True, she says, extracting down the cell pellet, versus cfDNA, will yield more total DNA. But because of the aforementioned difference in signal to noise, it won’t necessarily provide a larger fraction derived from tumor.
Dr. Bale and colleagues did a head-to-head comparison of methods during their validation (Bale TA, et al. J Mol Diagn. 2021;23[6]:742–752). “It’s very clear,” she says, “that the cell-free DNA component has a higher variant allele frequency. It’s much more informative,” with many more variants detected. “It’s just more sensitive.”
Early on, if a case were negative, it was not uncommon for colleagues to ask whether it would be useful to perform molecular profiling using the cell pellet. “We would say, ‘No, that’s not going to be more informative,’” she recalls. Fortunately, the laboratory had the validation data to make its case.
On the flip side, Dr. Bale says, “We are less likely to detect tumors that are low cellularity and low grade.” Tumors with those characteristics are the focus of ongoing research and development. “That’s our new frontier. We’re looking at approaches like tagging on and reflexing additional testing to our current pipeline, to be able to increase sensitivity for these types of tumors.” This frequently includes low-grade gliomas and glioneuronal tumors, which she calls “more genomically quiet tumors.” It is often helpful in these settings, she says, to make the diagnosis by CSF, know their drivers, guide therapy, and perhaps even bypass the need for immediate surgery.
For example, she and MSK colleagues are making headway with IDH-mutant gliomas, which span the spectrum from very low grade to malignant. When these tumors are in low-grade stage, or in watchful waiting after initial surgery, “we can still search for something like the IDH mutation as a marker, by reflexing the leftover libraries from a negative MSK-IMPACT test and reflexing to a Droplet Digital PCR for IDH mutations,” which is a more sensitive assay.
A bonus: Both the NGS and ddPCR “are living in the CLIA-certified environment,” she says. “We’ve clinically validated both tests, so it’s just being a little creative in how we go about answering a clinical question, using two different assays to maximize the sample.” In many institutions, she says, it’s not uncommon for patients with brain tumors to have a diagnostic lumbar puncture with the intent of sending the CSF for cytology, while the remaining supernatants can then be reflexed for extraction of cfDNA.
As a growing literature demonstrates, the interest in CSF extends beyond that at MSK, she says. “At this point, we’ve seen that almost any assay that is developed against genomic DNA, tumor DNA, can be adapted readily for cell-free DNA coming from CSF. People have even had success using methylation analysis.”
Standardization and best practices are emerging for the use of cfDNA from CSF samples for genomic profiling.
In the meantime, Dr. Bale says, laboratories that want to bring these assays online need a plan. “The real question is, what is it you’re trying to do?”
“What makes sense for your laboratory, for the patient population you serve?” For all the advantages afforded by CSF samples, there are real-world limitations to bear in mind.
Many preanalytical variables will influence the success of detection, including, for example, the proximity and direct accessibility of the tumor to the CSF space. Tumors that occupy the leptomeninges, the pia, and the subependymal spaces, for example, offer direct contact, she says, while tumors that are more intraparenchymal or otherwise walled off and circumscribed are less likely to yield useful CSF samples.
The contraindications for obtaining CSF are relatively few. “However, that’s a discussion that needs to happen with the clinical team, to understand what’s safe and how much sample you think you’re going to get.” Volume is a key preanalytical variable, and she recommends, in general, obtaining at least 2 to 3 mL. Moreover, it is important to have a sense of where this testing fits in with the overall timeline of patient care: Will they be having surgery? Is surgery even an option? Is tissue going to be forthcoming? “These are the types of discussions we’re always having.”