All right, we'll move forward with Rohan Ramakrishna, who is the winner of the Proust Research Award for his abstract entitled Whole Exome and Targeted Sequencing of Adult Infiltrating Astrocytomas Experience at a Single Institution. Thanks again for having me and thanks again to the award committee for honoring the work. I'm going to present our experience with whole exome sequencing for adult astrocytomas. This, of course, is a very large collaborative effort between us and the Englander Institute for Precision Medicine. I have no disclosures. So this talk is mostly about astrocytomas, not oligodendrogliomas, defined by 1P19Q co-deletion. As we know, histologic diagnostics for astrocytomas include a molecular heterogeneous group of tumors composing of diffuse astrocytomas, anaplastic lesions, and glioblastoma. And in terms of how we've thought of this historically, you know, we're all taught as residents of, you know, radiologically there's a progression with no contrast enhancement followed by contrast enhancement related to blood-brain barrier disruption, histologically increasing nuclear atypia, cellularity, microvascular proliferation, and eventually necrosis. And with these different buckets, you know, we get to see different survival times and different associations with molecular markers. In 2006, the TCGA reported this, you know, genomic landscape of gliomas, and that really set the stage for the more contemporary understanding for how we think of these things. And of course, the major pathways that are altered relate to receptor tyrosine kinases, P53 signaling, and the retinoblastoma pathway. And then there was further work done to try and really understand how the molecular diagnostics that we were seeing related to known histologic buckets and as well as transcriptomic patterns. So for example, we know that IDH, for example, tracks fairly well with, you know, so-called secondary glioblastoma and lower-grade lesions, but it also tracks well with transcriptomic patterns like the proneural subtype of glioblastoma, which again has a slightly better prognosis relative to other transcriptomal subtypes. And then great work by Eklipasau from the Mayo Group, and then this was a similar work was a companion article in the New England Journal of Medicine in 2015 looked at how can we rationally describe all these molecular alterations with a few simple signatures such as IDH, TIRT, and 1p19q. And what they were able to find is that you could indeed do that. I apologize if this doesn't project well, but basically they were very successful in class, subcategorizing these groups based on just these three mutations as defined by IDH, 1p19q, and TIRT. They basically segregated the different alterations and mutations that you see in transcriptomic patterns and were able to define things based on being triple negative, triple positive, or none, and get a much better sense of prognosis, particularly as it pertains to low-grade gliomas. So as you can see for lower-grade glioma, when you look at TIRT and IDH by itself, whether they're triple positive, they tend to have much better outcomes relative to just a TIRT mutation only. The segregation is a little less dramatic for high-grade lesions. This Kaplan-Meier curve on the right reflects the companion TCGA article, and you can see that oligodendrogliomas, again defined by 1p19q deletion, and low-grade gliomas with IDH mutation have much better prognostics than GBMs and anaplastic astrocytomas that are wild-type for IDH. But ultimately, that background I just gave you is meant to illustrate that while we have an understanding of the genomic landscape of gliomas, which of course is important for understanding how these behave, it doesn't give us a sense of how to necessarily individualize therapy, though we have things like IDH inhibitors coming down the pike. And we know for oligos, for example, that PCV and XRT are the way to go after surgery. Precision oncology for astrocytomas is still a moving target. So our goals for this study were to start and establish a precision medicine pipeline for patients with gliomas, compare the results against commercially available platforms like Foundation Medicine, and assess the value of whole exome sequencing over other platforms. So our age distribution of our cohort was, as you'd expect, with about 91 patients. Median age was in the late 50s. And just by using IDH as a marker or segregator, you know, most of our GBMs were wild-type, as you might expect, with a fairly even distribution of IDH mutation in anaplastic and lower-grade lesions. When you subclassify by IDH and TERT status, you can see that 10 percent had neither. And when you look at the cohort by commonly identified mutations, you know, you see this representation. What's interesting to me is that 38 percent didn't have any of these commonly reported mutations for astrocytomas like IDH, PDGFR, and EGFR. This, again, is a list of the genes that were most frequently altered, and the usual suspects were mutated, including missing mutations in IDH, ATRX inactivations through a variety of mechanisms, EGFR and PDGFR amplifications. But then our next goal was to see, could we identify – and that's all known, of course – but our next goal was to see, could we identify mutations in genes that were recurrent between cases that might be relevant to patient outcomes or treatment response? So we looked at genes that were less frequent but mutated in more than two cases and came up with this list, and most of them are not really described in cancer on any high level. And I'll go through a few of them. TSP, for example, is involved in desmosomes, which I'll talk about a little bit later, which are part of cell adhesion, also involved in differentiation and apoptosis, apparently, MTUS2, which is a microtubule protein, and MXRA5, which is associated with matrix remodeling. So if you look, again, at this expression plot of DSP in the TCGA data set, which is what we validated this against, you can see there's not really any specific method of – or range of expression for the different histologies, but that's not the point of precision medicine. That's the point – the point is to find it for this individual patient. So if you look at the – again, in the low-grade glioma data set from the TCGA, high DSP expression is associated with worse – with better outcome. And in lung cancer, it's been reported that decreased expression of this gene is associated with worse prognosis, so there seems to be some correlation in that regard. We then looked at, again, these various other genes that we found that were currently altered, and with MTUS and MXRA5, we were able to see that there was a survival benefit to either higher or lower expression, even in GBM for MXRA5. And so that begs the question, in terms of our studies where you have, you know, treatment responses, et cetera, related to new therapies, how much can we say are from the therapies versus some of these recurrent alterations, which may be low frequency, 3 to 5 percent, that may still impact survival in some of these cases. These are some of the other genes. QRICH2 is on chromosome 17, not really known to have a function. Von Willebrand factor – again, this doesn't seem that significant, but nonetheless, this is a – this gene is also found to be prognostic in lung cancer. And, you know, the other interesting aspect is if you take all these 15 genes, there are alternately 23 percent of TCGA GBM cases and 25 percent of low-grade glioma cases, just as if you just randomly pick 15 percent – 15 genes, you would never get that higher percentage. So I think it's approaching, you know, something valid to talk about and further study and validate. This is just a lolly plot showing that a lot of these mutations are probably loss of function because we don't really see them clustering. And these are some other genes that were a little less common that we found. SMARCA4, again, is commonly seen in ovarian and other cancers. Our second, just for time purposes, aim of this work was to try and figure out how often do we see germline changes in patients with malignancies. You know, when you look – when you get foundation medicine reports, for example, which are great, they don't report the germline. And so, for example, if you got that report and showed an ATM mutation and thought, oh, let's try a PARP inhibitor, that would not actually be potentially relevant for that patient because it's a germline somatic mutation. So in our cohort, you know, up to 50 percent of patients had mutations in cancer-related genes at the germline level. This just shows that whole exome sequencing is fairly similar in terms of the alterations that are found compared to foundation sequencing. The one place where whole exome, of course, falls short is with TERT, promoter methylation status, but you can, you know, modify your pipeline to detect that. So in conclusion, the majority of the most common recurrent genetic alterations are reliably picked up by whole exome sequencing. We're able to find some of these lower frequency alterations, and I think the next step is to validate those in the lab. Pipeline assessment is really important because that's going to determine how you interpret the mutations you find in the tumor. I think I didn't talk much about this, but data quality is incredibly important since all of these pipelines are homegrown for the most part, and so you really need to understand mapping errors versus true mutations, and then finally, again, to discover the functional significance of some of these mutations. So thank you very much. Thank you.

whole exome sequencing

adult infiltrating astrocytomas

molecular markers

genomic landscape

gliomas