Do you have lung cancer? Is it a carcinoma, sarcoma, melanoma, lymphoma or leukemia? Is your lung cancer small or non-small cell? What stage is your cancer, and is it recurrent? Perhaps you should try treating your lung cancer with Cyclodextrin-based polymer-camptothecin CRLX101 and Olaparib. Or ATR Kinase Inhibitor VX-970 and Topotecan Hydrochloride.
A cancer diagnosis itself can be life altering, but treatment can be just as overwhelming. That’s why a team at the National Cancer Institute is working to connect patients with clinical trials, to give them the best shot at a cure. And they’re doing it through an application programming interface (API).
“Cancer clinical trials are very important for developing new treatments and also improving existing treatments for cancer patients,” said Tony Kerlavage, chief of the cancer informatics branch at the NCI. “Unfortunately, historically there have been only about 5 percent of cancer patients who actually participate in clinical trials. There’s lots of different reasons people don’t enroll in clinical trials. They might already be too sick to get on a trial, maybe they don’t understand the idea or the intent of the clinical trial. But another important reason is that too few people actually know about clinical trials or whether there’s one that’s just right for their particular disease.”
The API is part of the overarching Cancer Moonshot initiative, a $1 billion effort spearheaded by the Vice President to speed up cancer research — and ultimately eliminate the disease.
The Moonshot has five goals: foster scientific breakthroughs, bring new therapies to patients faster, strengthen prevention, improve patient care.
Kerlavage said Vice President Joe Biden and the Cancer Moonshot team identified the low percentage of clinical trial applicants as “a critical gap,” and challenged NCI to make that trial information more accessible to patients.
“We’ve had for quite a while a clinical trials reporting program, or CTRP database,” Kerlavage said. “And this keeps track of all ongoing cancer clinical trials, but it wasn’t really designed for direct use by the patient community. So that was really the driver: how do we get this information out to the broadest set of patients and their advocates.”
Navigating through data
Work to close the gap between patients and trials began in earnest in May. The beta version of the API launched at the end of August. Kerlavage said his team , collaborated with the Presidential Innovation Fellows (PIF) program, and collected feedback from various stakeholders like patient advocacy groups, academic cancer centers and industry representatives.
“Each of them were doing their own work to make these data available in some fashion or another,” Kerlavage said. “In some cases they were downloading the data from our database, and then having to recreate their own database to house that … and their interfaces to those data, and everybody was doing it somewhat differently. So we realized that the whole process could be significantly streamlined if we had a better way of getting the data directly out to everybody in the same format.”
Sarah Kerruish, chief growth officer at Antidote — a patient advocacy group that matches patients with trial options — said her organization was pulling from ClinicalTrials.gov, Kerruish said. And while that site is very comprehensive in its collection of trials, “it’s really written for professionals, by professionals, and it’s not structured data.”
“It means it’s very hard to understand,” Kerruish added. “Some of the titles are very lengthy and full of medical jargon, which an ordinary patient wouldn’t understand. And secondly the data in every trial, there are lines and lines of exclusion criteria … and it’s just impossible to manually read through all of those exclusion criteria and match it to your disease without spending literally days.”
Martin Naley, founder and chief strategy officer at Cure Forward — a company that helps patients find and choose their cancer care — agreed that ClinicalTrials site is “very hard to navigate.”
“The information is not really structured in a way that enables effective searching,” Naley said. “The NCI knowledge actually is very searchable and we were thrilled when we learned about it. We are actually using the data provided by the NCI API, in our system, to help trial recruiters set up their searches and find more patients more quickly through their trials.”
The API works as a software middle man, Kerlavage explained. It allows programmers to access particular data from a database or other source, in this case the NCI’s clinical trial reporting program, and then through the API, can integrate that data with their own application.
Similar to traffic alerts and weather forecasts, programs are pulling from the same information database, but sharing that information through their own unique platforms — such as the Weather Channel app versus AccuWeather, or Waze versus HERE WeGO.
NCI has its own interface where patients, doctors and advocates can go to learn more about nearly 5,000 clinical trials.
Through feedback, the NCI team learned that it’s important to keep questions simple and the display streamlined.
Users type in a cancer or keyword, being as specific or generic as they want. Adding in their ZIP code and age, the user is then taken to a screen with a list of trials that correspond to their answers.
“When the website actually calls that API, the API delivers all of this information back,” Kerlavage said. “The API delivers all the information and then the web interface can filter that for you in a number of different ways. You can put all sorts of bells and whistles on your website to filter this in a lot of different ways.”
Thanks to the API, clients don’t need to store all of this data on their own computers, nor do they have to worry about keeping the data up to date.
That timeliness is one of the goals Naley said his organization is working on as the API matures.
“I think the next step will be for real-time updating of clinical trial sites activation and deactivation, so that patients don’t have that frustrating moment of finding a trial that looks like a good match for them, only to learn that that particular site or that particular arm has been discontinued,” Naley said. “So we’re happy to continue to work with NCI and with others to make sure that the information that’s available in these knowledge sites is really up to date all the time.”
Kerlavage said his team is working on improving the back end of the database by restructuring eligibility criteria to be less of a giant block of text.
“If we can make that much more structured, that will be very helpful as well,” Kerlavage said. “Then you can envision in your search interface you can be much more specific. Imagine a world where you’ve got your electronic health record and all of this information is in there, and now this can all be done in software, where all that information from your electronic health record can be interfaced with this API and really give you the smallest sets of trials that are the absolute best for your particular situation.”
Kerruish said going forward, it’s all about having a collaborative mindset.
“It’s not a problem to solve as an individual group, it’s a problem we can solve together,” Kerruish said. “Creating those standards, making things machine readable, making them accessible and open through APIs, that’s the key to really helping solve the problem of patients finding trials. Cancer medicine is incredibly complicated and different and there’s a lot of very smart people working on that. This piece of it is a data problem and access problem and information problem, so in that sense it’s tremendously exciting because we can solve it, we can make sure patients get the information they need.”