NEW YORK – Investigators hopeful about bringing liquid biopsy into the clinical care of brain cancer patients have published new data showing that imaging methods can help predict which glioblastoma patients can be successfully genotyped using a blood-based assay, and which cannot.
A team led by researchers from the University of Pennsylvania's Perelman School of Medicine described the method in a study last month in Neuro-Oncology Advances. Although authors wrote that the current data is only a first step and future studies will be need to validate the results, the groundwork laid could hopefully help doctors identify those who can be monitored or genotyped non-invasively, implement this testing for that subset, and eventually personalize treatments to improve outcomes.
As enthusiasm for non-invasive cancer testing grows, researchers are gaining a better understanding of which patients stand to benefit most from this paradigm shift and which pose more of a challenge.
Brain tumors, because they are blockaded by the blood-brain barrier, are one example of the latter. But despite the difficulty, clinical researchers still hold hope that they can find non-invasive solutions given the difficulty in access and the morbidity associated with surgical biopsies of the brain.
Luckily, as data has accumulated, researchers have found that using liquid biopsy in brain tumor patients isn't impossible; there are subsets of individuals whose tumors do manage to shed genetic material through the blood brain barrier into circulation.
The same UPenn team reported last year, for example, that they could see changes in cfDNA levels that matched up to or preceded changes in tumor burden seen on imaging in a small cohort of glioblastoma patients. The group was also able to identify ctDNA mutations in about half the group, with plasma-detected mutations occurring across several potentially therapeutically relevant genes, including TP53, EGFR, ERBB2, PDGFRA, and NF1.
In their new study published last month, members of the same group presented an analysis of samples from the same cohort, comparing cfDNA measures, MRI scans, as well as post-surgery tumor histopathology. They then used linear regression analyses to assess any relationships between plasma cfDNA concentration, and in some cases mutation detection, and both imaging and histopathologic characteristics.
Having a better understanding of the factors that drive differences in how DNA from a brain tumor enters the blood in one patient versus another, the group hypothesized, could lead to more efficient use of liquid biopsy by allowing clinicians to identify patients that actually stand to benefit from being monitored or genotyped with liquid non-invasive assays.
Genomic analyses are expensive and take time, Seyed Ali Nabavizadeh, the study's first author, said in an interview this week. "So if ahead of time you have a measure of ... which subset of patients you can potentially find mutations in because of the tumor microenvironment that you characterize by imaging, then it would help to try to find the patients in whom liquid biopsy is reliable versus who [needs] a biopsy or surgery to get tissue."
Based on their analyses of the MRI, tissue pathology, and cfDNA patterns in the cohort, the researchers were able to identify certain MRI metrics that reflect a disruption of the blood-brain barrier, and also correlated to certain histopathological patterns in the tumor microenvironmental. The presence of both these imaging and histology patterns matched closely with the presence of higher levels of plasma cfDNA and ctDNA.
According to the authors, there was no statistically significant correlation between plasma cfDNA and overall tumor volume on MRI, but the group did find that an increase in certain specific volume parameters, "elevated Kep and Ktrans," did match up well with the presence of plasma cfDNA.
This suggests, the authors wrote, that shedding of cfDNA into the blood is dependent on a sufficient level of disruption to the blood-brain barrier, "but not necessarily influenced by further elevations in BBB permeability beyond this threshold."
When they analyzed patients' tumor tissue material, the team also saw that increased macrophage density and tumor vessel size were both also associated with increased cfDNA concentration.
In contrast "high perivascular CD68+ macrophage density was associated with lower [MRI-based] BBB permeability measures and low detection rate of somatic mutations.
These post-surgical correlates likely have less relevance to the goal of determining which patients are most likely to see success with liquid biopsy, because the ideal situation would be to apply liquid biopsy in lieu of, or before obtaining tissue. They do however provide a more comprehensive picture of the biological complexity underlying differences in ctDNA passage from one brain cancer patient to another, which could potentially help build toward more precise and individualized treatment strategies, authors argued.
"The more information we have about a tumor, the better," the study's senior author Stephen Bagley, an assistant professor of hematology-oncology at Penn Medicine, said in a statement. "The combination of being able to measure the integrity of the blood brain barrier, understanding the density of macrophages, and tracking the tumor through liquid biopsy may be able to help us tailor our treatment decisions so that each patient is getting precision therapy that gives them the best chance of seeing a benefit."
According to Nabavizadeh, because tissue analysis is so dangerous and difficult in brain tumors, liquid biopsy is still in its infancy compared to what has been done for solid tumors elsewhere in the body. As a result these hoped-for clinical implications may be, and how to realize them still remains very unclear.
"At this point we don't have a good treatment for GBM [at all]," he said. "The standard treatment regimen is radiation plus chemo with temozolomide ... but a lot of patients don't respond to that initially."
"The hope is that we can stratify patients into groups and find out which patients can be targeted for clinical trials ... [so] these are basically the questions that a lot of people are looking into now, how to use precision medicine and how to tease out all patients with different tumor behaviors," he added. "But at this point, there is no evidence ... because until recently most people thought that [liquid biopsy was] not feasible."
That there appears to be an interplay between cfDNA, imaging, and histopathology also has some potentially useful implications, Nabavizadeh added. "Marcophages are a big player in glioblastoma, so [seeing this connection suggests] you might be able to get information on the tumor microenvironment and its composition by looking at liquid biopsy [results]," he said.
"It's very preliminary data in that regard, but it gives us some initial clues that we can [potentially] understand more about the tumor microenvironment by having the liquid biopsy results and imaging at the same time," he added.
For example, since cfDNA elevation seems to go hand in hand with the presence of macrophages, it could signal that a patient might be sensitive to treatments targeted against macrophages, or for immunotherapy, he suggested in a statement accompanying the study.
The investigators currently have an ongoing study to validate their initial results in an independent and larger cohort.
Nabavizadeh said he and his colleagues, like many other groups, are also hoping to study liquid biopsy techniques in cerebral spinal fluid, aiming to close gaps left for those patients whose tumors don't shed DNA into the blood.
"The problem with CSF is that ... the whole premise of liquid biopsy using plasma is that it's non-invasive," he said. "Obtaining CSF is invasive. Also doing lumbar puncture [can often be] contraindicated ... because of a risk of herniation, so it's not feasible in all the patients."
That said, the studies that have done it have had promising results, with higher mutation yields than plasma "so that is an area that we are going to explore," Nabavizadeh said.