NEW YORK – In a first publication of comprehensive prospective trial data, multi-cancer early detection firm Grail has produced one of the first glimpses of what the healthcare landscape might look like with the broader implementation of blood-based screening tests like its flagship Galleri.
In The Lancet last week, investigators from the company and clinical collaborators described the ultimate findings of the trial, which recruited more than 6,000 individuals at least 50 years of age with no current suspicion of cancer to have their blood samples tested with the Galleri assay. Designed primarily as a feasibility and safety study, Pathfinder is the first to employ a locked classifier, prospectively, in what Grail views as the intended use population for its test.
Galleri, clinically available since 2021, is one of several blood-based multi-cancer screening or early detection tests that supporters believe could dramatically shift the cancer death rate by increasing early diagnoses and opportunities for curative treatment. But scepticism remains regarding the potential for this type of test to prove clinical utility let alone show that it offers benefits that outweigh its costs.
The Pathfinder study doesn't answer these questions directly, but it has painted an early picture of how some of these clinical impacts could play out on a broader scale.
Eric Klein, a study investigator and a distinguished scientist at Grail, said in a statement that he and his colleagues were pleased with what the study achieved both in terms of feasibility and safety.
"There was widespread interest both at the investigator and the participant level. We had no problem recruiting more than 6,000 patients in the middle of COVID-19, and we were able, despite the constraints that the pandemic put on every industry out there, to deliver test results in a very reasonable period of time," he said this week.
Discussions of safety for screening tests often center on the impact of negative and positive results. As a test tuned for maximum specificity, Galleri isn't intended to rule out all possibility of cancer. Negative results aren't meant be taken to mean an individual is cancer free. As such, safety in that context largely pertains to how well this reality is communicated to patients and ordering physicians.
For positive test results, assessing safety has to do with measuring subsequent healthcare utilization and overall outcomes. Klein said that Grail has previously presented data from Pathfinder showing that an official diagnostic resolution was reached in 80 percent of individuals with positive test results as long as the Galleri assay was able to predict a physical origin in the body.
In the new publication, he and his colleagues went into more detail about the nature of the diagnostic interventions used by doctors for their patients with a positive result. According to Klein, there were "no surprises there."
"Almost everybody got blood tests and radiology exams, and there were very few surgeries," he said, adding that amongst the surgeries that were performed, only one was in an individual who was not found to have cancer.
That one patient was actually Klein's. "He had [what was later deemed] a false-positive PET CT scan suggesting a tumor in his left scrotum. At his age — he was in his seventies — I just took out his testicle and it turned out to be a benign inflammatory condition.
Overall, the study recruited 6,662 participants, for which 6,621 had analyzable results. A cancer signal was detected in 92 individuals, of which 35 were diagnosed with cancer (true positives) and 57 were not (false positives).
About 79 percent of the true positives and 88 percent of the false positives had various lab tests ordered by their physicians. More than 90 percent received imaging in both groups. Only 30 percent of false-positive individuals got other procedures compared to 82 percent of the true positives.
Klein argued that when evaluating this kind of safety signal, it's important to do so in the context of what prior screening trials using other technologies have achieved. For example, he cited the PLCO trial that screened hundreds of thousands of individuals for prostate, lung, colon, and ovarian cancer.
"In the ovarian cancer group, there were more women who did not have cancer that actually ended up having surgery than women who had cancer. Plus, 15 percent of them had had surgical complications," Klein said. "That makes [our] safety results all the more impressive to me."
Although it wasn't included in the current paper, Grail also collected patient-reported outcome information through surveys focused on anxiety, other measures of distress, and overall quality of life.
In previous presentations at scientific meetings, the firm has shared some of that data, which demonstrated that while individuals who had a positive cancer signal "certainly had more anxiety than those who didn't," the anxiety and distress levels largely returned to baseline at the end of the study, reflecting what has also been seen "with any other screening test," Klein said.
Other important details reported this week included a read out of the time to diagnostic resolution — either the detection of a tumor or being cleared of cancer — for trial participants who had a positive blood test result.
Excluding two individuals whose diagnostic assessments began before their MCED test results were reported, the median time to diagnostic resolution was 79 days, with a range of 33 to 143 days in true-positives, and 162 days, ranging from 44 to 248 days, in false-positive participants.
Klein said a few factors at play may have skewed the time to diagnosis toward the longer end compared to what might be seen in ongoing larger trials and potentially in general clinical implementation.
"Again, we did this in the middle of COVID when access to testing was restricted, and my personal opinion, having been a Pathfinder investigator when I was working at the Cleveland Clinic and managing these patients, is that I don't have any doubt that we can turn this around a lot faster than we did," he said.
Another reality of the trial is that Grail did not dictate what the diagnostic evaluation should be, but it did pay for the entirety without any restrictions. "My sense is that in the real world, with this test, the diagnostic evaluations may not be as extensive or as prolonged. Where you're talking about insurance coverage and access and co-pays and all of that, it may influence what [that] looks like," Klein said.
By its very nature, a screening test designed to identify minute genomic signals of a cancer that hasn't yet caused any symptoms should lead to a potentially complex follow-up. Klein said that in the study and in Grail's commercial testing database, there are examples of patients with what the company calls "apparent false positives," where a cancer signal and putative source are detected but follow-up evaluation can't find anything, at least initially.
"In Pathfinder, there was one man with a small intestine cancer, which is incredibly rare. His [origin] prediction was lower GI and upper GI. Appropriately, he got a colonoscopy and upper endoscopy, but they stopped in the stomach. It wasn't until, through various discussions, that the physicians went back and did a second scope. That could easily have been called a false positive, when the reality is that we have a learning curve about what the intensity and extent of the diagnostic evaluation needs to be to find the right answer," he argued. "That's really what Pathfinder was designed to do."
He also cited cases of patients whose cancer was not evident in an initial evaluation, but who clinicians still suspected highly of having a malignancy. In the study, the protocol for these patients was to offer a second blood test after a period of six months or so. "If it came back positive three to six months later, we did another diagnostic evaluation," Klein said.
He cited the case of another study participant, also his patient, who had had breast cancer 11 years prior. She had a negative mammogram before joining Pathfinder and no symptoms, but she had a cancer signal detected with an origin prediction of the breast. "We did the appropriate radiologic evaluation. All negative. And so, we waited, I think, four to six months, repeated the blood test, which was still positive and still indicating breast."
"A few weeks later, over the Christmas holidays, she presented to the emergency room with an upper respiratory infection and had a chest X-ray. It showed a very minor abnormality, a little fluid effusion around one of the lungs. I was so suspicious because of her history and the two positive cancer signals detected that we repeated the CT scan. The pleural effusion had grown, and it turned out to be recurrent breast cancer," Klein said.
In Pathfinder, investigators didn't calculate test sensitivity by stage the way Grail has done in its previous case-control data. But the team did see that about half the patients who had new, non-recurring cancers detected were at stage I and stage II.
"We also detected 35 cancers by the blood test that were not otherwise detected during the course of the study, so … we actually more than doubled the number of cancers that were detected in this population compared to standard-of-care screening," Klein said. In addition, over 70 percent of the cancers detected in the study currently have no alternative standard screening modalities.
Although are certainly differing opinions, Klein, and what he said are many other medical oncologists, also believe that there is an advantage to earlier, asymptomatic detection regardless of stage. Compared to a patient who is symptomatic — whether they are just feeling poorly, not able to work, or presenting to the emergency room with a serious obstruction — asymptomatic patients have a better ability to tolerate treatment and don't require as many procedures that might delay them from getting either potentially curative or palliative therapy that would extend and/or improve their lives.
To a certain extent, the very existence of new tests like Galleri opens the door for studies of the impact of earlier detection on quality of life, outcomes, and even healthcare costs, which simply haven't been possible before, Klein said.
In that light, he posited that some of the discussions around the potential costs of broad MCED testing may be mistakenly weighing the cost of screening and subsequent diagnostics without considering the yet-unknown positive impacts of earlier detection.
"Folks might say right now, 'no pancreatic cancer is curable,' but that's true using current detection and staging modalities. We don't really know what the world looks like if you pick up a true stage I pancreatic cancer because we don't see many of them," Klein said.
"If we can really shift detection to earlier stages, and I believe that we can, it's less costly to treat any early-stage cancer than it is any late-stage cancer across the board. There are also life-prolonging and social benefits; patients can return to the workforce and are still economically productive, and so I think the cost equation is more complicated than some people think."
Grail currently has several ongoing follow-up studies that are extending this prospective data to larger and more demographically representative cohorts. These include Pathfinder 2 in the US and the UK's National Health Service Galleri study, which both targeting asymptomatic adults. They are joined by the University of Oxford's SYMPLIFY study recruiting individuals presenting to primary care doctors with non-specific symptoms such as weight loss or fatigue.
A greater wealth of data expected from these trials and others will hopefully provide a basis for new modelling and longitudinal outcome research that can begin to better answer the questions circulating around the utility and impact of MCED platforms, Klein said.