All right, so I'm a rheumatologist, not a hepatologist. You can't ask me about the liver, but I'm happy to talk about immune cells. So we're going to talk specifically about CD8 T cells today and what role they might be playing in pediatric acute liver failure. I've got no disclosures. So how did we sort of get to this point in the first place? So there actually has been, if you look in the literature, a decent amount of evidence suggests that the slice of the pie that is pediatric acute liver failure that has been labeled indeterminate might actually have an immune dysregulation component. And that comes both looking at markers that are in the blood, including soluble IL-2 receptor, which is a marker of lymphocyte activation. And the numbers of CD8 T cells in the peripheral blood. There have been sort of mathematical models looking at multi-parameter analysis that suggests that there's a dysregulation of immune feedback in indeterminate pediatric acute liver failure. And then there was this report which suggested in a series of nine patients that the histology was consistent with an impressive CD8 T cell infiltrate that's associated specifically with indeterminate pediatric acute liver failure. So the question became to our group, maybe this indeterminate, or at least some portion of indeterminate PALF is not so indeterminate, but is rather characterized by an immune dysregulation process. Now I recognize that not everyone on the crown is an immunologist. We love our flow cytometry and our cell surface marker alphabet soup. And the point of this slide isn't to turn you into a CD8 biologist, but to just kind of introduce a few terms that we're going to need to kind of make our way through the rest of this talk. So CD8 T cells, like all lymphocytes, exist in an IU state, which is prior to seeing antigen, they are kind of unprimed. And then after they see antigen, they acquire memory so that the next time that they see antigen, they respond much faster, right? So this is that hallmark of adaptive immunity we all learned in medical school. But not all memory is memory. So there's different forms of CD8 T cell memory. They go by names of T central memory, TCM, T effector memory, TEMRA, TRM, TFH. We're not going to worry about all of those details. There's a whole bunch of markers on the slide. We're not going to worry about all of those details. You're going to trust me because otherwise I could spend the next three hours just talking about markerology. As we go through the talk, this is how I identify different kinds of CD8 T cells. I can use proteins that they express on their surface to put them into one of the boxes on this slide. So we, in collaboration with the group at Low Reach Children's, had the opportunity to assess a large cohort of banked specimens from patients that either had biopsies or explants with indeterminate PALF. And so the first thing that we did was to just ask the simple question, what's in there? What is infiltrating into those biopsies? And we were able to reproduce the report that I had showed you previously, which is that there is a massive CD8 T cell infiltration of these livers. It is both portal and lobular, really sort of widespread throughout. And in all of these studies, I'm going to show you specimens that come from patients that were diagnosed with indeterminate PALF versus patients that were diagnosed with PALF from some known cause, whether it was metabolic, mitochondrial, toxin, what have you. And so the first thing you'll notice is that patients with a known diagnosis associated with their PALF were all CD8 negative. Patients that were indeterminate had florid CD8 T cell infiltration. You can quantitate this using some imaging software, and you can clearly see that the groups separate out. And we looked at a bunch of other different immune cell types, and none of the other immune cell types differentiated indeterminate PALF from other forms of PALF. So CD4 T cells did not differentiate, B cells did not differentiate, NK cells did not differentiate, macrophages did not differentiate. So there's something unique about this thing that we're calling indeterminate, but actually seems to be well characterized by a CD8 T cell infiltrate. So now the immunologist in me says, okay, well what about those CD8 T cells? What can we learn about them to get more information? And that's when we start staining for all the things in flow cytometry to put things into these boxes. And we actually had the opportunity to do this because along the way while we were doing our study, we were able to get some fresh specimens, and the advantage of fresh specimens is you can get the cells out of them and do flow cytometry. So the first thing we can do by flow cytometry is, let's see, does this arrow, how does this work? Does that show up over there? I can't tell. All right, so I'll just do it descriptively. So if you look at your left, you're looking at CD3 by CD19, so B cells by T cells. And what you'll notice is that the patient with a known diagnosis PALF has a mixture of B cells and T cells and non-lymphocytes infiltrating their explant. But the three indeterminate PALF patients all have a major predominance of T cells. And on the right-hand side, what I'm showing you is that those T cells are, again, predominantly CD8 T cells. So these livers, even by flow cytometry, are marked by a CD8 T cell predominance. And then you can dive deeper and ask, okay, now what other markers do those CD8 T cells express? And it turns out that, again, using this markerology, so the two markers are CCR7 and CD45RA. The details don't matter. The point is that each of these quadrants are a different flavor of CD8 T cell. And you'll notice that the T cells are all piling up in the lower left quadrant. So that lower left quadrant represents effector memory CD8 T cells. So the CD8 T cells that are in these livers, they're not naive. They're effector memory in their phenotype. And then in the tissue, there's another marker you can use in that effector memory subset called CD103, which marks a special kind of memory cell called a resident memory or tissue resident memory cell. Tissue resident CD8 T cells are T cells that live in the tissue that are waiting for a challenge. They're sort of like our first responders. They're almost innate-like in a way, because they respond very quickly to a challenge while living in the tissue and not in a central lymphoid organ like the lymph node of the spleen. And indeed, the CD8 T cells in indeterminate PALF are remarkably almost all CD103 positive. Now we only had those three specimen, fresh specimens by flow cytometry to look at, but we were able to go back to the archived cohort and do this same kind of staining by immunohistochemistry. And what we were able to show is that, indeed, the CD8 T cells in the whole cohort were marked by the CD103 marker. And so what you're looking at here is a double-labeling stain where CD8 is in pink, CD103 is in brown. And what you'll notice is that you see a lot of brown. You see a lot of CD103 positive cells, and I don't know how well it projects or not, but almost all of those cells have some pink in them. There are a few pink cells that don't have brown, and remember I showed you about 30% of the CD8s were 103 negative. And so sort of consistent with that, most of the CD8s in the archived specimens were also 103 positive. Okay, there's a zoomed-in picture of it. You can probably make out the double-labeling better in that picture. So CD8, in the whole cohort, CD8 T cells bear the 103 marker. The other thing we can ask about these cells is, do they express effector functions? And so the major effector function of CD8 T cells are its perforin, which is used essentially to blow up and lyse a target cell. So that's what CD8 T cells do is they kill, and they kill using this perforin molecule. And the important thing is that actually most CD8 T cells are perforin negative. In general, the only cells that are perforin positive are the so-called CTLs, or the temra compartment, the compartment that is acutely activated and angry and wants to kill. And that makes sense because it's probably a bad idea to just be dumping out perforin indiscriminately. That's a function you want to reserve for when you really need it. So most memory CD8s don't express perforin, but when we did this by flow on the samples that we had, indeed, the CD8 T cells from the indeterminate PALF patients were perforin positive. And you can see as a control, our CD4 T cells were not perforin positive. So the stain is specific. And if we then went back to the cohort by immunochemistry, the indeterminate PALF cases are floridly perforin positive. The diagnosis, even though there are CD8 T cells, there are fewer, but there are some there, the diagnosed PALF cases were perforin negative. So this is a peculiar phenotype of CD8 T cells infiltrating these livers that bear markers of tissue-resonant memory. And even sort of more peculiar are tissue-resonant memory, but they're angry, they're expressing perforin, they want to kill. And this was statistically significant when we looked over the whole cohort. So that you can actually identify on, even if you wanted to do biopsy tissue, by staining for the combination of CD8 perforin and CD103, a specimen that would fit into what we're calling indeterminate PALF, but maybe might be better named immune-mediated PALF, or we can debate what the name should be. But it's clearly not indeterminate in my estimation anymore. This seems to me to be an immune disease. And in fact, these are stains that Children's Hospital of Philadelphia can actually do clinically. We've got them validated. In fact, that's part of our practice now for working up these cases, is we stain for all three of these markers. So that's what's there. Memory tissue-resonant CD8 memory T-cells. That brings up another question. Why are they there? What are they responding to? So memory T-cells can be responding to basically one of two things. They can be responding to other cytokines that might turn them on. These come in a lot of different flavors, IL-12, IL-18, so on and so forth. Or they could be responding to the things that T-cells classically respond to, which is antigen. We can ask that question by doing an assessment of the clonality of the T-cell receptors in the specimens. Because if you're responding to antigen, one of the things that CD8 T-cells do when they respond to antigen is they divide and make other copies of themselves. Well, if they divide and make other copies of themselves, they're increasing the numbers of cells in the specimen that all share the exact same TCR. So you can take your specimen, and you can just, using next-gen sequencing platforms, sequence all of the TCRs that are in that sample. And remember, the T-cell receptors of T-cells are all generated at random, right? So if I just take your blood, and I do this kind of assay, what I'll get is hundreds, thousands of different T-cell receptors with no unique single population rising to the top. They'll just be sort of evenly distributed. I guess unless you got your flu vaccine yesterday. If you got your flu vaccine yesterday, then when I do this, yes, I'll find a clonal population of anti-flu T-cells in your blood. But any other random given day, the distribution of the different types of T-cell receptors should be sort of flat. They shouldn't be piling up for any one particular T-cell receptor. If you indeed see the piling up of one particular T-cell receptor, that then suggests that that population of T-cells that you're looking at must be responding to some specific antigen. It doesn't tell you the nature of that antigen. You can't learn that from this assay. It just tells you that there is a antigen. So we performed that assay on our specimens. And there are a couple different metrics you can use to measure this clonality score. I don't want to, again, spend 30 minutes getting into the weeds on that. So we'll just try to give you a gross overview. If there are detailed questions, I can certainly answer those. So what I'm showing you on the left-hand part of the slide is the sort of raw data, what you get from an individual sample. And I'm showing an example of indeterminate PALF, autoimmune hepatitis, and a known diagnosis PALF. And what you'll notice is that you kind of get a flat distribution in diagnosis PALF. There's maybe just the beginnings of a slight piling up to the left-hand side of that histogram in the autoimmune PALF. But you'll see that the sort of top two or three clones of an indeterminate PALF sample are disproportionately higher in representation than all the other clones. And so that's an example of this clonal pile up. That's an example of a non-uniform distribution of T cell receptors, suggesting that there's an antigen driving the presence of those T cells in that population. And the one metric we use is we say, well, what percentage of all of the T cell receptors in my specimen fall within the top 20? So sort of like how many T cell receptors pile up in the count of the top 20? And the more that fall in the top 20, the more clonal your specimen is. And you'll see in the raw data that that indeterminate case was more clonal. And indeed, in the upper right-hand slide is the summary data for the cohort. And in fact, the indeterminate PALF cases had a statistically significant increased amount of clonality in their specimens compared to both the autoimmune hepatitis group and the known diagnosis group. There's another metric we use called percent clone overlap. This is, I don't want to get into the technical details of how we do this. It's an orthogonal way of asking about clonality that, again, tries to assess, are clones repeated in these samples in an overrepresented way more than what you would expect? And by that measurement as well, you can see that the indeterminate PALF cases have an increased amount of clonality compared to the known diagnosis cases. So if I summarize what I just told you, this group of indeterminate PALF cases are characterized by a CD8, CD103 perforin positive T-cell population that I think likely represents tissue-resident CD8 T-cells that are responding to antigen because there seems to be an increased clonality. Now, that antigen, as I said, I can't tell you what it is. Is it a pathogen? Is it a virus we have yet to identify? Is this an autoimmune disease? We would need to do more work to try and sort those things out. And the fact that the perforin is present suggests that these cells are on, angry, and they want to kill something, and that something is presumably the liver. So these data have been replicated quite nicely by Katherine Chapin, who will be presenting this work in an oral abstract at this meeting, validating these results in a much larger cohort. So it's curious to me, this association with aplastic anemia and indeterminate PALF, and what's curious is that aplastic anemia is another CD8-mediated immune disease. So that's been clearly established, that those CD8 T-cells are going after hematopoietic stem cell clones. And in fact, not only is it CD8 T-cell mediated, but it's got an MHC class I allele association, which also suggests antigen, right? So there's something about having the risk class I allele that makes you susceptible to aplastic anemia, and your CD8 T-cells are then angry at that, whatever the antigen that's being presented in that class I allele, and they're going after it. And so I think, you know, an interesting question is, is the same true about indeterminate PALF? We don't know the answer to that yet. That's a question that could be tested, that our group is certainly thinking about. So then, you know, the final question is, can we even do this in a less invasive specimen? Can we find examples of dysregulated CD8 T-cell immunity in the peripheral blood of these patients? And I don't know the answer to that question yet, but I'd like to suggest that it might be the case. We've begun to look. So we can get peripheral blood specimens from these patients as well, and essentially do the same kinds of flow cytometry studies, right? So you can stain with all the same markers, CD3, CD4, CD8, perfrin, CD103, what have you. And we're collecting samples. My graduate student is convinced that he can tell you with 100% accuracy who's who. I think we need to do a few more before I'm willing to put that out there, but I just thought I would show you a few examples. So this is an acetaminophen overdose. And again, in this case, you can see there are a few CD8 T-cells in the blood. They exist in a normal CD4 to CD8 ratio. That plot in the upper right that you're looking at, which is the CD45RA by CCR7 plot, which is, again, the different baskets of CD4 T-cells is a normal sort of distribution of naive and memory and temoras. And if you look at the CD103 and DR markers as a measurement of activation, they all look pretty normal in this patient's blood. But if we flip to a... And so actually, the thing I want you to pay attention to is a CD103 DR positive subset, which is essentially zero in this sample. So that would be an activated, angry CD8 T-cell expression tissue residency markers. Different story when we look at indeterminate PALF cases. Now you have a much more CD8 predominance in the peripheral blood. They express a much higher effector memory marker-like phenotype. Again, you see the cells in the upper right piling in the lower left quadrant. And when you stain for that 103 positive population, they're greater in number and they're DR positive as if they're angry, as if they're activated. So I think it may be possible that we can even find these cells in the peripheral blood, which would be a much less invasive way of getting after this patient population. So perforin makes these cells angry. The other major effector function of CD8 T-cells is interferon gamma. And these really aren't my data to share, but I'm just sort of free advertising. You should go check out this poster also by Catherine Chapin at the meeting, which shows these data, which suggests that there is an interferon gamma response that you can find in these specimens, which is the other major effector function of CD8 T-cells, perforin and interferon gamma. So I think these data also argue these CD8 T-cells are angry, they're doing something, they are generating an effector program that is quite likely damaging these livers. So to summarize, indeterminate PALF can be associated with a tissue resonant memory phenotype that appears to be responding to antigen. These cells have effector functions in the form of both perforin and interferon gamma. And I think that likely marks them as a likely pathogenic immune cell population, which is targetable. I mean, the nice thing about being a rheumatologist in the year 2019 is there's a bazillion drugs that one can think of to go after these guys, including just deleting T-cells themselves. You can block interferon gamma, there's an antibody that does that in clinic. There's a whole bunch of things you can do, but it suggests that perhaps we can move beyond this idea of indeterminate to, okay, we actually maybe have identified a pathogenic mechanism that is druggable. The remaining number of questions, what is this antigen that's driving this clonality? Can we identify a class one risk allele? Is it the same as the risk alleles that we know for aplastic anemia? And of course, we talked about whether we can target these cells therapeutically, both in what we're calling indeterminate PALF, as well as perhaps other forms of inflammatory liver disease. So, I'll just finish up with an acknowledgment. Tom Byrne was the graduate student in my lab that did a lot of the immunological work for the data that I showed you. Kathy and Portia are the hepatologists and pathologists at CHOP that did the work. And then you also see the Lurie team as well, that we couldn't have done this without them as well. And then that's it. I can take questions at the panel.

CD8 T cells

pediatric acute liver failure

immune dysregulation

effector memory cells

tissue-resident memory

perforin