Plasma Genotyping Platforms: Strengths and Limitations

Transcript:

Geoffrey R. Oxnard, MD: When I think about 3 broad categories of current liquid biopsy tests, those are the simple allele-specific PCR tests, the digital PCR tests, and the NGS panels. The best validated assay is the cobas allele-specific PCR assay, and that has now received FDA approval for looking for EGFR mutations at diagnosis when you don’t have tissue and looking for EGFR resistance mutations like T790M when you don’t have an opportunity to do a biopsy.

Now, this is great. The evidence that pharmacology plus biotechnology can come together and get one of these new technologies approved by the FDA is really important. What we know is the characteristics of this assay: it’s reliable, it can find EGFR mutations, but the sensitivity is not perfect. The sensitivity may be in the range of 70%. It will miss some cases, but it is an established and validated option.

I personally prefer digital PCR assays to the allele-specific PCR. And those digital PCR assays are the Bio-Rad assay—the ddPCR platform, Biodesix is offering this—or the BEAMing platform offered by Sysmex Inostics. These look for EGFR mutations like exon 19 deletion, L858R, T790M, and KRAS mutations. And they quantify how much they find. So, it’s a really rigorous way of measuring how much EGFR mutation you’re seeing.

For me, that gives me some more insight into the character of the cancer. If I’m seeing a low level, 0.5%, of EGFR mutations, it’s a less aggressive cancer. If I’m seeing 50% allele fraction floating around in the blood, that’s a more aggressive cancer—really, really dangerous. And indeed, I can get a sense of response as well, right? If a mutation has dropped 100-fold, and I’m following a patient on treatment, I’m not quite sure what’s going on, but the blood test tells me there’s a tremendous decrease. That gives me confidence in my treatment being effective. And so, I use a ddPCR assay that we have launched at our local hospital. We validated ours, I just can’t comment that every assay out there has been as validated. This is the catch, because we know FDA-approved assays have completed exceptional validation, and we can’t be necessarily as sure that the commercially available alternatives have completed the same validation.

The thing we should be careful about are low-level positives. With low levels below 0.5%, perhaps, there can be some noise. When you’re looking for tiny levels of these mutations, there is a risk that an assay will find a false-positive. And so, across all these assays, low-level positives need to be taken with the grain of salt.

The last category of assays are the next generation sequencing assays. These assays look for those recurring EGFR mutations, look for recurring KRAS and BRAF mutations, but also sequence large parts of the gene so confined. Rare variants can find a complex combination of mutations perhaps you didn’t expect: you’re not only finding the L858R, but you’re also finding the rare exon 18 mutation in EGFR and the exon 20 insertion. You’re not only finding T790M, but you’re also finding rare C797S and other EGFR resistance mutations.

Then you have these other panels add other genes in, like ALK and ROS and MET and HER2. You get a lot more information with 1 test. The downside I see with the panel-based assays is that they take weeks to come back, not days. And so, if you have a patient where you need an answer, T790M positive or not, in days because they are sick, the quickest way to get that is to send a ddPCR assay, a digital PCR approach, that can turn around quickly. But if you have a patient with whom you have a little more time flexibility and you can wait a couple of weeks, that’s a space where an NGS assay can be used.

This is not new to us. We’ve been doing this for our patients over the years figuring out which diagnostic tests gives us the right turnaround time. When can we wait, and when can we not? So, this is just one more set of tests we need to build into our armamentarium as we try to pick targeted therapies.

Transcript Edited for Clarity