 Great, so I think we've kind of slid into the general discussion. Maybe we'll let you sit down so that you don't have to stand up the entire time. But we've heard three talks about various aspects of case-finding and surveillance. And I'm wondering if there are some commonalities across these very different systems that or at least ways that they might be able to interact. I think, you know, Mark Williams and Josh talked about the potential for our HMO network potentially to provide data to the Sentinel system, specifically on this particular condition. I'm kind of curious, what would it take to get that to happen? Is that just sort of you, Mark goes to them and says, do this and it happens? That's the way I work. You go to me and you say, Mark, would you make sure that this happens? This is just recapitulation. So I know that from the HMORN perspective that the Sentinel thing has really just kind of gotten going. And so I, and Geisinger has not been an integral partner, is that the lead within HMORN related to the Sentinel activities. But we have a board meeting next week, yeah, next week, that we could potentially bring this up. So if there's a recommendation that comes out of this group, it would be relatively straightforward for me to take that to the board and just begin to explore some interest. I was curious, I had sent Josh an email about who you were specifically working with within the HMORN since I know Vanderbilt's not a member. And I had replied, but hasn't made it through cyberspace from here to there, apparently. I actually paid attention and turned my computer off. But Bob Davis gave me those numbers. So one thing I probably didn't make as clear in talking as we could is there can be multiple use cases. You can develop a set of characteristics that would give you valid reporting numbers from something like many Sentinel or other groups that could be involved, such as Emerge, that may or may not have biological data attached to them. And such as the highly specific codes in patients admitted for a long time or patients admitted to a shorter time in a burn unit that have those SJSTN codes. You could do that kind of stuff as well as sort of getting deeper where you don't lose any cases if you want to do incidents and exposure ratings and things like that. I mean, I think one really low energy thing to do would be if we could create a replicable phenotyping algorithm to apply to EHRs and get that stood up in phenodb that would allow others that are interested. I'm sorry, I'm sorry, what? Phi Kb, sorry. There's too many Kb's and Db's and everything else for me to keep track of at least. So that would be something that would probably lower the energy threshold for people to get going with it. So I did want to maybe ask Munir to comment. I believe you had said or written that it was very difficult retrospectively to validate a case that many of the cases that you look back on if you don't have pathology reports, you can't examine the patient, you can't look at blister fluid or whatever that those are difficult. Now I may be misquoting you, but I wondered if you could comment in general on the issue of trying to validate a case that's identified, you know, potentially when a patient leaves the hospital and contacts Andrea. So the gold standard would obviously be a prospective collection, but given the rarity and the expense of following patients up prospectively, it's not always possible to do that. And obviously in terms of the processes we've set up, if prospective cases do come up, then we do follow them and so on and try to recruit them. But because these cases are rare, and it is important therefore not to miss them, so we do go retrospective assessment. But we do undertake specific, we do need specific criteria to be present in the case notes and sometimes going back to the patient or the primary care practitioner, ambilatory care practitioner to get as much detail possible of the drug at the time the drug was started and also in terms of what was found in the hospital. And we go back to the pathology labs as well to get the reports of skin biopsy and so on. When we have all that data with retrospective case ascertainment, we do then include that as a case within our collection. When some of those issues are missing, then maybe it could go either way, then we particularly rely on our adjudication panel to be able to tell us whether they feel that that should be included as a case or not. And as I said, Neil has been on the adjudication panel and Peter Friedman as well. And what they do is that if they don't feel that there isn't enough data, they have thrown some cases out. Do you have a feel for the proportion of cases that you either accept or reject? Obviously it depends on the net that you cast. So the number of cases that were thrown out by Neil and Peter were a small number, largely because there were other cases at the beginning which were so bad they were thrown out anyway. So I can't tell you exactly what the proportion is, but from electronic health records, when I looked at it, it was 0% recruitment. But from kind of ascertainment via referral, then we are getting probably about 20% to 30% who are actually true cases based on the referral patterns retrospectively. Yeah, I'm not an expert in SGS by any means, speaking as an internist. But as I look through some of the records I have, and I'm just be curious as you feel whether or not these would qualify as cases. But I see dermatology assessments, I see pathology reports, things like that that say SJS. Are you talking about, you saw a lot of that in your EHR or? A dermatology assessment of SJS and so on, and I didn't believe it. Sorry. Pathology reports wrong too, usually, are incomplete. And pathology reports together with a good assessment knowing which mucus membranes have been affected and particularly if they've got evidence of how much of the skin area was affected. With a diagram, hopefully, in the case notes, that kind of starts to convince me that there may be a case, but a pathology report then adds to it. But unfortunately, many dermatologists get it wrong and you have more experience of that. Yeah, I mean, dermatologists, not everyone is a specialist in these conditions, for first thing. When we ask for case reports to our networking colleagues, we ask them that they call us or send an email as soon as they suspect a case, because we are the ones to exclude. It's better not to miss, so they report. And then we take like a one sheet, one page on the phone with some specific questions to try to find out what disease that is. And then if certain criteria are met, then someone goes and travels. But when it comes to that point that the patient is really interviewed and visited, more than half of the case notifications considered to be a severe reaction, including Dresden, Aegeb and SJSTN, can be discarded. It's not only reports by dermatologists, by pediatricians, by intensive care physicians, and others. So half of them, or more than half of them, fortunately, do not meet the criteria to be interviewed. By those that are interviewed, when we do this systematic case validation, maybe 10% are excluded. And some of them we already know, but the rule is whenever a case was interviewed, blood has been taken, the case has to be clinically validated. So that's about the right. Sorry, one more point to make is that more recently, what has helped us is actually patients' relatives taking pictures themselves. Because clinicians often forget to get photographs, medical photographs. But the patient's relatives are actually taking pictures and they sometimes take longitudinal pictures. And you can see the worsening and then the improvement. I agree with that. Pictures, good clinical photographs are better than anything else. Better than a histopathology, although we collect them and re-review them. And in the registrar's study, a case can only be labeled as a definite case. Photographs are there, otherwise not, no more than probable. And that has been proven very helpful. And sometimes I wonder, everyone has these mobile phones except myself, and can take photographs, and they're no, and they're not taken. I just want to mention that the probability of being a case in Thailand and in Europe is totally different. We have a lot of cases reported and our pharmacists and the clinicians have seen a lot of cases like in Taiwan. I think almost all dermatologists know this condition, I think. And in Thailand, I think almost all dermatologists in the U.C. hospital, they know how to diagnose these conditions. And sometimes in our setting, pharmacists know better because they see in a lot of adverse drug reaction cases and they can tell the non-dermatologist clinicians whether this is the Steven Thames or not. And I think if we use the criteria like a dermatologist diagnose criteria in Thai setting, then we will have a lot less case. But it's not, I mean, the patient might actually have Steven Thames, but it's not certified criteria. They're just afraid that we will use two stick criteria for the setting like Thailand or the others. And we exclude a lot of real case because of, I mean, the criteria was too strict. I mean, in some ways, I mean, this is a classic prior probability issue in the sense that the prior probability of it being SGS-10s in Thailand is orders of magnitude higher than it would be in Europe or the United States. And so to assume that diagnostic criteria will give the same yield in those two populations I think is fairly unrealistic. And I think to Josh's point, it would probably mean much more manual adjudication of cases to be certain that it's a true case because we're gonna have way more false positives based on the fact that our prior probability is very low. But it still sounds as though even in Europe, you may be getting for every case identified 30 to 40%, maybe 25% or so, turn out to be real cases, which is not horrible. I mean, it's not one in a hundred. The ones that make it as far to the study that the inclusion criteria are met, for those 90% are validated, yeah? But before that step, we have like a case notification and more than half of them are no case. But we try to get a discharge letter, photographs or any documentation to prove that it's really not a case. Not to miss anything. And those who meet the inclusion criteria, they are seen. They got a seven digit interview number, they are entered into the system. So it's like a two step procedure. So I would be interested in knowing maybe just around the room in terms of the causality assessment. Like I think we've had a lot of discussion about identifying the phenotype. But around the various cohorts, who is actually doing the drug causality assessment and is anyone actually using ancillary in vivo, in vitro or ex vivo information in complex cases where there's multiple drugs that could be causal? So if there are local cases and we have managed to get the T cells and undertaken work associated with them and proven the immunological etiology, then that helps in terms of the causality assessment that we do. But the majority of cases we've been recruiting have been from all over the world and therefore what we use are typical causality assessment tools and we've developed one ourselves in Liverpool, not specific. It's not case specific in terms of SJS TE entire, like the olden. But it can go for any adverse drug reaction. So we use that causality assessment tool and that's done sometimes independently by two people to see whether they come up with the same pathways. It was like researchers, pharmacists, MDs. What kind of category are the people doing the causality? So the causality assessment is done by experienced people who know what they're supposed to be doing. After I did 19,000 causality assessments on one particular study, then I can be called to experience it doing it, I think. So I am often the one who's doing the causality assessment, but somebody else will be doing that independently as well. And have you ever changed the causative drug in your itch or other state studies based on an in vitro or a vivo? On in vitro, yes, on in vitro assessment from T cells and so on. Yes, we have changed the causality assessment. But when you do have multiple drugs start at the same time, then obviously, as you know, it is very difficult to ascribe causality to one particular drug rather than the combination. And sometimes it's a combination, anti TB drugs are a typical example where combination therapy is probably causative rather than one drug by itself because there's well-known interactions between the drugs. Elizabeth, you're talking about just even clinically, not with a study, but just clinically whether you use causality? It varies widely from center to center who's doing it. And it's an important point because actually we're just about to do something where we go back on SJSTN patients based on biopsies and identify patients to see actually if the causality was actually correctly stated in the electronic health record. Because I think it's high risk that I've seen several cases as you, Neil, I'm sure where like the vignette that you gave about the woman who'd been labeled as penicillin, you know, TEN. And I had a similar case that was the mother of medical student as well. And she'd been labeled as penicillin TEN. And then, you know, I said, you know, basically symptoms within 24 hours of penicillin and, you know, she'd had this experience, you know, 20 years ago in Zimbabwe. And I said, well, have her go through all the records. And the next day she came back and she was so excited. She was waving this paper around that said, oh, my mother had back pain 20 years ago and she tracked down what they had given her. And it was like Tegretol, you know. So it was the same story and there's a million of those types of stories. So I just wondering how retrospectively I think the most difficult thing can be actually verifying the causality assessment, which is obviously huge in terms of drug specific studies. I think our strategy has been, of course, you want to find out the cause. I mean, it's a big question. And you'd like to be as certain as possible about it. And there are several scenarios. So if it's obvious, like it's a really high risk drug, the timing is perfect, the story is pretty clean, it would probably stop right there. If there's a question that might be 50-50, then still Mike Reeder will do the lymphocyte toxicity assay that we used to do in Solar Manuel and Newman. And so we might have people have that. And we published a paper with Claudio Naranjo years ago looking at if you had a high probability or low probability, what would the impact of a positive in vitro test be? If you already had a high probability, it's not going to change anything pro or con. If it's positive or negative, you still say it's likely. So just stick with the clinical. If it's a low probability, you might increase your respect for that drug if you see a positive test and then try and figure out why you thought it was a low probability. If it's in the middle, of course, the test could move you up and down a bit. The point with such a severe reaction though is, even if another drug is there, if you have say two or three drugs and you're not sure which, you have to respect all of them. It doesn't matter what a test says. You say, I think it's this drug, it might be that drug, it could be this drug. Well, you wouldn't want to expose the person to even the third one on the list because you're afraid of such a severe reaction. So with a reaction like this, it's not just a question of probabilities. If the probability is more than zero, you respect it. And it comes down to Advil too. We can say all we want, but would a person take Advil again if we didn't have another cause on the line? And so the ones we have the most problems with, I think are one we just saw recently where we really can't figure out what the person was exposed to. There may be an answer. I don't know if we have to do it like on the TV show House and break into their home and find out. But I remember seeing a patient and I don't know why we were looking at this patient in the burn unit. Everybody's all gowned up and the patient's all wrapped up. I'm listening to this story and I said, is this person an allopurinol? And the plastic surgery resident said, no, we went through the history completely, they're not allopurinol. So you call the pharmacy, call the family, call the pharmacy, find out. And it ends up that the cephalosporin that they thought cost it, actually it was allopurinol. And they had just got a prescription a couple of weeks ago, nobody knew, and they were in no shape to talk. So it can require a lot of digging. But I still have to say that even if you didn't, the tragedy is if you don't have that. If you have a list of two or three things, maybe, and you respect them, that's fine. But if you're really missing the elephant in the room, then it's really a tragedy because the person could well get re-exposed when they go back home. They just say, well, I'll go back on my allopurinol, because nobody ever mentioned it. There may be a bit of a bias. I mean, as physicians, physicians have to be much better at recognizing the disease early. But there may also be, and I'm not sure if any study has ever been able to capture, look at this. But in terms of the patients that actually tend to present a bit later and don't know their drugs or aren't as in tune with their medical care. So there may be, there's a bit of that gestalt when it comes to toxic epidermal necrolysis in particular, and it makes it difficult. And I don't think it changes what we do clinically, but I think that we've got the capacity now to build on in vitro and testing and in vitro approaches to actually improve the specificity of drug ascertainment. So again, it's another plug for collecting spent samples. I just want to say, you brought up a very good point about what people will remember. And I'm amazed that doctors, even when doctors are patients taking a drug, they will not think that's something they started two weeks ago or three weeks ago caused this. And I've warned people when I've given them a drug that could cause this, here's what it'll look like. When are you starting it? I'm starting it tomorrow. Okay, so a couple of weeks from now, what are you doing? Well, we're gonna be on March break or something. Well, when you're on March break and you start getting a fever and a rash, you better stop that drug. It could be that and try and point it out. And then I see them in the hall the next day and they took one pill and they say, I'm okay. I don't even have words for the recording here, but it's like, what? And you can't, and these are smart people, but they just don't get that there's this time lag that it isn't gonna be the next minute. I don't know what they're thinking, but it's tough. And I've seen patients too that, I've blown it. They said they were on carbamase apine and somehow I thought they'd been on it for years. And it actually ends up, they were really on it for three weeks. They'd been on another anti-epileptic before that. It was changed. So you can, it's tough, but you really do have to dig. I think you're right. So is it fair to say that better in vitro testing, I mean, I guess being a clinician and not a basic scientist, I assume the basic science has already solved all these problems. So I'll turn to Dr. Tripenyer and maybe some others and just say, isn't there a way, could we find an assay for what sort of sets this off? Is there a final common pathway by which the epithelium separates and that? Is that not possible to do? Boy, you're putting me on the spot. I mean, there is the in vitro cytotoxicity assay, which is, you know, it has reasonable performance across a bunch of different drugs, but we really don't know what the mechanism is behind that. So that's what we're trying to figure out, but I think that's the holy grail to have something that not only helps you identify patients, confirm patients, but helps you understand actually what's going on. And I don't know, Meena, what you think about the in vitro cytotoxicity assay? We don't use it anymore, the cytotoxicity assay. I did my PhD on it and have fond memories of looking down the microscope for many hours. But because of the labor-intensive nature, the variability in the micro-sum preparations that you have to use and so on, we felt that in terms of positive predictive values, et cetera, it wasn't worthwhile as continuing. So Dean Nesbitt particularly in Liverpool has been developing the lymphocyte transformation testing and we do, you know, with certain classes of compounds, such as anti-apileptics, we get very high positive response rates when you have a good history and so on. So, you know, with carb masaline, for example, almost 90% of our cases are positive with the lymphocyte transformation test. But this is mostly with dress, though. But when we get to the SGSTN, the lymphocyte transformation test positivity, goes down again. And again, that's related to the CD8 cells and that's much more difficult to proliferate. So he's been using cytotoxicity assays and so on, but you still don't get positives in every case. So to use that as a di... To be able to change your causality assessment at the moment is quite difficult. And we certainly wouldn't use it for diagnosis in a clinical case. We do sometimes say, yes, we will do this. If it's positive, it will be of use to you. But if it's negative, you still have to follow your clinical diagnosis in terms of stopping the drug or not. I think what's been striking is just how long these responses will persist after the reaction. And that's very striking with SGSTN. And there's now actually newer approaches looking at different cytokine outputs like at least spots with granulicin output. You can drive cells forward. There's beautiful flow cytometry techniques. I really think we can actually capture the elegant science from areas outside of drug research and really bring that in. I mean, there is a huge amount of work being done in infectious diseases and immunology and other areas. And all of those are directly related to drug research. So this is changing the topic a little bit. But I think maybe before you, just before you do that, did you want to make another comment, Laura? Oh, just that what we're doing with sulfonamide allergy, these are for immunocompetent people is we're using a panel of the in vitro cytotoxicity assay plus the LTT plus actually like an antibodies, particularly if they have thrombocytopenia and wanting at least one of those to be positive, right? Because none of them are highly sensitive. Well, and maybe something I might mention since we have both the National Center for Advanced and Translational Science and CATS and our NIAMS skin colleagues here, it seems like development of better in vitro assays would be like right up their alley and something that would be useful clinically. I think one of the things that we try to do, we talk a lot about going from bed to bedside in, sorry, from bench to bedside in translation, but we also have to make the other loop and go back and say, we really need a good in vitro assay for this and so I think perhaps that could be a recommendation from this group if you all agree. So I was gonna go back, I think I was sort of struck with the discussion of the power of having the visuals and a lot of the potential to think now how much easier it is to think of capturing pictures, but as you think about this, is there other sort of key information that would be relatively easy to capture that sort of missing to think about what are other things that, oh, if only we always had this, it would really help with some of this surveillance and identification sort of in retrospective situations. Did my question not make sense for all? Okay, so I think what my question is, if you think about trying to do sort of some of this pharmacosurveillance and retrospective identification of cases and what is the information that you wish was there that's often not there? And we were talking about the value of having photographs and information to sort of look at that visually. Is there other key information that if it was more consistently there would be helpful? So there are all of us who are doing work in this area, we'll have a case record form which we hope that we can fill every field and you can fill every field then you know that you've got everything you need. And but in most cases you can't because the data's just not there in case notes and case notes are very much dependent on which clinicians are writing it, what time of the day it is, et cetera. But I think one of the issues that will happen in the future with more electronic health records coming in, a more standardized assessment of patients that people will have to fill in specific fields. Now, at the moment in the UK, what's going on is a generic assessment of what kind of fields we need but not specific to SJST and also other conditions. And I think that's the next step from that in terms of how you go from that kind of generic data you can get on every patient coming through and in a standardized format but to more specialized kind of assessment and the recording of those kind of information. So I think we have a question in the back if you'd identify yourself. Yeah, hi, Jim Witter from NIAMS. Formerly of the FDA. Just a question, this is really for my FDA colleagues. I should know the answer to this but looking at the alparenol label, I can't remember that there's anything specifically that talks about HLA testing but my broader question is are there efforts underway? For example, I know the pregnancy label is being redone, rethought. Are there efforts underway to get this kind of knowledge into labels so that clinicians that are prescribing things could go to the specific site in the label and say we should be testing for this, this, and this? Mike, do you want to answer that, Mike Papinowski? Sure, there's currently no labeling related to HLA in the alparenol labeling. We'll talk a bit more about this tomorrow but it is somewhat of a complex issue to sort of communicate some of these testing recommendations because on the one hand you could recommend that all patients be tested but that place is obviously a very huge burden on prescribers if that type of language is in labeling and we don't necessarily know what the unintended consequences of such testing are. Also, even when there is information included in labeling it can sometimes be misinterpreted. So for example, in your presentation, I believe there was a number of drugs for which there's labeling related to pharmacogenetics but of the 150 or so that we know exist, not all of them are necessarily actionable and we don't necessarily provide recommendations around the test result per se. There is a lot of challenge and nuance to communicating this type of information in labeling but certainly I think most of the divisions at FDA do generally favor providing more transparency and information in communicating risk factors for serious toxicities than not. I mean, and even in the alparenol labeling, for example, there is reference to dose and renal function and other factors that could contribute to Steven Schottson risk. Yeah, I wanted to come back to Carolyn's question. To me, and I hate to continually be the wet blanket here but at least in this country, case reform forms are a dead end. We cannot get people to fill them out. We don't even have pharmacists like our friends in Thailand do that will diligently fill these things out. And while we always would like more data, I think the more relevant questions and perhaps the thing that can be the group three might think about would be what are the ways that we can identify the pieces of information that contain the most discriminatory value in terms of making a case designation. We've heard something about the role of pictures and visuals, are there other things that provide more discrimination versus other things that would be nice to have but aren't helpful? So that would be one possibility. The second possibility that we've talked about in the context of trying to gather more information around genomic variants for the ClinGen project and others is the fact that the most invested people are the ones that we generally never go to for information, which is the folks that are affected. And it sounds like we have a group here that's really invested in actually supplying a lot more information, particularly some of this longitudinal information that's going to be critically important to understand some of the long-term sequelae. And so I think one of the other things we should think about is how can we systematically obtain data from our patients who hopefully would be our partners in trying to collect data that would allow us to answer questions? I think the other thing that I would just highlight here is that it would be really nice if we could standardize the information collected across all of their different projects. Because my guess is, is that your collection form looks different from Maya's collection form, which certainly would look different probably from a collection form that Emerge might come up with. And if we could consolidate that, say, okay, here's what everybody's gonna collect, here's the minimum data set, that would allow us again to combine data much more easily. Okay, we're going to need to end up this session, which has been an excellent discussion and move into the working groups. Caroline is going to give the working groups their charge once again. The working group chairs have heard this, but what you'll be expected to do and that and where you will go. Just one note on dinner this evening. If you haven't let Jennifer know that you'd like to come to dinner, it should be very nice. And it's not terribly expensive to be given Bethesda prices. And actually, it turns out our frozen rain event is not turning out to be quite the disaster that we thought it would be. So the weather shouldn't be too, too terrible and we do hope you'll join us. So before I have you all dispersed, we're not gonna reconvene today after we break up into our groups. If you are taking the shuttle, however, at the end of the groups, if you come meet in this room, I'll be going back on the hotel with you.