Identifying leptomeningeal metastases is an especially compelling use. “It’s a devastating but not uncommon complication of other tumors originating elsewhere in the body,” Dr. Bale says, typically appearing in late-stage disease after multiple therapies have failed. Being able to sample the component that has made its way to the CNS is important for developing therapies and targeting that specific disease. “Because often it’s undergone clonal evolution, with the emergence of resistance mechanisms that are unique to the CNS compartment. By sampling the CSF, we’re able to see those and direct therapy appropriately.”
In addition to using results to diagnose recurrent disease, “We’ve been able to detect completely new primaries,” she says, even without having a genotype or specific target for a specific alteration. “The molecular findings will tell a different story, and you’ll be able to diagnose a completely unexpected primary.”
The emergence of liquid biopsies has generated excitement and hope for earlier diagnosis and more accurate prognosis for cancer patients. Many of the advantages of liquid biopsies as a whole pertain to CSF as well. “There are some aspects of the conversation across liquid biopsies that are always going to be the same,” says Dr. Bale. Yet there are also key differences between plasma and CSF samples.
While plasma cell-free DNA is more readily available via routine blood draw, the blood-brain barrier makes it a less useful option for primary brain tumors and metastatic disease that is confined to the CNS.
Enter CSF, says Dr. Bale, which differs fundamentally from plasma. “CSF can be safely obtained in the majority of cases.”
In any liquid biopsy, the sum total of cfDNA is derived from both normal (i.e. background) nonneoplastic cells and tumor cells. But in plasma, the tumor-derived component constitutes only a small part of the total cfDNA. The result is a signal-to-noise ratio that requires many special considerations in order to detect the signal of interest at clinically relevant levels, she says.
This is much less of an issue with CSF samples, since the majority of signal comes from the tumor. “Cerebrospinal fluid is really acellular in normal circumstances,” Dr. Bale says. Because of the reduced background contamination, “we tend to be very successful at detecting tumor alterations, even though the total DNA that one is able to extract from the CSF is often very low—it’s coming almost exclusively from the tumor itself. So the variant allele frequencies we’re detecting tend to be very, very high. It’s very sensitive, you could say, for the tumor signal.”
What has been most surprising for Dr. Bale and her colleagues? “It is startling how low the DNA yields can be sometimes and yet we are able to detect tumor variants.”
At MSK, she and her colleagues primarily analyze CSF samples using the MSK-IMPACT assay—a tumor-matched normal hybridization-capture–based next-generation sequencing panel targeting more than 500 genes, which is essentially the same NGS assay they use for typical tumor samples.
Without significant modifications, they can use the same workflow for cell-free DNA that’s been extracted from the CSF. Thus, without making changes to their pipeline or chemistry, she says, they’ve been able to readily identify tumors with high turnover rates, so to speak: metastatic tumors, and high-grade primary brain tumors like glioblastomas, which are very cellular and frequently apoptotic, with high-proliferative indices and therefore more consistent release of cfDNA. In discussing the tumor types that are most amenable to cfDNA CSF samples, Dr. Bale says, “At the end of the day, this is really about the signal to noise.”
Because CSF cytology has been such a stalwart in the care of patients with brain tumors, it might be surprising to some that cfDNA is so informative. As Dr. Bale has observed, “It actually was not all that intuitive that the cell-free component of DNA is more informative and is in fact the superior component of the CSF compared to, say, the cell pellet.”