NEW YORK – Many US veterans from across the country with advanced cancers are treated at small community centers or in rural settings that may not seem especially conducive to precision oncology approaches.
But researchers with the Department of Veterans Affairs, Duke University Medical Center, Argonne National Laboratory, and Duke Cancer Institute have demonstrated that it is feasible to systematically find and interpret actionable tumor mutations in this large patient population, even within a cost-conscious government health system.
In JCO Precision Oncology this month, they reviewed results for the first two years of the Veterans Health Administration (VHA)'s National Precision Oncology Program (NPOP). That molecular tumor profiling program was launched in 2016 for patients with advanced cancer being treated through the VHA, an integrated health care system administered by the US Department of Veterans Affairs.
"It was a goal of our program that this very sophisticated, cutting-edge program would be available equally to every veteran in the system," Michael Kelley, a medical oncologist at Duke University and the Duke Cancer Institute, who also serves as the national program director for oncology at the Department of Veterans Affairs. "We made it a point that we would ensure that we look at the rurality of patients being served to overcome the barriers."
The NPOP "is a clinical program designed to improve the options for treatment of patients with advanced cancer," Kelley explains.
In addition to gene panel-based testing options for eligible advanced cancer patients, the program includes provider education and telephone consultation services, as well as a virtual molecular tumor board that meets monthly to discuss options for individuals who do have strong evidence of actionable mutations in their tumors — from on- or off-label treatment options to relevant clinical trials.
Because of the program, precision oncology approaches have become a relatively routine part of cancer care for many veterans treated through VHA, Kelley said.
"They don't really have to go anywhere to do anything. Their doctors, nurses, or other providers are trained in how to access this," he explained. "There's a partnership between the clinicians and pathologists who actually look at the tumor tissue to make sure it looks like it's good enough to be [genetically] tested, and send it off."
There have been some challenges, however, particularly at rural sites where pathologists may not have as much experience preparing samples that will be subjected to DNA testing, according to first author Pradeep Poonnen, an oncologist affiliated with Duke and the Department of Veterans Affairs.
The NPOP team has been working to educate and support providers working outside of large academic centers to help them feel comfortable interpreting the genetic mutations detected by testing and pursuing compatible treatments.
"Our hope is that, as time goes on, folks get better at being able to interpret the results and prescribe the appropriate drug based on the results, or not prescribe a drug," Poonnen said.
Indeed, translating tumor mutation data to changes in treatment and improved patient outcomes has been a perennial problem in precision oncology over the field's relatively short history.
"When we're talking about molecular findings like this … the testing has really outpaced the treatments," Steven Powell, a medical oncologist and physician scientist with Sanford Research. "We've been able to really expand the testing and make it really easy to get access to the testing, but our treatments are years behind that."
Powell is not affiliated with the VHA or NPOP, though he has participated in precision oncology efforts in community cancer settings that span sparsely populated areas and small centers in South Dakota and North Dakota.
Still, the VHA team has seen an uptick in the number of samples submitted for testing through NPOP over the program's first two years, suggesting more VHA providers are becoming aware of the program and convinced of its potential. And Poonnen noted that the VA is working to improve access to clinical trials for NPOP participants.
During the first two years of NPOP, from July 2016 to June 2018, the team collected nearly 3,700 tumor or blood samples from veterans with advanced cancer who were treated at 72 VHA facilities in the US — part of a network of more than 150 medical centers and 1,400 or so outpatient clinics that the VHA runs across the country.
Some 97 percent of the samples came from male patients, on par with the proportion of men treated for cancer at VHA centers in general, the researchers noted. And more than one third of participants came from rural areas. Consistent with their prevalence in the broader population, lung, prostate, or colon adenocarcinomas were tested most frequently.
"The breakdown of the types of cancer seems in line with the general veteran population," Poonnen said. "Our patients tend to be more male, so we see some of the male-predominant cancers."
Samples included in the JCO PO study were tested with gene panels at Personalis or Personal Genome Diagnostics labs, though VHA has since awarded a nationwide tumor genomic testing contract to Roche's Foundation Medicine to continue profiling tumors for NPOP.Â
After quality control steps, the researchers were left with genomic profiles for 3,182 of the samples, identifying one or more actionable mutation in some 70 percent of the samples, particularly in common cancer culprit genes such as TP53, ATM, and KRAS.
In each case, results from the tumor tests go back to the center where the patient is treated and to a central database for the program, Kelley said. Some of those cases may be discussed by the program's molecular tumor board, though providers can submit their cases for consideration as well.
Across the subset of cases with actionable mutations that were analyzed in the current study, for example, the investigators identified potential clinical trials for 52 percent, while 9 percent of tumors had a mutation that could be targeted by an on-label, US Food and Drug Administration-approved drug, and another 9 percent of mutations appeared potentially susceptible to targeted treatments or immunotherapy drugs given off label.
In his experience, Sanford's Powell notes that "what the companies will say are 'actionable' may not necessarily be actionable." In other words, a potential treatment target or pathway may jump out, but "whether or not we can get a drug for that, that's the challenge."
"Mutations that are meaningfully actionable — ones you can get treatments for that are actually going to work — are usually much lower," he suggested "We're looking more in the 20 to 30 percent range."
While they did not dig into the number of patients who did enroll in clinical trials, the co-authors reported that ultimately 136 patients did receive an FDA-approved treatment either on or off label.
Within the VHA system, the VA is the payor and makes coverage decision, Kelley said. Many of the drugs prescribed in a precision oncology setting are not included on the VA's formulary, requiring a non-formulary request and review.
"There's a process of ranking the strength of the evidence suggesting a particular treatment be used for a patient with a genetic mutation in their tumor," he explained. "If they're above that level, we would endorse that, and that will result in the patient getting the drug."
Still, Kelley emphasized that in the team's initial review did not reveal any pharmacy- or insurance-related problems with drug access for treatment recommendations with the highest levels of evidence. A small subset of VHA patients also have their own insurance, which the VA would bill firsthand, when applicable, he added.
The study outlined some of the treatment strategies used for NPOP-profiled patients, and offered still other insights into mutation interpretation and tumor sampling. Perhaps not surprisingly, for example, bits of tumor that were biopsied with larger core needles contained more usable DNA than samples collected by fine-needle biopsy.
Members of the NPOP team are continuing to look at strategies to increase awareness of testing and decrease barriers to care, including methods for making molecular tumor data more accessible for clinicians in the electronic health record setting.
"We'd like to make sure that that data is integrated more closely so that clinicians can get the results and we can see what they're doing more readily," Kelley said.
Along with those clinical informatics, they are considering clinical pathways looking at who to test, what kinds of tests to offer, and so on. At the moment, it appears that certain cancer types are more apt to yield actionable mutations that would impact therapy, making them more obvious choices for molecular profiling.