrollment can expand access to precision oncology to more patients. Throughout the course of the study, they will also collect serial liquid biopsies using Foundation Medicine's FoundationOne Liquid tests to explore how circulating tumor DNA (ctDNA) levels might predict patients' response to treatment.
Within Flatiron Health-Foundation Medicine's clinico-genomics database, researchers already have access to deidentified electronic health records from 54,000 patients treated at more than 280 US cancer centers and results from genomic tests performed by Foundation Medicine. But these real-world data are retrospectively collected, and as Genentech's researchers acknowledge, "retrospective analyses may be limited by lack of availability to certain data, data quality management, and other confounding factors; however, prospective, meaningful data at scale may address these limitations."
The resulting comprehensive data and biorepository built through the PCG study will include overall survival data with four-year longitudinal follow-up from both structured fields, such as laboratory results, and unstructured fields, such as physicians' notes in electronic health records abstracted via Flatiron's technology. The repository, the researchers hope, will ultimately shed light on predictive biomarkers of response to standard treatments patients are taking — an exploration that has been set as the study's secondary aim. The PCG study's primary aim, however, is to demonstrate the feasibility of this type of prospective real-world-data collection in the context of large-scale clinical research.
"The study is really an examination of the scalability of this approach," said Mark Lee, Genentech's global head for personalized healthcare and product development. "It's an estimate of how well an EMR technology-enabled study can be run in real-world clinical practices … and the percentage of participation should be a great indicator of how scalable this can be."
The endpoints defined for the PCG study — in this case, the percentage of patients eligible to participate who end up enrolling and subsequently submitting blood samples for liquid biopsies at pre-defined intervals — will ultimately inform the companies' assessment of whether this approach to research is indeed feasible and scalable. The study is still in early stages, but Lee noted that he and his colleagues have been encouraged with the enrollment rate so far.
According to the digital poster accompanying the companies' presentation of the PCG study during the 2020 American Society of Clinical Oncology's virtual annual meeting, the trial expects to complete enrollment by early 2021 and to assess final outcomes after the last patient enrolled has completed four years of treatment.
Seeking predictive markers of treatment response
Although the primary goal is an assessment of the scalability of this research approach, the PCG study's secondary aim is to inform how genomic changes in a patient's tumor may predict response or impact resistance to treatment. This is where Foundation's serial liquid biopsy assay will be key.
Patients will be tested at study enrollment, at first tumor assessment, and at progression or end of treatment to track how their total levels of ctDNA change over time. When absolute ctDNA levels drop, Lee said, "it's quite possible that … can be indicative of how long that individual can benefit from treatment." On the other hand, if the ctDNA levels don't drop over time, that could indicate that the patient should switch to another treatment, he noted.
Patients' tumor tissue samples will be submitted at baseline for genomic profiling using FoundationOne CDx. Throughout the course of treatment, the study will then also use the liquid biopsies to identify specific molecular alterations driving the tumor, and whether these mutations or alterations change over time. This information together with absolute ctDNA levels will comprise the PCG study's predictive biomarker investigation.
"Collecting liquid biopsies for blood-based genomic assessments using cell-free DNA, and doing it over time longitudinally … allows us to study the biology, the evolution of the biology, and the dynamics of this new signal in the blood in a way we've never been able to before," said Lee. "So [PCG] is not purely a standard-of-care assessment alone. It also includes this new scientific question as part of the program."
Improving participation, reducing burden
Patients enrolled in the PCG study will receive the standard-of-care treatment for their specific cancers, which in this pilot study includes metastatic non-small cell lung cancer and extensive-stage small cell lung cancer. However, the companies hope to expand this approach to many other cancer types.
Since the trial sites will be treating patients just as they would have had they not been enrolled in the trial, the companies are collectively referring to the PCG study as "low-interventional".
Furthermore, the study will use the Flatiron database to identify those patients eligible for participation, which Lee expects will minimize the burden on the trial sites. "One of the real advantages of the Flatiron Health platform is the ability to look in an automated way at which patients might be eligible for a given program like this one," he said. "We know a patient can't participate in a trial if they are never offered the option, and in this case, since you've automated the identification of those who might be eligible, you're far more likely to get them to consider the study and ultimately enroll."
The companies are hoping that this enrollment method in the PCG study will improve upon the oft-cited 3 percent clinical trial participation rate among cancer patients.
So far, patients have been enrolled in the trial from 14 sites within the Flatiron Health Network, all of which are community oncology practices, a designation that Lee emphasized as a crucial component of the PCG's goal of expanding access to precision medicine trials. The majority of patients receive their care in the community setting, and yet precision oncology trials are often conducted at major academic medical centers.
"This study really begins with the idea that clinical research is not accessible to most patients, at least not in the US, because of a variety of reasons," Lee said, citing access barriers such as trials not being available at certain cancer centers or trial protocols requiring infrastructure and workload that many sites do not have the bandwidth to conduct.
"Our thinking was we have the technology today to collect data on the way patients are managed, their laboratory results, and clinical information that is collected in the medical record," Lee said. "Could we combine that with our clinical research aspirations in a way that really makes clinical research possible at a much larger number of sites with a much larger number of investigators, and essentially make that accessible to more patients than ever before?"
The Roche companies behind the PCG study have emphasized the importance of upholding high standards with this data collection and keeping the information reliable and usable. Regarding the latter task, Lee explained that the Flatiron Health technology is capable of abstracting data from medical records, including unstructured data that might be included in clinical notes. The software identifies and streamlines these data, after which a centralized review ensures the quality of this information.
"The challenge is that very often this real-world data doesn't look like case report form data, but if it goes through this kind of technology-enabled approach, you can begin to approximate that kind of quality," said Lee.