NEW YORK – Clinicians at Dana-Farber Cancer Institute are using an internally developed decision support program to identify biomarker-guided therapy or clinical trial options for the majority of gastrointestinal cancer patients who receive molecular testing.
Although a recent study showed that a minority of patients ultimately ended up on recommended treatments due to various reasons, such as not being able to partake in trials, researchers involved with the program, called Gastrointestinal Treatment Assistance Regarding Genomic Evaluation of Tumors, or GI TARGET, found that it still improved patients' access to therapies and provided much-needed support to oncologists struggling to keep up with precision oncology advances.
Oncologists at Dana-Farber can molecularly profile patients' tumors using its in-house next-generation sequencing platform OncoPanel, but the reports from such testing tend to be long, complicated, and hard to decipher. Through the GI TARGET program, which was developed by experts in the Dana-Farber Gastrointestinal Cancer Center, the Brigham & Women's Hospital Center for Advanced Molecular Diagnostics, and the Dana-Farber Knowledge Systems Group, oncologists can have experts review patients' tumor profiling results and receive support in identifying biomarker-targeted therapy options or clinical trials.
In JCO Precision Oncology this month, researchers involved with this program reported on the impact it has had to date based on a review of 500 patient cases. The published results included data from patients seen over a six-month period in 2019 by 19 medical oncologists in the Gastrointestinal Cancer Center.
Marios Giannakis, a medical oncologist and clinical investigator at the Dana-Farber Gastrointestinal Cancer Center who worked on GI TARGET, said the project brought together two longstanding programs at Dana-Farber to better support oncologists: the PROFILE initiative, in which every patient with advanced cancer undergoes molecular profiling with OncoPanel, and MatchMiner, the institute's in-house clinical trial matching platform. It then added an automatic review of the OncoPanel results either by a full molecular tumor board or through a resource dubbed Molecular On-Call, where an oncologist with expertise in genomics and a clinical genomic scientist review test reports.
Dana-Farber's program comes as the use of genomic profiling is increasing, more and more precision oncology medicines are coming to market, and new predictive biomarkers are being adopted. Leading cancer institutions like Dana-Farber that are providing tumor profiling to patients are recognizing that doctors need help keeping up with the rapidly changing precision oncology treatment landscape. Some are investing to build internal MTBs and decisions support systems, while others are outsourcing the expertise.
"We realize that clinicians themselves may not have the time or expertise to handle and interpret genomic results in the clinic," Giannakis said. "The need has to do with both an expanding amount of genomic information that oncologists are called upon to understand and busy clinical practices where you may not have the time to actually go through the results, interpret them, talk to your colleagues, and relay recommendations back to the patients in real time."
To provide more support, Giannakis and colleagues developed GI TARGET to automatically provide expert reviews of OncoPanel results and recommend interventions.
OncoPanel interrogates 447 cancer-associated genes allowing for the identification of single-nucleotide variants, insertions, deletions, and copy number variants as well as structural rearrangement variants of select genes. The panel has been integrated into the routine care of patients with certain tumor types at Dana-Farber since 2014. The majority of cases reviewed through GI TARGET were colorectal and pancreatic cancers, followed by esophagogastric, biliary, and other GI tumors.
Most GI TARGET reports (90 percent) included at least one recommendation for the treating oncologist with 81 percent of patients being recommended for treatment with one or more molecularly guided therapies. Among those with a treatment recommendation, 80 percent ultimately ended up enrolling in a precision medicine clinical trial. On average, each report contained two clinical trial recommendations. Through the GI TARGET program, 6 percent of patients received recommendations for on-label treatments with a molecularly targeted agent and 25 percent were suggested off-label treatment with a molecularly targeted agent.
The turnaround time for GI TARGET recommendations varied based on which workflow the report went through. The more straightforward cases, such as those with typical molecular profiles or few actionable alterations, were reviewed through Molecular On-Call. The more complex cases, such as tumors with many actionable alterations or those requiring input from experts from other departments like pathologists or surgeons, were sent for discussion to the MTB, which meets weekly.
For the 33 percent of cases reviewed by the MTB, treatment recommendations were typically available in four days, and for the remaining 67 percent of cases reviewed through Molecular On-Call, the turnaround time for recommendations was nine days. The MTB and Molecular On-Call programs reviewed an average of 20 cases per week.
To further expedite patients' ability to get on recommended treatments, the researchers automated some of the process, such as using MatchMiner to recommend clinical trials, and they hired a dedicated clinical genomic scientist to help with the program.
"This program was a tour de force to evaluate and interpret genomic results on all patients as they were being generated," Giannakis said. "That necessitated some of these steps like the automation of algorithms and databases and having these specific workflows so that we could do many, many cases per week."
One area the researchers said showed "modest" results was the rate of clinicians who acted upon the recommendations. Within the study period of six months, 16 percent of patients with available follow-up data had clinical action taken based on GI TARGET recommendations.
Giannakis noted there are many potential reasons why a clinician would not take the recommendation, noting that the study was not designed to measure the follow-up action and the short study time-frame may not have captured all follow-up decisions.
Another factor for the low uptake of the recommendations may be that treatment suggestions based on tumor profiling were generated for patients regardless of whether they were actively looking to get on a new treatment. "Reports were generated on all patients with tumor profiling results, regardless of whether they needed a trial or whether the provider had requested it," Giannakis said. "A patient could have a report generated, but they could be doing perfectly fine on first-line chemotherapy, for example, for colon cancer."
Even if clinicians did not use the reports immediately, they may be useful down the road if a patient's cancer recurs after initial treatment. The report could still be used to guide the next line of treatment by providing insight into the alterations in the patient's tumor and matching to clinical trials, Giannakis added.
Clinical trial enrollment is also a hurdle for patients, which may have affected the rate of clinical action. The GI TARGET report may recommend a study but doesn't account for the availability of slots that would allow for patient enrollment. Then there are the logistics of enrolling in a study, such as the location of the trial site, how frequently the patient has to come in for tests, and associated costs that can further make it hard for patients to participate in research, Giannakis said.
"Compared historically to the enrollment on clinical trials in the Gastrointestinal Cancer Center, we did see a doubling in the rate of enrollment since GI TARGET has been rolled out," he noted.
The program has continued beyond this study and has now reviewed more than 2,700 cases. Going forward, Giannakis' team hopes to continue improving and expanding GI TARGET, such as adding data on immune biomarkers or from functional screening with organoids or mouse models to make recommendations more personalized. They are also exploring another study, in which they hope to evaluate whether a program like GI TARGET actually improves patient outcomes.
"The challenge upon us is to further demonstrate the benefit of these programs so that we can further convince academic and community oncology practices to pursue them universally," Giannakis said.