While drug developers continue to develop promising investigational cancer drugs, conducting clinical research in oncology remains difficult. Here’s how AI-enabled software can help.
The statistics on inadequate trial recruitment and endemic challenges in oncology clinical trials are well known. They have only gotten worse over the past 20 years. While the number of cancer treatments has nearly quadrupled in that time period from 421 to 1,489, cancer drugs take 30–40% longer than other indications to gain approval and 80% of oncology clinical trials fail to meet enrollment timelines. Over this period, trial complexity has also increased due to more comprehensive trial designs (e.g., multi-cohort, basket and umbrella studies), precision medicine studies requiring gene, RNA or protein biomarker assays and the increasing quantity and sophistication of desired endpoints.
Additionally, drug development has evolved tremendously — specifically, it has moved away from systemic chemotoxic agents and radiation to a focus on the development of drugs tailored to attack specific genetic defects present in cancer cells. Harnessing the human immune system via cell and gene therapies also provides game-changing promise for cancer patients. These are included in a new class of treatments commonly referred to as ‘personalized medicine’; these therapies represent 73% of the oncology drug pipeline. The pace of innovation shows no sign of slowing down. While these treatments present an incredible opportunity for patients, they simultaneously create challenges for care teams responsible for conducting the clinical research required to get the drugs through the approval process.
The vast majority of patients enrolled in clinical trials are recruited from academic medical centers (AMCs) and comprehensive cancer centers (CCCs) with a prominent focus on running clinical research programs. The 50-years-old CCC model is a byproduct of the National Cancer Act in 1971. Today, these sites are the anchors of most cancer research efforts, even though their patient recruitment methods are essentially the same (and wrought with the same hurdles). Recent data from an observational trial (EFACCT) suggests that cultural and organizational factors that limit the potential to support growing trial complexity are vital challenges for these traditional trial sites.
Historically, many patients have seen clinical trials as a last option (before palliative care). However, with the publicity of promising new personalized treatments, patients are interested in clinical research as a care option (CRACCO) and have a positive perception of clinical trials. In fact, in a 2017 survey, fewer than 5% of patients report having a negative impression of clinical trials. While patient privacy will always be critical, many patients understand the importance of their data and the use of information garnered in clinical trials to improve treatment options. In a study in the New England Journal of Medicine, 82% of 771 survey respondents said that the benefits of data-sharing outweighed the potential negative attributes. Only 8% thought that potential negative outcomes were more important to consider than the positive aspects.
Continued inadequate patient recruitment, the complexity of trials, failure of traditional institutions and changing perceptions of patients create the perfect opportunity to change the clinical trial recruitment. Two critical elements are needed to improve the outcomes in trials: new sources of patients and technology to assist providers in keeping current on the fast-moving oncology trial pipeline.
Recently, a more targeted focus on the community oncology setting aims to solve some of these problems. More than 80% of cancer patients receive treatment in the community oncology setting. Aside from a few progressive practices, these thousands of community sites have gone largely unutilized for cancer research (particularly clinical trials). In general, the rationale for the underutilization of community oncology sites is as follows: The practices are generally smaller with fewer patients, a limited number of oncologists, and a streamlined infrastructure that does not have an AMC/CCC staff support a traditional trial program. However, with advances in technology, the evolution of better processes, the coordination of these practices via large networks, and the regulatory guidelines that modern practices can leverage, community oncology practices can assist in finding patients for studies and may be critical for success. Democratizing trials to the community will also help with other challenges such as diversity and demographic goals that go unmet in most trials.
Technology is the great equalizer. Finding patients who qualify for clinical trials is largely a matter of triaging large patient volumes to find a small handful of patients that may potentially be eligible for a specific study. Traditional methods involve a physician, nurse or other care team member sifting through millions of data points within thousands of patient records to find scores of patients in a process that can take as much as nine hours per eligible patient. This task is virtually impossible for even a large institution with a dedicated research support staff, and this process has been historically beyond consideration for a small community oncology practice.
Fortunately, technology has started to disrupt this model. Deep Lens is a company that has developed software and processes to assist with this specific challenge — and is currently focusing its efforts on helping community oncology hospitals and practices participate more effectively in clinical research.
The Deep Lens VIPER software has built-in AI technology deployed in hundreds of practices to help care teams perform this triage process more effectively. By looking through every cancer record that comes through a practice and focusing on several key elements that qualify or disqualify a patient for a clinical trial, the AI technology pulls information from many different sources in different formats. It can surface all patients for all studies at a particular site. Furthermore, this process is being performed around the clock. As a patient progresses through the disease, any updates are reviewed in real-time. This practice eliminates the problem many sites have in finding patients that partially qualify for a study but need to be tracked for disease progression, specific treatment regimens or metastasis of disease. These qualifiers are responsible for many patients falling through the cracks, but technology that tracks a patient and surfaces every change in status mitigates the issue. Additionally, for sites that lack the personnel to perform recruitment, Deep Lens provides resources to assist in identifying patients through the consent and enrollment process.
“Community oncology practices have traditionally run leaner than larger academic institutions, creating resource and staff challenges that make it difficult to effectively run complex clinical trials on-site. Technology can change that,” said Barry Russo, chief executive officer of The Center for Cancer and Blood Disorders in Texas. “With automated tools in place like VIPER, we are now able to compete with larger academic institutions on attracting patients and generating patient interest in trial participation, and we can bring an increased number of precision medicine trials to our practice. We look forward to the evolution of technology making the clinical research playing field more level, as this will ultimately result in better outcomes for the cancer patients we serve.”
In published research at the American Society of Clinical Oncology (ASCO) meeting in 2020 — as well as in additional unpublished research— Deep Lens has shown that they can effectively triage dozens of studies for providers and effectively reduce staff burden by more than 95% on complex precision medicine trials. “Prior to using Deep Lens, we barely had time to screen any patients, now we can screen all qualified patients,” said Deborah Friedman, the director of research at a Community Cancer Center in Southern California. In addition, they have shown that by surfacing more patients, they can increase accrual rates at a site two-fold or more.
AI-based software solutions can assist practices eager to participate in research because they reduce the need to hire staff that they can’t afford. By opening the doors for trial sponsors to find patients in new markets, they can also increase revenue, offer patients more and potentially better treatment options and improve their ability to become leaders in an evolving move to democratize clinical trials.
The addition of support frameworks such as just-in-time site activation programs where sites are pre-qualified for a study and only activated once patients are pre-identified for a study are also being introduced alongside technology, which helps reduce the cost burden and time-to-activate. These tools are deployed to provider networks free of charge, creating the ability to find patients where they are being treated in real-time, allowing faster site activation due to lower bureaucratic burden, quicker and more effective patient recruitment and eventually, the hope of drugs reaching approval faster.
“Historically, the community oncology setting has been largely overlooked by sponsors and CROs in the clinical research arena. Unfortunately, this has contributed to significant missed opportunities for patients, who may have otherwise had the chance to possibly change the course of their disease by receiving access to novel, cutting-edge therapies in development,” said Jeff Hunnicutt, chief executive officer at Highlands Oncology Group in Arkansas. “Embracing tools and technology like VIPER will start to bring important research to the community setting, ensuring that more clinical trials progress to approval, and it will also broaden access to a larger, more diverse population of patients.”
Pressure is mounting on clinical research organizations (CROs) from industry partners to help them solve this problem. Despite COVID-19 disrupting most industries, including clinical trial participation, CRO earnings exceed projections while the drug developers struggle to meet expected timeframes for development goals. With AI-based technology and new areas of recruitment providing access to untapped patient pools, the traditional recruitment methods that rely on a handful of large sites and empirical evidence to support their trial programs should be re-evaluated. The times are changing for the good of patients, providers and biopharma, and the next decade will be an exciting opportunity for everyone involved in cancer care.
T.J. Bowen is the chief scientific officer and co-founder of Deep Lens (Columbus, Ohio). Prior to joining Deep Lens, Bowen’s career has spanned from cancer research to software development and strategy and management consulting. His research focused on pathological identification of tumor differentiation in breast cancers derived from p53, BRCA1/2, ATM and other mutations. Bowen has also held operating roles as the general manager of the world’s premium biology and pharmaceutical patent search software at CAS. More recently, Bowen was a founding leader of the software innovation team at Fuse by Cardinal Health where his teams developed products for healthcare providers, pharmaceutical companies and patients. Bowen was a Regents Scholar at the University of California, San Diego where he received his doctorate in biomedical sciences.
Filed Under: clinical trials, Drug Discovery