The first slide has to hit with the mouse. It's like a activation thing, I don't know why they do that. Because this was an issue. Right, yes. No one told me. You're the second one, yes. Yeah, so whoever comes up, they hit their name, click the first one with the mouse, then they can walk around the room and use the advances. Okay, got it. You have to hit the first one with the mouse. Okay, great. I love it. And as far as, I'm just going to see, as far as time, they're showing 15 minutes per person. Yes, so I think it's about usually 10 minutes per talk and 5 minutes per Q&A or something like that. Okay, so what I'll do here is... And that's pretty much it. I mean, I can leave. Do you want the first slide up or the walk-in slides or what do you... Yeah, let's put the first slide up. Okay. So we help that person. Okay. Oh, you know what? Put it in the walk-in this way. It's easy. So yeah, go back. Okay. Because then I have to announce them, so this way I can say his or her name and figure out... And like I said, it's just after you hit start, one click with the mouse first, and then they can either use their finger. Yeah, it's just the regular click. Okay. And then they can use their finger or this. Okay. Because either one won't get activated until you hit it with the mouse first. I should have had you yesterday. Any problems, we'll be out in the hall. Oh, you're awesome. Thank you. Thank you. Go Bucs! Go! Thank you for watching. Thank you for watching my video. I did, actually. I've been working on it, so I'm sure that's something that needs to change. Thank you for joining us, this has been a wonderful experience. Thank you for watching. Thank you for watching. and th th Monday, Tuesday, Wednesday, and Thursday. Okay, good morning, everyone. Welcome. My name is Dr. Stephanie Seminera. I'm from Mass General Hospital in Boston, and I welcome you to this late-breaking abstract session. We are all here because we want to know the latest science and clinical medicine in endocrinology, so this is going to be a very exciting session. Just remember, we're all going to be good colleagues here today. Please don't take photographs of any of the slides or content. We have great methods to contact any of the speakers, so if you need to see their content again or review it or have questions that we're not able to cover, just contact them directly. Our format's going to be traditional today, about 10 minutes or so for the material and then 5 minutes or so for Q&A, so 15 minutes total for the speakers. When you ask a question, as always, just come to the microphone, state your name and your institution. And in addition to the late-breaking science, my back is also late-breaking, so if you see me wincing up here, it's not in any relevance to the presentations. It's just a little bit hurting today, so just ignore me. Okay, so with that, we are excited to begin, and our first speaker is Thais Rocha. The title of the talk is Machine Learning-Based Steroid Metabolome Analysis in Women with Polycystic Ovary Syndrome, reveals three distinct androgen excess subtypes with different metabolic risk profiles. Good morning, everyone. I would like to thank the Annual Meetings Terry Committee for the opportunity to present our work. So polycystic ovary syndrome affects 10 percent of women and is associated with an increased risk of metabolic conditions such as obesity, type 2 diabetes, and fatty liver disease. Androgen excess is a defining feature in PCOS and has been implicated as a driver of metabolic risk. Pathways to androgen biosynthesis in women involve the classic androgen pathway with DHT as the most potent active androgen. However, androsinidione is also converted into 11-hydroxyandrosinidione in the adrenal glands, which is a precursor for 11-ketoandrosinidione and for the active androgen 11-ketotestosterone. Moreover, recent studies have shown that the 11 oxygenated are implicated in PCOS. Therefore, here we aim to identify PCOS subtypes with distinct androgen metabolomes and determine whether they differ in metabolic risk. This analysis is part of the DAISY PCOS study, which is an acronym for Dissecting Androgen Access and Metabolic Dysfunction, an Integrated Systems Approach to Polycystic Ovary Syndrome. It involves 488 women with PCOS, prospectively recruited in Austria, the UK, Ireland, and Brazil. It's an ethnically diverse cohort of approximately 30 percent of non-white women. The recruitment treatment naive adult women with PCOS fulfilling Rotterdam diagnostic criteria. Patients were referred to us due to difficulties getting pregnant, clinical signs of androgen access, or self-referred. Laboratory analysis including the serum androgen metabolome by LC-MS, including the 11 oxygenated androgens. Androgen metabolome results were then analyzed by machine learning-based unsupervised k-means clustering and a comparison of metabolic parameters between the identified clusters, including surrogate markers of insulin resistance, was performed. As we can see here on the left, machine learning-based unsupervised cluster analysis identified three distinct androgen metabolomes. Here on the right, we have a 3D representation of the clustering analysis showing how the three clusters are grouped in three clear, very different directions. This is a heat map where we can see the androgen metabolome of the red cluster. Oh, sorry, yeah, the red cluster driven mainly by ovarian-derived testosterone and DHT. Androgen metabolome of the green cluster, it's mainly driven by adrenal-derived androgens, such as DHEAS, DHEA, and the 11 oxygenated androgens. And finally, in the blue cluster, we can find that all androgens are lower than in the previous groups, characterizing a mild androgen access cluster. So our next question is, do these three clusters differ in the phenotype presentation? So with regards to age and BMI, the three groups are very similar. However, higher fasting insulin is significantly increased in the adrenal androgen access cluster, as we can see here. And we've also found higher OGTT 120-minute serum insulin in the adrenal cluster. We've also find higher OMA-IR in the adrenal cluster. And Matsuda insulin sensitivity index was also significantly lower in the adrenal cluster, as we can see here on the right. Finally, the adrenal androgen access cluster had a two- to three-fold higher prevalence of unpaired glucose tolerance and also newly diagnosed type 2 diabetes than the other two groups. To sum up, machine learning and androgen metabolomics identified three distinct clusters of PCOS. Androgen within the adrenal androgen access cluster appeared to be at the highest metabolic risk, as they presented with a significantly higher prevalence of insulin resistance, unpaired glucose tolerance, and type 2 diabetes. These results implicate 11 oxygenated androgens as drivers of metabolic risk in PCOS and provide a proof of principle for an androgen-based stratification tool to guide preventive and therapeutic strategies in PCOS. I would like to thank the PCOS, the DAISY PCOS collaborators from Brazil, Austria, Ireland, and the UK. And I would also like to give a special thanks to the amazing colleagues who made this study happen, some of them present here in the audience. And thank you all for your attention. Thank you. Okay, so this is now open for discussion. Question. Hi, Jenny Visser from Rotterdam. Really beautiful work, Thijs, and congratulations to your colleagues as well. What I was wondering of, I was quite intrigued that you were mentioning that the adrenal component in 11-keto androgens are more related to a metabolic profile, whereas you also have your ovarian component with high androgens. And to my knowledge, the testosterone and 11-keto testosterone have kind of similar potency. So how do you explain the differences in a metabolic component? Well, this is a question that is still to be answered. This is because the 11-oxygenated have been recently, not recently discovered, but recently, we have recently found out that they have an impact on the metabolic features of PCOS. So we still don't know why the metabolic phenotype is different, given, I know what you mean, like we have both potent androgens, but why this phenotype is different from the other. Also, there is some interesting data here, which I haven't presented for the sake of time, but we also found differences, not only in the metabolic profile, but also in the androgen access clinical symptoms. So we found that the androgen adrenocluster, hirsutism is more common, while in the ovarian cluster, alopecia, female pattern alopecia is more common. So those are questions that are still to be answered, but it's a great question, thank you. Andrea Denef, Mount Sinai. So a lovely, lovely study. And we certainly agree that PCOS is not a single disorder, and there are subtypes, and that this is a way to do modern disease classification, instead of hanging out in the smoke-filled room arguing. But you didn't present, or I may have missed it, what are the phenotypic differences in terms of BMI, LH, FSH, in your three groups? Do they differ there? Yeah, so the BMI, actually, age and BMI didn't differ between the three groups. And regarding, pardon, your second question was BMI and? LH. LH, FSH, yes. So we did analyze those, and we have an increased LH, FSH ratio for the ovarian and the adrenal. Oh, for both? Yeah, both. Is it the same? Yeah. So it's just for the sake of time, we have so much data, then we focus on the metabolic, but yes, we do have this. Okay, thank you. You're welcome. I think if I remember your data, a little over half of the patients were in your mild androgen group, which I think, when I put on my clinical hat, is always part of the frustration of taking care of patients with PCOS. They can be sitting in front of you with clear, hirsutism, clear phenotypic burden, and yet their androgens come back normal. So do you have any thoughts or hypotheses as to what is going on with that subgroup? Yes, so actually, keeping in mind that 50% of the active androgens are activated by the peripheral tissue. So, and the measurement of DHT and 11-keto-DHT, it's quite challenging, because of the very low concentrations. That might be one of the hypotheses, is that there is a peripheral activation out there, and that we cannot, we're still not able to capture with our methodological tools. Okay, and I noticed that your subjects were treatment-naive, not even treatment-washout, but treatment-naive. Yeah. Was there a, might you wanna elaborate on why the study design was that way? Because we know that in clinical practice, we've seen that some women, when they have been in a long-term treatment for PCOS, OCP, or metformin, that might impact on a long-term, let's say, treatment effect. So we opted out for treatment-naive patients to be more sure that we don't have the impact of any medications or whatever. So what happens, we also excluded patients on inhalers, on steroids, using steroids via any VIA, so topical steroids inhalers, which makes it more challenging, the recruitment. But yeah, so we wanted to make sure that no interferences were out there. Yeah, wonderful. Any other questions? Great. Well, thank you so much, terrific. Thank you. Okay, so our next speaker is Dr. Kelly Brewer. She will be speaking. She's from Mount Sinai, Icahn School of Medicine, and she will be speaking on the transethnic analysis of PCOS subtypes. Long-wide association signals reveal three shared subtype-specific loci. Now start. OK. Thank you. And first. Okay, so we know that polycystic ovary syndrome is a highly heritable and heterogeneous and complex disorder. It's characterized by three domains of disruption. We have the neuroendocrine abnormalities in gonadotropin secretion and action. We have reproductive abnormalities with increased androgen production and granulosis cell function. And then we also have metabolic abnormalities and insulin resistance in BMI. One of the ways that one of the approaches to resolving heterogeneity and complex traits, which is becoming quite popular as of late because of larger data sets that we have access to, is cluster analysis, as you saw in the last presentation as well. So the cluster analysis is used to detect patterns in your data. It allows you to find relationships in the variables. So here we performed unsupervised hierarchical clustering. We used the traits in the bullet points here on the left that are metabolic and reproductive traits that are characteristic of PCOS. We collected these traits and assayed them as part of standard protocols that we have for our genetic analyses in order to have precise phenotyping. So what we found was discernible patterns in our clustering. So the group that is labeled here in blue we called reproductive. It's characterized by a higher SHBG, LH, and FSH. The group here in red we called metabolic. It's characterized by higher BMI, fasting insulin, and fasting glucose. And then we also clustered a third group called background. We called background. It didn't meet the criteria for clustering. So we found these discernible patterns, and we were able to replicate these as well in other European ancestry cohorts and also in an East Asian ancestry cohort. So one of the things that we know about the different PCOS diagnostic criteria is that there's genetic similarity. So the paper that I have highlighted here on the left was a publication in 2018 where they looked at loci that were associated with PCOS. They looked at 14 different loci and found that 13 of them were not significantly different for patients that were diagnosed under NIH criteria, non-NIH Rotterdam, and also by self-report. So the general consensus was that there was similar genetic architecture regardless of the diagnostic criteria. But we know that when we cluster and we identify these subtypes, we do find that there are distinct loci that are associated with each of our subtypes. So how we do this is by genome-wide association analysis. So we have our discovery cohort. We use a separate group of individuals in a replication cohort. And then we also have a Korean cohort. So the publication for this Korean cohort was done in 2015. It was a PCOS GWAS where they did not find any signals that were significant in a full GWAS. So we took our three cohorts and we imputed onto the TopMed Imputation Panel. It is the best panel that we currently have. It gives the most genetic resolution to our data. And then we do this phenotypic clustering, and we do it by ethnicity. So our discovery and our replication cohort were clustered together because they're of European ancestry. And then we did the Korean cohort separately. And then we do a GWAS on each of our subtypes, on the reproductive, the metabolic, and then the background group. So what I have to show is our combined meta-analysis of our discovery and replication cohorts. This is actually an update to a previously reported study that we published in 2020. In that publication, we saw six different loci in our discovery cohort, but we were unable to replicate four of them in our replication cohort because of the panel, the imputation panel we were using. And so now we have an updated panel, and we're able to have more genetic resolution. After this, I'll combine all three together so that we can look at the discovery replication and the Korean cohorts together in a trans-ethnic subtyping meta-analysis. So this is the results of the European ancestry, the discovery, and the replication cohort. So what you see here are three different Manhattan plots. I'll explain those just briefly. For each plot, you see on the X axis the chromosomes, and then on the Y axis, it's the significance level. So the red line in each of these plots gives the point at which we reach genome-wide significance. So what we see here from this meta-analysis is four loci that are associated in our, with PCOS in our reproductive subtype. We don't know of any PCOS association with these genes in particular. The metabolic subtype, we see two loci, including C9-ORF3. C9-ORF3 is previously reported in Chinese and European GWAS studies. And the background group, we see two loci, including FSH-beta, which has also been reported in previous European GWAS, PCOS GWAS. And for that Korean cohort that we have that didn't see anything in their full GWAS, when we cluster and we do subtypes, we do actually see distinct loci that are associated with each of the subtypes. So combining that together with the two European cohorts that we have and do a trans-ethnic meta-analysis, we see that there are three different loci. There's one for each subtype. We see EPH-A6 in the reproductive subtype. EPH-A6 is a receptor tyrosine kinase. It's associated with neural development pathways. In the metabolic subtype, we see that C9-ORF3, again, as I mentioned, it has already been reported in Chinese and in European ancestry PCOS GWAS. C9-ORF3 has been associated in a Chinese population with ovarian morphology. It's also come up in European ancestry GWAS associated with some cardiovascular measures. Most recently, it's been associated with hypertension in a diabetes population. And then we have the background group. The gene that come the loci that comes up on this is FSH-beta, which as I said has already been found in European GWAS. So FSH-beta codes for the beta subunit of follicle stimulating hormone, which we know is important for general fertility status. Each of the SNPs that's identified in the C9-ORF3 and the FSH-beta are in LD with previous SNPs that have already been reported. Each of these that I'm reporting here that are genome-wide significant were significant in both cohorts. They had to have a p-value of at least .05. The lead SNPs are in the same direction. And we limited our analysis to a minor allele frequency of 1 percent or greater. So all of these results require a follow-up of functional and biological validation to explain their role in causal pathways of PCOS. So to conclude, we've demonstrated replication of eight distinct subtype loci in our European NIH PCOS. Three subtype loci are present in our transethnic meta-analysis. We believe that PCOS subtypes capture biological differences, unlike Rotterdam phenotypes, which have been shown to be genetically similar. And finally, the shared loci in the European and East Asian NIH PCOS is consistent with the presence of these subtypes in early population history. Thank you. is now open for discussion. Kelly, let me ask you, a big subgroup of that cohort were people with the background phenotype. They didn't correlate with your reproductive or your metabolic. So what do those patients look like, do you think, phenotypically? Kelly Edwards Yes, so phenotypically, I mean, they're sort of in the middle usually when we have our, with the cluster analysis. So as, you know, the reproductive has the higher SHBG, LHFSH. They sort of fall in the middle there. So we're not sure if that is what happens when you pull out the reproductive patients and the metabolic patients. It's sort of what's left, which is kind of why we're referring to it as background. But they, they're, phenotypically, they sort of fall, fall in the middle in a lot of those levels that we're testing. Okay, we don't have any ultrasound data on those patients? No, we don't. Okay, let me just repeat that for everyone in the back of the room. Dr. Dinaev was just saying that the reproductive subgroup, there is data presented in another part of the meeting showing higher follicle count. And so she believes that that captures some of the traditional ovarian morphology considerations that we traditionally think of with PCOS. I noticed, Kelly, that so you have this trans-ethnic comparison with Korean population. Will there be opportunities to compare your data, let's say, with other ethnic groups? Yeah. We're actually collecting Hispanic and African-Americans. So we do see the same type of clustering. And we've just submitted, actually, the genotyping for analysis for those groups. So we should have that data soon, actually. Do you think some of the positive signals that you initially had when they fell out, sort of, and you were left with the three that crossed across, do you think that's a numbers issue? You know, GWAS studies, there's always power in numbers. And do you think that if your numbers were higher, you could maybe get greater concordance across the ethnic groups? It's certainly possible. I mean, more numbers in genetics is always often better. I mean, we are very particular about who we're including because we believe a more homogeneous population will give us more power anyway. So we're very strict about our phenotyping in our PCOS anyway. So we do see, you know, these signals are quite robust. But yes, I do think that maybe we could find additional loci with more people that are well-phenotyped. Okay. Great. Any other questions? Okay. Terrific. Well, thank you, Kelly, so much. Thank you. Okay. So, our third speaker is Jewel. Yes. And so, they'll be speaking on leptin-mediated regulation of gene regulatory networks in gonadotropes. And Jewel hails from the University of Arkansas for Medical Sciences. Good afternoon, everyone. My name is Jewel Banik, and I have nothing to disclose today. So, today I'm going to talk about how leptin regulates the gene regulatory networks in gonadotropes. So, we all know that reproduction is an energy-expansive process, and it requires the maintenance of a healthy nutritional state to carry out a proper or normal reproduction. And previous research have shown that a failure to maintain a healthy nutritional status severely harms the reproduction, particularly in females. So leptin, which is an adipokine hormone, can sense the nutritional status of a body and can send those signals to the brain, particularly these HPG eggs to inform the brain that how to maintain that nutritional status in a body. Leptin is an, as I said, adipokine hormone, which is proportionately produced and secreted in the adipose tissues, and leptin usually binds to its corresponding leptin receptor, and by binding cell, it can exert its function in many tissues throughout the body, including every tiers of these HPG or reproduction eggs. And I want to draw your attention here to the pituitary gland. Gonadotropes, which is a cell type in the anterior pituitary, they specifically express a leptin receptor, and leptin can bind to those receptors and can exert its function in the gonadotropes and in the overall pituitary. So in this study, we wanted to understand how leptin influences the reproduction throughout the gonadotropes. And to test this hypothesis, we specifically deleted a leptin receptor from the gonadotropes in the pituitary in females, and we have found that those females become sulfatol. Then we also looked at the gonadotrope population, and we compared the cell population between the control and mutants, and we found that, indeed, there is a significant decrease in the gonadotrope cell population than the control. Then to understand what's going on at the transcriptome level, we decided to perform a single-son sequencing from both control and mutant whole pituitaries. And as you can see here in the right panel, which represents the different clusters that represent each cell type in the pituitary based on the canonical mRNA transcript expression. And I want to draw your attention here to these two cell population. So the gonadotrope population is highlighted by these green circles, and the lactotrope populations are highlighted by these blue circles. So when we looked at the gonadotrope population, we have found that there is a significant decrease in the number of gonadotrope cells in the laparonoid mutants, as opposed to the control. And according to our differential expression gene analysis, we have found that there is a significant decrease in the FSH beta expression in the mutants, but not in the controls. When we analyzed the lactotropes, we have found that there wasn't any significant change in terms of lactotrope cell population. However, there was a significant change or alteration in the lactotrope profile, transcriptome profile, in the mutants than the controls. So then it begs the question, what's regulating this FSH beta and how leptin is involved in this process? And why in the absence of leptin signaling in the gonadotropes, altering the transcriptome profile in the lactotropes? So to answer these questions, we took the advantage of CINIC, which is a bioinformatics tool. By using that, you can develop the gene regulatory networks from the single-sided sequencing data set. So the way it works is that it takes the gene expression profile from the single-sided sequencing data set as the input, and by using that, it calculates the co-expression transcription factors and its targeted genes. And based on this co-expression modules, you can calculate or identify the gene regulatory networks, which are known as the regulons. And briefly, the regulons are a set of genes that share binding motifs for a common transcription factors. And based on those regulon scores, you can classify those cells in different cell types. And again, I just wanna inform you that this classification based on regulon is not based on the single-sided sequencing gene expression. So we did this CINIC analysis with our single-sided sequencing data set, and indeed, we could classify all those cells into different endocrine cell types based on the regulon activity. So in the left panel, it represents the control cells, and in the right panel represents the mutant cells. And I want to draw your attention here to these gonadotrope-specific regulons, and we have found that if you look at the control gonadotrope regulons, indeed, some of the regulons are driven by some of the canonical transcription factors which are specific to gonadotropes, such as NR5A1 and GATA2. But when we looked at the mutants, we have found that in addition to a canonical marker of gonadotropes, NR5A1, there's some non-canonical transcription factors that drive some of the most highly active regulons in the mutants, and of course, it's happening in the absence of leptin. Then we measured the number of total genes that are targeted by some of the common transcription factors that represent some of the top regulons in both control and mutant. So here, the red bars represent the mutant groups, and the green bars represent the control groups. And we found that, as you can see here, in the absence of leptin signaling, these regulons and the number of genes that are targeted by each of the transcription factors indeed differ because of the absence of leptin signaling. Then we wanted to, as I mentioned earlier, that there is a significant decrease in the aphasis beta expression in the mutants. We wanted to understand which transcription factors and which regulons these aphasis beta is targeted by. And we have found that there are four transcription factors that belong to these four regulons, such as NIR1, GATA2, NHLH2, and SOX11, that specifically target aphasis beta in the controls but not in the mutants, even though if you focus on these number of targeted genes of those transcription factors, there are more number of genes that are targeted by these transcription factors but that don't target aphasis beta in the mutants, which might be explaining, or we think that leptin might be controlling these aphasis beta through these four transcription factors in the controls, which is absent in the mutants. We also looked at the leptin, lactotropin-specific regulons as well, and we have found that CRAB3L1 is one of the highly active regulon in the mutant lactotropes. And again, we also measured the number of genes that are targeted by each of the transcription factors in those common regulons in both control and mutant. And as you can see here, the red birds represent the mutants and the green birds represent the controls. In the absence of leptin signaling, there is indeed a drastic difference in the number of genes that are targeted by each of the transcription factors in those common regulons. And when we zoomed into CRAB3L, which is highly active in lactotropes, there was a three times higher number of genes that are targeted by CRAB3L1 in the mutants as opposed to the control, which might explain that in the absence of leptin signaling, this high activity of CRAB3L1 transcription factor might contribute to that alteration of the lactotrope transcriptome profile that I've showed you earlier in the beginning. So to conclude, I've shown you that in the absence of leptin signaling in the gonadotropes, there is a significant decrease in the alpha-sub-beta expression. And in the lactotropes, the lactotrope transcriptome profile is completely remodeled in the absence of leptin signaling. I've also shown you that leptin actually influences the gene regulatory networks in both gonadotropes and lactotropes. So in the future, we plan to understand how those gene regulatory network changes in both gonadotropes and lactotropes can contribute to the paracrine signaling that might better explain what's going on in the absence of leptin signaling in the gonadotropes and between those two populations. We also plan to assess the proteome profile of those lactotropes in the leptin mutant pituitaries and see if that correlate with that transcriptome profile in the mutants. So with that, I want to acknowledge and thank our lab members and our collaborators. And I want to also thank and acknowledge our funding resources. And with that, I'll be more than happy to answer any questions you may have. Thank you very much. Thank you. Thank you so much, Jewel. We're open for discussion. Hi, Dan Bernard-McGill. Thanks for the presentation. Just to clarify for my first of what's probably gonna be many questions, I'm joking, don't worry. Yeah, I know. I expect it. These data were all from females? I'm sorry? The data were all from female mice? All from females, yes. Okay, so I'm just curious, why is your first hypothesis that you're affecting paracrine signaling, right? Because you told us that FSH is reduced. So you would think estradiol would also be reduced and that could actually explain the effects that you're seeing in lactotropes. Is that something that you've ruled in or out? Well, so first of all, the changes that I've noticed was that FSH beta decreased like the down regulation. We didn't think about the paracrine signaling at a front. And then there was pretty surprised to us that according to the differential expression analysis, we found that the lactotrope transcription profile is totally altered and shifted. So we were thinking, we are deleting leptin in gonadotropes, but why we're seeing the changes in the lactotropes? So again, we looked at some of the genes, for example, CHA, SCG2, CHA, GA, which are the secretogranin and chromogranin, they also changed drastically in the lactotropes. So we believe that since we are making the changes in gonadotropes, and there was a significant decrease in LH as well, so there might be a feedback, endocrine feedback, a disruption which is causing the disruption of the paracrine signaling. So that's we are trying to connect to the paracrine signaling, but that wasn't, we haven't thought about that at the first point, honestly. Okay, so. Since I'm here. But we'll alternate, Dan. Okay, go ahead. So just take me back, I might've missed it in your talk, I think you said that these animals were subfertile. Yes. Yes. Were there any other phenotypes in the animals in terms of their, you know, any recycling or, you know. Yes, so the serum FSH was significantly down. The basal LH was normal in both groups, like control and mutant, but the LH surge was totally blunted. There was a 70% decrease in the LH surge. So it wasn't normal at all. In terms of, in addition to the subfertile phenotype, we also seen that the females were having less number of pups per liter, and the period for the onset of pregnancy was delayed as well. So yes, we have seen those LH surge changes. Yeah. Okay, good. Dan. So when you were comparing your mutants to the controls for your sequencing, did you stage the animals? I'm sorry? Did you stage the animals? I believe there were, all the females were at diastereous stage. In diastereous? In diastereous, yes. Okay. Yeah. Well, and the reason I'm asking is that, you know, based on what we've seen, GATA2 really isn't expressed in the gonadotrope in female mice. It's expressed in males, but estradiol actually inhibits GATA2 expression, and actually the locus is tightly compacted, right? So it's actually not even capable of being expressed. If you ovariectomize, then you see an increase in GATA2 expression. So it could be cycle dependent. I mean, I was just curious that GATA2 came up in females. So I'm just, if you looked at a stage where estradiol was low, maybe that would be more, you could explain that, yeah. Yeah, so Dr. Childs, there were, they're Allie, estrous cycle? They were in pro-estrous. Oh, pro-estrous, okay. So paramyces. Yeah, so it would be high. So I actually would still expect the locus to be closed. Like we actually showed, if you treat males with estradiol, you actually can inhibit GATA2 expression. So, yeah. Yes, I know the Gramman one study, yes. So we thought about it, and I think it's because of the disruption of the LH surge, which has already disrupted the endocrine feedback, which is causing these GATA2 changes. So this is from the physiological point of view. And from the technical point of view, the CINIC analysis is a predictive model. So we want to state the caveat as well. So the GATA2 that are found in the gonadotrope as a regulon, the caveat is that the model is not 100% accurate. It's a machine learning based model, and it's a predictive model, so. Yeah, fair enough. Yeah, thanks. Thank you very much. Great, any other questions? All right, terrific. Thank you so much. Thank you. Thank you. Okay, so our next speaker is Tiziana Silva, coming to us from the Brigham and Women's Hospital. And the title of the talk that she'll be presenting is Genetic Profiling Using a Gene Panel Identifies Pathogenic Variants and Copy Number Variations in Pituitary Adenomas. So welcome. Here we go. Let's get you started. Okay, so hello everyone. So first of all, I would like to thank you, the organizing committee for selecting this study to be presented today. So pituitary adenomas are the third most common to cranial neoplasm. They arise from different cell lineages. They are usually defined as benign tumors. However, their clinical behavior vary from a stable lesion, indolent lesion, to very aggressive, invasive, causing morbidity and mortality to patients. The great majority of them, they are sporadic. In just about 5%, they are familial in origin. Several genetic alterations have been implicated in pituitary adenoma tumorigenesis, such as NAS mutation and somatotrophic adenomas, and USP8 mutation in Cushing's disease. However, in the great majority of them, their genetic profile are not completely understood. So our objective is to identify mutations associated with pituitary adenomas using gene panel called Oncopanel and to detect copy number variations associated with specific adenomas subtype and their behavior. We included all patients who underwent surgery at our institution from 2013 to 2020 and have consented to participate in this study and also have a good quality of tumor DNA. We exclude these tumors when the pathologists report a normal pituitary tissue instead of an adenoma and also when the pathology diagnosis was inconsistent with the clinical diagnosis. So we used the most recent WHO pituitary adenoma classification to categorize our tumors in three main families using transcription factors. So T-PIT for corticotroph and then we divide it in silent SHH and Cushing's disease. PIT1 for somatotroph, lactotroph and tyrotroph and also we have fully differentiated PIT1 adenomas and SF1 for gonadotroph. Oncopanel was a program launched in 2013 for genomic profiling of tumor DNA. It is a hybrid massively parallel sequencing panel to detect single nucleotide variants and copy number variations in the list of genes and those genes are important genes for oncogenesis. There were three versions of the oncopanel with increasing number of genes across the timeframe. I would like to point out that USP8 was included just in the third version and AIP-GPR101 which are important genes for pituitary adenomas, they were not included in the oncopanel. In order to understand the clinical significance of each variant, we classified them into tiers based on databases such as COSMIC and CLINVAR and also in predictive tools. Using this classification, we could differentiate variants that are known to be pathogenic and could be a driver from variants that are unlikely to be pathogenic and not associated with disease. Here, I would like to point out that tier five which are likely benign or benign variants, we did not include those in our analysis. So we used transcription factor to categorize our tumors as we can see here. And here, this is how they were classified before 2017 WHO classification. So our non-functioning tumors based on transcription factors, they were reclassified as a silent ACTH, fully differentiated P21 adenomas, new cell and gonadotroph. We included 171 adenomas, mostly gonadotroph, 18 cushion disease, 29 somatotroph adenomas, 14 silent ACTH, 21 prolactinomas, four fully differentiated P21 adenomas and eight new cells. So in this plot, we can see the number of variants per sample for each adenoma subtypes. What you can see here is that the mutation of load was similar across adenoma subtype. We detected that some genes were overrepresented in our cohort and they were enriched in specific adenoma subtypes such as GNAS for somatotroph adenomas, USP8, NF2, NF3, and FGFR4 for cushion disease, MEK1 for silent ACTH, and NLT2 for prolactinomas. This oncoprint shows, you can see here the list of genes that we detect mutation. Here we can see the frequency, the tier classification color coded, and this part we can see the tumor subtype was indicated by color. I'm not going in detail about each gene, but I would like to say that we identified common and private mutations in a cancer-associated genes affect approximately 40% of our cohort. This box plot shows the number of CNV copy number variation of each adenoma subtype. What you can see here is that fully differentiated P21 adenomas, prolactinomas, and somatotroph, they were the most disrupted tumors compared to cushion disease, gonadotroph, and no cell. I brought here two cases to illustrate copy number variations in pituitary adenomas. On the top, we can see a CNV plot. So here, you can see the chromosomes, which are color-coded. Above zero, this is an additional copy of the chromosome. Below zero is copy loss. In this case, we see a quiet genome. It is a case of cushion disease with USP8 mutation. On the bottom, what you can see here is a genetic instability with numerous copy gains and copy losses, which is in a silent ACTH with TP53 mutation. In this plot, we can see the number of CNV events across chromosomes. The subtypes are indicated by color, as you can see here. Above zero, copy gain. Below zero, copy loss. What you can observe is that the chromosome 1 and the chromosome 11 were the most affected in terms of copy loss. In terms of copy gain, we observed chromosome 9, 8, 7, and 5 were the most affected. We detected loss of chromosome 16 and the loss of chromosome 1q in different types of adenoma. However, the combination of both was detected just on somatotroph adenomas, and just those adenomas that are negative for JNAS mutation. So here we can see the genetic alterations of the GH adenomas. We identified in about one-third of our cohort JNAS mutation. In about 17 percent, we identified this combination of loss of chromosome 16 with the loss of chromosome 1q in our cohort. We analyzed CNV burden in terms of recurrence and tumor size. We didn't find any association, but MIP1 proliferation index above 3 percent was associated with high number of CNV events. I'd like to make some correlation with genetic features with clinical data and adenoma characteristics. So in Cushing disease, we identified USP8 mutations in 50 percent of Cushing disease samples screening for USP8. Samples with USP8 mutations, they were diagnosed at younger age compared to wild type. We didn't see any difference in terms of recurrence, tumor size, and MIP1. In GH adenomas, as I mentioned before, we identified JNAS mutation in 27 percent of our cohort. In JNAS mutated tumors, we detected they had fewer CNV events compared to wild type, as we can see here. However, it was not statistically significant. We did not see any difference in terms of recurrence, tumor size, and MIP1. In prolactinomas, in two macro adenomas resistant to dopamine agonists, we detected MEN1 mutation, and here we can observe the two-hit model with inactivated mutation and allele deletion. Also, we identified a recurrent macro adenoma with pretty high MIP1, BRCA1 mutation. Interestingly, in prolactinomas in MEN, they present significantly higher CNV events compared to human. In silent ACTH, we identified, as I mentioned before, somatic TP53 mutation with loss of heterozygosity in a malignant tumor with maybe one of 12 percent. So as a conclusion, we identified in this study recurrent genetic alterations, which was enriched in specific adenoma subtypes, and also we identified subtype-specific patterns of genomic instability. The association between high CNV burden and MIP1 proliferation index may help to predict adenoma behavior, and the genomic profile using a gene panel like this one has the potential to better classify pituitary adenomas and to contribute to clinical management. Thank you very much. It was just terrific. While we're waiting for people to come to the microphone, just a question or two. You know, sometimes as a clinician, pituitary adenomas secrete bad things, and we have to take care of those endocrinopathies. But other times, patients can present with non-functioning adenomas that are quite large where the considerations are really based on the size of the tumor, the impact on adjacent anatomic structures. Did you have any sub-analyses or look at, in your non-functioning group, the consideration of size as itself, sort of an aggressive phenotype, and whether or not there were any genes that associated with that? Just in gonadotrophs, you mean, in non-functioning adenomas. Yes, yes. Like a macro, you know, doing a separate macro. Yeah, I mean, the great majority of our tumors are micro adenomas, and yeah, especially non-functioning adenomas, they got a surgery. But we didn't identify any particular mutations associated, like recurrent mutations associated with gonadotrophs. And also, they don't have many CNV events. Okay. Hi. Maybe I missed this from the presentation, but do you know how many of your patients had also germline variants, and whether they were related to the somatic changes you found? Sorry, I didn't. Whether your patients had germline variants as well. We don't know. I mean, we didn't take the blood to identify germline mutations, but based on allele fraction, we may suspect which are germline, like MN1, and which are like somatic, like TP53, but just based on allele fraction. We didn't take the blood to confirm this. It seemed in your Cushing's adenomas, you had this very large signal in USP8. Can you describe the landscape of single nucleotide versus copy number variation for that gene? Well, actually, USP8 mutated tumors, they have fewer CNV events compared to wild type, but it was not statistically significant. They had fewer? Fewer. Wow. Yeah. Okay. So not enriched for copy number variation? Yeah. Okay. Just SNVs. Okay. Yes. SNV. Yeah. All right. Terrific. Well, thank you so much. Yeah, let's just close this out. Okay, so our next presenter is Dr. Yana Zavros. Yana comes to us from the University of Arizona, and she will be speaking today on the generation of CDH23-mutated pituitary neuroendocrine tumor organoids from induced pluripotent stem cells to model Cushing's. Thank you. Sorry, how do I stop it? And we should be good to go. Thank you for the introduction, and I'd like to thank the organizing committee for giving us the opportunity to present our late-breaking abstract. So Cushing's disease is a rare and debilitating disease that's caused by an ACTH-secreting pituitary neuroendocrine tumor, or as abbreviated here, PITNIT. The chronic exposure to excess cortisol results in the increased risk of stroke, diabetes, obesity, cognitive impairment, anxiety, and depression. The first-line treatment for Cushing's disease is transphenoidal surgery, which is followed by typically disease recurrence in approximately 56% of patients during their 10-year follow-up period. Cushing's disease continues to be a medical therapeutic challenge due to the lack of specificity of current standard-of-care treatments, and these treatments typically exhibit low efficacy and tolerability by the patient. And biochemical control by these treatments is achieved in only approximately 50% of patients. So the intratumor, or PITNIT, cell heterogeneity and genomic landscape is critical to understand because it can lead to therapeutic failure, tumor adaptation, and subsequently disease recurrence. However, the complexity of the PITNIT microenvironment and tumorigenesis leading to Cushing's disease is remarkably understudied. So the genetics that underlie the pathogenesis of Cushing's disease are typically caused by somatic or germline mutations. The somatic mutations that underlie the development of Cushing's disease, pituitary neuroendocrine tumors, are associated with mutations in USP8 that have been identified in 35 to 62% of sporadic corticotroph PITNITs, and USP48 and BRAF, mutations that are known to enhance promoter activity of POMC. In some instances, however, Cushing's disease, and these are rare, is a manifestation of genetic mutation syndromes. And some commonly associated germline mutations are MEN1, FIPA, CARNICOMPLEX, and the focus of today's presentation is CDH23, encoding for coherent-related 23. So the presentation today will—we've focused on developing induced pluripotent stem cell derived organoids. And the reason for this is that it is challenging to find an efficient way to investigate the direct effects of alterations in genes in humans in the development of PITNITs. iPSCs gives us this advantage since it allows us to control the genetic pre-tumor pathways leading to PITNIT development. Moreover, because we are looking at a germline mutation, in this case CDH23, this is maintained within the induced pluripotent stem cell line. So CDH23 syndrome was first identified to be clinically associated with the development of Usher syndrome, deafness, and vestibular dysfunction. However, in 2017, Zhang and team published in the American Journal of Human Genetics that the germline mutation in CDH23 is associated with both familial and sporadic PITNITs. So in our initial screen of induced pluripotent stem cell lines that were generated from patients with Cushing's disease, we found, interestingly, that one of these lines carried a CDH23 mutation that was very similar to the one that was reported in the 2017 paper. Therefore, we used this induced pluripotent stem cell line and, as a control, an iPSC germline that was generated from a healthy volunteer to generate iPSC-derived pituitary neuroendocrine tumor organoids. Our differentiation schedule is as follows, and this is work that we have published recently. So we begin with developing the iPSCs from the PBMCs from a Cushing's disease patient carrying CDH23 germline mutation. We then follow a differentiation schedule based on specific growth factors that are known to drive the development of the pituitary gland. And then by day 15, the iPSCs are harvested and embedded into matrogel, where they are further differentiated through days 30 and 60. The video that I am showing here is a confocal video where it's showing the sectioning along the Z-plane to demonstrate the complexity of these organoids. So these are not spheroids. So this is an organoid that was developed from induced pluripotent stem cells at day 60 in culture. So what you can see here is the three-dimensional structure and the complexity of these cultures. So we have published this previously, where we did compare the function of these organoids in culture. And what you can see here is that over the differentiation schedule of the iPSCs carrying the germline mutation CDH23, there is significantly greater secretion of ACTH compared to our normal control. And the immunofluorescence image that you see here is a 3D rendering confocal image of the video that I just showed previously, demonstrating the expression of ACTH and also proliferative zones within the organoid as marked by EDU. So we harvested these cultures from control and CDH23 iPSCs and organoids at days 15, 30, and 60 along the differentiation schedule. And we performed single-cell RNA sequencing at each of those time points. What you're seeing here is a combined U-map between the controls and the CDH23 over the 15- to 60-day differentiation schedule. And what I want to highlight is the differences in the cell populations that were detected by single-cell RNA sequencing between the controls, where by day 60, we found the presence of all of the major pituitary cell lineages. However, in the iPSCs that were generated from the patient with the CDH23 mutation, the environment or the cell composition that we detected within that culture was remarkably different in that there was a high abundance of this stem or progenitor cell population that I will go into a little more detail in the next slide. There was also poorly differentiated corticotrophs as well in these cultures. And then we also found fibroblasts in particular indicative of the myofibroblast cancer-associated phenotype. So this was interesting to us because early on, we did perform extensive immunofluorescence. But what I'm showing here is a representative image where at day 15, we clearly saw, even early on in these cultures, a loss of the PIT1 expression and also an increase in ACTH expression in these cultures. So from here, from this UMAP analysis, we were able to identify what we call a base cluster. So the base cluster, in this case here for the control, is cluster 3, which is shown here in the arrow. And then in the mutant line, cluster 21 was identified as our base cluster. And the base cluster is what we call the stem cell. So this was the cell population that highly expressed both SOX2 and S100B. So from there, we performed a trajectory analysis in order to identify whether there was any dysregulation in the pituitary cell lineage differentiation. So this is just to remind you that this analysis was done in cultures that were collected at days 15, 30, and 60 of the differentiation schedule. And what you can see here, this is a pseudotime analysis. So the pseudotime, the more orange the cell populations you will see here, the more differentiated they are further away from the stem cell. So if we begin here in our controls, which is the base cluster, these were cells that were highly expressing SOX2 and S100B. As they differentiate away from that base cluster, we obtain differentiation of all of the major pituitary cell lineages. However, interestingly, when we look at a similar type of analysis in the cultures from the iPSC CDH23 mutants, we find that there is a remarkable dysregulation of the differentiation schedule. So if we start here, which is our base cluster, again, a stem cell population that's high in SOX2 and S100B, we find that in this case, there is a differentiation towards a myofibroblastic phenotype. And in addition to that, there seems to be a skewed, poorly differentiated corticotroph subtype. When we look closely at those clusters and we compare them to individual single-cell RNA sequencing data that was collected from tumor-derived organoids, in this case, from individual patients, we found that in the mutant line that we developed, in addition to tissue-derived organoids, there was an increase in the CD44 XCT antioxidant system within those highly proliferative corticotroph cell populations, a disruption in normal cell cycle, and an increase in hedgehog and notch signaling. So we wanted to compare what we were observing in culture to the patient's tumor tissue. So in this case, the CDH23 mutant iPSCs were generated from a patient with a sparsely granulated pit net. And so we went back and we looked at the patient tissue with similar subtype. And what we found, and I'm sorry, so let me take a step back. So, the way we did this analysis was we wanted to compare the single-cell RNA sequencing from the cultures directly at the single-cell transcriptomic level using COSMIC's spatial transcriptomic imaging. So, to compare single-cell to single-cell on the tissue. The way this approach works is that we start with the FFPE section of the patient's tissue and in consultation with a neuropathologist, we select fields of view. From there, and what you can see here, the fields of view overlaid on that FFPE tissue are color-coded. So, each area that you see here is a single-cell color-coded that is identified based on the gene signature of the tissue. So, if we take a higher magnification of that representative, in this case, field of view 20, again, what you can see here is each of these colored areas that you're seeing is a single-cell. And we identify the cell populations based on the gene profile of these. And what we find is in the sparsely granulated subtype, we have tumor cells that are highly proliferative, an infiltration of Tregs, mycuffs, and endothelial cells. And this is consistent with the presence of these myofibroblastic cuffs. We know from other tumor microenvironments that these fibroblasts typically are associated with driving angiogenesis and a more invasive and proliferative phenotype. So therefore, we did some initial immunofluorescence to look more closely at the proliferation and the cancer-associated fibroblasts that were present. And what you can see here is an immunofluorescence of that same tumor tissue that we performed the spatial transcriptomic analysis that were highly proliferative and expressing fibroblast marker vimentin. And when we looked at vimentin more closely, these cells were highly expressing alpha-smooth muscle actin, but expressing low IL-6. And this is how we concluded that these were more mycuffs. Interestingly, comparing the clinical data directly to this patient, it was found that this patient did have an invasive macroadenoma tumor. Therefore, in conclusion, we find that PIDNET organoids generated from iPSCs of Cushing's disease patients, specifically carrying germline mutations, reveal a dysregulation in the cell differentiation pathways. With the case of CDH23, this led to a skewed development of corticotrophs. And then understanding the molecular mechanisms of the pathogenesis of Cushing's disease, and in particular, the complexity of the PIDNET microenvironment, will potentially identify therapeutic targets for these patients. Thank you. This paper is open for discussion. One thing I think I noticed at one of your early slides, is it true that this patient who gave these iPS cells, or from where you derived these iPS cells, also had an MEN1 mutation? No. We do have another line, separate, from a patient that does have an MEN1 mutation. Okay. Great. And then, when you showed the beautiful picture of the organoid, it appeared, very qualitatively, that your ACTH staining in red was sort of at one end of that structure, and that the proliferative elements were at the other end. Is it possible that, as you're looking at the three-dimensionality of how these things grow, that there's parts of them that are, we're going to grow and expand, and other parts, we're going to do the job of secreting hormones? Yeah. Yeah. That was an interesting observation, because you'll notice that the proliferative zones are also localized, seem to be localized. So yes, I think, spatially, that's also the advantage of using organoids. You can potentially track the differentiation of the cell lineages, as well. Hi. Hi. I'm curious about how the germline variants were detected, what sort of test was used, and also whether any genetic test was carried out in the tumor. So we performed whole exome sequencing. So that's how the CDH23 mutation was found. And then, I'm sorry, what was your? Whether any genetic test was done in the tumor, at the somatic level. Oh, yes, we did. We did do that, as well. So on the tumor tissue? Yes. And, I mean, there were, I believe it might have been USP8. That was the mutation that we picked up at the tumor level. In which patient? In the one with CDH? In the CDH23. Yes. Yes. Okay. Yeah. Thank you. Okay. Terrific. Thank you. Okay. So I think we're down to our last late-breaking from Dr. Julie Reffart from the University Hospital in Basel. And she'll be discussing diagnosing central diabetes insipidus using copeptin upon hypertonic saline versus arginine stimulation. So I'll just click here. Ah, here. I'm going to start. Okay, thank you very much for the kind introduction. I'm very honored to have the opportunity today to present the results of our multicentric diagnostic study. So if you have a patient that comes to your outpatient clinic with hypertonic polyuria polydipsia syndrome, you have three main entities in mind you want to test for. So one is AVP deficiency, which is caused, as the name already nicely states, by insufficient AVP secretion from the posterior pituitary. The second one is AVP resistance, which is characterized by renal insensitivity to AVP. And the third one is primary polydipsia, where patients drink an excessive amount of fluid with consecutive polyuria despite initial adequate AVP secretion and renal sensitivity to it. So our group went to great lengths to show that we can use copeptin, the stable surrogate marker for the difficult-to-measure vasopressin, as a diagnostic tool in these different entities. For instance, if you measure a random copeptin above 21.6 picomoles per liter, you already have the diagnosis of AVP resistance and do not have to test any further. However, for AVP deficiency and primary polydipsia, we do not have such an unstimulated value. Five years ago here at Endo in Chicago, I had the great pleasure to show you this data. We were able to show that using the hypertonic saline stimulation test leads to stimulation of copeptin in patients with primary polydipsia, here shown in blue, while the copeptin levels in patients with AVP deficiency remain low, leading to a high diagnostic accuracy of 96.5% for a copeptin level of 4.9 picomoles per liter. However, a year later, we were able to show that arginine infusion, you know this as a test for growth hormone deficiency, also stimulates copeptin and could be used as a diagnostic test for these patients. Here you see again nicely the patients in blue with primary polydipsia and the patients with AVP deficiency in red. So arginine-stimulated copeptin had a diagnostic accuracy of 93% for a copeptin of 3.8 picomoles per liter. But which test is now the better test to do to our patients? So to answer this question, we designed the multicentric CARGO-X study where seven tertiary hospitals participated. So patients underwent two diagnostic tests. One was hypertonic saline where they received an initial 250 milliliter bolus over 15 minutes, followed by a body weight adapted infusion. As soon as sodium levels reached 149 minimal per liter, we measured copeptin, the infusion was stopped, and patients were rehydrated orally and parenteral. The second test, arginine stimulation, is a bit simpler. So there patients receive a body weight adapted infusion with L-arginine hydrochloride, 21% diluted in normal saline, given over 30 minutes. 60 minutes after the start, you measure copeptin using the cutoff of 3.8 picomoles per liter to make your diagnosis. So in our study, patients were randomized to either receive first hypertonic saline stimulation or arginine stimulation or the other way around. After they had both tests, they received a preliminary diagnosis and according treatment. And we saw them again after three months and evaluated the treatment response and also asked them out the test preference. At the end of the study, two endocrine experts looked at all the patients. They looked at all the patients' characteristics, the test outcome, and also treatment response and made the final diagnosis. Importantly, the experts were blinded to the arginine-stimulated copeptin values. So our primary outcome was the overall diagnostic accuracy of the tests, meaning the correctly number of diagnosed patients to all tested patients to differentiate AVP deficiency from primary polydipsia. Our hypothesis was that arginine-stimulated copeptin is non-inferior to hypertonic saline-stimulated copeptin. We recruited patients that were referred with polyuria polydipsia syndrome or patients with known AVP deficiency. Here you see our patient flow. So 164 patients underwent diagnostic testing, six with true consent after the first diagnostic test, which left us with a total of 158 patients with a final diagnosis, which went into the later analysis. Here I would like to show you some baseline characteristics. In orange, the patients with AVP deficiency, 69 in total, and in blue, the patients with primary polydipsia, 89 in total. And I would like to draw your attention to the partial AVP deficiency patients compared to the primary polydipsia patients. And you see, it's very hard to differentiate them due to the clinical symptoms, because they have the same amount of polyuria polydipsia, the same occurrence of nocturia, and also drinking at night. And that makes it so difficult, and that's why we need a simulation test. So how did the tests do? So here you see the results of the hypertonic saline stimulation test. And we had, again, a very nice high diagnostic accuracy of 95.6%, with a sensitivity of 91.3 and a specificity of 98.9%, differentiating the patients with primary polydipsia, here shown in blue, from the patients with AVP deficiency in orange with a partial, and in brown with a complete defect. And what about the arginine stimulation? So here you can also see that the patients with primary polydipsia were nicely differentiated from the patients with complete AVP deficiency. However, there was an overlap between the partial AVP deficiency patients and the primary polydipsia patients, leading to a diagnostic accuracy of only 74.4%, with a sensitivity of 75.4 and a specificity of 73.6%. So we had to conclude that arginine-stimulated copatrin is inferior to hypertonic-stimulated copatrin. However, if you look at the results closely, although we did not find one copatrin cutoff, you see if you take the cutoff of 5.2 and the cutoff of 3, you can diagnose over 50% of the patients here, the AVP deficient, and here, the patients with the primary polydipsia. So we believe we can still use this test as a diagnostic measure, just if your patient has a value in between, you have to do additional stimulation with hypertonic saline. Then what about safety? Here you see the adverse effects and adverse events in red for the arginine infusion, in blue for the hypertonic saline stimulation test. So generally, the adverse effects were mild. However, the symptom burden was higher for the hypertonic saline test, and that also was mirrored by patient's preference, with 72% of patients preferring the arginine stimulation test. So to conclude, I hope I was able to show you that, unfortunately, arginine-stimulated copatrin is inferior to hypertonic saline-stimulated copatrin in the diagnosis of AVP deficiency. However, we can still use it as a first, simple, well-tolerated diagnostic test using a lower cutoff of 3-pricamol and a higher cutoff of 5.2-pricamol per liter to diagnose AVP deficiency and primary polydipsia. Hypertonic saline-stimulated copatrin remains the gold standard to diagnose AVP deficiency with a high diagnostic accuracy, and we were again able to confirm the safety of this test. With this, I would like to thank all our amazing collaborators in Switzerland, Germany, Italy, the Netherlands, Brazil, and the United Kingdom, and I'd be happy to answer any questions. Thank you. The paper's open for discussion. I noticed in the design of the study, you had slightly different cutoffs for, I think, the level of copeptin in each of the different arms. So could you maybe elaborate for us how those cutoffs were decided upon? Yeah. So the one for the hypertonic saline stimulation, that was already the one we had in the New England paper, so we had already validated that. So we went with that one. And the 3.8, that was the copeptin value that was calculated in the monocentric study we did in the Lancet paper, which showed the best differentiation between patients with primary polydipsia and patients with AVP deficiency. So we chose to go with this previously described copeptin cutoffs. Ming Chan, Boston Children's Hospital. Do you have information on why the participants seem to prefer one test over the other? Yes. It's, if you see the patients, hypertonic saline infusion is not a pleasant test. So when you get hypertonic, you really can, I can visually tell you when the patients reach the sodium level of 150. So they get a little headache, they're very thirsty, it goes away as soon as they're able to drink, but it's quite, it's not a pleasant feeling. And also you don't know how long the test goes, so you have to do rapid blood measurements in between. But with the arginine infusion, it's quite straightforward. You can do it in the outpatient clinic, you just start the infusion, 60 minutes later you measure copeptin, so that's quite a simple test. And the people who didn't return for the second test, I don't remember, was it different between the two groups? Yeah. So there were five people that dropped out after the arginine infusion and one after the hypertonic saline. And it could be that maybe the study team told them now the worst test comes, so maybe that's why they decided not to go further. Hi, Luigi Garibaldi, Pittsburgh. Very interesting paper. I have a general question. Have you figured out why stimuli for anteropituitary stimulation stimulate copeptin, except for maybe insulin, hypoglycemia, you can consider stress, but glucagon and arginine, they really, why should they stimulate the posterior pituitary? That's a very good question, and I don't have a final answer on this. As you probably know, there's some older data showing that arginine may work over the nitric oxide pathway, that that could be a stimulation. We also think that maybe because there's a slight decrease in glucose that that could trigger then a copeptin release, but we don't have a good answer yet. But we hope we'll be able to give that in a few years. Thank you. Great. Well, I think that concludes our questions. Thank you so much. Thank you so much. Before we leave the room, we do have a note from Dr. Jenny Vissler. There is a basic science reception this afternoon, so if you're in that community or you want to learn more about it, please look that up, and you're welcome to join basic science reception this afternoon from the Endocrine Society. Thank you all very much. Okay, so here's what we're going to do, and you've got to do that. So... Thank you. Thank you very much. Thank you for watching. So, for example, Sauce 2, even though it's highly atomized diocese, it doesn't get involved in the process. What happens after? So, it's either the excess of the threshold, the non-threshold, or the excess of the threshold. And then all of the other ingredients, so you get a cross out there.

central diabetes insipidus

copeptin levels

hypertonic saline stimulation test

arginine stimulation test

diagnostic tests

polyuria-polydipsia syndrome

AVP deficiency

diagnostic accuracy

primary polydipsia

overlap

partial AVP deficiency

copeptin level cutoff values

safety

symptom burden

gold standard