So for this last lecture, we'll be talking about solid tumors. And ultimately, a couple of things that we want to go over the course of this lecture. One is just basics. This is probably late in the game at this point, but basics of cancer definitions, overview and review of some of the mechanisms of tumor genesis with a set of genetics point in mind. We'll talk about some of the classic solid tumor set of genetics, high-level overview of the main findings. And then we'll end with sort of actionable alterations and sort of what findings can we currently use from a therapeutic perspective. And did everyone get the email? So this is going to be a two-hour lecture, if they decided to extend. No. No. All right. So cancer, obviously, I think everyone's familiar at this point. Again, it's a multi-step process in which acquired aberrations lead to uncontrolled cell growth. And probably the best example of this is initially the work that Bert Vogelstein did in the Vogelgram. The acquisition of mutation initially changes leads to a selective advantage. And specifically, these APC mutations lead into these polyps, ultimately the acquisition of further mutations, in this case, in the context of colorectal cancer, which ultimately then is subsequently followed on with chromosomal aberrations. But ultimately, I think the point to consider here, again, is that sort of progressive model from carcinogenesis, or tumorigenesis, I should say. And obviously, in this process, again, it's through either the elimination or the inactivation of tumor suppressor genes or the activation of oncogenes. So again, find a way to basically deregulate the normal cell cycle controls, either cut the brakes or you step on the gas. And between those two models, again, you have an ability to sort of accelerate and make sure that cancer is non-responsive to the normal environmental and cellular cues. Now interesting, how many people have heard of Boveri before? There's like two people who still have the energy to put their hands up. So one of the things I think that's fascinating about him as an individual is in terms of his impact specifically and first really postulating on the impact of chromosomal aberrations in disease. And this is actually work that came from studying nematodes and was looking at basically fertilization. It was examining an event in which an egg was fertilized by two sperms. And there was a formation of these multipoles that formed here. So rather than have just two poles and two daughter cells through replication, these four poles formed. And ultimately, it led to this unequal or improper segregation of chromosomes. And what he evaluated or what he postulated is that through the unequal sort of sharing of chromosomes, he felt that this was leading to changes to growth. And so he first thought of basically the notion that chromosomes could basically lead to the development of cancer. And certainly, this is a theory that I think that now we're seeing hold true much, much later on. So cancer is a disease of the genome. And at this point in time, again, we know that there's many genomic mechanisms that have been cited, genetic mechanisms, whatever you want to call it, that can lead to basically changes in the genome that can basically accelerate or lead to the development of cancer. And so obviously, we have things like aneuploidy. So again, gains or losses of chromosome. Polyploidy, the entire chromosome or the entire genome complement doubles, by example. And then events like reciprocal translocations. And I've led there that, again, that can either lead to gene fusions. So again, the hyperactivation, bringing together two genes, we talked about enhancer hijacking, we talked about in-frame gene fusions. But also, reciprocal translocations can also lead to inactivation. The translocation occurs at a location that causes a disruption of a gene of interest. And then similarly, gene amplifications. And it's probably worth taking a moment to talk about homogeneously staining regions and double-minute. So who's heard those terms? Awesome. Perfect. And so in general, again, this is when we think of an amplification event if we're looking by fish, or let's say, by array or by sequencing. We're doing a sequence panel. We're looking at 450 genes. We see one gene that's way up in terms of its overall copy number. Well, all we know is the overall copy number from that type of analysis, from array or for sequencing. We don't know where it is, typically, in the genome. And certainly, what we've learned from traditional cytogenetic studies originally is, again, amplifications can either happen in situ. So again, the same gene reinserts itself again, again, again, again, and amplifies into a particular region. And that region, therefore, stains homogeneously and produces this sort of characteristic appearance. Or often can lead to the formation of these double-minutes. And actually, this is a whole interesting aside that, for sure, in the interest of time, we'll not get into. But why would it be advantageous to be double-minutes? So who doesn't know about that? So we have HSRs and WUs. The end product is the same. Lots of copies of a gene. Through the WU. So why would I want to be a double-minute? Any thoughts? I can wait. So it's actually, it has to do with, if you're trying to, so there's been two sort of models in the postulate. One is that double-minutes form through sort of the unstable nature of HSRs. They can fragment and shatter out. And so that's one thing. Formation double-minutes can resolve from HSRs. And what then happens is that they don't segregate equally. And so suddenly, you can have unequal segregation in terms of how many double-minutes go to a single cell. That can lead to then further selective advantage of that cell type. But that in and of itself following subsequent cell divisions isn't stable. So what's also been shown more recently is that they can reintegrate back into the genome. It's basically a way in which the cancer can basically cause these massive events in terms of acceleration, its growth, and development in these cases. So these really complex mechanisms that we've appreciated that now, through the combination of both visualization and empirical observation from cytogenetics and now more bioinformatics studies, sort of tracking clones and evaluating breakpoints, we can sort of understand a little bit more sort of the nature of why we're seeing the changes we're seeing. Yeah. So for those of you, there's some sort of scratching. We either have a huge rat on the other side of this wall or we're doing some painting. But some scratching in case we hear that in the background. So cytogenetics of solid tumors. So I think that while there was incredible amount of success in hematologic malignancies, and again, we talked about in the 60s and 70s and first identifying path-hormonic characteristic changes, there was actually a challenging road before ultimately we were able to develop techniques that sufficiently allowed solid tumors to basically grow in culture. And again, rather than grow in suspension, these typically grow in substrate. And for those of you who've done cell culture analysis in primary tumors, they're notoriously difficult. Where initially a lot of the success came from was actually through the growth of sarcomas. And those tumors typically have very aggressive phenotypes. And so not surprisingly, they're highly mitotic cells. And they were sufficiently mitotic that in vitro, we're able to capture enough of them, able to basically evaluate them, work with them, and then start to identify specific recurrent changes. So this is now no different than what happens in the identification of the 922 in CML. Initially, we find a specific disease, either a tumor type that occurs in a specific location or that has a specific histopathology or morphological appearance. And what they found is that through characterization of increasingly larger number of tumor types, they kept on saying, wait a second, we're seeing the same thing. One of the big challenges was the signal to noise. So with sarcomas, great, they grow, incredibly active. But at the same time, again, there's an incredible amount of genomic instability. And so in looking at individual tumor types, again, the challenging part, well, what's the driver? What's not? You know, what's the signal? What's the noise? But ultimately, through enough analysis of numerous tumors, again, both for, in the case of myxoid liposarcomas and for synovial sarcomas, among others, it was the identification, again, of these recurrent breakpoints initially. And so again, going back to what we talked about earlier, you have a specific morphological or histopathology diagnosis. You then find a recurrent finding. Now the ability to begin to use and sort of expand upon some of the molecular techniques that existed at the time, and then try to find out what actually is the gene? What's driving this? And so that led to the DDiT3 phosphusion in the context of myxoid liposarcomas and the SS18-SSX in the context of synovial sarcomas. We're gonna go through all these one by one. So Ewing, no, no, no. So Ewing said there's a lot, this is just sort of a slew of different sarcomas and bone tumors, and now sort of the pathognomonic rearrangements that have been identified, or the characteristic rearrangements. Again, worth noting that even within different tumor types, you have Ewing sarcoma, as well as these angiomatoid fibrous histiocytomas. Both of them have EWSR1 rearrangements. So again, different tumor types would still have, again, the same, similar drivers in terms of their pathogenesis. The same thing is for many of these, actually, sarcomas listed here, same for clear cell sarcoma and desmoid round cell tumors in general. But so again, there was a rapid expansion in this time, again, of improvements in terms of the way that tissues and solid tumors could be grown in vitro. And ultimately, that led to, again, this rapid characterization of a number of different genes of interest. Now, in the absence of a characteristic structural rearrangement, and again, that, to some extent, that's the picture-perfect thing that you're looking for. That if I wanna say what's driving this tumor, I see a characteristic rearrangement, I'm thinking, again, that's likely going to be an enhancer hijacker or an in-frame gene infusion. This is what's gonna be driving the tumor. That's ideally from a diagnostic and even potentially from a therapeutic perspective because it likely will give me a chance to identify something that maybe I can treat. Maybe there's something about this oncoprotein, oncofusion that's generated that's targetable. In the absence of that, there's still a potential of sort of using the information that we can leverage from cytogenetics initially at the time, the GTG banding analysis, and that's by looking at numerical changes. And so for diseases like renal cell carcinoma, we know that there's a number of different histologies or subgroups that exist in this disease. Clear cell that's particularly aggressive versus papillary, chromophobe, and then oncocytoma and some other more recently described events. But in the absence of having rearrangements, which is often not observed, it's not entirely absent, there is this XP1.2, these TFE3 rearranged renal cells. The vast majority of renal cell carcinomas seem to be driven from a cytogenics perspective from these copy number changes, numerical changes. So no rearrangements, just losses, gains, but it's the pattern. And so certainly that ended up being seminal in terms of the way that we evaluated these tumor types. When you look at a GTG banding analysis, when you look at metaphase spread, what's the pattern you see? So rather than focus on a specific aberration, a specific rearrangement, we're saying across the genome, is it like the pattern of odd number chromosomes? Or is it the gains? In this case, in the case of clear cells, renal cell in the clear cell type specifically is loss of three exactly ends up being the characteristic findings. And in fact, what's interesting too is that there's a gain of five. Well, the reason there's a five gain is actually oftentimes the way three is lost is through an unbalanced rearrangement. So chromosome three and five are rearranged. It results not in an oncogenic fusion, nor does it result in a direct gene inactivation, but it results in a gene dosage change. Three P is lost and three P is gained. So it's a common sort of mechanism that leads to that specific finding. And then again, you can differentiate. And again, I would say what's useful is remembering the overall patterns, not necessarily the specific chromosome numbers, but it's a clear cell, you have a three P loss, papillaries typically gains, seven, 17, et cetera. Chromophobe is notable for the losses you see. And again, so fairly clear differentiators between these various subgroups. And also cytomas is relatively bland type with one 14 and Y that's missing. So cytogenetics and specifically solid tumor, cytogenetics of solid tumors really ended up being a powerful technique. But we don't use it so much anymore for solid tumors. So why? What's the downside to doing karyotype in this cell? What would be a reason that we wanna move away? Should we move away? Pardon me? Growing. Yeah, growing, that's for sure. That's a big problem. So just are the cells dividing, problem number one. Any other problems? For sure, no one look at the next slide. Tumor loss? Yeah, tumor could be that. So, although to some extent, part of that is what's your comparator, right? So yes, that could be true if the cells are dividing slowly, but at least they're dividing, but it would then be what's your alternative? Is it array? Well, it depends on how long that takes. But that could be a concern, absolutely. Any other points? Pardon? Yeah, absolutely, normal tissue. So either the tumor doesn't grow or the normal, the reactive tissue outgrows it, absolutely. And then another concern is contamination. So all of those are pretty significant challenges. Not to mention also that there's tumor necrosis. So again, what's the quality of the specimen you're getting? Highly necrotic sample, again, the likelihood that's gonna divide in culture is certainly reduced at that point. And so again, all these are factors that basically sort of stack up, to some extent, against karyotyping in solid tumors. Not to say it's impossible, it absolutely is possible. That's how we led the identification of some of these findings, but it's not always ideal. And certainly, there has been a shift, necessarily, in part because when we only had karyotyping, if we didn't send a significant amount of tissue to cytogenetics, you weren't gonna get a result, right? There wasn't molecular testing, there wasn't fish, there wasn't an array, this was your chance. So send us a big gob of tissue, and we'll do what we can. Increasingly now, we have other options. And I think other challenges that are, again, just generic, I think, to karyotyping in general, again, is that the resolution, to some extent, too. This is this 3-5 translocation that I was talking about. So the end here of CRMS is actually a 5q. So there's a break right around here on the chromosome three and it typically takes out VHL, 3P21, being sort of the critical gene of interest, and then this last bit is all 5q. So there was a significant advantage, or there was, fortunately, I would say, I think we've learned from the experiences that we've had. So we did all this work with cytogenetics that led to the identification, or I shouldn't say cytogenetics, GTG banding, and that led to the identification of recurrent aberrations. Well, the minute you have a recurrent aberration, through the advent of FISH, we also have a more targeted way we can specifically ask that question. So great, we know the Ewing sarcoma are characterized by EWSR1 rearrangements to a number of different genes, most commonly FLY1, shown here. There's 1122 rearrangement. But now that we know that EWSR1 is the expected rearrangement, again, let's just do a break apart strategy. And now we have another advantage, which is that suddenly I'm no longer limited by the necessity to actually do this in fresh material. So now, in histology, again, or in the frozen room, or initially after resection, again, rather than immediately sending something, we can embed this, this can be further reviewed, there can be discussion among histopathology, and they can say, yeah, we have a really good feeling this is Ewing sarcoma, let's move ahead. And that picture is not very helpful. This is, no, there's nothing there, there's nothing. So that's not a great fish. But ultimately, it's a rearrangement. We know, what's a classic rearrangement pattern? For a break apart, what would we expect to see? We can picture this now, we've done this a few times. So break apart probe. You still have phone a friend option, and we have a, that's. Yeah, exactly, yeah, exactly. So one fused, one green, one red, that'd be the classic. One is gonna be an intact EWS1 locus from the normal chromosome 11. And the other one is gonna be the rearranged. Because this is just a break apart, not a dual color, dual fusion. Also, again, we know that, again, other sort of events that we typically were looking for, for the liposarcoma, again, it's very frequently to have these sort of large, these are giant ring chromosomes. And those ring chromosomes were largely contained of chromosome 12p material, and specifically MDM2. And so when we culture liposarcoma, we were looking for these giant marker or ring chromosomes, and that was really the sign. You saw that, thank you very much. It's a liposarcoma. We could differentiate that from a lipoma, which has HMGA2 rearrangement, or other sort of other benign variants. But again, now in this case, rather than have to culture, we can go straight to fish. Rule in, rule out. Do a quick fish for MDM2, and again, we expect to see, again, just a slew of probes leading to the amplification of the specific genes. Oh, and there's, there we go, there's a fish. That's. But fish also has its own challenges. So I'd say fish, for the most part, again, for many cases, in the context of solitary humans, is exactly what we want to do. I think the C. ewing sarcoma, ask a specific question, get a specific answer. But we have to know what that question is. So the challenging part of that is if I have an undifferentiated sarcoma, and I don't have a clue what I'm looking at, I don't know at the time, and I don't set any material aside for GTG banding, I've lost that opportunity. If I don't know what to ask for fish with, now what? What are my options? Yeah, exactly. So now we have, now, if we're gonna do an array, though, that, what are we then assuming about what we expect to see from a copy number perspective? What are we expecting, what are we hoping the tumor's gonna be driven by? I think that's a better way to ask it, maybe. I'd say not a structural rearrangement, is the answer you're looking for. So it has to be, again, then a copy number imbalance is to that point. If it's going to be a synovial sarcoma, we know driven by that X18 rearrangement, I'm not gonna find that biarray, I'm out of luck. So I've lost that opportunity at this point in time. It's just something to keep in mind. Another point here, I probably should follow the slides a little bit, here's the, we've talked about this briefly, and I think I wanna make sure I just reiterate this point. So again, the currency that we use traditionally in pathology is these five micron sections. It's exactly that, it's exactly what I say. You take a block, the solitum has been embedded in paraffin, we're then cutting off the sections that are five microns, which means necessarily we're cutting through the cells. And so that means that in those instances, there's the potential that again, if we're looking for deletions, those cells, that medium of testing may skew negatively the amount of deletions we're seeing. So if it's a, if we have a tumor specimen, it's infiltrative, only 30% of the specimen is tumor, the other 70% is normal, and we're cutting through, is it possible that we're gonna mask the deletion? It's absolutely possible. And so just a caveat and a concern that we have to be aware of if we're specifically looking for patterns of deletions. So by example, if it's chromophobe, I'll go back for a second. With this chromophobe, we must look at carcinomas to look for these aneuploids by fish, again, if it's a low tumor content, that potentially could be challenged. Okay. So. Great, we did, you know, credit where credit's due. Karyotyping led us to the identification of these pathognomonic findings, huge. We then took those findings and we said, well, if this is the specific kind I'm looking for, do a targeted test. Amazing, the right thing to do. In certain situations, you may need to couple this with additional studies, and I think that's sort of the next phase of this that I want to spend a bit of time talking about. So let's go back to renal cell carcinoma, and we've seen the patterns of losses. This is, I would say, a perfect example of the type of tumor that you would want to look at by array. And the reason you want to look at by array is that, do you want to grow it? You could, you absolutely could. You could put it in culture, you could see what grows. Maybe your growth rate's fantastic. Maybe you get the results you need. Absolutely do that. Having said that, it's also possible that after histopathology review, someone says, this is a clear cell. I don't need any other test. So why did you say, said something in the first place? Don't send it. Better yet, do the histopathology review. Have a chance to say, is this a case where I do have a differential diagnosis, or I'm not certain, or I do need to find, or use orthogonal testing? At which point in time, now you're in a great place to use array. What's the limitation, though? What am I not gonna be able to detect here? Exactly, the structural units. Specifically, that TFE3 subgroup is one I will not be able to capture. So what I will say about that is that certainly, again, what we've done locally is that in these types of situations is that if morphologically there's a suspicion for TFE3 rearrangement, do that. Start with the fish for TFE3. If negative, proceed to a ray. And again, try to figure out, use the assays to their strengths, use algorithms to your advantage, and then try to make sure that we can manage both our resources and the resources of the system in general. So there's a large, long preamble to get to what an array looks like in this particular case. But really, the point is that it doesn't matter what you're doing, whether you're doing it by karyotype or you're doing it by array, or fish for that matter. Fish are a little more tough to make that argument. But it's looking for patterns and interpreting them for many of these solid tumors. So rather than say this specific finding, ah, that's it. That drives the tumor. That happens occasionally. Oftentimes, it's saying, I'm looking at the whole gestalt of the tumor and say, that pattern fits with this. And certainly, from a microarray perspective, that's oftentimes how we interpret them. There is a recent ACMG and CGC guideline that just came out about sort of suggestions and recommendations for interpreting microarrays for cancer. And largely, that has to do with how we go through the interpretation. So there's actually been a tiering system that's been described, basically, to say, is it a pattern of events that collectively means something diagnostically, prognostically? Is it a specific finding that we really need to highlight? Are some of the changes a change that we don't know the significance? And so trying to make sure that we can introduce some stability and similarities in the field in terms of how we handle microarrays. But all these, again, clear cell, 3P loss, and other changes remain. Papillary 7 and 17. In this particular case, it looks like there's probably a two or multiple copies of chromosome 17. This is a snip array. I've just cut off the bottom track showing the allele peaks. And then a chromophobe characterized by losses. So again, fairly easy to distinguish one from the other. The same is absolutely true for tumors of the central nervous system. This is a little where there's perhaps differences here is that CNS tumors were notoriously difficult, even more so than renal tumors to grow in culture. And so our understanding initially through karyotypic or GTG banding analysis was rather primitive. We knew that gliomas often had gain of seven. We eventually found that there were some amplifications, AGFR being one of them. But the level of resolution was just insufficient to be able to necessarily parse out the differences among these subgroups. Now this is just some of the gliomas, both pediatric, if we look at pilocytica, astrocytoma, and adult, in this case, oligodendroglioma, astrocytoma, and glioblastoma. But again, these tumor types are largely driven by copy number imbalances that, again, are easy and therefore appropriate to evaluate by array. Pediatric brain tumors are a little bit different in the sense that oftentimes pediatric tumors and certain pediatric CNS tumors are driven by rearrangements. In fact, this first one is a rearrangement. So how come I'm pointing to a copy number imbalance? And how do you detect a rearrangement on an array? So let's play that out. So how does the SNP help you detect it? I don't know. I looked at the data, but I know you can target the SNP to a certain region. So you can, you can't. You can't in the sense that typically with any platform, for the most part, the platforms are correct. And an individual SNP in this case will be less relevant than actually the boundaries of a specific rearrangement. Where actually the two do sort of meet hand in hand is that when you have a copy number change that's intragenic to a gene, it's highly suggestive, suggestive, note the word, of a possibility of a rearrangement. There's about five asterisks on the top there. Having said that, we know through, again, extensive characterization of many of these genes that actually for pilocytic gastrocytomas, that's exactly how they're formed. It's a 1.99 megabase tandem duplication. You basically gain part of the middle of BRAF and KI 1549, and it re-inserts itself back into the same locus. But it does it in a way that basically the three prime, the kinase domain of BRAF is now being driven by the five prime region of KI 1549. If you did this by fish, and you fished right in the middle of that locus, you'd basically see one dot being the normal BRAF, another dot just adjacent to it. And you'd have that positional information too that would suggest there's something intrachromosomal that's going on. You'd have two red dots that always follow each other. At this point, because it's been so well-established, you're done. This is a slam dunk. You drop the mic, and you leave. There's definitely plenty of cases that we'll come across where you'll see a copy of re-imbalances that partially encompasses a gene we've seen that with MET, with ALK, with ROS, all of them brain tumors, but all of them are genes in which we've seen, and ROS is a little different because it's a recurrent finding. ROS is a deletion on six, chromosome six, that takes out part of GOPC and part of ROS. And actually, that's been identified a number of cancer types as a recurrent copy number imbalance that leads to a fusion. But for ones that are relatively novel, the most we can say is we suggest and absolutely perform orthogonal testing and sequencing studies in those cases. And almost inevitably, those end up being pediatric cases in which, to some extent, you could maybe make the argument, unless you had a strong reason to suspect it was something else with a likely copy number imbalance, just go straight to the sequencing. Look for the rearrangements at that point. Look for the mutations. Now, the other point I wanted to try to raise is that, again, I think one of the principles that I think that we've tried to impart quite a bit on the fellows or residents, the people we're training, is that, irrespective of what type of data you're looking at, think about mechanism. Think about the way you would if you're looking at a karyotype. Try to think about how do you get these patterns? We've talked about how a gain of one end of a chromosome and loss of the other end of a chromosome in the same chromosome in the context of a constitutional case is likely suggestive of recombination of an inversion. In the same way, loss of one end of a chromosome and gain of a different end of a chromosome that includes the part way through till the telomeric end, those end up being suggestive of an unbalanced rearrangement. So think about the patterns, but think how do you get to those patterns? The same is exactly true for some brain tumors. So for 100% of all oligodendrogliomas, and this is now understood, I think there's some historical context we'll go back to maybe after, but in 100% of all oligodendrogliomas, you always have one P, all of one P from the centromere all the way down to the telomere, and always 19Q, same thing. Always delete it, always delete it at the same level. How does that happen? Yeah, exactly. So it's a derivative of chromosome, and that's exactly it. Again, we can go back to our historical understanding, going back to being able to karyotype, and what you produce is exactly that. This unbalanced rearrangement, so there's DIR-119, this is the breakpoints here, but the long and short of it is basically you have one Q here, 19P on the top, and the other half of that rearrangement is lost, gone, always deleted. It's always that entire thing. So somehow the formation of this chromosome also leads to the loss of those two, and it's believed that those two events together creates enough of a haploinsufficiency that basically drives the tumor down this oligo lineage at this point. This is an interesting aside, but for a long time, this is again what happened in the absence of sort of robust molecular data. So historically, 80% of oligodendrogliomas had a 1P19Q whole arm co-deletion, and 20% didn't. And now what we understand is through sequencing studies, actually, all these tumors have an IDH mutation, and in addition to that, they all have a 1P19Q. Now looking back historically at data and trying to clean up the literature, what we've understood is that 100% of oligodendrogliomas have 1P19Q, and 20% of the time we were wrong. It was a different tumor type. And that's almost a challenge that we're seeing happening in many different diseases. We're going back and saying, well, wait a second. This was our appreciation of what this meant. Well, this subgroup that we identified was actually just a misdiagnosis in the first place, and now we can sort of make this more robust through molecular copy number or epigenetic data. What type of IDH1 or IDH2 mutation is the most common? An R132H is the most common. R132H is the most common mutation that occurs. That's half of it. What's the common number? It's a mutation. It's a mutation. Yeah, yeah. So in gliomas, adult gliomas, there's sort of two pathways. One is that you have an IDH mutation, at which point in time you have one of two options. You lose 1P19Q, and you become an oligodendroglioma, or you have a mutation in ATRX and P53, and you become an astrocytoma with a very different copy number, and very different prognosis for that matter, too. All those are relatively indolent. They do reasonably well, 10 to 15 years survival. It's five-ish years for astrocytomas, depending on the grade of the tumor. And in the absence of an IDH mutation, in a glioma, they become glioblastomas, essentially. They're particularly aggressive, have EGFR amplification. So, strengths in the array, absolutely. This is actually all from fixed material. This is actually, I think, was a relatively recent, in the last, say, five years, advent that happened. For the longest time, SNP microarrays were useless when it came to fixed material. And then there's been developments that have happened, technological developments, and frankly, just a concerted effort that was put on the part of many, a few commercial entities that ultimately led to sort of increasing the robustness of the assay. So, we can get good quality data from fixed material. Again, great. I mean, someone can look at it first, say this is a tumor that's worth profiling before we start those studies. But it's important to also highlight the limitations of this approach also. So, here I'm showing, this is actually a microarray study, two microarray studies that were done on a patient with mesothelioma. So, I'm not sure how many of you are familiar with this disease. Mesothelioma is a particularly aggressive disease, asbestos-related. It basically cakes the line, cakes the outside lining of the lung, and it's typically a fatal disease. It's characterized by recurrent copy number and mutational changes. So, recurrent copy number changes, it's typically loss of secant 2A and 9P, loss of 22 and loss of 3P, or a combination of one or multiple of those things. And we've been doing microarray studies on these, on patients with mesothelioma, in part because we have a program and some sort of aggressive therapeutic and surgical approaches at our hospital. We had a time point, it's a pleural biopsy, so they basically drain the pleura, self-suspension, we spin it down, isolate data, run the array. And on the top array they said, do you see cancer? And the reality is we saw some waves in the data, but we couldn't definitively say. And basically, microarray studies, I would say, certainly the platform we use, I would say, your ability to detect reliably and confidently changes less than 15%, 10%, I would say is about as good as a coin flip, which is exactly what you don't want in diagnostic space. So we said we saw changes, it's suggestive, couldn't do more. We did FISH. FISH was actually negative in this case, it's unclear why. Later on, a few months later, the patient re-presented for a second pleural biopsy, and now this is the same patient. Now you can appreciate the higher tumor content, now your ability to more visually appreciate and ultimately make calls about the underlying cognitive changes that exist. So two factors ended up being critical determinants about the success of microarrays, and I would say the same is absolutely true for sequence-based technologies too, is that what's the quality of your data, and what's the input? But if either of those factors is off, it can significantly impact your ability to make a call. And so make, again, this comes down to sort of how did you validate your assay, how much did you push it? I can tell you we validated our assay down to, we felt confident making calls about 30%, we knew visually we could appreciate changes down to about 15%, this fell below that. We're still looking at ways, and we've adapted some ways to sort of try to help sort of compensate for that, but it's, I think highlights in certain situations, again, microarray studies aren't really appropriate. Okay. Strengths and weaknesses, briefly we go over. Karyotype fish, microarray. Karyotype, I think we've gone over this. Copy number, CN and SV for copy number. Structural variance, so great. Whole genome-wide, single resolution, single-cell genome-wide assay, a little tongue-in-cheek comment, but nonetheless, true. Resolution's not great, requires in vitro growth. Fish is rapid, rapid-ish. If we're talking about paraffin material, we're still talking four to 10 days, realistically, because we have to go through the process of taking fixed material, deparaffinizing it, basically making the DNA accessible, and then going through the fixation process and hybridization. And it's targeted, which is either a strength or a limitation. Limitation if you don't know what you're after. If you have a great question, fantastic. If you don't know your question, you're out of luck. And then array, obviously, we just went over, so genome-wide. Obviously, only copy number. Occasionally, you will find a change which is suggestive of structural rearrangement. We shouldn't kid ourselves to thinking that it can detect them. Certainly, there are situations and specific scenarios in which we know a structural variant occurs as a result of a defined copy number aberration. That's the minority of cases. Oftentimes, they're balanced rearrangements. You're not going to detect anything by array. Oh, wow. And we're done. Okay. Last couple moments for this. Just want to end, so this is one to sort of really highlight the different assets, different possibilities. All of that is really discussing solitum-assertive genetics as a tool for diagnostics, for the most part. Can we support or refute the histopathology diagnosis? You said it was clear cell. You think it's a renal cell carcinoma. Can we tell you if we agree or not that it's a clear cell or not? That's satisfying, but certainly from a patient perspective, I think certainly not as satisfying as the identification of events that are actionable. This change leads to this drug. This may change management. This may change outcome. And so two such examples, just to end on briefly. One is ALK. So it was initially identified in anaplastic large cell lymphomas, but Jim Downing now at St. Jude's, and again this is the ALK NPN1 at 2, 5 translocation that was identified. It ultimately brings this 5 prime region of NPN1 in front of the 3 prime end, again the business end, of an ALK rearrangement. And so for all kinases, it's the same. The 3 prime region ends up being where the kinase domain is, and that's basically the functional component of that gene. You need that 3 prime end for the gene to be functional. ALK we now know is actually present in much more than just these anaplastic large cell lymphomas. In fact, again, this is the host of other tumors, inflammatory myofibrastic tumors, non-small cell lung carcinomas, and you can see again there's a diverse array of partners that ALK rearranges to. But here at the top, again, the 3 prime end ends up being consistent. You need that kinase domain for something to work, for the gene to work, I should say. Now why is that so important? Well, it's important because ultimately we had a protein structure that again does not exist in normal development. It means there's an opportunity, not always a possibility, but an opportunity to try to crystallize it, to try to identify areas of its structure that potentially can be targeted through a pharmacological agent. In this case, this has sort of led to crizodinib and sort of demonstrated in these seminal studies that the use of crizodinib ultimately could improve in this blue line the overall survival of patients who had an ALK-positive non-small cell lung carcinoma, about 2 percent of all cases. And just sort of going back a little bit sort of from a diagnostic testing perspective, so we've talked about FISH before. This is just to sort of end with sort of thinking from this part of it. You can see what this break apart probe strategy looks like in this case. So again, the 3 prime region, we really want to see where does this end up going in this particular case. Normal, 2 fused as expected, rearranged, we see a break. But don't forget that just because these are sort of the two outcomes we're hoping to get, either normal or abnormal, there's many other possibilities that can occur. So again, polysomy or polyploidy, what the heck does that mean? Or this. So now you have one fused, one red. So the doctor calls and says, is the patient rearranged? What's the answer? Well, there's a fused. So a fused is intact, so that's a normal. Oh, I see. So it's a location. So, or. Or another one. So. Or a geometric. So it ends up being, again, the. If we have a rearrangement, so again, the 3 prime is the business end of ALIC. Where this goes is relevant. Where this goes is not. I can't definitively tell you this last one is rearranged. But that pattern is suggestive of either a deletion or an unbalanced rearrangement. You had a rearrangement. You lost the derivative of chromosome with the 5 prime end of ALIC that no one cared about anyways. In fact, this has been studied by the College of American Pathologists. There's been a few studies, additional studies that have looked at this specifically. And actually, about 70% of the time, this actually results in a rearrangement. It's exactly that, an unbalanced rearrangement that happens. Now, how good do you feel about 70%? That's not great. But this also can be supported by immunohistochemistry and certainly by sequencing studies these days too. And still provides a rapid way to see the screen. Okay. And in the last five minutes, three minutes probably, HER2. So the other, when we talk about sort of actionable alterations, HER2 obviously is another one that certainly I think that I'm sure everyone in this room has heard about HER2 testing and its sort of implications of breast cancer. So it's another, this is another, this is a receptor tyrosine kinase. It's amplified by 15 to 20% of breast cancer. And signaling occurs through dimerization. Again, this HER2 molecule will dimerize ultimately with other receptors. And ultimately, what's odd about HER2 is that, again, we don't actually know what the ligand is that activates it. It seems to sort of auto-activate in these instances and cause activation of downstream signaling. But what's been particularly notable is that in instances of HER2 amplification, they've been able to develop monoclonal antibodies that can specifically bind to the receptor and can, again, can basically prevent downstream signaling and lead to improved survival. I feel like I should have ended the last two with a sexier slide, but this is, so this is a mouthful then. The outcome, amplified or not, that distinction should be fairly easy. Having said that, there's been pretty significant, there's been development of fairly rigorous algorithms that have been set up. So, you know, we talked about in different cancer types, what do you have to, how many copies do you have to have to be amplified? That's been reasonably well established in HER2, and they keep on making updates to the guidelines. This is the most recent guidelines, what happens. So currently, again, if on the invasive component of a breast cancer, if a patient's going to be tested for HER2, they typically undergo HER2 immunohistochemistry. Coupled with, then, FISH studies. And across a cohort, again, the region is selected by a pathologist and scored typically by two technologists, and ultimately they'll count 30 nuclei and say, what's the ratio in each individual nuclei of HER2 to a control, to CEP17, centromere for chromosome 17? If I have a ratio greater than two, and an average number of signals of HER2 greater than four per cell, thank you very much, we've got an amplified. If the ratio is less than two, and the number of copies is less than four, again, if it's ish negative, thank you very much, it's negative. And then we've got this big bin in the middle, which basically requires additional workup. And so all this to say that, you know, the testing itself is straightforward, but the algorithms are becoming increasingly complex, and I would say maybe appropriately complex. It's that we're now, we're taking, and we're integrating the copy number state together with in situ hybridization, what we presume to be then the functional output, is it producing protein or not, taking both of those pieces of information together, and trying to define basically an amalgamated score for the likelihood that this patient will have a HER2 positive cancer that we feel will likely benefit from treatment. Okay. That was three hours. I think that was a good run. So I just wanted to end this slide, and these lectures, I think, with, hopefully I think what we tried to cover today, or with this lecture specifically, is that really highlighting that cancer is a disease of the genome, and thinking about the ways in which you can either activate an oncogene or disrupt a tumor suppressor gene. I hope you leave here thinking a little bit about, you know, the way you would think about cytogenetics. Don't think about it as any other assay. Think about it as a field as a whole, and it's a tool. Apply that tool to any assay you're using, whether it's NGS, whether it's array, whether it's fish, whether it's karyotype. Think those same mechanisms, think with those same principles, and it will make you a better geneticist. All right. With that, I'll end. Thank you.

solid tumors

genetic mechanisms

cancer definitions

tumor genesis

genetics

actionable alterations

genetic changes

diagnosis