Transcript [EN]: WDPLQW_GG7U

So welcome, everyone to today's Drug  Hunter flash talk. And I'm your host, Dennis Koester, Director of Industry  Research and Relations here at Drug Hunter. We're happy to have you here  today. And we're glad to discuss DILI risk today drug induced liver injury  risk with Anh Tran from Amgen welcome Anh Morning Dennis good morning, everyone. Oh, good afternoon. Good evening,  wherever you're joining from, Thank you so much for the opportunity.  I really appreciate it. Awesome. So without further  ado, I want to introduce our guest Anh Tran today Anh obtained her Ph.D. in organic chemistry from  the University of Sydney in Australia in 2014. And followed that up with a postdoc in Jin-Quan Yu's  lab at Scripps then Anh started her career in the industry at Arcus Bioscience here in the Bay Area  in the East Bay and significantly contributed to their immuno-oncology portfolio and has actually  two compounds in phase II clinical trials right now. So really amazing achievements in the  beginning of Anh's career. She then in late 2020 joined Amgen in Thousand Oaks and there she  is currently a Principal Scientist and a Medicinal Chemistry Team Leader. And she's working in the  space of early drug discovery projects mainly in the targeted protein degrader space. Anh is also  a very active writer. She writes for example for the ACS, medicinal chemistry reviews and is also  a regular contributor to The ACS MEDI division, and the GRC medicinal chemistry  future planning efforts. So Anh, really nice to have you today. Welcome to  the webinar, and thanks for educating the audience on DILI today. I already learned  so much from your book chapter, actually. Yeah. In the good old Drug Hunter  tradition. Before we start, before we start launching in the presentation,  I just wanted to ask you a few icebreaker questions. So what actually inspired  you to start a career in drug discovery? Yeah, I would say like most people, I do have a  personal story of why I went into drug discovery. And for me, it went back to when I was eight years  old. That was when my dad almost had a fatal heart attack. So he survive and thrive today, thanks to  really advance in modern medicine, right. So he's literally take every possible cardio metabolic  drugs you can imagine, to survive to this day. And for me and my family, we've been on the receiving  end for so many years. And at that day in time, it really inspired me to be a drug Hunter. So  one day I can be on the giving end there's a way to pay forward to the drug hunters will help  bring those medicines to save my dad's life. Wow, that's super inspiring story. And yeah,  we're glad your dad is is with us today. So thanks for sharing that Anh. What are you  most excited about in drug discovery today? I'm actually really excited about like  new screening technology. Right. One of my favorite at the moment is the cell painting  technology, which is a type of high content and imaging kind of assays that allow us  to investigate like the modern molecular mechanism of multiple organelles at any one  time. I'm really excited to see how it can combine with artificial intelligence and machine  learning to process those multi-dimentional data to really advance this this new paradigm  of drug drug screening technology. Oh, no. There's so many like super cool screening  technologies out there these days. Yeah, cell painting is actually one really cool one. We  we looked into that earlier my career too. So it's super interesting. What's your favorite molecule  and why we love to ask this question because like, like gets like to so diverse, different molecules.  So just curious, what is your favorite one? My favorite one is actually Lenacapavir, which  is HIV-1 inhibitor from Gilead, which is the 2022 Drug Hunter molecule of the year. Not only  because it has a crazy structure. So I think every medicinal chemists love that structure,  right? But my friend from Gilead told me it's nicknamed the unicorn, affectionately by their  Gilead chemist's and I think I was super inspired by this the story that was told by John link two  years ago at the GRC is really a story of grit, raight, the perseverance of this compound  take 10 years from the discovery to the final approval is super, super cool and  inspiring story for every drug hunter. Oh, yeah. Anh, I love this molecule, I have  to tell you, it's like a once in a half year injection to HIV, first in class capsid inhibitor.  It's like an amazing compound, the structure is absolutely wild. I've heard people from Gilead  talking about it. It's absolutely amazing story. So very good choice. I can totally relate to that.  Next question is if you could go back in to early in your career, and you could give yourself some  career advice. What do you think it would be? I actually so you introduced me as  having a PhD in Organic Chemistry, I would elaborate on that I have a Ph.D. in  Medicinal Chemistry. Little bit. Little did I know that earned me the nickname of the jack  of all trades, which is actually equivalent to knowing a little bit of everything  and not knowing very well anything. So I at the beginning of my career had fought  really hard again, this prejudice, right? So if I would go back in time and tell my younger  self, is that actually a jack of all trades, not a liability is definitely an advantage  and like the unique perspective and voice will definitely be heard and valued and do  not let other people perceptions define what is possible for you, which is what I did. I  limit myself a lot of opportunities, because I actually believed in this label. And it's  really hindered my career development early on. Interesting. That's, that's kind  of surprising, actually. Yeah, but I know like Medicinal Chemistry Ph.D.s are  sometimes perceived weird in the industry. But yeah, so yeah, thank you for that advice.  I think that's, that's good. Good for our audience to know. So final question. What  do you like most about Drug Hunter? Yes, Yeah, thank you so much. Those are kind of my  favorites too. For the conference, you can't like find this information anywhere else. And the  patent highlights always good to kind of keep up with something. And sometimes it doesn't even  hit the primary literature at all right? Yeah. Cool. Thank you. Anh so before Anh gets started,  I quickly wanted to share because Anh is talking about DILI risk today. And a compound class where  this is a particular risk is for BTK inhibitors for MS. So I just wanted to very quickly screen  through a case study that we have here on Drug Hunter. So here you can see our one reversible  BTK inhibitor for the treatment of MS. So you can see the competitive landscape, you can see  we're talking about this DILI risk, we're talking about how the target relates to the disease.  And for the hit finding and lead optimization, you can find very many transferable insights that  you can potentially apply to your project. So you know, like this has many, many transferable  insights, how you can solve common problems in drug discovery, and just you know, gives you  a profile, a preclinical profile and clinical profile of the compound and also some synthesis  highlights that are really interesting about this particular project. So with that, I'll open it  up for Anh to share her presentation with us. Yes, yes. Perfect. I can see it fullscreen. Okay, perfect. All right. So hello everyone. So thank you for joining today.  And actually true to my jack of all trades label. Today I'm going to tell you a story on drug  induced liver injury and from here on in abbreviate as DILI. Even though it's a toxicology  subject, it has high relevance to medicinal chemistry. And I hope I'll be able to convince  you about that at the end of the presentation. Sorry, this problem with slide transition. First  is an outline of today's presentation, we'll go through an introduction on what is DILI, then  we'll delve into the multiple mechanistic drivers that underlying DILI, and also a classification  based method integrating all of these risk factors to really assess DILI risk in the clinic, we'll  follow up with a series of case study, outlines successful DILI mitigation strategy we'll begin  to conclude the talk by balancing between DILI risk and therapeutic opportunities in the context  of recent FDA approval. And finally, with a final conclusion, but before that, so what is DILI?  DILI manifests in the clinic in a really diverse number of way as you can see on this diagram here,  probably the most common manifestation of DILI in the clinic is cholestasis, which is the bile flow  obstruction resulting in bile acid accumulation and subsequent liver injury. DILI can also  result in acute liver failure, hepatic necrosis, as well as autoimmune hepatitis. So currently DILI  is being assessed in the clinic through a series of liver injury biomarkers such as alanine amino  transferase, ALT, aspartate, amino transferase, AST, or total bilirubin. But as you can see  on this diagram here, DILI is a result of a very complex interplay between multiple DILI  mechanisms, the patient genetics factors, as well as environmental stimuli in order to  give rise to the complete DILI in the clinic finally. So there is a very strong business  case on why we should care about DILI. In general adverse toxicological findings account  for 40% of program termination. And this is a survey that's being carried out across multiple  pharmaceutical companies. So with cardiac toxicity being the number one factor, liver toxicity is  a close second, and DILI made up a significant proportion of liver toxicity. Actually though, in  terms of frequency DILI is quite rare. It occurs in about two in 10,000 cases. But despite this low  frequency, it gives rise to more than 50% of acute liver failure cases, because of this low frequency  as well as the many case idiosyncratic nature of DILI it's actually escaped detection until really  late stage clinical trial like phase II and III, where it's already incur very significant R&D  costs. And really, the highlight in this case are the BTK inhibitors with Dennis had really briefly  mentioned about that had been put on partial FDA clinical hold, BTK inhibitors are intended for the  treatment of multiple sclerosis. As you can see, these are molecules from different companies  representing different chemo types, and they will put on clinical hold at both phase II  or phase III, three of these molecules are actually covalent BTK inhibitors. The last  molecule fenebrutinib that was highlighted by in Dennis just then from Genentech is a  reversible BTK inhibitor all show DILI risk, resulting in this partial clinical hold. So as  mentioned, the underlying mechanism of DILI is very complex. And this is complicated further by  the historically very poor correlation between animal and human pathophysiology. So you'll  see in the subsequent section of this talk, the way in which DILI risk is being currently  assessed in the preclinical space is through a suite of in vitro assay and not through animal  toxicology studies like we would do with cardiac toxicity for an assessment for an example. So  the first and probably most well characterized risk factor for DILI is inhibition of hepatic  transporters. And hepatic transporters regulate the hepatic circulation of bile salts. Bile  salts are important not only for the metabolism of Xenobiotics, but also critical for the  metabolism of nutrients such as lipids. However, because bile salts being detergent like and also  hydrophobic, they actually really cytotoxic and so their level has to be tightly regulated,  whilst also being taken up from the blood by transporters that is NTCP and OATP into the  hepatocytes, whereby it will be efflux by a series of transporters expressed on the surface  of the canalicular membrane into the bowel. And the key transporter here is bile salt export  pumped or BSEP which regulates the rate limiting export of bile salts into the bowel. So you can  easily imagine if you have a BSEP inhibitor, you have a consequence of accumulate bile salts,  resulting in liver inflammation and subsequent liver injury. At the moment, there is no consensus  on a safe in vitro BSEP inhibition value in order to mitigate DILI risk in the clinic. And that  is because as previously mentioned, DILI is a consequence of multiple risk factors and not  just BSEP inhibition alone. So in order to more effectively translate the in vitro BSEP inhibition  to an in vivo BSEP disruption risk, the Merck team has developed a two tier testing cascade in order  to assess this risk. And the tier one assay is probably the most well established BSEP inhibition  assay utilizing an inverted membrane vesicle made up of insect cells, and inserted to this membrane  vesicle is a human recombinant BSEP. So, what this assay tracked is the uptake of a tritiated BSEP  substrate taurocholic acid or TCA as you imagine, if you have a BSEP inhibitor, you have a reduced  uptake of the substrate into this inverted vesicle membrane. This will allow you to determine an  IC50 value. And if the IC50 value is less than five micromolar, these compounds will move into a  tier two assays, it's important to emphasize that this tier 1 assay can only capture the effect of  the parent compound in terms of BSEP inhibition in vitro, it does not take into account the  effect of metabolites on BSEP inhibition, because this system is not metabolically  competent. And this shortcoming is addressed in the tier two assay, the tier two assay utilizing  HepatoPAC, which is the micropatterned coculture system between primary human hepatocytes and 3T3  murine fibroblasts. So, this cold culture system is a practically mimicking the liver physiology  it also has a really high metabolic activity and stable phenotype over time. This assay then  allow you to determine what the team call a biliary excretion index IC 50 value and if this IC  50 value is less than 10 fold the maximum unbound concentration at the lever inlet, so there is now  a potential for this compound to inhibit BSEP in vivo. So then the project team has to consider to  reduce this liability either through an SAR effort or if they decide to nominate this compounds  series always clinical nomination, they have to monitor for cholestasis in the clinic. So highly  recommend you check out this paper from Hafey et. al. for the description of this two tier assay  to monitor BSEP inhibition risk in vivo. Another risk factors that go in hand in hand with  BSEP inhibition is mitochondrial dysfunction, mitochondrial toxicity account for about 40% of  hepatotoxicity that relate adverse drug reaction, as you can see and appreciate on this slide, the  structure of mitochondrial toxicants are extremely diverse, they also very diverse in terms of their  lipophilicity range, they either can be reaching SP2 center or SP3 center pretty much there is no  way you can structurally de-risk mitochondrial dysfunction risk. And you can also see here that  in many cases, these mitochondrial toxins ends are also BSEP inhibitors. So if you have compounds  that pose both mitochondrial dysfunction risk and BSEP inhibition, there is a very high chance they  will exceed the DILI in the clinic. And in fact, many of these compounds are all DILI positive,  at least at high dose in the clinic. So there are multiple mechanisms in which mitochondrial  dysfunction can take place as you can see here can either a complex one inhibitor, electron  transport chain inhibitor, or the incorporation of nucleoside drugs into mitochondrial DNA.  But regardless of the microscopic mechanism, ultimately what happened in mitochondrial  dysfunction is that you will have impair ATP synthesis because mitochondria is an ATP  powerhouse. And another effect is that you have an increase in the production of reactive  oxygen species. So currently, the way in which mitochondrial toxicity is being assessed in vitro  is through two different high throughput assays. The first one is a cytotoxicity differential  IC50 assay where HepG2 cells are being fed with two different nutrients source, galactose  or glucose. So the main difference between these two nutrients source is that for cells on  galactose, they can only rely on mitochondria to produce ATP. Cells with glucose however, it can  utilize alternative pathways such as glycolysis to produce ATP when the mitochondrial function  is being disrupted. As a result, galactose fat cell will be way more sensitive to a mitochondrial  toxants and compared to a glucose fat cells giving rise to an IC 50 ratio. So if this IC 50 ratio is  greater than three, there is a pretty good chance to compound showing in vitro mitochondrial tox.  Another way to assess mitochondrial toxicity is an assay co respiratory screening technology  and this assay utilizing oxygen consumption as a proxy for mitochondrial respiration,  utilizing a fluorescent oxygen sensitive probes. So because the fact that mitochondrial  toxicity and BSEP inhibition tend to go hand in hand and when they do they elevate DILI risk in  the clinic. The current best practice right now is to incorporate mito toxicity screening in  really early stage optimization programs just so we're more informed of the potential DILI risk  moving forward, however, to be the order to detect mitochondrial toxicity in the clinic is extremely  challenging because mitochondrial toxicity it is very well known for having a delayed presentation  in vivo due to multiple innate compensatory mechanism. And the best example here has to be  Fialuridine. Fialuridine was actually is an HBV inhibitor. And it's a mitochondrial toxic and  but it was not detected in clinical trial until the 13 week mark where it resulted in delayed,  fatal liver failure, ultimately result in five patient deaths in late stage clinical trial.  So it's really highlighting the challenge of translating from the early risk assessment to the  risk that will eventually be posed in the clinic. Even though both of the bile salts, transporter  inhibition as well as mitochondrial dysfunction, really offer the medicinal chemists very little  opportunity for structure based DILI mitigation strategy, reactive metabolite formation give  us a lot of chances, because this is really a structure base here. So the overall goal of  xenobiotic metabolism is to produce soluble and benign metabolites, but during that process, it  also generates electrophilic intermediates, that in many case can be captured by nucleophiles in  endogenous protein. So, this can result in either direct toxicity or hapten formation that will give  rise to immune mediated DILI injury, we will go into in the next slide and both mechanisms are  operative in DILI. So the current gold standard for detection of reactive metabolite formation  is covalent binding study of radio labeled drugs to primary human hepatocytes. However, as you can  easily appreciate access to radio label drugs is restricted. And so this process only applicable  for late stage optimization program. However, advancing MSMS instrumentation over the years  have meant that right now, many companies are shifting reactive metabolite detection to an in  vitro MetID format of non radio labeled compounds incubated with either HLM or primary human  hepatocytes in the presence of soft nucleophile. So that cysteine and GSH or hard nucleophile,  like semicarbazones and potassium cyanide, and so you're looking for the trapping adducts  with these nucleophiles. But it's important to emphasize that while this technique is is is  great, it's so sensitive that in many cases you will detect trapping adducts and the presence of  trapping adducts does not correlate itself with the risk for DILI in the clinic is just provides  a flag and so you have to evaluate with other risk factors at the same time, you cannot use the  presence of trapping adducts as a go or no go decision for chemotypes. In terms of evaluation  of overall risk in the clinic, sometimes it can provide a good prediction of DILI risk and that  is the case of Fasiglifam, for example, Fasiglifam generates a GSH adduct and it was positive in the  GSH trapping assay also shown to be DILI positive, that result from the formation of the acyl  glucoronide within being trapped with GSH to generate this adduct here. Sometimes, though,  having no adduct formation does not mean that you are safe that the case of Ximelagatran, whereby  there was no adduct formation at all with GSH or any other nucleophile. However, this compound  was shown to be DILI positive, possibly by an immune mediated mechanism that's still not very  well understood at this point in time. And the last two example really highlight the elephant  in the room. Both Troglitazone and Pioglitazone have exactly the same bio activation mechanism  as they both have the same problematic motif, that is this thiazolidinedione, that being formed  through sulfonium ion then being trapped with GSH. However, only Troglitazone is DILI positive.  And that is because the enormously high dose of Troglitazone compared to Pioglitazone. So  it's really important that you assess DILI risk, also in the context of dose, and we'll come  back to this point multiple times throughout the presentation. So in recent years, there  really have been multiple advances in order to look at reactive metabolite risk beyond just  these covalent binding study or trapping adducts. And that's been the work done by the Merck team.  So the Merck team has developed both an in vivo, and in vitro rat transcriptional signature  assay, which were they call a bioactivation liver response assay. So here they look for  transcriptional activation of a prioritized set of genes in the Nrf2/KEAP1 and the Nrf1  pathway. So the workflow is right here. So a compound is either dosed in rats at two different  dose, low dose and high dose in a four day study or being treated with the culture of HepatoPac  cells repeated dosing at three different dose for a period of nine days. So either the liver  or HepatoPac and being collect their RNA level of the prioritized set of genes in this oxidative  stress pathway is then being analyzed and compare with the control level. And then this allowed them  to determine what they call a bio activation liver response assay score BA-LRA score and it was  determined that if this BA-LRA score is greater than 0.2 at a liver exposure index, that's more  than three, compounds the showing potentially high DILI risk in the clinic. And this kind of the  BA-LRA score and liver exposure index, when used together has a much better predictive ability of  differentiating high and low DILI risk compared to those that have been previously considered gold  standard before. So that's the covalent binding study is it because this assay looking at more  holistically the bio activation risk and not just the target formation of the reactive metabolite  species that you're looking for. So, I really highly recommend checking out these two paper  for the in vivo and in vitro BA-LRA workflow it's really incredible work. So related to the reactive  metabolite mechanism is immune system mediated liver injury. So, this mechanism is really rare  and complex, but it has gained appreciation in recent years, mostly due to the increased  usage of checkpoint inhibitors. So as expected, it involved both the innate and adaptive immune  system. So it's been appreciated that in DILI positive patients and DILI negative patients, they  differ by the fact that DILI positive patients actually have drug specific T cells. It has also  been noticed that there is certain human leukocyte antigen alleles that are present in certain  individuals that making them more susceptible to immune mediated DILI risk and the the pathways  go through here. So the initial cascade is kind of similar to the reactive metabolite risk where you  form an adduct between the reactive metabolite and the protein, with events being digests further  into peptide conjugates. This represents itself as an antigen to antigen presenting cell and  this count as signal one, there's need to be to signal for this cascade to proceed signal two  can come from a variety of sources in many cases that come from the innate immune system with to  signal in hand what will subsequently happen is the maturation of naive CD4+ and CD8+ T cell into  drug specific effector T cells, we can now go on and interact with MHC-1 molecules on the surface  of hepatocytes, and this trigger hepatocyte cell death and liver injury. So this mechanism is  pretty relevant for checkpoint inhibitors, such as anti-PD1 and anti PDL-1, which are  increasingly being approved for multiple cancer indication and the use of these checkpoint  inhibitors are associated with increased risk of all-grade and high-grade hepatitis. So  it's important for us drug hunters to really appreciate the fact that if we plan to combine our  new chemical entities with a checkpoint inhibitor, we have to have a lower predicted human dose than  when we just want to use it as a mono therapy in order to really reduce this immune mediated DILI  risk. An important factor to consider is because as this axis of liver injury is being increasingly  appreciated, there's been a call for need for new screening technologies to really detect this risk  in vitro because right now, there's not really a good in vitro system. So this has been trigger  moved towards long-term 3-D hepatocyte culture or a co-culture system that consisting of hepatocytes  and dendritic cells to more effectively evaluate the stress in vitro. So as you can imagine, with  all of these risk factors happening at any one time, there have been multiple attempts to really  incorporate all these risks in one place. And one of the really successful avenue is through  the software called DILIsym. So DILIsym is a quantitative system toxicology model, and itself  is a model that consists of multiple sub models, each sub model representing an individual  DILI mechanism as shown here, whether it's BA transport, inhibition, metal toxicity or  reactive oxygen species, and the sub models are linked together. And they also incorporate  additional factors like population variability, which is as previously mentioned, the introduction  is very key to manifestation of DILI and through a PBPK approach, as predict the hepatic  concentration of both the parents as well as the metabolites in hepatocytes. And with that  information, it can allow for the calculation of the concentration as well as predicting the  timing of the release of liver injury biomarkers, and it's ALT, AST, and bilirubin and the  timing of release into the serum. Ultimately, it will allow for the determination of the  key mechanistic drivers underlying DILI. So DILIsym was a result of a partnership between  academia and major pharmaceutical companies. And so it's best to illustrate really this  through a case study and these are the CGRP antagonists in turn for the treatment of migraine.  So the first generation and the second generation compound Telcagepant and MK-3207 from work was  discontinued because they were DILI positive. So with that, these two molecules actually have  triggered the Merck team to develop both the in-vitro and in-vivo BA-LRA score that I  previously disclosed, actually, but what happened is the simulation plus the company who  developed a DILIsym also put in the ALT and AST value into this model as well, and they also come  and predict that these two molecules are actually predict to be DILI positive if you're using this  workflow. And they also predict that Ubrogepant the second generation compound is DILI negative.  And that's actually true in the clinic expanding out and its also DILI negative at the same time.  So even though it looks like this software, is really capable of predicting whether a compound  is positive or negative, and you can basically put in the IC50 of each DILI mechanistic assay and  predict whether your compounds posing a risk or not, it does have additional room for improvement,  because actually, it predicts that Telcageoant and MK-3207. Driving DILI through BA transporter  inhibition and the Merck team demonstrate through the experimental study that the main factors  underlying DILI for these two compounds are actually reactive metabolite formation. So there's  room for improvement. This final step in analysis of mechanism is not quite accurate for for these  two compounds, but it does have the ability to predict the final outcome pretty well. And I'm  sure there will be improvement over the years for the software as well as people increasingly use it  for understanding potential DILI risk translating into the clinic. Another way of incorporating DILI  risk is through a classification based method. And this is through work from multiple company and  the first one is GSK and Merck combination. So in 2004, Merck published a paper where they  underline the risk of reactive metabolite and how it's correlated to DILI, and GSK took it  one step further, and combining this reactive metabolite risk with the daily dose predicted  daily dose and arrived at the decision tree model as you can see here, so if a compound has  both reactive metabolites, and it's being dosed as greater than 100mgs per day, there is a pretty  high chance that that compound will exhibit DILI risk in the clinic, so you have to consider it  for termination. If the dose is less than 100mgs per day, perhaps you can progress the compound to  candidate selection with precaution. AstraZeneca took this one step further, when they actually  introduced a zonal classification of DILI risk here, we're using two different axes, the x  axis is called covalent binding body burden, you see the unit here mg per day because actually  calibrated for dose predicted daily dose and the y axis is the performance of compound across  five DILI mechanistic assay obviously, the higher the number the more individual toxic  the compounds are. So this allowed them with this 1mg per day limit for the CVB burden established  four different zones for compound where zone one is no hazard, zone two is medium risk and zone  three and four are high risk for different reasons and probably should be considered for  termination. And the current classification of DILI risk will incorporate the work from GSK,  AstraZeneca. And we can also put in the recent transcriptional readout work from Merck. So if you  have a compound with high transcriptional readout, in the Nrf2 / Nrf1 pathway and high BA-LRA score  and also score in DILI mechanistic assay and also predict to have a high dose and probably the  compound have to be considered for termination. So it's everything has to be considered in the  context of the projected daily dose. So you can see though with many of these risks have really  been focused on the latest stage of optimization, but as predicted, there is a lot of interest  to move this risk screening to an earlier phase and that's work done by Pfizer, where  they incorporate a machine learning approach using a random forest model. They incorporate  the predicted daily dose with the caveat that daily dose prediction at early phases highly  uncertain. But at least regardless of that, they combined with lipophilicity and ionization  class and come up with an ROC AUC score that allows them to really differentiate between high  DILI risk and low DILI risk chemotype so they can prioritize the early discovery efforts. So I  recommend checking this paper out in case of you interested in a more early phase screening  for DILI risk via machine learning approach. So with that, we'll we'll switch gears into the  final section of this talk, we talked about the DILI mitigation strategy. And the first one is  the LPA-1 antagonists from BMS. So LPA-1 is a key mediator of lung fibrosis, which is the underlying  driver for idiopathic pulmonary fibrosis, and BMS-986020 is the first generation compound  it demonstrate proof of concept efficacy in a six months phase two trial, however, at a really  high dose 600mgs BID. So it came as no surprise that it gave rise to elevated level of ALT and  AST with three times the upper limit of normal and also result in cholestasis in three out of 96  patients with this being the common manifestation of DILI. So with that BMS voluntarily terminated  this compound however, they've been able to utilize this level of ALT and AST in the clinic  and fit them into the DILIsym model and DILIsym predict that bile acid transporter inhibition and  mitochondrial dysfunction as the key contributing factors to clinical DILI. So using this  information, the BMS team were able to set up all these in-vitro assay to monitor risk  in these two particular areas. As you can see, the first generation compound clearly show  micromolar inhibition for multiple transporters, not just BSEP so clearly represent a risk here.  So the way in which they mitigate this risk is to reduce lipophilicity, as you can see, by one log  unit, increase fraction, SP-3, but probably most importantly, decrease the maximum daily dose. So  it Golding from 600mgs BID to a final 125mg BID in the final compound. And you'll also see the  final compound also stay clear of all the bile acid transporters inhibition capacity. So actually  this 986278 exhibit no hepatobiliary toxicity. And he's now in phase III for IPF. So really highlight  the successful DILI mitigation strategy utilizing this weight of in vitro assays and as well  as in silico risk analysis with DILIsym the second case study is the BACE1 inhibitor  from Janssen and here they also take lessons from a failed clinical example atabecestat. So  atabecestat actually terminate in phase II due to elevated level of ALT and AST in 11% of patients.  This is thought through an immune mediated delay, they recognize that one of the key problem motif  here is an aminothazine group that will remove in the second generation into a dihydropyrimidine.  But still this compound was showing up to be positive in a cyanide trapping assay which they  reason is due to this present of this dioxane motif here and through a series of SAR effort  they were able to replace this 1,4- dioxane with this fluorinated 1,3-dioxalate, would block  reactive metabolite formation completely there was no trapping out of being detected synthesis, the  DILI risk is significantly reduced. This compound was not progressed further, but it was for other  reasons not toxicity and not DILI toxicity. So the recommend practice here is try to stick to this  rule of two as much as possible by lowering the dose of less than 100mgs per day and keeping  lipophilicity as low as reasonable preferably below three, to really reduce DILI risk. However,  as with any risk, we really have to balance with therapeutic opportunities. And nowhere can  it be demonstrated as well as this case study here with Omaveloxolone. So Omaveloxolone is a  Nrf2 activator is approved for the treatment of this really debilitating neurodegenerative disease  called Friedreich's ataxia. So before the approval of Omaveloxolone, there was no approved treatment  for this disease indication. And this disease is so devastating, that patient only survive up  to about 35 years of age because of the highly unmet medical need. FDA has exercised regulatory  flexibility. So even though Omaveloxolone, will give rise to elevated liver injury biomarkers,  not surprisingly, because it's a Nrf2 activator. And you might remember, Nrf2 is one of the key  regulator of oxidative stress. We mentioned a few slides ago when we talked about the  transcriptional signature work from Merck. And so it's not surprising this is definitely on  mechanism that it gives rise to elevated level ALT and AST. Because of the highly unmet medical  needs, though it still went through approval with recommendation from FDA to really test these liver  injury biomarkers before and during treatment. However, not all stories turned out this well.  Vadadustat iis a story that doesn't turn out as well. So in 2022, FDA issue a complete response  letter to Akebia for this molecule, which is a HIF prolyl hydroxylase inhibitor for the treatment of  anemia in chronic kidney disease patients. So the complete response letter was due to increased risk  for thromboembolic events as well as DILI however, Vadadustat was already approved in Japan for use  in June 2020. And with that, post marketing safety data generate in Japanese population, Akebia being  able to resubmit the information to the FDA to highlight that there's really no DILI risk at all,  they haven't seen anything during these post two years marketing period in the Japan population  that eventually have allowed FDA to approve this compound in March this year. However, this delay  in approval two years means that Vadadustat is no longer the first in class HIF prolyl hydroxylase  inhibitor in the US market because in 2023, a similar compound from GSK, because of no DILI  risk was already approved. So it's really in the best interest of company to really mitigate DILI  risk to avoid these costly regulatory delay. So with that, I would like to have a conclusion is  that DILI Hopefully you appreciate it by now arise from a really complex interplay between multiple  mechanism but the consideration in the preclinical space have to be on a recent therapeutic benefit  analysis and this case by case and it's very, very different depending on which therapeutic  area you operate in it. Of course of you in the oncology space, there's more flexibility. If you  in the cardiometabolic space, you are not going to have any flexibility at all, unfortunately,  and the successful mitigation strategy, so far have been involved in the reduction of  maximum daily dose and compound lipophilicity. But it's only possible because these companies  have increased look at the integrating the readouts for multiple DILI risk factors in vitro  assays. And there's been multiple recent advances in technology such as the HepatoPac culture that  I talked about, as well as the 3D organoids, I did not have the time to talk about today,  transcriptome profiling, as well as in silico risk prediction with DILIsym. If anyone interested  in these additional development, I really highly recommend check out this two part ACS webinar  that's really highlight very well. And this is actually co-sponsored between the toxicology  division and the MEDI division. Some of you may have recognized that today's presentation is based  on a book chapter that I led the cross company writing effort of last year in ACS Medicinal  Chemistry Reviews that I would like to take this opportunity to thank all my co-authors who  have contributed to the writing effort, I would also like to thank the MCR section areas, Brad and  Matt, who truly champion and believed in me in my ability to deliver on this complex topic at the  very beginning. And the two of really inspiring and wonderful editors in Joanne and Joachim for  their really attention to detail and constructive feedback that really increase the quality of this  book chapter. And thank you all for listening today. I'm happy to take any questions. And  I'll hand it over to Dennis for hosting the Q&A. Thank you so much. Anh that was really, really  amazing. I mean, I learned so much about the different assays that you presented for, you know,  mito toxicity, and it was really awesome. And we already have a lot of questions in the audience.  So for the audience, I see a few raised hands. If you have any questions, please submit them  in the Q&A panel, and we'll get to it. So yeah, let me just start here with with some audience  questions. We have Ben Taft asking is BSEP the only transporter linked to DILI risk. I mean,  I think you started already talking a little bit about MDR4 and other transporters that may  be involved. So that that question, I think, is Yeah, I think BSEP, as I mentioned, is a rate  limiting transporter, right. But it's definitely not the only one. As you can see from the BMS  LPA1 antagonists case they look at multiple other transporters, right. And it's not just  the transporter on the canalicular membrane, which is where BSEP is, but also the transporter  on the basolateral membrane also play a role. So I think while BSEP is the main one, just looking  at BSEP alone is probably not a good idea. And I think that's why having the two tiered assay  from Merck the tier II assays looked more than just BSEP so a little bit more holistic  assessment. So I think it's also taking into account of some of the other not so well known  as well, BSEP is definitely the key factor. But I would recommend maybe look at both the Tier  1 and Tier 2 assay in a hepatocyte culture, I think it gives a much more idea than just a  BSEP in vitro assay, like the invert membrane vesicle is actually giving a lot of false  positive, which that's not what you want. Right. Right. Right, totally. And I think in that vein,  we have another question about the properties of compounds that are typical BSEP inhibitors. So is  there anything, maybe any guidelines, like any, you know, properties that medicinal chemists can  modulate like a log D or pKa? That, you know, makes compounds more, you know, that increases  the probability of BSEP inhibition? Yeah, I don't think there is like a complete structure.  But I think there are some association actually, if your compounds is like an OATP substrate  transporters which many acids are the chance was being a BSEP inhibitor is also pretty decent,  right? So I guess then we should try and minimize like, OATP transporter liability, which will also  help with BSEP inhibition as well. And I think if anything that resemble a BSEP substrate, right,  if you have working in the steroid field, if it resembles taurocholic acid or something, right.  But if that is the pharmacophore that you need, then you have to try and see if you can change  the structure a little bit to stay away from it. But structure based study for the BSEP inhibition  is, I don't want to like sell that idea. Because I think that's very dangerous. I think the  only way where you can really mitigate by structure is in reactive metabolite formation  actually, and with the others, you basically have to set up the assays and put through as  many compounds from those assets as possible. I mean, a question that I had was where in the  in the cascade, do you actually start to assess DILI risk, right? Like, is it like and what do  you start with, like, what assays would you kind of start with like, let's say you have a form,  do you start at the candidates stage or even early in late lead optimization and kind of what  assays would you recommend using in that stage? Yeah, I think this is actually a difficult  question, right. I think different companies probably have different factors. I guess  it depends on where they've been burned by DILI before. For, for the companies who have been  burned by DILI, they probably put it as the early optimization phase, right? Like, no, I don't want  to risk that, because it was a nightmare for me years ago. So and but I've, I've seen so far that  people do put BSEP inhibition and mitotoxicity risk pretty early in the like the early  optimization phase, and if you start seeing risk, especially in both of those two assays, almost a  flag that you need to look into it further, right. And I think maybe BSEP, mitotoxicity, and reactive  metabolites in terms of like just covalent adduct formation, because it's a pretty standard and not  too expensive assay to set up. So maybe with those three, it can inform you with some kind of whether  you are in a high risk zone or not. And I guess this question also related to what, where your  target is, right? If actually your target is in the liver, like GLP1, for example, then you're  already, by your target being that location, at a higher risk of DILI, in general, if you  are in that space, then you obviously have to evaluate this risk way earlier than a program with  a target that is expressed elsewhere, like in the lung or in the heart. Right. And I think it's  also very much case by case dependent. Yeah, makes total sense. Thank you. Here's another  question from Hendrick. He's wondering how do teams interpret ms ms detections of  covalent adducts? Is there a threshold that is commonly used as a  flag for DILI risk? Yeah, this is a this is a difficult question, because it just  the problem is it's so sensitive, right? Yeah, I actually heard a talk from Janssen about  this, they were like, right now they literally can see covalent adduct in every possible  compound that they put through. And truly, that's not true that every compound has a DILI  risk, right. So it's already showing how many false positives that you're gonna get. So I  don't think that is like a threshold that you can actually use, as I said, it's probably  used as a flag. So if you start seeing that, say, maybe if you start seeing a lot of those  covalent adducts for particular sub series, perhaps it's now time to consider  investing into more elaborate assay, like the covalent binding study with like radio  labeled drugs, right? Or if you are willing to utilize like the in vitro BA-LRA score system  from to more holistically assess this risk, but I don't think there is a threshold per se.  Unfortunately, just because it's so sensitive. Yeah, makes sense. Makes sense. We had another  question from anonymous. Are there specific liver markers, eg, Hy's law cases or bilirubin,  suggestive of particular mechanisms of DILI? No, I don't think so. Those are pretty much like a  global final readouts. Right. So it's not because that you are seeing certain biomarkers, that means  they will be indicator of certain mechanism, that probably if there's one I would say, for if you  have an immune mediated DILI, what they have seen, but the problem is this is happening in patients,  it's not an animal, anything immune mediated DILI, they tend to see that sometimes for some patients,  you just have either drug specific T cells, or the presence of certain HLA alleles, actually,  that only detected in DILI positive patients versus DILI negative. That's the only thing that I  can think of that in the future, when we have more data, those HLA alleles, there is an attempt to  use them as biomarkers to predict the high risk patient from the low risk patient, even from  like a patient selection standpoint as well. Yeah. Thank you. And then there's a  question about organoids. How do you feel about the potential of them  being used? Or do you think they are currently overhyped for  the assessment of DILI risk? Oh, they already being used. Actually, I just  didn't have time to talk about them. That's why I kind of tried to mention it at the at the end,  because I want to pay tribute to that development, too. Actually GSK has already moved their  cytotoxicity because that's the another mechanism that I didn't talk about, because  there's too general, right. So there's, of course, the compounds would just be cytotoxic  to hepatocytes, generally, cytotoxic and GSK have moved their cytotoxicity assay for hepatocyte to  the 3d hepatocyte culture from the 2d because it's significantly more sensitive as well as specific.  So there's already work in that area. I don't think it's overhyped, to be honest, because I  think the problem with DILI is that is multiple parameters, right? And I think the primary human  hepatocyte is really only good for short term toxicity. The problem when you have something  like DILI with so many mechanism is sometimes just need time to manifest because many times need  to interact together. So you're in vitro system need to be stable over a long period of time  and actually primary human hepatocytes are not. And that's why you have culture like hepatocyte  HEPATOPAC and 3d organoids, which are way more stable over time that act more like a liver. So as  a result, you have a much better chance to capture this risk. A lot of this risk is time related.  If you leave these compounds for a longer period of time in vitro, you eventually will detect this  risk. But you can't really do that if your system is not stable over time. Right. So what you  see is actually the system itself is degrading, not the compounds doing anything. And and  that's really complicate result interpretation. Yeah, no, that makes total sense. Yeah. Thank  you. So I guess now we have two more really specific questions on specific programs. So  Gagan is wondering if the reason the failed BTK inhibitors or you know, halted BTK  inhibitors and clinical trials have been looked at in the new generation DILI assays? And  what the results of those readouts where if you I am not sure if they have been I mean,  not that I'm aware of, through my reading, actually, the one from Merck KGaA is actually  terminate completely. Yeah. Yeah. So I think it's just the company just deemed too much  risk to take it forward. So I don't think it has been looked at unfortunately, maybe if  we look at now, with the new set of assay, maybe we can detect it way better. But one  thing I want to emphasize that for all those four compounds that been on actually three  out of those four compounds that have been on partial clinical hold for BTK inhibitors,  they're all at pretty high dose, they all more than 100 mgs per day. So maybe it's just also  that the dose is playing a role here as well, as we mentioned, right? It has to be  considered in the context of dose. Yeah, makes total sense. And then another  specific question, the same Ubrelvy. So you briefly brought up that case study of  ubrogepant? And so people are wondering, how did the Merck team dial out the DILI  risk in that second generation? I think it's a CGPR. Yes, yes. It's a  CGRP antagonist, Exactly. So Telcagepant and MK 3207, which are the  first and, generation 2.1 compounds, I guess, they both showing DILI positive in the  clinic. Right. And I think the Merck team has determined that is true reactive metabolite  formation. So they use that information and they use that assay cascade to guide all  the subsequent compounds and to basically de-risk in that aspect. And Ubrogepant score  negative in those be in vitro BA-LRA era score, and also in vivo BA-LRA. That's allowed them  to have a safe compounds towards the end. Yeah, cool. That's, that's always cool to  know, like different strategies that people have used. And yeah, another question towards  the properties here is like, when considering lipophilicity, like, I think you mentioned  that Pfizer paper that had multiple inputs, people are wondering about is using log D  and log P something useful for DILI risk? Or, you know, should people were just  looking at, you know, mainly dose? Hey, I would say look at dose for sure. But, like,  in general, minimizing lipophilicity never hurts, right? It's I mean, even, even if it's not helping  in DILI risk, it probably helps elsewhere anyway. So I would say consider both at the same time,  there is actually some mild correlation. I didn't want to flag that up too much. But there  is a mild correlation between lipophilicity and mitochondrial dysfunction. In general, the  lipophilic compounds have just have a higher risk for mitochondrial dysfunction as well. So there's  multiple reasons for trying to keep a reasonable logP or logD. Sometimes it's not possible it  just because of your target, but try the best to basically keep reasonable lipophilicity.  And in the context of the dose as well. Yeah, makes total sense. And  then people are wondering about the throughput of these next generation  assays are. Are there any ones that are particularly high throughput? I think  you mentioned a few in the Mitotox, assessment. And then people are also  wondering about false positives in those assays. Maybe you can talk a little bit about  the high throughput aspect. And then is there a chance to like confirm false positives, I  guess, with orthogonal assays, I'm assuming? Yep. So high throughput, I think the  mitochondrial toxicity assays are really high throughput. The BSEPs assays also very  high throughput, at least the tier one assays; I think the tier two assay is reasonable. The,  actually the in vitro BA-LRA score to look at the reactive metabolite formation from Merck. I  mean, two years ago, it was described as medium throughput, and it's not really done in 96  well, plates actually. It's really not bad, right? So I think right now, they'll probably  have increased throughput. So I would describe many of these in vitro assays as medium to high  throughput, there's not really one that's super low throughput, right. And I'm so glad that  someone brought the questions of false positive; didn't have time to go through that. So I have  to emphasize that something in toxicology, when I started reading, I start to appreciate how  different they think, compared to how we think as medicinal chemists in the screening campaign.  They actually not concerned about false positive, they are concerned about false negative, which  is understandable, because you assess DILI many times super late stage, you don't want to throw  out good compounds. And you will explain if, if someone asked me, why would they not worry  about false positive? That is because you have so many different assays, right to really  reassess that risk, you have so many chance to re-correct even if one assay is giving you  a wrong false positive result. So minimizing false negative has always been the go to for  any DILI mechanistic assay, and they I think they happily trade with the false positive on the  equation. So hopefully that answer your question, if you're interested, we have this section in the  book that talks about how this is being assessed. Interesting for medicinal chemists, you don't  want to give up your compound, because it's the final minute when you can't deny  any further right. So basically, yeah, Been there done that. Another question  here is associated with a PK. So can you actually minimize DILI risk by  BID dosing? So basically, you know, like lowering the dose, but then  increasing the frequency of dosing, or like without giving, giving, like the  medicine without food, with or without food. I actually haven't seen the food effects, or  the dosing. I'm not sure if people actually have played around with that. But overall, I've seen  that it seems that the overall cumulative effect, I don't think you can play that trick,  unfortunately. Because again, it's it is a mechanism that takes time to develop. So I don't  think playing around with a dosing schedule will actually help that much. You eventually you  really have to solve the problem at its core, is your compound just has to be so much  better PK wise and reduce the dose. Yeah, makes sense, then I think you're  going to like this question. Since you're a fan of cell painting.  Do you think cell painting and high content imaging has a role in  predicting and understanding DILI? That's very interesting, actually. I'm gonna be  excited to see, I think it does have the chance to to move in that direction. For sure.  I think so far, it's mainly been used for like in the in the screening stage, right.  But I will not be surprised in the future, especially with all AI and machine learning  can help us understanding so much what these methodical change will be. Actually that's,  that's, that's a very good point. Because a lot of time, you cannot really see through  histopathology DILI manifestation actually. Even in animal model, right. So I think  that's a pretty good potential there. So nice. Now, well look forward to  seeing that the data. Yeah, I know, I know. Another property that people  are wondering about, as, you know, besides lipophilicity, et cetera, et cetera, is  fraction SP3, is there any, you know, correlation that has been seen with fraction SP3 and DILI  risk? Or, you know, nothing that comes to? Yeah, there's some mild correlation, right?  Again, there's really never been any strong correlation with DILI. But I think it does  trend in the positive direction. So if you increase your fraction SP3, you also decrease  your DILI risk to a certain extent. And that's been the case for the LPA-1 antagonists  as well, where they increase fraction SP3, so it's definitely helping the  same way with lipophilicity, probably a little bit less, but if if you  can do so it's probably helpful. Yeah, I mean, you showed this great slide with  a structural diversity of the mitotox, right? And there was really nothing  to be three like everything was gonna be all over the place, right? Yeah. Yeah. Yeah, it makes total sense. I guess we have one more  minute. We have a lot more questions. So Anh like really great, great presentation. So maybe,  maybe we'll do one more to wrap it up. And then the other questions, we might share answers  online later, but really, it speaks to what, how great this presentation was. And so thank  you so much. One more last question. I'll ask is, could route of administration make a  big difference in DILI tox profile? It's an interesting question. Actually. All of  the DILI positive compounds I come across so far orally dosed, actually. Yeah, I actually I have  to get back to this question. I don't know the answer to this question on the top of my head,  but I'll look into it. I'll get back to them. Awesome. Yeah, no worries about this. Yeah. And  so thank you so much. We're in the top of the hour now, it was really, really amazing. And thank you  also for being here to answering all the audience questions. I mean, you know, you really rarely  get the opportunity to talk to a medicinal chemist expert in DILI, and you're one of the few ones out  there. So we really appreciate you and you being here giving that presentation and answering all  the audience questions. Thank you so much, Anh. thank you so much for hosting. And thank you  everyone for joining today. Yeah, thanks, everyone for joining and, you know, you'll catch  us at the next event. Thank you so much, everyone. We look forward to seeing you. Bye bye.