Johann de Bono: And my privilege to introduce Kim Chi from Vancouver, who's going to talk about a really key area, how to read a genomic testing report. Kim.

Kim Chi: Thank you, Johann. And thank you, Silke and Aurelius, for inviting me to another fantastic meeting. I feel like I'm emerging from the COVID hibernation. It's great to see all the colleagues and have further discussions around science.

And thank you for asking me to talk about how to read a genomics report, which is at its fundamental level, relatively a mundane task. You just read the report. But I think of course behind it is a lot of complexity, which I think we all need to appreciate. And this is why I'm being asked to specifically address this topic.

These are my disclosures. And then I also wanted to thank Dr. Tracy Tucker and Dr. Intan Schrader. They're with our genomics... our medical genetics department at my institution. And I also wanted to highlight them just to say that this is... We're expanding our multidisciplinary prostate cancer care team. We've got nuclear medicine. We've got pathologists. But more and more as we do more genomic testing for our patients, more widespread and also mainstreamed where the oncologists order the genetic testing and are responsible for the test, we need the assistance of our colleagues.

And it is important because as oncologists ordering this test, we actually admit to really limited formal training in the interpretation of tumor genomic profiling results and the communication of those risks, especially the hereditary cancer risks. And so again, this is why I think Silke and Aurelius chose this as a topic to discuss.

So, what I've listed here is what do I need to know? What do I want to get out of a genomics report, specifically a tumor testing report. So, first off we have to understand what are the genes and the regions covered within those genes and the types of mutations that are detectable in the limits of those detection. Because it has to be relevant for the cancer that you're treating, in this case prostate cancer. So we want to make sure those prostate cancer-relevant genes are covered in the assay that you're ordering. And not all assays are going to detect all alterations. Most will, of course, report on mutations. But as we heard earlier, rearrangements, copy number variants may not be reported by the assay or the assay pipelines.

Then I want to know what the mutation status is of the genes of interest, not just whether they're mutated, but also whether they're normal or not. And this was alluded to earlier in Colin Pritchard's talk around variant allele frequency. So the variant allele frequency is, in simple terms, the number of times a variant is seen versus the number of times that region was sequenced. And in part reflects how many cancer cells within that biopsy that's being analyzed carry that gene. Why is this important? Because if you have a very high variant allele frequency, this suggests that it's truncal, that it's driver, and therefore a relevant therapeutic target. If you have something that has a very low variant allele frequency compared to the other VFs that you're seeing in the mutation, it suggests it's a passenger or it's subclonal and therefore not as relevant as a therapeutic target. And of course, I want to also see from the report whether, not assays do this, whether there is signatures of microsatellite instability that I can look at and total mutational burden, because this, of course, identifies patients for eligibility for immune checkpoint blockade.

So, I'm going to walk through a FoundationOne CDx report just to show this as an example. And of course, you can't really see this in the audience, I'm sure. But basically, you can go through the gene list and there are relevant genes here for prostate cancer, including androgen receptor, mismatch repair genes, and genes associated with homologous for combination repair. Specifically, they also list the genes where they're looking at introns, which are important. You need to have introns that are sequenced in order to identify rearrangements, which is of course important for BRCA2 identifying alterations there. And then they have genomic signatures. And you heard of this before at Colin's talk around loss of heterozygosity, microsatellite status, and tumor mutational burden.

And then I'm just going to walk through what a report looks like. And FoundationOne provides a handy dandy guide to look at that. First of all, they show the genomic findings that have FDA-approved companion diagnostic claims. So BRCA2 would be listed here. And then they list other important alterations and biomarkers that are identified. They also include pertinent negatives. And then helping you to action on these genomic alterations, they list therapies of clinical benefits, levels of evidence, and clinical trials.

And then finally, in the back pages, you get the details around each of the mutations. In this case, we see BRCA2 and it includes also the type of alteration they're seeing in BRCA2 and, importantly as I mentioned, the variant allele frequency. So you do have to go into the back to find that.

So, what are the limitations? So really we want to know about the false negatives. And as I mentioned earlier, it's really important to understand that not seeing an alteration doesn't mean that there isn't one there. So panels will only test for certain types of variants. And we really should know what is covered if you're ordering a test. As an example, one of the local validated accredited tests and my institution doesn't report BRCA2 deletions. And as we heard earlier from Johann, those are the patients that benefit the most from PARP inhibitors. And so of course, I'm not going to want to use that panel or that assay for my patients.

Copy number changes in structural rearrangements may not be detected particularly with lower tumor cellularity, less than 30%. And although there's a pathologic review before a sequencing is done, this is an estimate. And so even with a pathologic review, there can still be issues with tumor cellularity. And as was discussed earlier, tumor mutational burden calculations differ from different assays and small panels will not be accurate. So again, you have to know what the limitations of your assay is in order to interpret the results.

Another issue is variance of unknown clinical significance. So usually what you get in a report are the Tier I and the Tier II alterations. These are variants of strong clinical significance or potential clinical significance. Variants of unknown clinical significance are known or presumed to alter the normal gene function. However, there's no evidence that they actually have any clinical consequence. And these are going to be more and more picked up. As technology advances, we've got larger gene panels. We're going to go to whole genome or whole exome sequencing. And I suspect in the next five years, we're going to be doing a lot more whole genome sequencing.

And therefore, the likelihood of finding these VUSs is going to increase and it's as high as one in three already in some series. The key issue though is that not all labs report these. They're subject to reclassification. So what is a VUS now may not be a VUS later. And they're also potentially classified differently in different labs. So this is something that, again, you have to work with your medical genetics team, and your molecular pathologists, to understand how this is evolving over time.

So, you got your test, but this is a tumor test. And we don't know whether it's acquired or of germline of origin. And so an important consideration is that if you do detect a variant from your tumor testing, that you may need to order next germline testing. And again, mainline mainstreaming as oncologists, we're ordering this test. It's our responsibility to follow up on this.

So, when do we do this? We should do this when there is a variant of stronger potential clinical significance, that is those class one and two variants, in a known cancer susceptibility gene. And so this may be germline pathogenic or likely pathogenic, especially if it's a known founder mutation like in BRCA2. And then there are recommendations for which genes you need to be looking out for. And that's what's in this box. These are ESMO guidelines showing if you see an alteration in these genes that should in your tumor testing, this should trigger germline testing. If you see MSI high, that should also trigger germline testing for mismatch repair gene mutations.

Now often quoted is if you see a variant allele frequency of greater than 50%, that this suggests germline and this could be a germline mutation. However, this is not a good rule of thumb. You should not use this because you can get a VAF of greater than 50% in somatic. Let's say you have copy number changes, high tumor cellularity, loss of heterozygosity. And you can get VAFs of less than 40% that are germline.

Finally, regardless of the tumor testing result, if there's a strong individual or family history that would lead to normally germline testing, you should do it anyhow. And why is that? That's because if you get large chromosomal changes in the tumor that may mask a germline mutation. And as well, certain panels will not cover the relevant germline mutations. A good example of this is HOXB13. HOXB13 has no therapeutic implications. It's not covered in many panels. But it's actually a prostate cancer hereditary gene and therefore has implications for the patient's family.

So, for the last half of my talk... I'm not sure how much time I have left... Hopefully enough. I was going to talk about cell free DNA and circulating tumor DNA, because increasingly we are using this as a source to genomically profile our patients, because it's easy, minimally invasive, and we get it. So cell free DNA. We all have cell-free DNA in our blood. It comes from cells that are apoptosing, mostly from white blood cells. In patients with cancer, a proportion comes from the cancer, and that's what we call circulating tumor DNA.

And there's a bit of a discussion about ctDNA fraction so on. So ctDNA fraction is an estimate of the proportion of cell free DNA that is derived from the cancer. So in tumor testing, you can see that from the pathology slides. But we actually don't know ctDNA fraction until we actually do the sequencing analysis. And it's highly variable. So in our series of 201 patients, this one series of ours, 40% of patients had a ctDNA fraction that was undetectable, so less than 1%. This is equivalent to a variant allele fraction of 0.5%, which is about the limit of detection of most large panel assays doing ctDNA testing.

So right off the bat, 40% of patients, not enough ctDNA. The median was 17% ranged between one and 89%, which is 90% ctDNA in the blood. That's pretty cool, but that's another discussion. ctDNA fraction is associated with metastatic burden. So for example, liver metastases, those patients have high ctDNA. Lymph nodes, low ctDNA, high LFOS, high LDH, and so on. So it's all correlated with metastatic burden and perhaps also proliferation of the disease.

But ctDNA fraction is also dynamic. So this is unpublished data from a study that's led by Drs. Van Erp and Mehra from the Netherlands and they allowed me to show this data where they took ctDNA samples four weeks after starting an AR pathway inhibitor. And as you can see, the majority of patients, or a large proportion, actually go down to a 0% ctDNA four weeks after treatment. And the majority of patients all have significant declines.

So why is this important? Well in the context of this talk, ctDNA fraction limits your ability to detect alterations. So with the ctDNA fraction that is typical, we're probably picking up those mutations. Just give me one more minute. And then for the other alterations, and we heard this before, those, for example, biallelic loss, you need a ctDNA fraction that's 30% or higher. And as I showed earlier, only the minority of patients have that kind of ctDNA. So if you're going to do a ctDNA test, you have to make sure your patient is actually appropriate for that ctDNA. If they're on an AR pathway inhibitor and their PSA's going down, don't do a ctDNA test. And then the reports, as we said, we need to know what the report should determine, tell you what the ctDNA fraction is so you can interpret the results.

Further example of this. This is from the FoundationOne liquid CDX assay for BRCA2. Limit of detection, you need a 48% tumor fraction or ctDNA fraction. This is from the PROFOUND study showing how ctDNA can miss those homozygous deletions. And this is an important component of patients. It does detect other variants, but this goes back to the caution and as already discussed before around false positives and CHIP. And this is not insignificant. So this was already shown by Colin. So 19% of patients in this series had CHIP interference close, and of those 10%, half, could have been prescribed Olaparib, which would've been based on a false positive test, because you're misattributing that mutation to the patient's cancer.

So last slide summary, tumor genomics is now a standard part of the workup for patients with advanced prostate cancer. And we need to bring in and build our team with medical genetics and molecular pathologists. We need to understand the genomics report and its limitations in order to bring this into action. Germline testing needs to be considered depending on the result, but also independent of the result if there's a strong personal or family history. And cell free DNA analysis has limitations. It's a great technology, but we have to understand its limitations and use it appropriately as we move forward.

Thanks very much.