Identification of Patients for Germline Testing

Mark Robson, MD: But can we talk a little bit more about this? How do you decide who to send for germline testing? Dr Ross and Dr Henson sort of fixed the problem because they were doing simultaneous tumor normal. But if you do tumor-only, and you find a variation, how do you decide who to send for confirmation? How is that?

Dana Farengo Clark, MS, LCGC: It's been my work for 4 years, so I will say the process continues to be refined. When we started, there wasn't much out there on any sort of algorithm We don't love to use allele frequency because obviously, that has its own problems. You can get ones that are very low that confirm. You can get ones that look around 50% that don't.We kind of threw that out the window. We looked at TCGA [The Cancer Genome Atlas] data of the genes that seem to have a high germline to somatic ratio, so we didn't want things like p53 that you find in every tumor. We also used a lot of the current guidelines put out by ACMG [American College of Medical Genetics and Genomics]. There was a paper by ESMO [European Society for Medical Oncology] last year.We do try to look at what everybody else is doing in terms of what the standards are. We kind of basically decided that the [Penn Center for Personalized Diagnostics] CPD would send a notification email to me and a notification email to the ordering provider that something that is potentially germline may have been found. It's clunky, but that's what we've been doing. And so, the obvious ones; anything BRCA was an automatic referral. Any Lynch was an automatic referral. And then we really tried to use some of the, as kind of a playing ground, some of the ones that have NCCN [National Comprehensive Cancer Network] guidelines that we thought, "Someone with a RAD51 CD might have their ovaries out," so something that was clinically actionable in our minds. And of course, that's coming from a clinical genetic standpoint as actionability. So that's kind of how we started. We just reconvened, and now we're adding things that, phenotypically, you would expect to see in the germline, things like BAP1. And we really just are using our clinical judgment a little bit, and that's kind of how we've bumbled through it a little bit. But we keep adding genes as things evolve in terms of literature. And again, the high-risk genes, such as Lynch syndrome and BRCA, those seem to confirm much more often than not the majority of the time, so we're continuing to look at those. Some of them come up so infrequently, but things like p53 now, per ESMO, we're only doing people with a diagnosis under 30 years of age. So, there's a lot more caveats now, I would say it's not easy and doing what you guys are doing seems a little bit less messy. But we're trying to follow published guidelines and then kind of go with our clinical gut.

Mark Robson, MD: It's a little less messy in some ways, but we'll talk about process and flow later on. It introduces complexities there as well.I'm curious, either Dr Ross or Dr Henson, if you do tumor-only sequencing, and you don't find a pathogenic variant, let's just say in BRCA1 or BRCA2, do you feel like that's good enough that they're off the table?

Ashley Ross, MD, PhD: That is a difficult question. For me, I would say that practically, my heart would say no, and I'm mostly doing prostate cancer where germline testing for the high-risk in metastatic disease is endorsed and often covered. When I see no pathogenic variant, I still sometimes push for the actual dedicated germline sequencing. I think that the NGS [next-generation sequencing] reads a 600-plus gene panel. I sometimes don't trust that they're actually identifying all the variants are there. And then I can't remember exact specifics, but I think there's been a couple times where there has been something in the germline that has been missed. There are tricky genes, pNS2, some other ones that are tricky to identify by standardized NGS, and so I guess it's not enough for me. If there's a strong indication for germline, I'm going to a dedicated one. As Dr Henson said, X-Gene and GeneDX makes that kind of easy, but again, I'm the least genetic savvy on this call, so I'll see what Dr Henson says. I don't know if I'm doing it the right way or not, but I think I have to be belt and suspenders when it's indicated.

John Henson, MD: For me, I think if I get a 500-, 600-gene panel back on NGS from the tumor, and there's actually nothing there that fits the patient's expected profile of potential alterations, I do relax a little bit. There are weird things that can happen; for example, in a treated tumor, there's a small percentage of cases which will have revertant mutations, which is a really fascinating phenomenon where the initial driver event goes away because you're successfully treating those cells.There are things that can happen, but I still like the follow-up of the germline. It just makes me feel more comfortable, particularly if it's a tumor where I'm really anxious about the presence of a germline change. We had a case today in clinic where there was a large tumor panel that was completely negative, and we looked at the family history. It was very quiet. Big family. Not much to actually look at. And we actually accepted that as evidence that we weren't going to do germline. But most of the time, we do it.

Ashley Ross, MD, PhD:It always strikes me when I get these reports. There's a lot of data that I'm either not looking for or is maybe not being brought up to the forefront as much. For this NGS on its own, for example, sometimes, they'll report out the low-coverage amplicons from that particular sequencing run. And then I never, as a clinician, take the time to look at, "Well, is one of my 46 germline genes there? And maybe I had lower coverage than I wanted, and that's why." And so, the solution to some of these problems is actually doing the dedicated panel so you don't have to do that. But we were talking a little bit before about nonsense-mediated [mRNA] decay and these VUS [variants of uncertain significance], and are they important or not? There's a lot of computational stuff that's also missing that could be added on top, even for categorizing the VUS, not just from genetic data banks, but also just from biology. I'd love to know if they're doing an RNA sequence at the same time, if my mutation was expressed or if it was a nonsense? I'd love to know, where is this mutation? If it's the amino acid level, where is it? Is it in the active site of the enzyme? Does it affect tertiary, quaternary structure? Or is it just nothing? I think there's a lot of rich data. I know it's off topic, but just grabbing the soapbox, I would say the solution to some of these problems is adding in the more testing. But I think that there's so much rich data that is not at the forefront of someone who is not a geneticist like you guys who's just trying to go through, with the patient, the report. And it's a blind spot for us sometimes.

Dr. Henson: The RNA is very powerful. I think if you get normal RNA on the tumor, both in terms of expression levels and sequence, you're safer in that case, unless there's something really weird going on in the tumor. But germline testing is still, I think, very, very valuable.

Mark Robson, MD:We have a fairly wide experience with our simultaneous tumor-normal platform, and I don't think we've published this yet. Dana and I were talking about this yesterday. But we took a look at our patients who had germline alterations and tried to figure out how many of them were identifiable in the tumor sequencing. And it's about a 10% miss rate, and most of it is due to usually homozygous deletion.You have LOH [loss of heterozygosity], and then you lose the mutant allele, or you delete out the variant, so particularly, when you're finding hom dels [Homozygous deletion]. And BRCA2 actually is a gene that has a reasonable number of hom dels. I think the take-home message for those of you who are listening is that tumor profiling alone does not exclude germline variation. If you have a normal tumor profile, or if you don't see a BRCA mutation or some other mutation on the tumor, it doesn't necessarily mean it's not there. And I do think it's worthwhile considering dedicated germline testing, either as part of a package or as a separate standalone test.

Dana Farengo Clark, MS, LCGC:Sorry, Mark. The last point in our bullet that we send to everybody in terms of tumor-only sequencing is, "If it smells like a mutation, send them anyway."I think sometimes, we have to go back to basics. And if it's a canonical tumor that looks suspicious, there's maybe family history at all, we always tell people, just use a little judgment as well. And I think that sometimes goes out the window. But basics do work.

Mark Robson, MD: The 30-year-old triple-negative patients with a family history of breast history, and they don't have a BRCA1 mutation on their tumor, you might want to send them anyway. It's just common sense.

Transcript Edited for Clarity