By Deborah Borfitz 

November 9, 2022 | An open-source computer platform developed at Dana-Farber Cancer Institute is adding ease and speed to the process of matching cancer patients to trials of therapies targeting genomic alterations. The matching tool, which links individual clinical and genomic information to trial eligibility data in real time, is designed to help overcome the stubbornly low precision medicine trial participation rate despite more common use of genomic profiling, according to genomics scientist Harry Klein, Ph.D. 

Researchers have previously found that as few as 10% to 15% of patients with profiled actionable mutations enroll in precision medicine trials. But in a five-year retrospective study newly published in npj Precision Oncology (DOI: 10.1038/s41698-022-00312-5), the Dana-Farber team used their MatchMiner platform to facilitate 20% of all possible consents and did so 55 days earlier than they otherwise could—accelerating the trial enrollment process by 22%. 

The researchers found 166 instances where the platform identified a potential match between a patient and a trial, and the trial team or the patient's oncologist viewed the match, leading to the patient’s consent to join the trial. The time-to-consent efficiency gain was measured against 353 consents obtained without the platform. 

MatchMiner links patients to trials by the molecular features of their tumor as well as by their age and tumor type, says Klein. Preliminary matches proposed by the platform get followed up to ensure patients meet all the other trial criteria, including tumor stage, previous treatment, and their overall health.  

“The goal is to help patients get on trials and help trials find patients... and improve the state of cancer care for everybody,” says computational biology scientist Tali Mazor, Ph.D., co-lead author of the paper with Klein. MatchMiner is being used to connect patients to all the precision medicine trials offered at Dana-Farber, now numbering over 450, and is being provided as a resource to other institutions wanting to do the same for their patients. 

The main holdup in getting the open-source informatics system more widely adopted is the upfront implementation cost, says Klein, notably for personnel to get clinical and trial eligibility data into MatchMiner. Resources would also be needed if other institutions wish to follow Dana-Farber’s lead and integrate the software with their clinical trial management system and electronic health record (EHR) to get the trial matches in real time. As new trials open, a curator would need to review them to see if they should be included in MatchMiner to keep the platform up to date. 

Meanwhile, in collaboration with Dana-Farber medical oncologist Kenneth Kehl, M.D., MatchMiner’s designers will be integrating natural language processing (NLP) into the platform to expand its recruitment horsepower. The team will be piloting NLP-based prediction of patients in need of a new therapeutic option, based on an analysis of the text associated with radiology scans.  

The Options 

Low participation in precision medicine trials has many causative factors, among them low clinician awareness, patient performance status, and patient attitudes and financial concerns, says Klein. MatchMiner is only addressing one of roadblocks.  

But the difficulty in matching patient genomic data to precision medicine trial eligibility data is a significant barrier, he adds. In the absence of an advanced trial matching system, oncologists must wade through hundreds of active clinical trials to find the few that may be relevant for any given patient. Further complicating the matching exercise is that many of these studies are basket trials looking for patients whose tumors have similar genomic changes and are being treated by a variety of different specialists. 

Existing trial-matching platforms are proprietary and thus cannot easily be adopted by other institutions, points out Klein. Molecular tumor board software has proven useful for patient enrollment but can be resource-intensive and is often focused on specific cancer types.  

As discussed in the published paper, data can be as readily inputted into MatchMiner as other popular open-source software platforms such as cBioPortal for Cancer Genomics. It can either be integrated into other clinical systems or used in a research context. Users include Princess Margaret Cancer Centre of University Health Network and Memorial Sloan Kettering Cancer Center. 

The platform was developed by the Knowledge Systems Group at Dana-Farber led by Ethan Cerami, Ph.D., and Michael Hassett, M.D., and draws on Dana-Farber’s extensive programs in genomic analysis and clinical research. Over the past decade, more than 40,000 patients at the Institute have had their tumor tissue analyzed for alterations in over 400 cancer-related genes. 

MatchMiner launched in 2016 (and won a Bio-IT World Best of Show Award in 2017). The intended users are referring clinicians and trial staff, including clinical research coordinators who are screening patients for trials, Klein says.  

“The tool is built to be used within a single institution,” says Mazor. “Our focus is on patients being treated at Dana-Farber, since the trials here are the first pass of [those] they are likely to go on.” 

The power of MatchMiner at Dana-Farber comes from presenting all the precision medicine trial options to oncologists at the point of care, Klein says. “[They would typically] hear about trial options through word of mouth and here we have this system where they can see everything available in one place.” 

How information gets displayed to clinicians and study teams may look different at other institutions that choose to adopt MatchMiner, adds Mazor. The software exists as a standalone website but was integrated into the EHR at Dana-Farber by physician request. Elsewhere, integration could theoretically be with a molecular tumor board or other software.