Next-Generation Cancer Therapeutics: Molecular Glues for Intractable Targets - James Bradner

Well, Jill, thank you so much for the invitation to return, albeit virtually, to talk with UMass and Dana-Farber's P54 program. I'm going to talk today about prototype cancer medicines and chemical tools to dissect the complex biology of gene activation and gene control. But truthfully, the fondest memories are working with all the students that came through on this remarkable program project grant that has really had a big impact for sure among those who participated, but an even more amplified impact on those of us on the faculty who, through this program, got to see just how having a greater diversity of background, of training, of perspective can really make a biomedical research project all the richer, all the more nuanced, and insightful. Full disclosure, there will be some small molecule structures, but I think it also highlights the way in which we need more ideas around a table, diversity in all dimensions, because the monosynaptic way that our field has approached the science of therapeutics for so many years—we have antibodies: what can I bind and remove from the serum? We have small molecules: where can I bind and inhibit?—really isn't going to get us where we need to go for most of the most severe diseases. We need a whole reconsideration of what it means to be a therapeutic in order ultimately to deliver on the expectations that society rightly has for us. I'll talk about the off-label use of medicines, if only to show you the myriad of possibilities that exist for a new therapeutic beyond the narrow and focused spectrum of the development path most medicines take. Well, I thought because it's a cancer symposium, to really begin with the cancer patient. This gentleman exemplifies the burden of cancer that is today still so impossible. Despite incredible insights, even atomic resolution into the cause of these diseases, and the unprecedented therapeutic modalities that we have before us, still at the bedside cancer remains an almost intractable problem. So here's a 53-year-old man with shortness of breath. This is a real patient. On both sides of the chest, these shadows—for those that aren't used to looking at CAT scans of the chest—are all tumors. Normally this would be dark, filled with healthy lung tissue. Here's a tumor on the contralateral side of the chest, as well as large lesions, cancer deposits in the liver and other distal sites. This gentleman's tumor is not surgically removable, can't be radiated owing to how extensive it is, and so systemic therapies are then considered. This gentleman receives really everything that modern medicine has to offer: chemotherapy, immune checkpoint therapy, combination chemotherapy and immune checkpoint therapy, and then a host of targeted therapies that are being investigated for their use. This gentleman received three months of G12C therapy with an excellent medicine, investigational today from Amgen, and then subsequently got what's called a SHP2 inhibitor—I'll talk about that more in a moment—combined with a MEK inhibitor, and then CDK4/6 inhibition. This fellow received a lot of therapy, and unfortunately still his disease persisted. So what will we do to treat this man? What types of medicines would need to be imagined, and then invented, and then investigated to have ready for him the medicine that patients like him expect from us in science? I can tell you, having devoted my career to thinking ultimately to address the profound burden of metastatic solid tumors, NIBR is one such environment that is fully dedicated to this science of therapeutics. These Novartis Institutes for Biomedical Research—just to introduce so you get a sense of where the project I'm about to present hails from and why it's a unique environment for that type of research, but it's also a shameless plug for those that might want to explore a job in biopharma—is 6,000 drug hunters. We have actually 270 drug discovery projects. We work in eight disease areas, and the portfolio, because we have been performing drug discovery now for almost 20 years in this way, is all focused on the concept of a drug, a target, its study, its validation, all the way through to the proof of concept of an investigational drug in a Phase 2 clinical trial. Then we hand off to the development organization within our Swiss pharmaceutical intergalactic giant Novartis AG in Switzerland, and our colleagues there perform the Phase 2 and Phase 3 studies needed to get a drug approved. So where I work is more focused on science and pathways, targets, invention, and early clinical investigation. Well, these are the disease areas where we work. You might know a medicine that Novartis made a few years back called imatinib or Gleevec. This was a medicine that targets the ABL tyrosine kinase that is mutated by a translocation in chronic myeloid leukemia, and it's a remarkable discovery that has really changed the lives of patients with chronic myeloid leukemia. It also opened our minds as drug hunters to the idea that you could bind and inhibit a kinase selectively by binding into this ADP binding site as shown here in blue. This little molecular key fits into a little macromolecular keyhole, displacing ADP and blocking the catalytic activity of the kinase. Of the medicines we want to make, definitive medicines for life-threatening diseases. When I was a stem cell transplant doctor starting out, about half the wards were full of CML patients, but with medicines like imatinib, most of these patients are home now, taking a once-a-day oral therapy. I'm no longer distracted by their disease. So we're very serious about small molecules in cancer, and the story I'll tell you today is a discovery chemistry story. But there are four other modalities that we use and deploy for cancer therapeutic discovery. One you might know about are called CAR T-cell therapies. These are stem progenitor cell therapies. T cells isolated from a patient, infected with a virus that expresses a gene that normally... the T cells are activated to both kill and to divide, and this is then like a living therapeutic inside of your body. The first such medicine is now approved from Novartis and the University of Pennsylvania collaborating together, called Kymriah, and we have a host of other medicines of this type for other blood cancers. The third platform that we use for cancer therapy is called radioligand therapy. These are little binders that we make either as peptides, proteins, natural products, or even small molecules, and we append in a little chemical cage radioactive lutetium or other radionuclides. Upon injection, these molecules swim through the... antibody drug conjugates, more on that another time. I will tell you a story of chemistry today which I delight to share because I myself am a cancer doctor but also a chemical biologist. The chemical biology mindset at NIBR is the idea that there need to be new types of therapeutics to address some of the most grand, historic challenges in cancer therapy today. This is a serious group of people. It might be 800 medicinal chemists and maybe 600 chemical biologists, but you can't beat cancer with scale, although we have an incredible human resource and an era-appropriate investment in drug hunting. We try to violate the dogma of science to make new types of medicines, definitive medicines for life-threatening diseases. We have no cosmetics branch, regrettably. Amidst this pandemic, we have no vaccines unit, which is totally focused on new types of medicines to reach this highest hanging fruit. So our story today begins with the oncogene K-RAS that I mentioned that gentleman had mutated in his tumor. This is a picture of K-RAS, shown here with a ribbon and a surface translucent overlay. It looks to have any number of pockets. There's a pocket right here, there's a pocket over here, there's a pocket down here, and you might think that a conventional drug molecule might bind and inhibit K-RAS. This is the way... complementarity, and then we match up the positive and the negative charges on the molecule with those on the protein. Remember, opposites attract. Then ultimately we have also electronic parity between the molecule and its target. Here's that molecule Gleevec or imatinib in this tiniest active site, and you can see this solubilizing piperazine sticking out of the pocket and the way these heterocycles kind of thread the needle and fit inside, shown sliced open as a cross-section. This is the work of the medicinal chemist to be a molecular locksmith, making small molecules that fit in small protein pockets to inhibit or to activate. To be clear, you can still make a lot of really important... yet, and there's a lot of proteins that haven't been drugged yet. There are great reviews on that subject. When I was a professor working on this P54, we were interested to make the first drugs for what I'll call bromodomain proteins, as shown in the lower left on this slide. The bromodomain protein here is an antiparallel bundle of four alpha helices that have these two interhelical loops in a pocket. Into that pocket we positioned thienodiazepines and many other chemical features that might bind. When we made the first inhibitors of the bromodomain BRD4, it sort of opened up the mindset of our field that, hey, maybe drug bromodomains are druggable, and the 50 to 60 other bromodomains... protein-protein interaction. So people think, well, maybe that's not druggable. It's not like an enzyme; it doesn't have a catalytic site. Unfortunately, that's still a way a lot of scientists think. But scientists at NIBR saw a little donut hole here in the middle, in this great region with our Dunkin' Donuts—this is quite thematic—and made first small molecules that bind this little hole in the middle of the WD repeat domain of a protein called EED, which is an allosteric modulator of the Polycomb Repressive Complex 2. If the bromodomains turn genes on, this one turns genes off. So we have these two molecules now that can switch cancer genes on or switch cancer genes off, or switch developmental genes on. And so all of our rules about drug discovery must be sort of thrown out the window, and we need a new way of thinking to go after undruggable targets. In cancer, they're everywhere. If all we did were to look at those genes that are mutated and activated to cause cancer—and here's a list of blood cancers, as I'm a blood cancers doctor—there's a long list on the right of genes that are altered to activate cancer, and most of these don't have a direct-acting therapeutic. Some of them, like MYC, like MIB, like ID3, like MATH, it's going to be very hard to drug them because they don't have a pocket. Some of them, like p53... see all the data. This is now published in 2017, but we had functionally mapped all of the Achilles' heels of two or three hundred different cancer cell lines—I think ultimately 500 cancer cell lines. We did it initially by hairpin RNAs, but then subsequently did it with CRISPR/Cas9. If you look here in the middle, every one of these little bars is a different cancer cell line, color-coded by the tissue it's from, and the vertical axis is whether they're alive up here or dead deflected down. You can see that ESR1, that's the estrogen receptor, so breast cancer cell lines and ovarian cancer cell lines will die when that gene is removed. Here's another one called MITF, and in green these are melanoma cell lines... development and growth, so it's no surprise that that would turn up. Here's the challenge: unlike the estrogen receptor, MITF doesn't bind estrogen or some other molecule. It has no pocket. When we looked at all of the genes that showed these types of asymmetries, suggesting possibly a therapeutic index with selective modulation of cancer cell line viability, all of the kinases had already been drugged, and so there wasn't much in the way of low-hanging fruit for these cell-autonomous cancer drivers. In fact, when you look at genes like MITF and the genes I studied as a professor—gene control or transcription factors—only two of them had been drugged: the estrogen receptor and the androgen receptor. We need to change the way we think about drug discovery to do this. The same is true for K-RAS. K-RAS has a pocket, but it's well occupied by GDP. You might say, let's just pry that out of there with a small molecule, like ADP in a kinase, but the GDP binds way too tightly, and at the concentrations of GDP that are so high in a cell, it would take a pretty special molecule to remove that ligand. So we needed a new way of thinking for K-RAS. The second concept I'm going to talk about is called synthetic lethality. If undruggables is the first concept, concept two is synthetic lethality or context-specific dependency. This is where a protein target is hard to drug, and so you go after its nearest... a total reconsideration of how small molecules work, called molecular glues. I don't know if this term is going to take on or not; this is what we've been calling it in the laboratories at work, so don't feel obliged to use that terminology. So what's a molecular glue? A molecular glue, as much as it sounds like, is a low molecular weight small molecule that glues two proteins together. It could be an intermolecular glue, like a protein that glues a target to the garbage disposal system for degradation. We had in my old lab at the time of the P54 invented the chemical solution for targeted protein degradation... restricts a single protein, a single polypeptide, by engaging different domains of the protein and gluing them together. A glue, biophysically defined—for the scientists on the line, we define it here at least in the parlance of how we use the term—as engendering non-saturable interactions. This is different from PROTACs or bifunctional molecules. Now, of course, we didn't invent this; nature did. There are some really brilliant examples from the natural world of small molecules that will bring two proteins that normally in human or eukaryotic biology don't bind each other but cause chemically induced proximity. Bring a cyclophilin in, and in the case of FK506, bring the protein FKBP12 in. This is sort of a short circuit for immune cell activation, which is why both of these medicines are used today so prevalently in solid organ transplantation and in bone marrow transplantation. If you're interested to go deeper into molecular glues, my postdoctoral mentor Stuart Schreiber has written a recent review earlier this year about molecular glues that goes all the way back to these earliest moments of characterizing the mechanism of action of these immunosuppressive macrocyclic natural products. At Novartis, we've been working on glues for a while, all the way back to... as interacting with two proteins, gluing FKBP12 and mTOR FRB domain together. All right, well, let's get back to intramolecular glues and the protein SHP2. So what is SHP2? It's also called PTPN11. It is a non-receptor, so cytosolic, tyrosine phosphatase. If kinases put phosphate groups on tyrosine, phosphatases remove them. There's really nothing that makes a kinase more interesting than a phosphatase, just that a kinase is druggable thanks to Zimmermann and friends. Now interestingly, with SHP2, there's a strong rationale in human genetics to go after it. People born with activating mutations of SHP2... and this would suggest maybe that for a fully developed human who has a cancer, maybe SHP2 inhibition could be tolerated, but this would remain to be seen. There's a huge field of study out there, and so this science builds on a very strong basic science foundation. It is known in the literature that SHP2 is required for signaling from receptor tyrosine kinases, and many cancers are driven by activating mutations in receptor tyrosine kinases such as EGFR, HER2, or HER2, ALK, and others. We found back in that big project I told you about that when you knock down SHP2, only a small number of cancer cell lines will die. It turns out SHP2 might be important for PD-1 immuno-oncology signaling, which would be pretty interesting. Now, not all literature validates when you really commit to repeating it in a professional environment. You know, trying to justify a big investment to make a medicine, but the SHP2 research did. We found within our laboratories that in these agar plates, these 2D growth plates, that knocking down SHP2 in breast, esophageal, gastric, lung, other cancers—all driven by mutations and known to be dependent upon receptor tyrosine kinases—that that growth was arrested. This worked in vivo with hairpin knockdown, and then a really important experiment was done by our cancer... but when you add back a version of SHP2 that has this cysteine 459 substituted with a serine—that cysteine is essential for its catalytic function—that these cells don't grow back. This means that the phosphatase function itself of SHP2 is needed for these cells to grow. So let's make a phosphatase inhibitor. Well, it's easier said than done. After the first tyrosine kinase inhibitors were made by Jurgen and colleagues in our Basel site, there's probably 50 or 60 tyrosine kinase inhibitors now in various stages of development. I'm sure it's over a hundred. There's actually no examples of a tyrosine phosphatase inhibitor, and it's not for lack of trying. We and others... we had crystal structures, we had 200 picomolar inhibitors, but the problem is that the active site of the phosphatase binds phosphate, which is very polar. So the potent molecules that we invented to be the keys for that keyhole were also very polar. Polar molecules don't get into cells very well, and polar molecules are not frequently orally bioavailable and long-lasting in blood. So interestingly, it's easy to make a potent biochemical inhibitor but hard to make a potent biological inhibitor. It would take a whole new way of thinking, and I'm going to share that way of thinking in this story of an intramolecular glue, the first-ever inhibitor of a protein tyrosine phosphatase. So... SH2 domains bind phosphotyrosine peptides. You can see the way that this IRS-1 substrate comes along, and it's like spring-loaded. As soon as it binds the SH2 domains, they release themselves off of this tyrosine phosphatase domain, and now the enzyme is active, shown here biochemically as we add this peptide, the activity of the phosphatase increases. So we did two high-throughput screens. First, we looked at the tyrosine phosphatase domain itself to look for inhibitors, and then next we would look and ask how many of those inhibitors only work when all the domains are present. We screened 1.5 million molecules... hits because they didn't work in the full-length assay. We ultimately stumbled upon a few classes of very special molecules that bind at this middle interface of all three domains—intramolecular glues. We thought, well, maybe these are molecules that conformationally restrict this big protein, ultimately causing it to be inhibited. The crystal structure—this is a model, and I'll tell you when we get to the crystal structure—right now the yellow small molecule prototype drug binds into this pore in the middle of the protein. When it binds into this pore, it brings the white and the green regulatory domains down, and this little... that creates a lattice in the dotted lines of electrostatic hydrogen bond interactions that connect to each of the three domains of the phosphatase. More on that in a minute. Well, I know that at least the work we did with Dr. Zhang and this P54 was all about small molecule lead optimization, so I know a handful of you are really interested in how do you take a hit from a screen and turn it into what might be real drugs someday. It's a real science actually, and I've surely learned more about that at NIBR than I came in the door with. It all starts with these assay positives, these molecules that hit in the primary screen. Shown here are five different chemical classes from which we had some starting point. Now you can see these are all micromolar... these two have this dichloro-aryl feature shown here and here, and most of these molecules have a basic amine over here on the east coast. We build what's called a pharmacophore model. The pharmacophore model is helpful because it gives you some immediate hypotheses to test, like let's get rid of these amines and see how these molecules do. Do they still work, or is that basic amine, that nitrogen, really important? So there was this basic feature, there was this more typically aliphatic spacer in blue, and then there was in green something a little bit greasy here on the end. All right, well then we solved high-resolution crystal structures of these molecules, and indeed the pharmacophore model proved correct... pyrimidine group as shown right here as a lead series because it was 12 micromolar. It didn't have a lot of rotational bonds, it looked nice and rigid for potentially a rigid pocket, and it was nice and small that we could work with. So we started off with this eastern piperazine here that was perceived not to be that pivotal and was surrounded by polar residues. So we worked to change the display vector of this primary amine, and in the process of bringing it extracyclic and stabilizing with this gem-dimethyl group, we picked up an order of magnitude of activity. Whoa. Then we started to explore this dichloro feature... central heterocycle, and you can see the way that this ring nitrogen has been spun out just one click to the left. This subtle movement of that ring nitrogen stabilized an interaction with arginine 111, and so we ended up learning that this amino pyrimidine was really preferable to the aminopyrazine, and the ligand efficiency was quite dramatically improved. Okay. So will this inhibit SHP1? Now this is really important because SHP2 is important for therapeutic use, but SHP1 could only lead to toxicity. In fact, there was no biochemical activity against SHP1, which was curious because the proteins are so similar, but... straight out, in fact, this proved to be a really potent and really selective chemical probe of SHP2. We characterize it in all sorts of ways: differential scanning fluorimetry for biochemical direct binding, surface plasmon resonance to get the on and off kinetics, isothermal titration calorimetry to confirm that it would be a stoichiometric one-to-one relationship. It all looked good. So we published it and we put the molecule out there to try to excite the scientific community to use this SHP-099, the first of its kind tyrosine phosphatase inhibitor, as a chemical tool and maybe tell us what would be the best use of such a molecule. But while we had a pretty good idea about how to use it in solid tumors, as I... toxicity, and it also causes phospholipidosis owing to vacuolization in hepatocytes. So we just couldn't develop this molecule. So what do you do? Well, we think, do we have to go back to the drawing board, or do we keep tinkering with this in medicinal chemistry? But there might be something about the relationship of these two aryl rings that would lead to the phototoxicity.

Because a lot of phototoxicity is transmitted by aromatic features. So we went back and ultimately settled on these thioethers, these diurothioethers, and made a bunch of derivatives of these molecules. Ultimately we found that when we separated this dichloro aromatic feature with this thioether and then installed the protein target, they didn't have any phototoxicity. Now we just needed to make these a little bit more potent. I'm going to avoid telling you too much medicinal chemistry. We ultimately were able to substitute out the dichloro feature for this heterocyclic feature that had a trifluoromethyl group, and that made the molecules quite a bit more soluble and drug-like. We swapped out that piperazine and substituted in a little bit more solubilizing, but there was nothing really wrong with that cyclo feature. Here again is that primary amine sticking up. Now this molecule looked great, it had no phototoxicity, but now we had a new problem: cardiotoxicity. So we had to work out this so-called inhibition of the hERG channel. We modified the molecule to avoid inhibition of this cardiac channel and find this Goldilocks molecule that was potent, selective, kills cancer cells, and doesn't harm the heart. That only took five minutes to explain, but it was years of medicinal chemistry over these last five years. Ultimately we arrived at TNO155, and this is the actual structure of our drug molecule for SHP2. Now we're getting excited to put it into a human, and you want to know a little bit about what that dose might be. So we did a lot of modeling in animal models to look at the exposure. The characteristic response in cells is to downregulate a protein called DUSP6. Here is the impact on DUSP6 at a concentration level. Ultimately we put these into tumor models, and there you can see cancers are regressing in response to TNO155 in animals.

How to bring this medicine to the clinic? We have a couple of ideas where this molecule could work on its own. You might know that biotech and pharma companies always look to see what the drug could do on its own, but we have this mantra at Novartis: medicines first for those who need them most. It struck us that maybe the best way first was to look at a blueprint for how to make a KRAS inhibitor for G12C. As we've known in our labs for a long time, there's a type of chemistry called electrophilic chemistry that can bind to cysteines because cysteines are a nucleophile and opposites attract. What Kevan Shokat brilliantly showed is that small molecules that have an electrophile combine with cysteine 12 of G12C in a manner that doesn't require prying out GDP. This was a revelation, a fresh new idea that might ultimately lead to drug molecules. With that information published, a lot of companies, ours included, made molecules that drug G12C. One such molecule is shown here. You can see the GDP right here, and then you can see the way in which it ligands onto the cysteine over here. This molecule looks pretty exciting in its ability to arrest the proliferation of G12C mutant cancers here with no effect on G12D or G12V cancers, so it's a mutant-specific KRAS inhibitor. We studied our own G12C inhibitor as well as those from other companies, always with our SHP2 inhibitor, and every time we observed true synergy. This is a color-coded plot, a heat map, of each drug given in combination: a G12C compound here, a chemical probe, and TNO155 here. Together these two molecules kill cancer cells. Our molecule will also work with tyrosine kinase inhibitors. Today we have open clinical trials studying TNO155 in combination with a number of tyrosine kinase inhibitors and with KRAS G12C inhibitors. It's early days and we hope to share the results of these studies over the coming few months.

Let me tell you about this 53-year-old man as just an example so you remember that he had all of that prior therapy: chemotherapy, immune therapy, and targeted therapy. He received Amgen's very excellent KRAS G12C inhibitor. He received a SHP2 inhibitor from Revolution Medicines that made a SHP2 inhibitor just after we did. He was out of options. At the very first dose level, he had a very dramatic response. We're just not used to seeing this in solid tumors, especially solid lung cancers. We've seen it with mutant EGFR inhibition, sometimes with mutant ALK inhibition, but we're not used to seeing this with KRAS mutations. It's early days and this study is ongoing, but we're so hopeful that patients like this man might benefit and maybe might even receive this therapy a little earlier in the course of their disease so perhaps they don't need to get so sick so fast. In my heart of hearts, I do not believe G12C inhibition will be curative for these tumors. Data presented this weekend at the AACR annual meeting suggests we need to treat these patients at earlier lines of disease. But I'm so inspired by this story because by reconsidering how to drug SHP2 with a new kind of molecule, an intramolecular glue, and by reconsidering how to drug KRAS G12C not to displace the substrate but to bind into it by engaging a cysteine nearby, these two new revolutionary ideas in molecular pharmacology—glues and covalent chemoproteomics—together could really help a gentleman like the one I shared with you a moment ago. We're just going crazy with these ideas at Novartis right now. That's not to say we have it all figured out. This is highly collaborative work with scientists such as at the Dana-Farber Cancer Institute. We're working on drugging undruggable proteins, and I didn't even tell you about the work we've undertaken to drug RNA in a sequence-specific way for severe neurologic diseases. I think we're pretty close to the end of this idea of undruggable.

In the closing few minutes before the Q&A, I was invited by Jill to share some reflections on the state of diversity and inclusivity in science. Surely it must be better than it was 20 or 30 years ago. I've only been practicing science for about 20 years, but it's still fair to say we have a really, really long way to go. I find in the natural and physical sciences the data, when it's sampled, to be very troublesome. Regarding gender diversity, statistics from our community show that a 50/50 workforce at the time of college science concentration selection ultimately becomes 24% female in tenured faculty in the sciences and 18% female on boards of directors in biotech. The Deerfield study recently reported that 48.5% of all biotech companies they surveyed have no female on the board of directors. Other types of diversity are equally concerning when we and others look at it in our community. We really have to do better. Society expects it. I'm the first here to tell you there's no altruism involved. It surely is a good thing to work on and it feels good to work on it, but the science demands it. I have seen the way groupthink around some of the principles of medicinal chemistry keeps scientists stuck in their old rabbit hole. I was trained to use a hammer and I'm going to hammer things. We need to be in a much more agile, inclusive, challenging, speak-up environment than poorly diverse communities often allow.

What are we doing about it at Novartis? We start by making it a major core focus of our strategic plan, a pillar five: invest in our people. I'm proud to say that across all of Novartis now—106,000 global associates—it's 50/50 men and women. We're almost to 50/50 in management, which is a stated goal of our executive committee. 150 nationalities are represented, and I believe four of 13 of our directors on the Novartis board of directors are women. At the executive committee, led by our CEO Vas Narasimhan, who is quite passionate about this subject in more than just gender dimensions, we have stated goals that we as an organization and each of our units are measured against. There are unintended forces at play that occur after recruitment. It means having a diverse final slate of candidates before extending an offer. It sometimes means working or networking with externals to be sure that a candidate list is adequately diverse. I did this just two weeks ago when we announced that our head of Novartis Oncology, Jeff Engelman, will be moving to the biotech sector. I made a list of all of my dream candidates to lead Novartis Oncology. To my taste, the list that I could produce even as a cancer scientist was inadequately diverse, so I scheduled three meetings with leaders in the US and in Europe to try to do better. Culture really matters. It's one thing to have diverse people around a table, a diverse mindset; it's another thing to have a culture that invites everybody to contribute. In a hierarchical work environment like companies can be, it's really important for leaders to be selected who know how to reduce this to practice, to really get the quietest voices in the room to contribute to a weighty scientific matter. I'm sure all of you have experienced this, but in my time in leadership these last five years, I can say there's no correlation between the loudest voice and the best idea.

Here's some data I hope it's okay to share. This is from a presentation our CEO gave on ESG dimensions and other dimensions like environmental dimensions. We're a big company that's $220 billion, and we want to be sure that everybody inclined to buy the stock can. We listen to these so-called ESG investors and share with them our own personal ambitions and leadership ambitions for the company. We get feedback and we turn in report cards as to how we're doing. I believe that if the academy were on the hook to deliver diversity, equity, and inclusivity, to close gender pay gaps, if they were as publicly on the hook as we are, we would make much better strides in academia and biotech than we have. But we also feel like we have an obligation to share a discipline that you really can't train in but for a few academic institutions, like medicinal chemistry at Ohio State University. There aren't a lot of great programs in Boston, maybe Northeastern. We open our labs to trainees wanting to be a part of a rising tide that can improve the ecosystem. It's not lost on us that when people leave Novartis, they sometimes go to be the CSO of a biotech company. If that's the pool that these venture capitalists are drawing from, we can't do this alone. There are a lot of companies very focused on this, but we have to walk the talk. I can tell you there are deals I did not do last year because we weren't happy with the diversity around the table.

We have these two internships: a summer internship, a 10-week internship that is thoroughly scientific but can focus on other areas of interest like entrepreneurialism and business development. It's an all-access pass to Novartis. At this stage, the summer internship is fully recruited because it's April, but we're going to open applications for 2022 in September. Secondly, we have a post-baccalaureate program for students who would enter graduate school. I myself did a period of research study after college and before I went back to medical school, and it was some of the best time in science I've ever had. Not just scientific mentorship, but these are really well-organized programs where you meet a lot of scientists with different career trajectories and network. This isn't altruistic either. We believe that if we really invest in some of these talents, maybe a few of them come back and work with us as professional scientists someday. As shown on this slide, there's a real focus to these programs for scientists that arise from underrepresented backgrounds. It's about making medicines. For every reason, having a big conference table with all the right minds working in a culture of total openness, bravery, and curiosity—I think this is the secret sauce of Novartis and what has made this for 20 years one of the most important and productive drug-hunting engines in the world. Thanks again. I'll gladly take any questions that you have.

Absolutely. I just want to be sure that we can see each other. Am I out of screen sharing mode?

Thanks for the question. Just as for our workforce, we measure and have targets and expectations for diversity not just in the individuals that we are able to recruit, but also in our peer group of pharmaceutical companies in clinical trial diversity. Novartis is really following the lead of our global development organization that does the much larger and more global Phase 2 and Phase 3 clinical trials, but we do measure for early clinical study to ensure that we are executing these studies in the most diverse way possible. It is important to do this as a first principle, but I can tell you scientifically and medically it's essential. We've seen many times molecules that would move through clinical study in one region and then not translate to an Asian population. We often do bridging studies in Japan to ensure that the pharmacology will read through to an East Asian population. Our society of drug discovery and the regulatory agencies are on it and held to account, but the pooled data would suggest that as in the early moments of the COVID-19 vaccination efforts, the studies can be inadequately diverse the way they're operationalized. John Psay, who's our Chief Medical Officer, is my partner that we hand off medicines to for Phase 3 trials. He has a leader, Audrey Swan or Hassan, who measures our diversity on a per-trial basis.

Our organization under Vas Narasimhan, our new CEO who started three or four years ago, has undergone a real transformation, a cultural transformation. I hope that doesn't sound silly. It's very serious to change the culture of a long-standing global company rooted in Basel, Switzerland, which has some very traditional and actually wonderfully traditional behaviors. It has been a heavy lift and it's ongoing. This has been a heavily invested, heavily communicated, thoroughly considered, and I think very inclusive process. Our efforts in diversity and inclusion are really cornerstones of that. We have a concept called 'unboss.' What does 'unboss' mean? We've gone very deep with that in our associates. I think that the D&I work streams and the equity work streams have dovetailed with that and have solicited and collected thoughts from the global organization. With Novartis, the 6,000 scientists that we have, we have done the same and we're even more organized. In the current strategic plan we're writing this year, Anastasia, I don't know if Dr. Anastasia is on the phone right now, is leading a very ambitious effort to collect exactly that information beyond the dimensions of gender, which for some legal restrictions and union expectations we're not able to measure everything in every country where we operate. My chief of staff, my executive administrator, and I are the strategy team writing the strategy. I think we are well connected to the organization that I hear from our associates in the labs all around the world. Yes, systematic, and the doors of communication are open. We hope coming out of our strategic planning process in two months to have some really tightly and well-articulated and also measurable targets in these dimensions as we've had for the last five years.

Regarding training, maybe those of you who are really expert in this can even hear in the way that I'm speaking of it that I don't have any training in this. I had a half-hour meeting with Dr. Goyo from the colorectal cancer group at Dana-Farber, and I learned so much in that half hour. I'm still in the midst of my own download on this. Thankfully, I'm supported by our People & Organization, what might be called HR leaders, like Lisa Naylor, who have studied this academically and practiced it professionally their whole career.

Thanks, Joe. I'm not ever distracted by these academic versus industry comparisons. The big eye-opener is just how similar they are. But I do think there's something to learn from the way professional organizations approach diversity, inclusivity, and equity. If you just look at the numbers at our beloved Dana-Farber or at our own Friedrich Miescher Institute, it's behind what I know that we can deliver. The luxury of no longer being a scholar is that when you're a scholar, you need to have the most authoritative opinion in the room. Now I just have to find the best people.