 Good morning. We'll get started. Thanks everyone for making it through the snow. It's my deep joy and privilege to introduce our speaker today who is both my mentor and friend for now 17 years. Anthony Adamus is a native of Illinois. He went to college at University of Illinois at Chicago in Urbana and then medical school at University of Chicago. His ophthalmology training was at University of Michigan, and then he did his cornea fellowship in the true spirit of the time with Dr. Klaus Stolman, who was probably the father of modern cornea. I met Tony when he was residency director in Boston in 1994. I still have my rejection letter from my application to residency at Mass Eye and Ear, and in the case of adolescent persistence I did not let the matter drop, and Tony was open-minded enough to reconsider my application, and somehow I managed to get in. It was a very fortuitous time. Boston at the time was home to Dr. Judah Folkman, who was the father of angiogenesis research from the 1970s, and I had the great opportunity to work with through Tony, key members of Judah Folkman's lab. Dr. Folkman always had two maxims that to this day carry forward. One, that angiogenesis owes a tremendous amount to the cornea of all things, in which both Tony and I trained, and two, from a research perspective, always wonder when you look at an interesting finding, if it's true, what does it mean, and Tony has exemplified that in his research. He has asked probing questions with great insight and patience. Without taking too much more time, Tony's great contributions to the field of ophthalmology include revolutionizing a field distant from his cornea training, that is, macular degeneration. He founded ITEC, which was the first company that launched an anti-vegetary antibody for any application, anti-vegetary aptamer for any application, and that was specifically in macular degeneration, and it was at that point that macular degeneration started shifting from the depressing entity that it was when I was a resident to the more hopeful entity that it is today. That was about eight years ago that ITEC was launched, and that paved the way for a vast and lucentus in future years, which have then taken AMD management forward. As a clinician scientist, Tony has made huge differences in diabetic retinopathy, corneal dystrophies, macular degeneration, of course. As an entrepreneur, he started ITEC, he started geridia ophthalmics, and is now vice president of ophthalmology at Genentech, the first vice president of ophthalmology at Genentech. With that, I'd like to invite Tony to share our, share what's new in AMD. Thank you very much for coming. Thank you, Bala, and it's a great pleasure to be here. I've always wanted to come to the Moran Eye Center. People told me how spectacular this place was, and I got my first glimpse today. It really is an impressive facility, an impressive faculty. I'll tell you one Bala story. So I ran the residency at Mass Eye and Ear, and we had Bala apply. We turned him down. He was persistent, and my former mentor, Judah Folkman, used to say there's a fine line between persistence and obstinance. He didn't cross it, but we said, okay, we'll take a look. The reason why we passed on his application originally was he was 17 years old. He's the youngest medical school graduate ever. He was the youngest ophthalmology applicant ever. So we invited him in against our better instincts, and I didn't tell the other interviewers that day. We had a dozen people come in. Which one was a 17-year-old? I just wanted them to guess. At the end of the day, we would guess. And only a minority could figure out it was you. He was so, he was so poised. He was so articulate that none of us could figure out that he was the, you know, the teenager in the room. And he's going on to do great things. So it's a great honor to be here with you again. And I realized just a moment ago, I've known you half your life, your adult life. Okay. All right. So I want to talk about Neovascular AMD. Greg Higginman tells me we shouldn't call it WED-AMD anymore, so I won't. And where we are today, and where can we go in the future to get better outcomes for patients? And first at my disclosures, I work at Genentech. So my paycheck comes from there. Genentech's part of Roche. Roche bought Genentech, so I own shares in Roche. I'm going to discuss studies, phase three studies and others that we've done, but we've had IRB approval for those. So VEGF inhibition in AMD started in 19, or 2004. 2004 is when Macugin was approved. That was the first one that I had the privilege of taking forward and getting FDA approved. But really, the transformational moment came in 2006 when Lucenus was approved. And at the same time, Avastin was being used. So AMD went from a surgical disease where you would treat with a laser or photosensitizing dyes to pharmacologically treated disease, where we started treating with drugs. At the same time, VEGF inhibition began in oncology. So in 2004, Avastin was first approved for colon cancer. So I'm showing you the pivotal results, the phase three results here. This is Marina, this is for AMD, and this is for colon cancer. Now let's look at colon cancer first. There was a median 4.4 month progression-free survival. So this is how long it took before the tumor started growing again. This was an overall survival, which is the holy grail. This is just progression-free survival. But even though it was only four months at the time, this was deemed the biggest advance in 30 years in colon cancer because the only other drugs used there were 5FU and other toxic chemotherapeutics. But the effect wears off. In this Kaplan-Meier graph, you can see at 20 months, almost all of these patients have progressed and unfortunately go on the dock. Because this was not first-line therapy, this was actually third-line therapy. Contrast that with macular degeneration. In macular degeneration, we saw an early and maintained increase in visual acuity. These are the two treatment arms, the two doses that we treated. And this is the last natural history arm you'll ever see for wet AMD, for neovascular AMD. So 10 years ago, this is what we dealt with. Patients would come in. The patients who had classic lesions on angiography would get photodynamic therapy. Everybody else got nothing if it was sub-fogial because nobody was lasering back then when the lesion was sub-fogial. So this really was a game-changer. When the data were announced in the summer of 2005 in Montreal, Gene DeWon, who many of you know, who was president of the American Society of Retinal Specialists stood up and he said, this is penicillin. It's not penicillin. But it really did change the way we treat AMD. So compare and contrast here. What's amazing is how durable the response is here. At 24 months, the vision is still maintained. At 24 months, most of the people here are on their way to death. Why is that? Well, the retina, fortunately, is a genetically stable tissue. It is not transformed. We're not treating a cancer that has this evolutionary pressure to mutate around the VEGF blockade. That's my hypothesis, but I think that's the principal reason why in ophthalmology we've gotten lucky with Lucentus and Avastin and Ilea now, all these anti-VEGFs appear to be durable. And so we're very fortunate there. And they seem to work for a relatively broad spectrum of patients. Now, this is a paper that came out in 2010, 2011. Shoshana Coleman's in our Health Outcomes Group. She worked with Neil Bressler, Rohit Varma, Paul Lee. A lot of people that you know, Paul just became chairman at University of Michigan, my alma mater. And they took the data from Marina Anchor and they applied that to the incident population with a new onset, a neovascular AMD. And then they took the results from Marina Anchor and they showed that with monthly injections of Ranibizumab, the incidence of legal blindness in two years is reduced by 72%. And legal blindness here is in the second eye. So the first eye is already legally blind. These are second eyes with wet AMD. They get the drug and you prevent people from going legal and blind. So this is what I mean by it's had a real positive impact, but enough commercial. We're not done. And the results are good, but there's still gigantic gaps for patients where we need to improve outcomes so that they do better. This is from Marina as well, all right, the 2006 paper. 60% of patients who were treated in that trial see 2050 or worse. So these patients are visually impaired. This is best corrected vision. So they can't read the text on the little handout that you have in front of you as a result of even being treated here with Ranibizumab. So we can certainly try to improve on these outcomes. And what I want to talk about the rest of the talk is what strategies can we use going forward? Some of the strategies we're attempting in San Francisco to get better outcomes for patients. So first is we can try to optimize the efficacy of the drug and the dosing. And by that I mean the following. In when the data came out of Marina and Anchor and in subsequent large pivotal trials, there was always a little bit of a difference between the 0.5 milligram dose, which is the commercial dose, and the 0.3 milligram dose. And if you look across trials, it was about 2.2 letters. And so the question in our mind was if we go higher, are we going to get better results? Are we at the top of the dose response curve? Or perhaps we can do better with the higher dose. The other thing we learned in trials that came shortly after Marina and Anchor, like peer where patients were treated with three monthly injections, but then they got quarterly injections. And similarly in Sailor is if you try to space out the injections too far, if you go quarterly, patients will start to gain vision with the monthly injections, but with the quarterly they lose. So is there an optimal dosing paradigm? Something between every month and every three months where we can get better results. And so we designed the Harbor trial. This is a large 1,100 patient randomized controlled trial for groups randomized one to one to one of the commercial dose. And this is what's in the label 0.5 milligrams monthly versus an arm where people got three loading doses, but now they were treated PRN. Something that Phil Rosenfeld with the pronto trial first tried. So patients would come in monthly, you do an OCT, check their vision. If there's evidence assigned, the disease is reactivating, they get a shot. So you sort of tailor the therapy to the patient's disease. And then we went to a higher dose four times higher than the commercial dose two milligrams monthly and did the same thing with the PRN, but this time using the two milligram dose. And what do we find? So these data just came out around the academy. What we found is all four doses and dosing regimens led to rapid increases in visual acuity. Patients start the plateau at four months, but the monthly regimens do a little better than the PRN regimens which are in the dotted lines here. This was a non inferiority trial. And for the biostatisticians, there was a four letter margin of non inferiority. And technically we did not meet our primary endpoint. So the PRN arms were not non inferior to the monthly arms, but you could see here the absolute letter differences are in the order of about one to two letters. So it's pretty close. It's like 80 90% of the way there, but it's not quite what you get with monthly. But if you go with the PRN doses, what was interesting is you need a lot fewer injections, which I'm sure patients appreciate. So you could get away with 6.9 to 7.7 injections versus, you know, 11 for the monthly treatment. Now pharmacodynamically, this was the first trial where we used spectral domain OCT. So you get the super high resolution OCT. All these patients dried out very quickly on OCT. And statistically, there was no difference between the treatment arms. They all dried out the same amount, even though the visual acuity results were a little different. And this is now breaking out the two PRN groups, the 0.5 milligram and the two milligram, by the number of injections that they needed over 12 months. And you see there's a very interesting spread. So the mean and median number of injections is shown here for both doses. It's around 7 to 7 plus. But you have some patients and not a lot who got three injections, 6% here, and they're done. They don't need any more injections for the rest of the year. Then at the other end of the spectrum, you have patients who needed a shot every month. They got 12 injections. And most of the people are somewhere in the middle, suggesting that we need to think about individualizing therapy more. It shouldn't be perhaps a set it or forget it. Everybody just gets some monthly injections. But understanding that, if you do this, PRN regimen, you're going to lose a letter or two from what your peak abilities are. The other thing I forgot to point out is we're at the top of the dose response curve. That was the other thing we were testing here. So the 0.5 milligram actually did a little better just numerically than the two milligrams. So going higher is not going to be leading to better results. We were lucky, in a sense, that the 0.5 milligram was the top of the dose response curve. Now, this is very busy and I apologize. This is called a mosaic plot. This is everybody in the 0.5 milligram PRN arm, okay? And it shows the number, the injections you got, month one, month two is the second column, month three. So the protocol said mandatory three injections. And then everybody else got, and then after that you got PRN. And what this shows is the PRN number of injections that you got, whether you were a loser, meaning you lost more than 15 letters, or whether you were a super gainer, you gained 25 or more letters, you got roughly the same number of injections. So the data do not segregate out by whether you gain vision or lose vision number of injections that you need. So that's one thing that we needed to do and we have done, but there's still more work to be done in that front. The second thing we can do to advance wet AMD treatment or neovascular AMD treatment is to answer these questions. And these are questions that every patient who's newly diagnosed with wet AMD has, and we can't answer yet in a definitive way. And that is, what's my prognosis, right? You just got diagnosed, you want to know, are you going to go blind or not? And when's my next shot? All right, if we're going to go this PRN paradigm or treat and extend, patients want to know, their caregivers want to know, how frequently are you going to be bringing them into the clinic? So we started to look at our data across the clinical trials, and it was kind of surprising to me, these are all monthly injection arms with randubizumab. The absolute visual acuity gains differ quite a bit from trial to trial. Now, statisticians will tell you it's not fair to do cross trial comparisons, but nonetheless, it was interesting, in Anchor patients gained over 11 letters, yet in Marina they only gained seven letters. Why is that? And we started to dig into that to see if there are any covariates that predict response. And one of the things that fell out is baseline lesion size. So if you have a big lesion, a larger lesion over four disc areas, you're going to gain less vision on average than if you have a smaller lesion. And that kind of makes sense, right? If you have a small lesion, there's less overlying retina that's going to be damaged, presumably from the leaking and the hemorrhage. And if you have a very large lesion, the large lesions are probably older lesions because it takes a finite time for those vessels to grow. So this is biologically plausible, and we wanted to look at it further. And we did it in Marina, Anchor, in and Harbor, and we split out the patients who had smallish lesions versus the patients who had large lesions. Small lesions are in red in all these trials, larger lesions are in blue. And the cutoff was, as you can see here, it was a computer program that did the best fit cutoff, this binary division of large and small. But again, across all these trials, small does better. So that's one of the things we think that leads to predicting prognosis for the individual patient. Ultimately, what we want to do is build a model here where you can plug in covariates and they will tell you for that individual patient what they can expect from a visual acuity standpoint. And some of the covariates in these things are in the literature. It doesn't look like polypoidal chloride on the vascularization responds as well to anti-veget. So this is something we're looking at and may plug into the model. Feter vessels as shown here. Feter vessels are mature vessels that are arterialized, do not respond well to anti-veget treatment. Vitromacular traction. So if you have a physical force, if you have a physical force pulling inward on the macular region, you're not going to get rid of this edema and there are papers that show that. In outer segment lengths, now that we have SDOCT, we can measure the outer segment lengths of photoreceptors and see if there's pre-existing dead retina. If there's pre-existing dead or damaged retina, it's very likely not to recover vision. So if we plug in these covariates into a model, there is a possibility we can get better ability to predict the visual outcomes. And not everyone gains vision. And this is a graph. It's a little bit complex, but it's color-coded. So patients in the blue here are the patients who lose more than 30 letters. And on the other end of the spectrum, the patients in the light blue here are the ones who gain more than three letters and everything in between. And what you can see here is there is a subset of patients who lose vision even with monthly looseness. This is from the Marina trial. Can we identify who these folks are? Another way. And the way we're trying to do that is genetically. So Rob Graham in our clinical group at Genitech have attempted over the last three years to identify polymorphisms or SNPs, single nucleotide polymorphisms, that are associated with treatment outcomes when you give the drug to patients. And the goal here is a more accurate prognosis again. And the technical term is can we come up with the predictive biomarker? So can we run a blood test, see if the patient have the SNPs, and say okay, you're going to respond to this drug, or for certain they're not going to respond, we're not going to subject that patient to monthly injections unnecessarily. So the trial looked at a number of large trials that we conducted. Anchor, Marina, and Focus. We collected bloods on these people. They went into this horizon extension study. And this genetic study is called DAWN. So we had 352 patients and we analyzed their genetic makeup. And in this GWAS analysis, genome-wide association scan, so this is a non, what's the word, prioritized scan of SNPs, six different SNPs fell out that had very significant P values as you can see here. So in this cohort, we identified six SNPs. And what was interesting in these SNPs is we created a genetic score. So if you had one SNP, two SNPs, three SNPs, six SNPs, if you had six SNPs, okay, it really predicted that you would gain vision with Lucenus or Renabisman. If you had one or zero SNPs, it predicted in this backtesting that you would lose vision. But genetics can be misleading. And I think any of you who does it who do genetics know that you can get really nice P values, but it's a function of data mining, right? If you look enough times, things will come out significant. So you have to do something called a validation cohort. You have to do a prospective validation of what you see in this backtesting model and confirm in a prospective test what you see here is real. So what's happening now is Rob Graham is analyzing all the bloods from the patients in Harbor to see if this is true. And if it is true, I think it is likely we'll come out with a diagnostic test that will give you some sense and give patients some sense of their prognosis. And the other key question is when's my next treatment? So here's the graph I showed you earlier. And I've characterized these patients into addicts. The addicts are over here. They're the ones that you give them an injection. They come back another month. There's still disease activity. You've got to give them another injection. They need it every month. And then at the other end of the spectrum are the people I call the cures. For some reason, anti-vegeta is disease modifying. You give them those three monthly injections and then you short circuit the disease process and they no longer need the drug. They're dry. Their visual acuity is stabilized and improved. I want to look at these cohorts because this is the first PRN trial that we've ever done of level one. You know, it's level one evidence that's large. And look at baseline characteristics, anatomic functional demographic, look at patient comorbidities, look at their concomitant meds, and also look at their pharmacokinetic data and see if any of those correlate and help you discern who's an addict and who's a cure. You're the addicts. You're the cures. So for instance, could it be that when you inject these drugs in the eye, some people just clear them very rapidly so they don't have enough anti-vegeta in the eye. Actually some of our pharmacokinetic data suggests that. So the average half-life for Lucentus in the eye is nine days. But some patients, it's 20 days. Other patients, it's two days. So if you get the drug in the eye and you rapidly dump it, are you an addict, therefore, because you need more frequent injections? You don't have a drug on board. So we'll do those analyses and try to plug them into a model that'll predict how often you get injected. The third thing we need to do, regardless of what those analyses show, for the majority of patients, even seven injections a year, frankly, is too much, is to reduce the treatment burden. So here's Marina, here's Anchor. We got those really nice data compared to the controls, the last natural history arms you'll ever see. But we got those data with monthly injections. That's a real burden. Not only on patients, but treating physicians tell me, they feel like robots. Patients come in and they're just injecting every month. And the logistics of their offices, especially with the new indications, DME and RVO and stuff, are just becoming untenable. But most importantly, for patients, we've got to come up with long-acting delivery for a number of reasons which I'll discuss. So this is something we've worked on even before I was a genetic. This is when I was an academic, when I was a mass pioneer. We created these microspheres. This is a polymer that contains pygaptinib, which was macrogens, still is macrogens, that slowly releases the drug. You inject it in the eye and it slowly releases it. These are very small, 40 microns in diameter, so they're syringeable. You can inject them through a 30-gauge needle. And what we saw in rabbits was you were able to release the drug consistently for 100 days. And if we extrapolate this curve, 120 days or six months. And you're above the minimum effective concentration that's needed to block VEGF in the eye, which is the dotted line here. So this gave us hope early on that we can maybe come up with a long-acting version of these anti-VEGF drugs. And this is really the in vivo experiment we did that convinced us that this was a possibility. So the rabbit retina, I think all of you know, is a merangeotic retina. Rabbits don't have a vascularized retina. They have these two wings of vessels coming off horizontally off the disc. And if you inject VEGF in these eyes, these vessels leak like crazy on fluorescent angiography. So that's what you're seeing here. So what we did in this experiment was we injected these microspheres that are releasing pigaptinib, the anti-VEGF, into the rabbit eyes. And then one week later, we came in and we injected some VEGF. And we blocked that leakage that you see here in the control. And 12 weeks later, we came in again and gave another bolus of VEGF, and again you blocked. So this suggested pretty strongly that we were having long-term, at least three-month blockade of VEGF in the eye using these microspheres. So the microspheres may give us this sustained release. The other thing I think they may do, and it's less appreciated, is with current intravitual injections, we give a bolus of the drug where we put so much drug in the eye. And the reason we do that is so that we can extend the treatment interval. So we're going to give orders of magnitude more VEGF inhibitor than you need to block VEGF in the eye. And you do that because as you go out in multiple half-lives, you're still above the effect of concentration that you need to block VEGF. But a consequence of that is you get a very high C-max or maximum concentration of the eye. And that VEGF has to go, or that antibody, or anti-VEGF drug has to go somewhere, where it goes when it exits the eye is the systemic circulation. And it turns out that when it exits at C-max and goes into the systemic circulation, you can actually get levels that block systemic VEGF. And there are papers that have been published that patients were treated in one eye for, let's say, diabetic retinopathy. And the retinopathy in the other eye magically gets better. Well, it's not magic. It's the drug seeps into the systemic circulation, and there's sufficient amounts that you can get to the contralatal eye, where you see an inhibition of the pathology. So I think if we do long-acting delivery in addition to the reduced treatment burden for patients, I think there's a potential safety positive outcome as a function of that as well. This is something we just announced in a press release a couple weeks ago with Genentech. This is an implantable reservoir. This is another approach to coming up with long-acting release. This is a reservoir that was designed by Gene DeWaan that's implanted in the PARS plana and stays there indefinitely. It has a chamber that you fill up with a highly concentrated form of drug, in this case, Lucenas or Radibizomab. And it's slowly released into the vitreous over a period of months. And the drug essentially just diffuses down its concentration gradient. It's solid state. It's got no moving parts. There's really nothing that complex about it. A phase one has been completed, but before it went into patients, in vitro testing showed that you could release Radibizomab or Lucenas for six months in a very nice and predictable manner using this device. But the device has been implanted in 20 patients in Latvia where the phase one was completed. It's been in eyes for a year. And when it's first implanted, it's about a 15 minute procedure. You do this now in the OR, but I think eventually it could be done in a minor procedure room. You use an MBR blade, make an incision. It snaps into place. And all you got to do is put a vicarol to close the contractiva. And you see the contractival injection goes away over time. This thing is flush with the surface of the sclera. And it's got a port where you can inject into the reservoir the drug every six months. So patients now will come in twice a year, not every month. Get refilled, not into the vitreous, but into the device. And the device will slowly release the drug into the eye. And when the patient looks up, actually at the slit lamp, you can see the edge of the device. It's visible as it protrudes into the center of the vitreous. So we're in the process of writing an IND. Hope to get FDA approval to go into phase two later this year. So this is just another approach. So I show you these two examples because at least where I work, we think that long-acting delivery is so important. We have to place multiple bets. It's early. We don't know which ones are going to work and which ones are going to fail. And these are two of the approaches that we're using. Oh, and I just wanted to show you, in the phase one, we got nice evidence on OCT of a real pharmacodynamic effect. So here's a patient who had the implant. It was filled with commercial Lucenas, so not the highly concentrated form. But nonetheless, you could see on OCT the patient drying out for three months, and then the device was refilled. And this is not just us at Genotech. It really is in the development side, something that a lot of people understand is going to be transformational for treating AMD and other retinal diseases, frankly. So there are multiple companies using erotable and non-erotable polymers. People are using adenoviruses or adeno-associated viruses that are expressing VEGF inhibitors long-term in the retina. Allergan has its Osardex platform. Neurotech has a very interesting platform. These are RPE cells that are genetically engineered to make a VEGF inhibitor, and they're surrounded by a plastic polymer so that they're not attacked by the immune system. This is implanted in the eye, and basically you have a little bioreactor that's making your anti-VEGF continuously in the inside of the eye. It's pretty cool technology. I showed you this before. The fourth thing we can do is to treat early and or prevent carot on neovascularization. In the latter, I think we can start to contemplate if we really come up with long-acting delivery. I don't think patients are going to tolerate monthly injections as a prevention strategy. So before a patient converts to neovascular AMD, they typically see 2020, 2025. I don't think they're going to come in and accept the risk of endoplamitis and the rest of the monthly injections. However, if you have a device that delivers just twice a year, I think that's a possibility. And maybe we could treat patients who are high risk, and I'll talk about that for a moment. But again, I'd just point out that 60% of patients see 2050 or worse at 24 months, even with monthly random-bismab injections. So we should also think about catching them earlier. And I think we already do. I think the results are probably better than what we saw in the original trials. But one of the things that told me treatment delay limits visual acuity recovery was the horizon trial. So the horizon trial was an extension trial, where in the initial study in Marina Anchor patients got the monthly injections for 24 months. But at month 24, we allowed a crossover of the sham treatment arms. So these are the control arms. Now, when you get newly diagnosed with AMD and you give somebody a VEGF inhibitor, they gain vision, substantial vision, as we saw repeatedly before. But if you've had a newly diagnosed with AMD, but you were randomized to the control arm, and you waited 24 months to get your first shot of an anti-VEGF, you don't gain any vision hardly at all. So delay means while we're waiting those 24 months, it strongly suggests the retinas dying. And the longer you wait, the worse it is. And we need to figure out ways to catch patients earlier. And I think, and this is a hypothetical curve I drew on the plane, I really think that if we are able to catch patients when those first vessels are poking through Brooks Membrane and start them on therapy at that point in time, I think there's a real chance we're going to recover all vision. And we're actually doing a trial where patients are getting quarterly angiograms. These are patients with second eyes. You know, show you data in a minute why this is so. So they've already lost vision in the first eye. And the second eye doesn't have neovascular AMD yet, but they're getting quarterly fluorescent angiograms. And as soon as we see the first hint of neovascularization, they go on therapy. So there's only eight patients in the trial, but all eight have gone to 2020. So it's suggesting it's not definitive that if we catch them early, they'll do better. I think the longer you wait, where they're able to go, here's 2020, right, is dotted line. You're going to gain less and less vision. And at 24 months, as we saw in Horizon, they gained little or no vision. So I think this is true. I don't know for certain yet. So I think what we need is better screening. And here's just one company. There are multiple companies working on this. We all know what the AMSLA grid is and the Yanozzi cards. But this is one I found on the web, My Vision Track. It's an app for your iPhone where they ask you progressively which of these three is different, but it becomes more and more subtle with each test. They say this is two times more sensitive than the Yanozzi card. So we need something sensitive and specific. Otherwise, the offices are going to be inundated with false negatives, right? So the test is going to be key. But I think if we catch patients earlier, we're going to save a lot of vision. And the reason why I think we can do prevention is the following. And I alluded to this a moment ago. When we did quarterly angiograms in the contralateralized of the patients of Marina and Anker, we saw 24 to 39% convert depending on the trial, Marina and Anker. So you can get up to 39% of patients convert in two years. That's a much higher rate than Air Reds. Air Reds was just using color photographs. But if you do an angiogram, it's much more sensitive. I don't know how sensitive it will be with OCT. It may be even more sensitive. So it could be that in patients with their first eye involved with neovascular AMD, and now you're looking at the second eye, the one you definitely don't want them to lose vision in, we may be able to prophylax this group. I think if we do a prevention study, it will be in this group of second eyes because of the relatively high conversion rate, assuming we can get long acting delivery to work. And the reason why I think prevention will work, the other reason is, we did experiments, Joan Miller and I, this is back in the late 90s, we published them in O2, where we gave rentabism app to monkeys at day zero. So I think all of you know, in the monkey model, you laser the macula. And when you laser these laser spots over time, over two weeks, develop corretal neovascularization. So what we did here is we started treating at day zero, the day they got lasered, they started getting injections of rentabism app, and we completely prevented neovascularization compared to the controls. This is mislabeled. This is the rentabism app treated harm, and these are the controls. So this suggests that if you treat with an anti-vedgeff, you can prevent CNV is what this monkey study seems to indicate. The fifth thing that's being tested right now in phase two is regressing the corretal neovascularization. So anti-vedgeff drugs are not good at regressing the vessels that are already there. They're very good at preventing further growth. And again, these are data from the New England Journal paper, the Marina trial. So this is the mean disc areas on fluorescing angiography of the corretal neovascularization in the sham arm, the .3 arm, and the .5 milligram arm. And you see at 12 months in the sham arm, the vessels grow, and rentabism app was very good at preventing further growth. You see here that it's absolutely flat. It stops the vessel growth dead in its tracks. What it doesn't do is it doesn't regress the vessels. The vessels are still there at 12 and at 24 months. And this is just another way to show it, change in baseline CNV, very little, a little bit of regression, but not a lot. So the anti-vedgeffs don't regress well. So when we were in the lab back in the late 90s, we wanted to come up with an eye model where we could test drugs and see if they could regress vessels. All the eye models that we were working with, whether it's the laser CNV model or the oxygen-induced retinopathy model, vessels grow and then they naturally go away in those models. So they don't replicate human disease. You don't have persistent neovascularization. So here, finally I'm going to talk about cornea a little bit. What we did in these eyes is we destroyed all the limbo stem cells of the cornea. And this was with dilute alkali, with dilute sodium hydroxide. This is basically an animal model of alkali injury to the ocular surface. You destroy the corneal stem cells. And what happens is just like in patients who've had an alkali injury, you get a conjunctivalization of the ocular surface and the vessels grow and they meet in the middle of the cornea in two weeks and they persist. They stay there forever. Now when we published this, we went out to four weeks, but we took these animals out to six months later and the vessels are there forever. So now we had a model of persistent vessels and we could see if we could regress them. And the first experiment we did was to do the injury, have the vessels grow, and start treating with an anti-vegeta for day 10, at day 20, and at day 30. And what you see is the amount that we're able to regress at day 10, pretty good. It's on the order of 70%. But by day 20, the vessels undergo a maturation event where you can no longer regress effectively. It's about 25, 20% here. And at day 30, none of the vessels hardly regress. But we were able to get adequate levels of the anti-vegeta drug to the cornea. That's what this graph shows here. So proving the point that the longer the vessels are there, the more resistant they become to the anti-vegeta. So the idea we had was the following. A new capillary is made of two types of cells. There's the endothelial cells which line the capillary and then very quickly, almost simultaneously as the endothelial cells are growing and sprouting a new capillary, they're enveloped by perisites. And these perisites are the other cellular element. So there's just two types of cells that make up new capillaries. And the thinking was the following. Anti-vegeps work on endothelial cells. They don't touch perisites. We need a drug that will do the same thing that anti-vegeps do to endothelial cells to perisites. And if we could do that, then the perisites would go away and if you have a perisite poor capillary, you hit them with an anti-vegeta and then the rest of the capillary goes away. So that's the hypothesis. And a guy named Doug Hanahan, a UCSF, was the first to publish a combination of an anti-vegeta and an anti-PDGF, platelet drive growth factor, and he showed in tumor models that vessels regressed. This was around the year 2000. That was a very exciting paper. So we poured over that and you look at the literature and you learn that PDGF is a misnomer. Platelet drive growth factor is actually the key growth factor, the B subtype that brings in the perisites. So PDGF is the perisites as VGF is to endothelial cells. And if we could block PDGF, the thinking was we could hit both the perisites and the endothelial cells and regress vessels. And what was interesting is in the cornea model, cornea is cut here so it's flat mounted. And here's the neovascularization. Like we said in the cornea, the vessels grow and they plateau and they persist. All the right molecules are being made. So not only was VGF being made in the cornea when we measured it, this is a messenger RNA, but PDGF's made and the PDGF receptor, which is PDGF receptor beta. So both ligand and receptor are being expressed in the right place at the right time. And when I was at ITEC, we licensed the only PDGF drug we could find at the time. The only one at the time was one being developed by a small company that we bought and brought in and started to develop. We wrote an IND, but first we did some clinical or preclinical experiments. It's an aptimer that specifically binds to PDGFB and blocks it. And in the animal studies when we gave animals this drug, perisites which are labeled in red on the corneal vessels here go away. So all we have on the new vessels is now just endothelial cells which are green, but the perisites go away. So sort of proving the point that PDGFB is the right target. It is addressing the perisites. And when we gave an anti-VEDGF drug monotherapy in this model of corneal neovascularization, the vessels do not go away. They don't grow any further, but they don't go away. But when we gave the combination of anti-PDGF and anti-VEDGF, they go away, but they leave the normal limbo vessels, which was kind of interesting. One of the things we're always concerned about is, all right, if you give this combo, are we going to be regressing normal vessels too? Unfortunately, we didn't see that. And what was interesting is when we stained those limbo vessels, the PDGF aptimer was stripping the perisites they're now labeled in this yellow-orange, was stripping the perisites from the abnormal vessels growing in the cornea, but was leaving the perisites in the limbus. And I don't know why that is, but it was a fortunate occurrence because those vessels do not go away. And in fact, we were able to regress vessels with the combination. So we wrote the IND. We're ready, getting our plans ready to start the phase one, and then ITEC was sold. We sold the company, and as happens in the business world, I lost control of that program. So OSI Pharmaceuticals bought ITEC, our company, and they put the program on a shelf until Sam Patel, a retina specialist and one of my co-founders at ITEC, bought this aptimer and started a company called Optotech. And he's taking this forward. So he did a phase one, and now he's doing a phase two trial. But in the phase one trial, they gave random abysmab with this anti-PDGF aptimer in patients. And here's a patient of baseline, a fluorescine in geography. This is 100% classic legion. You can see it's margins of the corneal neovascularization. And when they gave the combination of the two drugs, you could see some regression and a little more regression of two months. And the patient's vision improved to 2025. They gained 27 letters on the ETTRS chart. Here's another patient with a small lesion, goes away with the combination. The patient goes to 2025. Now this is an open label phase one, okay? And I tell all my students and everybody who works for me at Genotech, beware phase one data. Phase one data are open label, unmasked. And there's a huge placebo effect in AMD. Patients can will themselves to look around their scatomas. They learn how to read that ETTRS chart. They get better every time. And they're going to gain 5 to 10 letters with the placebo. Okay? So when people come in and they're shopping their new drug to us, and I see curves like this of, you know, percent of subjects who gain 15 letters always take it with a grain of salt. Barbara's smiling because she knows it's true. But this is objective data. This is fluorescein angiographic data. And this was a little more convincing to me. 100% of the CNV lesions regressed. They treated 21 patients. And 86% of the lesion size on average went away. So all that remained was 14% of the original CNV. So that's pretty impressive. This means you're seeing a pharmacodynamic effect. The drug is doing something. I don't know that it's going to lead to a better visual outcomes, but it's doing something. So this is being tested in a phase two, a large well-powered phase two. This one is randomized, masked. 450 patients in the phase two results will come out of optotech in May. But there's still a key question here, and that is, okay, if we're able to pharmacologically regress these vessels, does that mean patients are going to gain vision? Doesn't necessarily mean they're going to gain vision. It may be there's already a lot of preexisting injury in the vessels go away, and they're not going to see any better. But it stands to reason that the CNV shown here in this cartoon here is underneath the RPE, but if it's a lot of CNV, classic CNV is between the RPE and the photoreceptors. And we know how the visual cycle works, right? It's a very good cooperation between photoreceptors and RPE to cycle cis and trans retinol. If you can restore that anatomy, it stands to reason that some patients will see better, but we'll see when the trial data come out. So that's still out there. The sixth way we can improve outcomes, I think, is to treat inflammation. Some people have been doing this. They've been injecting kennelog and eyes. You inject kennelog in patients with CNV. We're not doing it routinely now, but they were five years ago. You'll see some CNV dry up, suggesting there may be an inflammatory component to this disease. This has worked from Hans Grossner-Klauss and Emery. So he looked at patients with CNV not too long ago. They used to operate on patients with CNV. They would go in physically and retract me and remove the net because there wasn't any other better way to treat. And there were a lot of good surgical specimens to study. And what Hans showed here is he did serial sections of these lesions, counted the number of macrophages and vessels, and showed very nicely that there are a lot of macrophages in coronia vascularization, and they co-localized with the vessels, suggesting they may play a role in CNV. This is an experiment we did back when I was at ITEC, where we looked at the normal developing retina. I think you know in the rodent postnatally, the retina grows out centripetally towards the aura. When the vessels, normal vessels are growing, they sprout. It's a normal sprouting angiogenesis. There are no inflammatory cells that we could see. But in the same rodent species, if you put those animals in high oxygen, so this is an ROP-like model, and then you take them out of oxygen and they get retinal neobascularization. Those abnormal retinal vessels are full of inflammatory cells. So normal growing vessels don't have inflammatory cells. Pathological growing vessels have lots of inflammatory cells. And now it's kind of interesting. It was a cellular differentiation between the two. So this is a work that I did with Susumu Ishida in the laboratory, but also J.M. Body, Bala's brother, where we treated with a drug, chlogenate liposomes that get rid of macrophages. And when you get rid of macrophages, you, to a large degree, get rid of the retinal neobascularization, suggesting that the macrophages were not only present, but were somehow operative in the neobascularization. And here's a model, again done with J.M. Body, Bala's brother that we published, where we selectively in these genetically modified animals got rid of the inflammatory cells. You're looking now, carotone neobascularization on FOS, it's stained in green, and here are cross-sections. And you can see if you get rid of the inflammatory cells again in carotone neobascularization, there's a 50 to 70 percent reduction in the area of CNV. So both of these experiments suggest that inflammation plays a role in the new vessel growth itself. And then Scott Cousins at Duke, many of you know him, he runs a retina service there, is really a fantastic experimentalist. Scott studied the laser model of carotone neobascularization, and here's the CNV four weeks after he induced it with the laser. And late in the process, he sees inflammatory cells coming into the inner retina, all right? And then they come into the inner retina, they're labeled here, and they disrupt synapses in the inner retina, between the bipolar cells and the ganglion cells. And the electrical signal is diminished. So there may be another role for inflammatory cells, and that is inflammatory cells may be disrupting neural signaling in addition to the role they'll be playing in carotone neobascularization. So now there's a new model of wet AMD that I'm very excited about, Neovascular AMD, good thing Greg's not here. And this model is a rodent model where at 12 weeks of age, these animals get spontaneous carotone neovascularization that's multifocal in nature in both eyes, with 100% penetrance. And we're sequencing the genome of these animals to find the gene. Dave Shima first discovered this at the Jackson labs. Dave Shima ran our research labs when I was at ITEC, so we're good friends, and we still collaborate together. So anyhow, this model I think is going to really blow open our ability to study CNV in the lab, because no longer are you going to have the artifact of laser injury, you're going to have spontaneous CNV without the laser injury as occurs in humans. But what's interesting is Dave is stained for inflammatory cells and just as the vessels are poking through Brooks membrane, they're surrounded by these dendritic like inflammatory cells, which actually come in right before the vessels break through the membrane. So arguing that inflammatory cells play a role very early in CNV. So this is a really unexplored area that we're looking at much more deeply back in San Francisco. And here's the last bit we can do, and that's protect the neural retina. Now this is not AMD, this is retinal vein occlusion. And back in the early 90s when Joan Miller and I were using the laser model of retinal vein occlusion, and what Joan did here is modify Heire's model. She used a dye yellow laser to laser shut all the veins coming off the disc into monkey. And when you do this, it mimics a human CRVO. Retina is ischemic. These animals go on to get neovascular glaucoma, iris neovascularization. When we looked at the retina three weeks later, it was really interesting to see not all that surprising, a lot of dead retina, right? So ischemic retina dies over time, clearly. But most of the inner and outer cellular elements are gone. Now we were doing this experiment originally, this was early in the 90s, when we were figuring out what VEGF was doing in the eye. And you can see in ischemic retina, VEGF levels are very high, and I'll talk about that later this afternoon at the research seminar. But it was also very interesting that there was a lot of dead retina. And we see this also in our human study. So this is our central retinal vein occlusion study, the one we used as we submitted to the FDA to get random bismap approved for this indication. In patients with nuanced vein occlusions who started treatment right away gained a lot of vision, and that vision gained persistent. But the controls who were allowed to cross over after six months didn't gain anywhere near as much vision as you can see. So again, this time we only waited six months, not 24 months like I showed you before. A lot of vision loss, and the OCT showed clearly the crossover patients dried out. In fact, they were just as dry as the patients who got random bismap from baseline. So it's not because they didn't dry out, they dried out. These data strongly suggest there's dead retina. So that's in vein occlusion, that's not surprising. The next day I'm going to show you our new unpublished data. Actually, no, the next slide after this. So here's showing the sham crossovers. When they cross over, a lot of patients lost vision. This is a waterfall graph. It shows every single patient in the trial. A lot of them lost vision. Some gained, all right, so it's very variable. Like we're learning in this disease, there are individual nuances. But you could see the patients who were treated at month zero did much better. There were fewer losers and more gainers. But here's the AMD data. So this is Harbor. And our biostatisticians just put this together for me last week. So Harbor was the first large phase three type trial where we used SDOCT. And we can very accurately measure retinal thickness. Now on SDOCT, central foveal thickness, normal central foveal thickness, is about 225 microns. Okay, so this is dead center in the macro. And look at this statistic. Central foveal thickness, and all the treatment arms are listed here, of patients with greater than 175 microns, all right, is 40 to 48%. That means more than half the patients have retinas that are less than 175 microns. They have a central foveal thickness that's much less than normal, arguing there's a lot of dead retina. And I think a lot of this dead retina existed at baseline when these patients were enrolled. Because remember the clinical trial results? They go up and then they stay at a plateau. If retinas dying during the treatment interval, you would expect the curve to go down, but it's not. I think a lot of these patients are being enrolled with a lot of dead retina. So again, it's to sort of, I think, further highlight the point that we need to get better with prevention and early diagnosis. I think we're going to have to pick up these patients earlier if we're going to save vision and get better outcomes. So just to summarize what we talked about, there's a lot of things, and these are all things we're working on. Optimize the efficacy in the dosing, the prognosis in the dosing frequency, particularly for patients and their caregivers. Come up with long acting delivery. Treat early and prevent coronary obfuscation. I think of all these things I have listed here, my bet is this is going to be the most important. Regressing coronary obfuscation, we're going to get good data in a couple months, whether that works. Treating inflammation and protect the neural retina. And again, I think we're going to protect it mostly by treating early. So I want to thank your attention, but before I end, I just want to acknowledge the people. There's a bunch of really talented people that I work with, and they're the ones who did the work, and I want to acknowledge them. Jill is a retina specialist. Phil's a glaucoma specialist. Jason came out of residency at Stanford, and Roman also out of Stanford. And this is our medical team. Gary worked with me at ITEC and made a move over from New Jersey to San Francisco. This is our research group. METAL runs our inflammation immunology unit in the eye, in the vision science group. Waylon and Jennifer run our vascular biology unit. Joe does the neural protection unit. And Rob did the genetic data that I showed you in the Dawn study. And then last but not least, are people toiling with the biostatistics piece in these clinical trials, and I certainly overutilized them, but I thank them. So thank you for your attention. It's really great to be here, and I'm happy to take any questions. So in addition to the GWAS, there was a targeted analysis of the known SNPs. And Rob was looking only at treatment response. And so when he looked at those known SNPs as a function of treatment response, they didn't fall out as being significant. I think where they're clearly seen, where we've gotten really lucky on, like other diseases, is how this seems to be effective for a broad swath of the population. I don't think that's going to be the case going forward in dry Andy, for instance, where you're going to have, I think in the future, according to Greg's work, patients segregated as to whether they have chromosome one or 10 disease and are going to be treated with different therapeutics. No, we weren't surprised. So they designed the trial in 06 actually with the FDA right after we were approved. And then they constructed after the trial was underway a pharmacokinetic model. So we did it backwards. The pharmacokinetic model said going up forex on the dose shouldn't increase the PRN. We shouldn't have much of an effect on the PRN. So even though to get an extra month in our models of treatment efficacy, you need to go up 10x in the dose. So we should have treated with five minutes. So in our hands, and I think I'll show data when I give the diabetic talk, we were able to regress. What's interesting is because there are pericyc poor vessels and diabetes, an anti-vegetable loan and that disease does regress. But what's interesting is the endothelial cells go away, but the basement membrane is still there. And when you stop the anti-veget, the vessels come immediately back and they're exactly the same architecture they were before because they grow along that basement membrane. So in our hands, at least we've seen regression. And we also saw it in rise and ride, the pivotal phase 3s in DMA. We saw regression there as well. I don't know why Avastin isn't regressing. Vegeptal is a monocytic chemo-attracted, but I'm not sure that it's the only chemo-attracted for inflammatory cells. I suspect other things are coming up. And people have seen in the case of diabetic retinopathy, IL-1, IL-6, TNF. Same thing in AMD, they're seeing TNF come up. So there are other pro-inflammatory cytokines that VEGF inhibitors are not getting in. Good. Well, thank you.