Liquid Biopsies and NSCLC
Genomic analysis of plasma cell free DNA (cfDNA) using liquid biopsies has been adopted widely in academic cancer centers. (It has not necessarily been adopted as widely in community settings). Liquid biopsies can be used for convenient genotyping of advanced NSCLC.
Indeed, the rapid uptake of this novel diagnostic (liquid biopsies) as part of the standard of care reflects the compelling nature of such a convenient molecular assay. Intuitively, many are now asking how these blood tests can be used for guiding cancer care once a treatment decision has been made.Geoffrey R. Oxnard, MD
For example, it would be valuable if we could use these liquid biopsies to monitor treatment outcomes in the same way that serum tumor markers are used for some cancer types. In contrast to proteins found in serum, plasma cfDNA has a short half-life; therefore, levels can change quickly. We and others have shown that mutation levels in plasma cfDNA can decrease dramatically during treatment and can increase in advance of disease progression. These dynamics can be detected either with focused assays like droplet digital polymerase chain reaction (ddPCR) or with multigene assays like next-generation sequencing.
Despite these observations, some major challenges remain before treatment monitoring using plasma cfDNA can be adopted clinically.
How Best to Quantify Mutation Levels with Liquid Biopsies and NSCLC
Early assays measured the absolute concentration of mutant DNA in plasma, quantified as copies/mL. With the emergence of NGS assays, it is now more common to quantify the relative prevalence of any given mutation compared with all mutant and wild-type sequencing reads, quantified as the allelic fraction (AF), a scale that most assays can use. AF calculations are highly correlated between NGS and ddPCR assays, but it is unknown whether benign processes could spontaneously alter the calculation of AF. For example, it is possible that a change in the levels of wild-type DNA (for example, an acute infection leading to white blood cell degranulation) could result in fluctuations in mutation AF without any corresponding change in tumor burden.
How to Track Multiple Variants
Earlier assays like ddPCR follow the single key driver mutation in plasma, whereas many newer assays use NGS, which can detect multiple mutations. Some of these mutations are cancer drivers, and some are subclonal mutations or resistance mutations. Furthermore, some mutations in cfDNA are germline; others are somatic mutations derived from clonal hematopoiesis and not from the tumor. How are these multiple variants best handled? Some investigators have tracked just the highest AF variant, although it may not be the driver mutation. Others have averaged the level of all variants at each timepoint. Consensus has not been clearly established.
How Much Change Is Meaningful?
We know that complete clearance of mutations from plasma cfDNA is a good prognostic sign, but it is unclear what lesser magnitude of response is the best marker of treatment effect. There are decades of historical precedent supporting specific objective criteria for response and progression on tumor imaging. No such literature exists yet for response in plasma cfDNA. Some degree of change is likely due to random variation, unrelated to treatment effect. This must be robustly quantified so that clinicians can know what degree of change instead might be clinically meaningful. It would be unfortunate to change therapy based on variations in a blood test that are, in fact, due to assay artifact or extrinsic, nonmalignant conditions.
What Turnaround Time Is Needed?
Many have shown that plasma cfDNA analysis is much faster than getting a biopsy; however, cfDNA testing remains slower than imaging or serum tumor markers. These tests involve multiple steps—spinning the plasma, extracting DNA, genomic analysis, and test interpretation—which can take days to weeks depending on the assay. This may be too slow for routine use because decisions about whether to continue treatment or to switch to a different regimen are usually made in a day or two based on imaging studies. Faster assays for cfDNA analysis are needed and are in development. In the meantime, turnaround time must be considered when building this testing into clinical decision making.
What Cost Will Be Scalable
Genomic analysis of plasma cfDNA can cost hundreds or thousands of dollars per specimen. This is likely too expensive for routine use every few months during therapy, unless clear clinical utility is established. For this reason, development of the cfDNA analysis as a monitoring assay should focus on key decision points where the cost–benefit ratio is more favorable. Of course, if cost decreases, it could be possible to run these tests more routinely.
Cancer monitoring using plasma cfDNA is compelling, but there is much work to be done.
I currently do not use plasma cfDNA analysis on its own to assess response or progression, but at times I will use it in combination with imaging and clinical evaluation to help understand if a treatment is failing. Indeed, this also is how serum tumor markers are used—not on their own, but as a complement to imaging to understand the clinical picture.Geoffrey R. Oxnard, MD
If scans seem mostly stable without response but symptoms are worsening, I might send plasma genotyping to assess for resistance. If I see high levels of the driver mutation, this is concerning because progression is likely brewing and makes me favor a change in treatment. However, if scans and symptoms are stable, a blood test on its own, especially one showing low AF mutations, is not enough to divert me from a treatment that is otherwise effective in the palliation of metastatic cancer.