OK, why don't we go ahead and get started? Thank you all for coming to Medical Ground Grand Rounds Day. Today as the Philip C. Liverman Lectureship. And today we have Barbara Murphy, who is our Liverman lecturer. And we've had a spectacular visit. Yesterday was a beautiful day. And we had a tour of the rotunda. And then we had a wonderful reception with the Liverman family. And we had one family member come in from Denver, which was very special. And then today Barbara gave a terrific talk to the transplant folks. And now we're here at Medical Grand Rounds. Before I introduce Barbara, I just want to tell you a little about Philip. So Philip was a fellow with us back in the '70s. He received his bachelor's degree from George Washington University, and then a medical degree at UVA. And then he went to Case Western Reserve to do his residency training. He then returned in the late '70s to do his fellowship in nephrology. And in addition to that, he did some research with the late [INAUDIBLE]. His wife at the time was Joan Liverman. He has a son Eric and Astrid. All three of them were at the dinner last night. Philip unfortunately, died very early in life. He died in 1984. And his family and friends established this lectureship. And as you can see, we've had a series of distinguished lecturers throughout the last couple of decades. And so Barbara is the Murray M. Rosenberg Professor of Medicine. She's a chairman department of medicine at Mount Sinai Health System. And she's also the dean for Clinical Integration and Population Health. She received her education and residency and fellowship training in Ireland. And then she came here to do additional nephrology fellowship training at the Brigham Women's Hospital, and then received her research training with Mo Sayegh and Charles Carpenter in the Laboratory of Immunogenetics and Transplantation at the Brigham Women's Hospital. She rose through the ranks at Mount Sinai after she became an assistant professor, and is now the chairman of medicine. Barbara has a number of honors and awards. She was the president of the American Society of Transplantation. She was elected fellow at the American College of Physicians, honorary fellowship, Royal College of Physicians in London. She is counselor of the American Study of Nephrology and will be the president in 2022. And she's on a number of editorial boards and an editor as well. Her research focus is in translational studies of kidney transplantation. And her work is funded by a number of different NIH grants. She is an active clinician and educator. And she has been invited to speak at over 145 events internationally. And she has a number of articles and chapters and reviews and editorials. And before I have her come up here, I just want to ask what all of you are doing tomorrow at 6:09, because I know what I'm doing. So without further ado, I want to I have Barbara come up and give us her lecture. So thank you Barbara, for coming. Thank you very much Mark, for inviting me and for your very kind introduction. It's really been a wonderful visit. As you can tell, I got to enjoy the flora and fauna yesterday out walking around and have a rapid acceleration of my allergies. So bear with me on this one. And it really was wonderful to meet the Liverman family yesterday, wonderful, lovely group of people, and honored to give this lecture. So I'm going to do today is talk about some of the research that we're doing in my lab in conjunction with several other centers. I just want to point out that I do have a conflict, a recent conflict, which is kind of exciting one, which is that we've formed a company out from Sinai to pursue some of the findings that we have, as a result of the research that we're doing. So I just want to start this case out, very simple case. A woman I've known for years, a 24-year-old woman of Italian descent, who had end-stage renal disease due to an unknown cause which started as a child. And she receives a living donor transplant from her brother, which is a haplotype match. She received induction therapy. And it was because it was many years ago, she received cyclosporine, prednisone, and MMF. She had no rejection episodes. I significantly doubt that she had any sorts of issues with compliance because I've never seen a more overseeing mother than this case. But she lost her kidney to a chronic allograft nephropathy after three years. And she was very fortunate that she had three brothers. So she received a second transplant. So she still has one in the bank that I hope she never needs. So she receives a second transplant from her brother. She never needed dialysis in the interval. Again, a haplotype match, received induction therapy, Tac, MMF, and prednisone, no ACR. And she still has this kidney working totally fine after 10 years. And we've no notion why one would not work and the other would work. And the questions are is it that HLA, was it a different haplotype? Was it the Tac? Was Tac so much better that she didn't develop damage? Or it was this because she was now a teenager and into her early 20s, and she had compliance issues? But the bottom line is we don't know till we know that there's damage. So what can we do to pick this up earlier and identify the course earlier? And that's something that's fascinated me for a while. Some of the people who were at the talk earlier on will recognize some of these slides. So this is a slide that is used very frequently in transplantation to demonstrate that we've done a wonderful job in improving one year graft survivals and decreasing acute rejection. But I'd like to point out two things. One, this is at the cost of these individuals here who didn't need all of this immunosuppression and did OK. There's a group of people at this stage that do OK, that didn't need to receive induction therapy and three maintenance immunosuppressive therapies. The other issue is although one year graft survival has improved, that we have not done a good job with regards to long term survival. In fact, if I show you data from prior to 2000, we'll show no improvement in one year survival, but we've slowly started making inroads onto long term graft survival. But it's really not anything to be proud of. And we still are losing grafts at an unacceptable rate. The most common cause for allograft loss is chronic allograft nephropathy. And here are-- for actually it's death with a functioning graft due to a cardiovascular disease. And the second is chronic allograft nephropathy. We in transplantation will discuss the name of what we call it. Is a chronic allograft nephropathy? Is it interstitial fibrosis and tubular atrophy? What is it? It doesn't really matter what we call it. We all know when we see it. And it's glomerulosclerosis, fibrosis, and tubular interstitial-- tubular interstitial fibrosis. And it is the final common pathway for all damage in the kidney, much as a similar picture is the final common pathway for damage within chronic kidney disease. And we have different ways of quantifying that and determining how to score it. We call it, as I said, tubular atrophy and interstitial fibrosis. And if you look at that score and you grade it, and the worst the IFTA is, the worst the graft loss. The other score that can be used is the CADI score, the Chronic Allograft Damage Index Score. And again, the worst that is, the worst the graft does. And the only difference is that this incorporates inflammation into the score. And now what we're doing is starting to look at IFTA, I IFTA, inflammation and scarring. And all of them are similar, will be found in graft cycle one, two failed. And this is important for many reasons. This is important not only because of the individual themselves that loses their allograft and has to go either through a second transplant, if they're fortunate, or who returns to dialysis with an unacceptably high mortality, but it is also an issue for everyone else, because it puts an additional burden on the transplant waiting list. You see roughly every year there's about 11,000 to 12,000 patients that are on the transplant list that have had a previous transplant that failed. And we've said that this is declining. But it's not the percentage is declining, but in fact, the number of transplants has increased. So therefore, it looks like it has declined, but the total number has remained the same. And at any given time, we have about 25% of patients are waiting over four to five years for a kidney transplant, nationally. With a mortality of 23% per year on dialysis, these are unacceptable data. They're unacceptable statistics. And there are many factors that impact this and contribute to this. And this is looking at the Kidney Donor Prediction Index, a KDPI. And the worst the donor score, the more likely you are to lose your kidney. You get a bad kidney, you do badly, you lose your kidney sooner. And so that's very clear here for a KDPI of greater than 85%. But you look here, anybody below 85 is bunched together. It doesn't spread out. So there's clearly other factors that contribute to why a kidney is lost. And that's many, as I said, there's donor factors. There's factors related to procurement and delayed graft function, cold ischemic time, toxicities due to the drugs we use it themselves, and as I discussed this morning, the issue of underlying inflammation within the graft that leads to long term damage. We do a phenomenally awful job in predicting who's going to lose their kidney, even though we know there are these factors, even though we know a person that receives a bad kidney is going to lose it sooner. This is data from our study. And it points out to two groups. One is the overall group, all comers, including kidneys that have fibrosis on their kidney already by three months, recipients of kidneys that were not great donors, extended criteria donors. And if we use those clinical criteria to predict who goes on to lose their kidney, you get AUC of about 0.73. If you take people who have great donors, whose kidneys look pristine at three months, there's no fibrosis on them, and you try to predict who will go on to develop fibrosis, and the AUC is 0.646-- and I'll show you a little bit more of this later on. So we do a bad job of projecting who's going on to lose their kidney, and therefore, I think we do a bad job of trying to protect those individuals. Our standard approach to this, our gold standard, is biopsy. However, it's invasive. And the trigger for a biopsy is an increase in creatinine, which as we all know, is not a sensitive marker of graft injury. By that stage, you already have fibrosis in there in the allograft. And we know from numerous papers-- these are just two papers-- there are numerous papers that show that if you do protocol biopsies, you can identify fibrosis within those allograft. So there's evidence of fibrosis in about 50% of patients at one year in individuals that have normal kidney function. And that you can see progression between protocol biopsies. And these dates back to 2002. And there's numerous papers in the interval to show that protocol biopsies identify individuals with fibrosis before you have allograft damage. So our current approach is not only invasive, it's reactive. It suffers from the issue of tissue sampling. So not only is it the biopsy tissue sampling, but we're only looking at a sliver of the biopsy. So we're not capturing truly what's going on. We determine arbitrary cutoffs for what's important and what's not important. And so we score-- and even with as I was discussing this morning, there's subclinical acute rejection, there's borderline. We decide borderline isn't important, but it's a continuum. And so if you're looking at injury and inflammation, it leads to further injury and inflammation unless you break that cycle. And there's also the issue of subjectivity of reporting. And we actually have data from our own goCAR study to show that a major discordance between local reporting and centralized reporting by three transplant nephrologists. And in the data that I showed with regards to subclinical acute rejection, it was missing about 50% of the cases. So we felt the approach using molecular approach to prognostication and diagnostics had several advantages. First, it's an unbiased examination of the biopsy. It aids with potentiates with appropriate classification. And there's some data from Phil Holleran to look at that. There's the potential to avoid a biopsy if we can actually identify gene expression in the periphery, and that you might actually identify the molecular changes before you actually identify the subclinical fibrotic or damage in the graft. And lastly-- not 30, clearly I can't count-- lastly, there's the opportunity to identify key drivers and potential therapeutic targets. So we took a different approach. There's a lot of papers out there-- there's a lot of papers out there that looked at gene expression within allografts with fibrosis. And what they did was-- that a lot of these are very small numbers, but they looked at kidneys that already had damage to look at the differential gene expression between those that had damage and those that did not have damage. And so what we wanted to do was look a differential gene expression in the biopsy before the damage occurred so that we could understand the pathogenesis of it and the potential for an interruption of that process. So what we looked at is here, if you look at fibrosis within the graft, you're looking at tissue injury, looking within the tubular glomerulus or the endothelium-- getting endothelitis-- activation of the residents and infiltrating white blood cells with psychocytosis of the regional debris that promoting pro-inflammatory molecules that further drive the process recruitment of macrophages and lymphocytes, and then this initiation of this pro-fibrotic process that leads to damage and matrix deposition. And so what we wanted to do was look at gene expression at this phase before that vicious cycle has occurred to see, could we identify targets that might help us interrupt that process or identify individuals that are at risk for that process to treat them differently and prevent the process. So our hypothesis was that chronic graft injury, and therefore, allograft loss can be predicted by gene expression profiles within the graft and or the periphery early post-transplant. And what I'm going to focus on today is within the graft. We enroll five-- there's actually six different centers, because one of our PIs moved in the middle of the study. And we followed them and enrolled another center. And as I mentioned, we had Bob Collins's group where the lead for the pathology corps. We also did immune phenotyping, and then this, the Bioinformatics and Genomics Court Sinai. And so for the entire cohort, we did a baseline and 24 month biopsy for about half of the patient population, because we limited it to two sites. And then additional sites that started doing it as protocol biopsies a standard of care, we also got their samples also. We did biopsies at 3 months and 12 months. So about half the cohort, about 240 patients had a biopsies done at baseline 3 month, 12 month, and 24 months. And overall, we had 588 patients enrolled in the study. And then we also had gene expression peripheral blood for gene expression. And we had a donor recipient SNPs. So it really is a very well-informed cohort. And we were actually able to follow these patients at long term, because we have their [INAUDIBLE] and [INAUDIBLE] data linked to the clinical ongoing registry. And so some of these patients we've been able to follow at now to nearly eight years. So for this study that I'm talking about today, what we did was we did a 3 month biopsy and a 12 month biopsy. We took the 12 month biopsy for a histology, and graded it by Banff and also looked at the CADI score, and delineated patients into high or low CADI. We then also took the 3 month biopsy and did microarray to look for differential gene expression that differentiated between high and low CADI. And so this is that initial 204 patients that we had, 159 patients that were for the initial cohort, and then a validation cohort using PCR. And then we went on to validate this in different ways that I will talk about. When you look at the patient demographics, important to note that these are patients that were not high in the logical risk. They did not have preformed donor specific antibodies. So we did not take patients that were highly sensitized, because we wanted to look at the contribution of de Novo donor specific antibodies to the development of graft injury. And we felt that highly sensitized patients where a different pathology, and that we should leave them out. So we're predominately looking at the patients that are not developing ABMR, Antibody Mediated Rejection. So there really was very little different between the group, apart from the anti-HLA, particularly class two antibodies, and including non-donor specific. This is just showing the difference in outcomes. This is just within a three year follow up that we had in patients. And this is born true going forward that CADI correlated with long term outcomes-- 12-month CADI. [COUGHS] Excuse me. And so what we did was we looked at genes that correlated with 3 months we looked at genes that correlated with 12 months. And we did this 100 times and removing one patient each time. So identify genes that occurred two or more times. And if it occurred two or more times, then we felt it was significant. [COUGHS] Excuse me. Then what we did was we looked at genes that were correlated to 3 months and 12 months. And we only took genes that correlated with 12 months. We didn't want to use genes that were like, collagen and other indicators that there was already fibrosis. We wanted to identify genes that led to the subsequent development of fibrosis. We didn't want to just identify a bad kidney. So we found 169 unique genes that correlated with 12 month fibrosis. Sorry about this. And then we adjusted for clinical characteristics that were associated with fibrosis. And then we did penalized logistic regression analysis. And then what we did was we broke it out into three groups, randomly into three groups. And we did this 100 times. And we analyzed one group to see did the genes work on the one group. So it's very rigorous statistical methods to ensure that we weren't overfitting. And then we went on to validate it in several external cohorts. When you look at the genes that are in the biopsy at 3 months associated with fibrosis at three months, versus the genes that are in the biopsy at 12 months and associated with fibrosis at 12 months, they're very different genes. The 3 months associated with fibrosis at 3 months of predominately immune response genes. And if you look at the genes that are associated with 12 months, there's a lot of cells signaling metabolism, and actually some B-cell related genes. This is just showing that me gene expression correlated nicely with high to low CADI. And if you do immune cell enrichment for the genes in the biopsy at 3 months associated with fibrosis at 3 months, it's predominately dendritic cells. The gene immune cell enrichment to identify cell types that are associated with fibrosis at 12 months, it's dendritic cells, macrophages, stromal cells, and CD4 T cells. So clearly, breaking out very nicely between the two endpoints despite the fact this is the one kidney at the same time. We identified 13 genes that predicted development of fibrosis. And these are the 13 genes. And this is a paper that was published in Lancet in 2016. And using those 13 genes-- this is the training set. And AUC of 0.967. If you use clinical factors only, at the AUC is 0.7, and clinical and pathological it's 0.8. But what's different here is AUC doesn't tell the whole story. If you look down here at sensitivity specificity NPP and PPV, the 13 genes clearly outperformed either clinical or clinical plus pathology for predicting who would go on to develop fibrosis. We then took a totally independent second cohort in goCAR and used R2 PCR with the 13 genes, and had an AUC of 0.991, and similarly very good performance with NPP and PPV. Obviously, if this is true for a CADI score of greater than 2, which was a pretty low cutoff, it should work if you change your end point. So we've done this in multiple ways to make sure that it still worked. And so if you took a CADI score of greater than or equal to 3 or greater than or equal to 4, it also worked with different cutoffs. I mentioned that there's this discussion around the endpoint. What is the right endpoint? Is it CADI? Is it IFTA? And to make sure know it wasn't just being driven by inflammation, we used the IFTA score. So this is taking out a CADI and using IFTA to make sure that-- and we also did it another way. We took out all biopsies that had any positive RI or T score, any sign of inflammation, and we reran the analysis, and found that we still had a very good MPV. The PPV here isn't as good, but it's a small population and a very low incidence of the endpoint. So then the important thing is if you take a kidney that's good does it still work? So it's interesting when you have longitudinal biopsies and you see what happens CADI score over time. You can see some people have a high CADI score here and it goes down. I don't think that they're recovering. I think it's probably an issue of sampling error. You have some people that are high that stay high, and then you've people that are relatively low CADI that stay low and some that go high. And we had to find criteria for identifying individuals that were progressors and non-progressors from this group. And when you look at these, that you can see here-- I hope you can see-- that progressors and non-progressors, there is no statistical difference with regards to any of their scores at 3 months, but they clearly broke out by 12 months. So when you look with these patients-- and we gave this patient to pathologists to try and predict. And they were totally unable to predict which would go on to develop fibrosis. The 12 month that you see for these patients was 0.916. If you look at that 24 month AUC is 0.846. And in fact, what's interesting, you go and look and see why does it drop off for 24 months? Well, there were patients in between that actually-- this is why sometimes I'd like reviewer questions-- they went why-- what was different about those patients that had fibrosis at 24 months that had a good score? Some of them actually developed PK. And so we couldn't have predicted that it was a factor that occurred afterwards. And then if you look at the clinical criteria, this is clinical plus pathology at 12 months for these patients, 0.75. And it gets worse for just clinical and alone for 24 months is 0.641 where for 24 months it's 0.584. We also went and validated this on publicly available data. There's two data sets. One that you see is 0.83 was a study that was very much structured like ours, very low number in the study, and it was for pediatric patients. And then there was a very large cohort, a large cohort, which actually what they did was they biopsy patients anywhere between and 6 months and 21 years and looked at graft loss. And we used our data to predict graft loss. And it worked very nicely. So that's taking a biopsy at any time point post-transplant. This is just showing that CADI correlates with EGFR-- inversely correlates with EGFR. And this is showing this prediction of the correlation or ability of the 13 gene set to predict EGFR at 12 and 24 months. And if you did-- I just blanked a name-- principal component analysis, you could stratify your patients into high and low risk using the 13 gene set for graft loss. And that cohort that I said was from anywhere from 6 months to 21 years. Again, very nicely breaks patients into high or low risk for graft loss. And this is just looking at this a different way, the AUC for two years at 0.84 and 0.844 for three years. That's from the time of the biopsy and here-- that's from three months. And for here in this other cohort, it's from the time of the biopsy. So in conclusion for this part of the study, I think we've identified a 13 gene set that nicely stratifies patients at risk for development of fibrosis and graft loss. And it's interesting potential applications potentially I think for modulating immunosuppression, identifying individuals for a more aggressive management of hypertension, hyperlipidemia, diabetes as we would a chronic kidney disease, but also if we understand the immunological risk of those patients, potentially switching them from a pro-fibrogenic agent, such as Tac [INAUDIBLE] potentially to another agent that will be less likely to cause fibrosis. The other point is that we've done a very bad job of stratifying our patients for risk and for studies that might potentially be useful for risk stratification in studies. But this data also has other potential uses. So we've been using it-- and I'm not going to talk today-- around immune stratification, detection of subclinical inflammation in the biopsies. And these are the three people actually here, [INAUDIBLE], and [INAUDIBLE] that have really done the main work around for the data that I've presented. We've also used this data looking at differential gene expression for identification of novel targets, drug development, and drug repurposing. And what's nice is we took the data on the 13 genes that we had and went to Nephroseq. This is patients with chronic kidney disease. These are patients without chronic kidney disease. And it turns out that this is a really nice model for chronic transplant. It's actually a very nice model for chronic kidney disease. In chronic kidney disease, obviously, we can't biopsy the patient before they have fibrosis, because we A we wouldn't be allowed. And we just don't ever get the opportunity, even if we were allowed. But it turns out that these genes correlate with development and progression of fibrosis in chronic kidney disease. And in fact, these are the genes that are top differentially regulator for CADI. And it actually correlates with-- this is the EGFR starting with an EGFR here of less than 60-- of 90 going down to those with end stage renal disease. We also noted that it differs depending on the disease. This is diabetic nephropathy. So diabetic nephropathy versus healthy living donors. It's clearly different. And it worked for pretty much all of the diseases we looked at apart from minimal change in hypertension. So it turns out it's a very nice model for chronic kidney disease. And so we've taken this and the data that we've seen and gone back into mouse models of chronic kidney disease. And we've identified novel targets in the data that we have from the transplant patients, and then examined several of these identified in different ways, in mouse models of chronic kidney disease. One of the genes that we identified early on was Shroom. It sort of popped up because it had an interesting name, but also because it turns out it is one of the genes that is most differentially-- that is most predictive in GWAS studies for decline of EGFR and CKD. And it turns out what was funny-- as we increased patients, it dropped off in significance. And it turns out that as we increased our patient pool and goCAR, we had more African-American patients. And it turns out it's important if the recipient receives a white donor. So this is looking at the odds ratio with regards to development of fibrosis if you take all donors. And it's particularly here actually, it's particularly associated in this case with living donors. So it turns out that Shroom is an interesting SNP. It correlates with development of fibrosis. The SNP is an intronic SNP that correlates with expression. And that expression obviously then correlates with development of fibrosis. Here, you can see the difference. This is either taken linearly or as a dichotomous variable. And this is just showing in a UUO model. So that was the data, the data that I showed in humans. And then moving here to UUO model, showing that Shroom is associated with increased fibrosis, and that if you knocked on Shroom it's protective for development of fibrosis. Now, [INAUDIBLE] in the lab has taken this to different heights. Because if you look at the data with regards to Shroom, it's actually associated-- if you knock it down it's associated with protein urea. So it turns out it's a different role in the tubular versus the glomerulus. It's protective in the glomerulus and it's damaging in the tubule. And it is a different-- [INAUDIBLE] has demonstrated that it has a different regulatory mechanism leading to fibrosis in the tubule versus being protective in the glomerulus. And he's dissected this out very nicely and just had this published. And Jason-- and very nicely, I'm happy to say that he has just got his first [INAUDIBLE] based on dissecting out this front row in the kidney. So Chen in the lab also has been working on this. He's taken our data he's taken publicly available data and did a meta analysis to identify key drivers to fibrosis and identified HCK. HCK is a star kinase inhibitor. And he's shown in Nephroseq that it is associated with IGA and nephropathy and lupus nephropathy. It's also associated with diabetic nephropathy. He's shown that in a mouse model, that it's associated with diabetic nephropathy. And you can see here that he's use in different-- multiple different models, including UUO and lupus model and diabetic kidney model to show increased expression of HCK associated with kidney disease, src kinases overall, but particularly at CK. You can see here from his data that in a UUO model using dasitinib to inhibit HCK. Now it's not a very precise way of inhibiting HCK, because dasitinib inhibits multiple Src kinases. And I'll get back to that. And this is more data just showing very nicely that again, that decision of decreases fibrosis in the UUO model, and correlating this with HCK expression. Now, as I mentioned, dasitinib is not necessarily a great agent, in fact it's used a lot in cancer. And we've now just made our Cancer Institute aware that it is actually associated with fibrosis. Everyone in our group has just-- it literally says impressed. We got a email yesterday to say it's up and going in Nature Communications. And he's shown that dasitinib is associated with [INAUDIBLE] cross effacement. So it's not a great agent to use to prevent fibrosis. And what we actually know, as I said, have got our colleagues in oncology monitoring the patients for protein urea and they're using this agent. So what we have done with [INAUDIBLE], who is a medicinal chemist, is we have start developing small inhibitors for HCK in the lab. He's refining this. Within, it's very hard to get specifically HCK, but he's got a scr-- he's got a src kinase inhibitors. And what we're trying to do is specifically get ones that are stronger for HCK inhibition, but decrease the fin inhibition, because we believe it's the impact on fin that's causing the protein urea. And so we are working through this. And now he just gave me another 12 new agents this week. What we've also find is that this also impacts-- with [INAUDIBLE] and Samira in the lab-- that it very nicely impacts T cell proliferation with spares proliferation of T regs. So these agents might be interesting for different reasons. We have other targets, other interesting molecules that we're working on. And in fact, Samira is-- the way we've picked them is in kind of this random way where we do searches, see that it's kind of interesting, definitely that it's novel. But now Samira has been working with the people in the Zebrafish lab to start screening the top 85 five genes to see which ones might be of most interest in the kidney. [INAUDIBLE] is a superb person who's working in the lab, and works with Joel Dudley, who had developed this major database around drug repurposing. So she took our data and did an analysis to identify drugs that already exist that might be beneficial in targeting as many pathways that lead to fibrosis as possible. A lot of these drugs interestingly were anti-convulsions antipsychotic agents. I had this postulated idea that it was because the synapses like the podocyte. It turns that they're just dirty drugs that impact lots of things. So we avoided them because they have lots of horrible side effects. And we picked two interesting agents, [INAUDIBLE] and [INAUDIBLE]. One impacts [INAUDIBLE] and the other actually impacts-- [INAUDIBLE] affects the retinoic acid pathways. And we use these two agents in mouse models. We did it several models. I'm just showing a UUO model here to show that they actually decreased fibrosis in a UUO model showing that we can take pre-existing drugs, use this novel mechanism-- Actually, Joel has identified a statin that is now being used as an chemotherapeutic agent using similar mechanisms. But it's a really interesting novel way to try and identify targets that are already approved to solve clinical problems that we're dealing now. I don't think these two agents are going to be game changers with regards to fibrosis, because they had effects, but they weren't overwhelming. But it was a nice demonstration that you can do this. What we've done now and are submitting is we've done similar things with acute rejection, and have shown that we can use minocycline in conjunction with low dose cyclosporine in a mouse to decrease acute rejection, with something that you could very nicely add into an acute rejection regimen or a [INAUDIBLE] immunosuppression regimen that has some benefits itself, and may actually impact the immune response as well. So it's a novel way of approaching this. And I show these different ways for people like the residents and that just to demonstrate the amount of different ways you can approach using data that is like this. So it's an incredibly rich resource for us. And they're a very well characterized group of patients that I think we can continue to explore and develop collaborations we're talking earlier on, but potentially looking at iron and iron metabolism. And we already have the data. So a great opportunity for collaboration. So in conclusion, I think we've identified key drivers and key pathways in the development of fibrosis. I think targeted interruption of these genes and pathways enable the identification of novel therapeutics for the prevention of fibrosis intervening before the fibrosis develops, and that I think very importantly that transplant is a wonderful model to leverage for further understanding of chronic kidney disease. So I've kind of outlined this. And again, these wonderful group of patients. John He, Cijiang He, tremendous collaborator. I am not a CKD investig-- I was not a CKD investigator. And he has helped me set up all of these models within my lab. All of these people who've been involved in doing all of this work, a superb group of people to work with. And I mentioned Christopher always, because he orchestrated thousands and thousands of samples around the world. He kept track of them beautifully thank goodness. And as I mentioned this morning, he's now doing forensic medicine. So there's also this wonderful group of people in all of the transplant centers that have really made this work, and the patients, an incredibly rich resource to the transplant community in the patients who went out of their way to have these biopsies. And as I mentioned to somewhere earlier on, I worked with a lot of people who don't do bioinformatics who don't understand-- I don't have-- I don't understand what they do. They don't necessarily know what we do. But by bringing this together, it's been wonderful and to have them understand what an incredible resource this was, where I can tell them what has happened to all of these patients. And they were shocked to discover how many of these patients have lost their kidneys, and how many of these patients are now dead. So we are beholden to these patients. And we have an obligation to really do good work with these samples. So thank you very much. [APPLAUSE] Thank you, Barbara for that terrific lecture. Are there any questions? [INAUDIBLE] So thank you very much. Of those 13 predictor genes, how many are [INAUDIBLE]? It's totally different. It's totally different. I actually showed a slide this morning. The peripheral blood at the same time is totally different. So a lot of the genes in the biopsy at 3 month are down regulated in the blood. In fact, the blood that most reflects the biopsy at 3 months is the recipient blood before they're transplanted. So that's most part [INAUDIBLE]. So yeah, they're not informative in the blood at all. We have a different group of genes that we're working on, in fact, MIRNA in the blood at 3 months that helped us identify those at risk for fibrosis later on. [INAUDIBLE] related question. [INAUDIBLE] maybe related is-- [AUDIO OUT] Yeah. We have looked at pathways. We've done the pathway analysis and looked at other factors that might be interesting in pursuing. And you are right. It is more up or down regulated. In those genes individually by themselves, not all of them are statistically significant. But when you take the combination, becomes significant. So they're not all significant. Obvious next thing I would think-- [INAUDIBLE]. Yeah, that's always-- that was the reaction I kept on getting. Everyone's like, so what do you do? But Bella is actually a really good drug. If we know that you can switch someone to Bella, you could switch them over to Bella quite nicely. And but understanding their immunological risk. And when you're at the same time, so you're not switching person. But a lot of them have declared-- so yeah, so you understand the immunological logical risk before you switch someone over to Bella as well. And a move to Bella is away from fibrosis or towards better [INAUDIBLE]? Could be both, but I think getting the CNI off might be helpful. But again, more questions to ask. But I think getting the CNI out of there would be beneficial if you're saying someone's pro-fibrotic. Other questions? Barbara, thank you very much for the wonderful-- [APPLAUSE] Are there other pathways that-- [INTERPOSING VOICES] So the interesting thing is we can also-- we haven't got very far with it yet, but we have identified in the recipient who before they're transplanted, we've identified pro-fibrotic people. So the question is, can you pick the person who isn't pro-fibrotic before they get the transplant and understand that biology as well. But there are clearly pro-fibrotic people. And so the nice thing with that is we don't have to wait for the transplant. We can go and study them, [INAUDIBLE] the dialysis we can study them and then go look at them in CKD. There's so many things. I just need to [INAUDIBLE]. [INAUDIBLE] cardiology. I just wanted to say that was a great talk. Thank you very, very much. I love your idea about that you really have to test these things [INAUDIBLE] their tolerance [INAUDIBLE] differences in outcomes and what we do [INAUDIBLE]. I think in medicine overall-- and this is something that I've been as chair of supporting in other areas-- is doing a better job of earlier detection, stratification, and just stop throwing drugs at people randomly to see what happens. They're not test tubes. Ready, aim, fire. Yeah, exactly. Oops, that didn't work. Anyway that was a really good talk. Thank you so much. Thank you. Thank you. --really lovely. Thank you very much. I wanted to follow up on his question with differential gene expression. --ask you a little bit more deeply about [AUDIO OUT] --in the context of the genes. --or in potentially other genes it might be [INAUDIBLE]. We've looked at that. We're working on that as well. Yeah, I can't give you the exact data, but we are working on it. There's just so much to do that we haven't-- and we've been trying to get the papers out for these certain number of papers-- but there are-- I mean, the Shroom is the example of where there's care-- and then so what [INAUDIBLE] has actually done is-- were you-- we're you at the talk? OK. So what [INAUDIBLE] has actually done-- I didn't mean that as [INAUDIBLE]-- No, I'm sorry. No, no, no. I meant because I don't want to repeat myself. What he has done is, because he's been able to look at-- we've done RNA seq on a smaller number of patients. Unfortunately, we had microarray because we started [INAUDIBLE]. But with the RNA seq on a certain number of patients in the biopsy, we've been able to-- and we have donor recipient SNPs. He's been able-- you can infer from the RNA seq the SNP. And so then we can tell in the graft whether the expression is donor or recipient within the graft. So we're doing stuff like that as well. But we have looked at the SNPs for these specific genes as well. We just haven't moved forward with doing anything major on it yet. [INAUDIBLE]. I was really thinking what you're going to do with the immune signatures. Are you puzzled by that? Or do you find that-- I think what it is is we've taken-- so we've taken all genes. We've had totally on supervised approach to this. And so when you're doing that, some of the immune signatures get drowned out by other things. So other thing-- But those other things are still differentially regulated. So we could narrow it down and just look at immune and identify a pattern. But we didn't do that. We wanted to look at all comers and realize that not all injury is immune mediated, that there is other ways for fibrosis-- there's other initiators. And just have a broader picture too. So when you do that, you're looking at everything. And so sometimes the immune genes get a little drowned out by those. And so even some of the genes are not necessarily genes that are a major role in the development of the fibrosis, the predictive genes are not necessarily pathogenic. So it's interesting, yeah. It's interesting. It's great to see you. [INTERPOSING VOICES] Good to see you. Thanks very much, and thanks very much for showing me around. --donor and recipient-- [AUDIO OUT] It's not even that. I think what it is is the immune response. The immune cells have moved in. They migrate from-- they've moved into the kidney. So they're actually down regulated in the periphery because they've moved into the kidney. So is what you're [INAUDIBLE] simply more of a donor-- --combination, depending on which gene you're looking at. So do you think [INAUDIBLE] best tests for donor tissue-- Potentially. Potentially and we do have-- we do have the baseline biopsies as well. So we can go and look at that. We haven't even got there yet. More things to do. Lots of great questions. Thank you very much. Take care. Thanks. Lovely to meet you.