NEW YORK – New data from a Canadian cancer early detection effort have added to a growing body of evidence supporting a path forward for a new class of multiomic blood-based screening and surveillance tests.
Investigators published a preprint study last month in MedRxiv focused on individuals with Li-Fraumeni syndrome (LFS) — a severe cancer predisposition condition caused by germline pathogenic variants in the tumor suppressor gene TP53. Analyzing patient blood plasma using a combined genomic/epigenomic approach, the group was able to detect cancers with more than 70 percent sensitivity, catching cancer-associated signals up to 16 months before the patient's tumor was detected by conventional clinical screening.
According to the authors, led by Trevor Pugh, genomics director at the Ontario Institute for Cancer Research, the study provides a framework for the integration of liquid biopsy into current surveillance methods for these patients, who face a near 100 percent lifetime risk of developing cancer and frequently manifest multiple, often diverse, tumors.
This population is already recommended to undergo intensive surveillance, but current tools are burdensome to both the patient and the healthcare system, the group wrote.
Although multi-cancer early detection blood tests are already being marketed clinically in the US — namely Grail's Galleri assay — there are significant unanswered questions about whether these tests would result in significant benefit to the general population. But for individuals at significant risk who already undergo surveillance, like LFS patients, the value proposition may be more persuasive.
Girish Putcha, an independent molecular pathologist formerly at early cancer detection firm Freenome, stressed that this question isn't answered, but that "intuition suggests that the benefit-harm ratio here should be more favorable than general population screening" given an elevated-risk population already undergoing intensive surveillance.
"This seems unlikely to be inexpensive but depends on what it could replace and how it may improve adherence, which should not be underappreciated or undervalued," he said in an email.
According to Pugh, the "tragic" stories of LFS patients was a significant inspiration for his research team's work. Some patients, he noted, are "on their fourth or their fifth tumor. They're fine one year, and they have full-blown disease the next year. And the chance to relieve some of that uncertainty is the big opportunity. … That's why we started with Li Fraumeni syndrome."
The OICR team's sequencing approach combines targeted sequencing, shallow whole-genome sequencing, and a genome-wide epigenetic analysis method, called cfMeDIP-seq, developed by Princess Margaret Cancer Center's Daniel De Carvalho and licensed last year by Adela, a spinout from Canada's University Health Network.
The study cohort included 196 blood samples from 89 patients, of which 26 were pediatric and 63 were adults. This included 27 patients with active cancer and 62 without (either never having had cancer or having been treated in the past). In the active cancer subset, investigators were able to detect a cancer-associated signal in nearly 80 percent of samples from patients with active cancer.
Putcha questioned the stage-distribution of detected cancers, which skewed toward later-stages, but said that the question of what constitutes a "late" diagnosis can be tumor-specific, and the impact would depend on what the stage-distribution looks like for LFS patients at diagnosis.
Among cancer-negative patients, 13 had a positive result that was subsequently confirmed as either a tumor or a suspicious imaging finding at follow-up, including one case where the signal was detected 16 months prior to diagnosis.
In another 20 individuals, a positive signal was not followed up by a suspicious finding or a cancer diagnosis, but the study authors cautioned that these should not necessarily be taken as false positives without longer-term follow up, given the high prevalence of cancer in the LFS population.
"These next-gen sequencing technologies often have much greater sensitivity than the existing technology, in this case a medical exam that we're benchmarking against, so it's a really common question," Pugh said. "We're keeping the study going because I suspect we are seeing real signal there, and we have some other genomic evidence that suggests it's there," he added.
Pugh also highlighted the fact that each aspect of the team's multiomic approach appeared to measure independent biological signals, achieving the best performance when combined.
TP53 variants were detected in under half of samples from patients with active cancer, for example, suggesting that a targeted sequencing-only approach largely fails the mark. Genome-wide copy number analysis also only picked up about 38 percent of active cancers.
The OICR team also explored circulating DNA fragmentation as a fourth cancer signal, but reported that fragmentomics was the least sensitive analyte, likely because of the diversity of fragmentomic profiles from patient to patient. Nevertheless, with more longitudinal samples, it might be possible to develop patient-specific fragmentomic baselines that would increase sensitivity compared to a general nonspecific cutoff.
Important for future development, this could also mean that different hereditary cancer syndromes would require either syndrome-specific or patient-specific approaches to achieve meaningful sensitivity.
Overall, Pugh and his coauthors concluded that their results suggest that future clinical tests will need to integrate multiple different biological targets.
"While there is still some debate about whether cfDNA fragmentation patterns and genomic alteration detection provide incremental, statistically and clinically significant orthogonal information over and above cfDNA methylation, increasingly a 'multiomic' approach, whether that’s focused on cfDNA alone or other biological molecules in the blood, seems to be favored and is supported by this paper," Putcha said.
Following on their initial LFS data, Pugh and others have expanded to a larger, ongoing trial called CHARM, which is recruiting cancer-naive individuals across a range of hereditary cancer predisposition syndromes.
"We have several projects ongoing in breast and ovarian cancer, Lynch syndrome. Li-Fraumeni … was the first to get out of the gate. But every cohort we do, we think about how this could be applied across all hereditary cancer," Pugh said.
Right now, the team is working to flip its protocol from retrospective to prospective testing. "Instead of looking back in time at banked samples we'd be … starting to report some of these results back and asking geneticists, can you act on this data? Is there clinical utility there?" Pugh said.
Putcha urged caution though in conflating cancer screening with cancer prognosis, "each of which may have a different clinical workflow downstream of the test." In other words, a cancer signal may not be a false positive, but it may also not be actionable if tumors only appear years down the line.
Pugh argued though that by continuing to follow patients with positive results, they might be able to tease out that immediate utility. "If we go back in time and see that we have a very high breast cancer score, we can look at whether having that cell-free DNA evidence might change how the imaging group is looking at their data, because now we've given them where we think the needle may be, and they can go and do that hard target search."
"It gives you a very high priority to look with a greater resolution or a more critical eye than you would [otherwise]. Especially for Li-Fraumeni syndrome, the current protocol is for whole-body MRI, where you have imaging groups going through full stacks of images. We want to be able to tell them which slice to really spend a lot of extra time on," Pugh said.