“One of the most difficult parts of my job is enrolling patients in studies,” says Nicholas Borys, chief medical officer of Lawrenceville, NJ, biotechnology company Celsion, which develops chemotherapy and immunotherapy agents from new generation for liver and ovarian cancers and certain types of brain tumors. Borys estimates that less than 10% of cancer patients participate in clinical trials. “If we could reach 20 or 30%, we probably could have beaten several cancers by now. “
Clinical trials test new drugs, devices and procedures to determine if they are safe and effective before they are approved for general use. But the road from study design to approval is long, winding, and costly. Today, researchers are using artificial intelligence and advanced data analysis to speed up the process, reduce costs, and deliver effective treatments to those who need them faster. And they’re tapping into an underutilized but rapidly growing resource: patient data from past trials
Build external controls
Clinical trials typically involve at least two groups, or “arms”: a test or experimental arm that receives study treatment and a control arm that does not. A control arm may not receive any treatment, placebo, or the current standard of care for the condition being treated, depending on the type of treatment being studied and what it is compared to under the study protocol. It’s easy to see the recruitment problem for researchers studying therapies for cancer and other life-threatening illnesses: Patients with life-threatening illnesses need help now. While they may be willing to take a risk with a new treatment, “the last thing they want is to be randomized into a control arm,” Borys says. Combine this reluctance with the need to recruit patients with relatively rare diseases – for example, a form of breast cancer characterized by a specific genetic marker – and the time to recruit enough people can stretch over months or even weeks. years. Nine out of ten clinical trials around the world, not only for cancer but for all types of pathologies, fail to recruit enough people on time. Some trials fail completely for lack of sufficient participants.
What if the researchers did not need to recruit a control group at all and could offer the experimental treatment to everyone who agreed to participate in the study? Celsion is exploring such an approach with New York-headquartered Medidata, which provides electronic data capture and management software for more than half of the world’s clinical trials, serving most of the major pharmaceutical companies and medical devices, as well as university medical centers. Acquired by French software company Dassault Systèmes in 2019, Medidata has compiled a huge “big data” resource: detailed information on more than 23,000 trials and nearly 7 million patients dating back about 10 years.
The idea is to reuse patient data from previous trials to create “external control arms”. These groups serve the same function as traditional control arms, but they can be used in settings where a control group is difficult to recruit: for extremely rare diseases, for example, or conditions such as cancer, which involve life threatening imminent. They can also be used effectively for “one-arm” trials, which make a control group impractical: for example, to measure the effectiveness of an implanted device or of a surgical procedure. Perhaps their most valuable immediate use is in rapid preliminary trials, to assess whether a treatment is worth pursuing until a full clinical trial.
Medidata uses artificial intelligence to probe its database and find patients who have served as witnesses in previous treatment trials for a certain condition to create its proprietary version of external control arms. “We can carefully select these historical patients and match the current experimental arm with historical test data,” says Arnaub Chatterjee, senior vice president of products, Acorn AI at Medidata. (Acorn AI is the data and analytics division of Medidata.) Trials and patients are matched for study objectives – so-called endpoints, such as reduction in mortality or how long patients stay cancer-free – and for other aspects of the study designs, such as the type of data collected at the start of the study and along the way.
When creating an external control arm, “We do everything we can to mimic an ideal randomized controlled trial,” says Ruthie Davi, vice president of data science, Acorn AI at Medidata. The first step is to search the database for potential candidates for the control arm using the main eligibility criteria of the investigational trial: for example, the type of cancer, the main characteristics of the disease and its stage of development. progress, and whether this is the patient’s first time to be treated. It’s essentially the same process used to select control patients in a standard clinical trial, except that data recorded at the start of the previous trial, rather than the current trial, is used to determine eligibility, says Davi. “We are finding historic patients who would be eligible for the trial if they existed today.”
Download the full report.
This content was produced by Insights, the personalized content arm of MIT Technology Review. It was not written by the editorial team of the MIT Technology Review.