Imatinib as a Paradigm of Targeted Cancer Therapies

Part 1: Imatinib (Gleevec): A Targeted Cancer Therapy

00:00:01.21 Hello, my name is Brian Druker from the Oregon Health and Science University Cancer Institute.
00:00:06.21 The topic of my talk today is Imatinib, or Gleevec, as a paradigm of targeted cancer therapies.
00:00:13.22 What I'd like to do in the first part of my talk is discuss chronic myeloid leukemia
00:00:19.10 and the development of imatinib as a specific targeted therapy for this disease.
00:00:23.09 I'll start with a clinical description of CML, then discuss the molecular pathogenesis of CML,
00:00:29.18 followed by how that understanding led to the development of imatinib,
00:00:33.12 both through its pre-clinical and clinical phases.
00:00:36.00 Now, for those of you who, that know chronic myeloid leukemia, you know that it in fact is relatively rare disease.
00:00:42.17 But I really need to have you think of this as a paradigm.
00:00:46.08 So, in order for you to think of this as a paradigm, I'd like to show you a movie clip from a movie titled "Dr. Ehrlich's Magic Bullet."
00:00:53.03 If the cause has been discovered, there's hope for a cure.
00:00:57.00 And when you think about what I'm talking about today, I want you to really think about your own favorite disease,
00:01:03.11 and if you understand the cause, then there's hope for specific, targeted therapies that can be quite effective.
00:01:09.13 Now, I'll use CML as our paradigm for today, but this really is a generalizable phenomena, as we'll discuss.
00:01:15.28 So, when we think about chronic myeloid leukemia, we first have to understand the molecular pathogenesis,
00:01:23.25 but before that, we also need to describe the clinical entity.
00:01:26.24 So, the clinical entity of CML was first described in 1845.
00:01:31.14 About 140 years later, the molecular pathogenesis was understood
00:01:35.23 and then in 2001, a specific therapy was developed.
00:01:39.27 Now, I actually view this slide as the heart and soul of translational research.
00:01:44.20 You describe a clinical entity, understand its molecular pathogenesis and use that knowledge to develop a specific therapy.
00:01:52.08 But one of the lessons we learned in the clinical trials of imatinib is that translational research doesn't end when you go into the clinic.
00:02:00.04 In fact, we need to understand how our therapies work,
00:02:03.06 why they work, who they work best for, so in other words, is, we get into the clinic,
00:02:08.01 we actually need to go back to the lab to understand some of those issues.
00:02:11.15 And, for example, why do some patients relapse, as we'll discuss.
00:02:15.17 As we begin to understand that, we can then go back into the clinic with even better therapies.
00:02:20.22 And we then begin a translational research cycle.
00:02:24.00 And I'll demonstrate how that worked in our clinical trials of imatinib as well.
00:02:29.02 So, it's absolutely clear that we can certainly do better than 150 years of discovery,
00:02:35.16 but I'd also just like to take you through this briefly.
00:02:38.14 And we'll start with a clinical description of chronic myeloid leukemia in 1845
00:02:44.06 by two pathologists, Dr. Bennett and Dr. Virchow.
00:02:49.05 Now, these two pathologists, without staining methods for blood,
00:02:53.13 described the clinical phenomena of chronic myeloid leukemia based on autopsy findings.
00:02:59.17 Now, in 1845, these two pathologists published these papers within 6 weeks of one another.
00:03:06.03 Now, as you might imagine, there ensued a vigorous debate about who was really first,
00:03:10.28 ultimately Virchow capitulated that Bennett has described CML six weeks prior to him.
00:03:17.27 Fast-forwarding, we now know that chronic myeloid leukemia makes up about 15 to 20 percent of all leukemias
00:03:25.25 and it is one of the four common types of leukemia.
00:03:28.28 There are about 1 to 2 cases per hundred thousand per year,
00:03:32.29 and the incidence of CML is relatively even world-wide.
00:03:36.17 In the United States, its incidence translates into about 5,000 new cases per year.
00:03:43.02 Although the disease can affect any age group, the average age of onset is in patients 50 to 60 years of age.
00:03:50.25 The disease itself has three recognizable phases.
00:03:55.22 The so-called chronic, or stable, phase, in which 95 percent of patients will present, hence the name, chronic myeloid leukemia.
00:04:03.26 However, the disease does progress over time through an accelerated phase
00:04:08.28 to a blast crisis, and it's these two phases that are collectively referred to as advanced phase disease.
00:04:15.15 In the chronic phase of the disease, it's characterized by a massive expansion of white blood cells.
00:04:23.22 So, for example, a white blood cell in a patient at diagnosis
00:04:27.17 will be anywhere between fifty to five hundred thousand.
00:04:31.08 For reference, a normal white blood count is between five to ten thousand,
00:04:35.25 so you can see the white blood count at diagnosis is between five to fifty times the upper limit of normal.
00:04:42.13 But the white cells mature completely normally and function normally.
00:04:47.13 They're just too many of them.
00:04:48.26 So, we don't see infections or susceptibility to infections as a manifestation of this type of leukemia.
00:04:55.22 If you look at a typical blood smear of a patient with leukemia,
00:04:59.22 you see a full spectrum of white blood cell lineages.
00:05:03.08 So, here you see some immature white blood cells,
00:05:05.24 here you can see we have basophils,
00:05:07.25 but you also see that there are lots and lots of mature and maturing white blood cells.
00:05:13.05 In fact, this almost looks like a bone marrow aspirate,
00:05:15.25 because of the maturation of white blood cells through the entire lineage.
00:05:20.08 So, in the stable phase of the disease, which historically has lasted 4 to 6 years,
00:05:27.10 we currently actually don't know how long the stable phase lasts with the current improved therapies,
00:05:33.21 like imatinib, which I'll cover very soon.
00:05:35.29 However, the problem with this disease, or historically the problem with this disease,
00:05:41.29 is that a malignant clone loses its capacity for terminal differentiation
00:05:46.28 and the disease progresses to a highly treatment refractory acute leukemia.
00:05:51.13 And by the time you get to this highly refractory blast crisis stage,
00:05:56.01 survival is measured in mere months.
00:05:58.06 So, typically no more than 3 to 6 months when a patient evolves into the blast crisis.
00:06:03.28 Now, if we're going to do anything or make an impact on this disease,
00:06:08.03 we need to understand the causes of this leukemia.
00:06:11.16 So, here we go back to, again, two pathologists: Peter Nowell and David Hungerford.
00:06:17.23 In 1960, they published this landmark paper in Science
00:06:21.27 that showed an abnormal chromosome in the blood cells and bone marrow of patients with chronic myeloid leukemia.
00:06:29.10 Now, this paper is the entire three paragraph Science paper.
00:06:34.02 Now, that tells you immediately that quality doesn't equate with, quantity doesn't equate with quality.
00:06:39.28 Now, as you know when you write a paper, in the last paragraph, you can sort of wax philosophical
00:06:45.27 about the broad implications of what your findings mean.
00:06:49.13 And if you look here at their very last paragraph, what they said was that the finding suggests a causal relationship
00:06:56.22 between the chromosome abnormality observed and chronic myeloid, or chronic granulocytic, leukemia.
00:07:03.00 Now, in 1960, nobody believed them.
00:07:06.15 Everyone was skeptical about this and they thought it was just an associated abnormality
00:07:11.03 that had nothing to do with the pathogenesis.
00:07:13.17 As it turns out, Nowell and Hungerford were well ahead of their time in recognizing how important their finding truly was.
00:07:20.15 Then in 1973, Janet Rowley, pictured here,
00:07:24.07 showed that this short chromosome 22
00:07:27.01 actually came about because a reciprocal translocation between the long arms of chromosome 9 and chromosome 22.
00:07:35.00 And as a result of that, you end up with a short chromosome 22,
00:07:38.26 as well as a slightly longer chromosome 9.
00:07:42.05 Now, by the 1980's, it became recognized that as a result of this chromosome translocation,
00:07:49.11 all of the Abelson tyrosine kinase, which is located on chromosome 9,
00:07:54.16 had been translocated into this break point cluster region,
00:07:58.05 or BCR, on chromosome 22. The chromosome breaks on chromosome 22
00:08:05.03 occurred between what's historically called the second and the third exon of the break point cluster region.
00:08:10.29 Now, as a result of that chromosome translocation, we end up with this fusion protein,
00:08:17.06 or RNA, called p210 BCR-ABL. And it includes either the second or third exons of BCR
00:08:24.18 fused in-frame to virtually all of the Abelson tyrosine kinase.
00:08:28.23 So, if we think now historically about chronic myeloid leukemia,
00:08:34.24 the reality is there's lots and lots of research that went into these discoveries.
00:08:40.06 So, if we think historically, actually, the discovery of a tyrosine kinase, like BCR-ABL,
00:08:47.00 actually had its roots from Rous sarcoma virus, which ultimately led to the discovery of the SRC oncogene
00:08:54.14 and somewhat later, the ABL oncogene.
00:08:57.17 And once the ABL oncogene was discovered,
00:09:00.16 it immediately was recognized that the BCR-ABL oncogene was a related event with oncogenic potential.
00:09:09.00 But in addition, as it also related, there's another thread that came from chromosome banding
00:09:15.20 that allowed gene mapping that allowed the assignment of BCR and ABL to the chromosomes 9 and 22.
00:09:23.07 But lastly, there was the entire field of protein phosphorylation.
00:09:27.24 So, beginning in the late 1930's, with the identification of protein kinases,
00:09:34.08 then in 1979 Tony Hunter discovered tyrosine kinases
00:09:38.29 and it was immediately recognized that SRC was the founding member of the tyrosine kinase family.
00:09:45.13 ABL was also identified as a tyrosine kinase and once BCR-ABL was identified,
00:09:51.00 it could be shown it was also a tyrosine kinase.
00:09:54.01 So, if we think about all of these events over the past 100 years,
00:09:58.20 there's a thread from tumor virology, a thread from protein kinases and phosphorylation,
00:10:04.07 and a thread from chromosome mapping and chromosome banding.
00:10:09.24 So, the points of this slide are several. First of all, we have a convergence of several different fields.
00:10:16.23 You can end up with remarkable breakthroughs that lead to specific therapies like imatinib.
00:10:22.09 But in addition, it tells you that there are literally thousands of researchers that are participating in this discovery process
00:10:30.01 over the past 100 years. Many, many investigators have made small incremental advances
00:10:36.22 that have led to a very big picture that allows major breakthroughs.
00:10:41.13 And so, for those of you that are working in laboratories or thinking about research careers,
00:10:47.00 the reality is there are many, many ways to make contributions.
00:10:50.23 I have the advantage of being able to look back over a hundred years now,
00:10:54.29 seeing how all of this has come together that's allowed an advance like imatinib.
00:10:59.19 So, now we're ready to talk about the development of imatinib
00:11:04.05 because we understand the molecular pathogenesis. So, here we have an ideal therapeutic target:
00:11:09.24 BCR-ABL. It comes from a fusion gene and protein that was generated by this 9 :22 chromosome translocation.
00:11:18.03 It's actually detected in all patients with chronic myeloid leukemia and we know from animal studies
00:11:24.12 that if you put BCR-ABL into mice, they develop leukemia.
00:11:27.24 This is an oncogenic event that causes leukemia.
00:11:32.20 We also know that the protein BCR-ABL functions as a constitutively activated intracellular tyrosine kinase.
00:11:40.27 And this kinase activity is absolutely required for its function.
00:11:45.09 So, we think about this schematically, and for those of you that don't think about tyrosine kinases every day
00:11:51.00 like I have for the past 20 years, this is a schematic of a tyrosine kinase.
00:11:55.14 So, what does a kinase do? In the case of any kinase, they bind ATP
00:12:00.26 and in the case of tyrosine kinases, they transfer phosphate from ATP onto specific tyrosine residues of substrate proteins.
00:12:10.14 And it's these substrate proteins that induce all the phenotypic abnormalities we see in CML.
00:12:16.20 The high white count, the excessive proliferation, presumably also protection from apoptosis,
00:12:23.04 and possibly even the genetic instability that leads to blast crisis.
00:12:28.01 So, if you could imagine, if you could block binding of ATP to this specific tyrosine kinase,
00:12:35.22 you'd have an ideal therapeutic agent for this disease.
00:12:38.21 And that's exactly what we did in the STI, or imatinib program.
00:12:44.05 So, beginning in the late 1980's, in collaboration with scientists at what was then Ciba-Geigy
00:12:50.14 and ultimately became Novartis pharmaceuticals, they were interested in developing kinase inhibitors.
00:12:56.24 Now, in fact, they weren't interested in chronic myeloid leukemia because of the relatively small market,
00:13:01.27 but they had other major kinases, like the epidermal growth factor receptor, or the platelet derived growth factor receptor,
00:13:09.08 and they had large libraries that they could run through robotic screens of their favorite tyrosine kinase.
00:13:15.06 So, for example, looking at the PDGF receptor, EGF receptor, they could run large library screens,
00:13:22.25 and end up with a lead inhibitor.
00:13:24.24 Now, this lead inhibitor was relatively impotent, not terribly potent,
00:13:29.25 and really not very specific.
00:13:31.24 But what they could then do with this lead compound is they could synthesize related compounds,
00:13:38.26 go back and screen against their favorite tyrosine kinase,
00:13:42.22 find some inhibitors that were better, some inhibitors that were worse.
00:13:46.25 But then by looking at what features allowed their inhibitors to be better or worse,
00:13:52.02 they could then go back and synthesize related compounds again.
00:13:55.19 And then by repeating this cycle several times,
00:13:58.29 they actually ended up with very potent and selective inhibitors of their favorite tyrosine kinases.
00:14:06.22 Now, in the original project, the kinase that they were looking at was a platelet derived growth factor receptor,
00:14:12.17 but as they optimized against the PDGF receptor, they carried along ABL inhibitory activity.
00:14:18.13 Then, in 1993, Nick Lydon from Ciba-Geigy sent me the best of their dual PDGF receptor ABL inhibitors
00:14:26.17 and our laboratory showed that the best compound they had at specifically killing CML cells
00:14:32.03 was this drug we now know as imatinib, or in the old days I knew it affectionately as CGP 57148.
00:14:39.25 It then picked up a name of STI 571 and interestingly enough,
00:14:45.03 it's Gleevec with two e's in the United States and Glivec with an i in the rest of the world.
00:14:50.21 And that's actually a pretty funny story about how that happened.
00:14:53.22 When Gleevec went to the Food and Drug Administration in the United States,
00:14:59.02 as well as all the similar regulatory agencies around the world, it went as Glivec with an i.
00:15:04.25 Now, as the story goes, the FDA wrote back to Novartis that they had a problem:
00:15:09.17 that Gleevec was going to get mispronounced as "gly-vec"
00:15:12.27 and that meant that it was going to get confused with a number of diabetes drugs that were on the market.
00:15:18.02 Like glyburide and others. So, Novartis wrote back to the FDA, how about if we spell it phonetically in the United States
00:15:25.00 with two e's, Gleevec. With two e's, the FDA accepted that
00:15:29.03 and despite that, Gleevec still made it through in less than six weeks,
00:15:33.13 and still holds the record for the fastest drug that made it through the FDA.
00:15:38.23 But, nonetheless, it's still Gleevec with two e's in the United States,
00:15:42.05 Glivec with an i throughout the rest of the world.
00:15:45.24 Now, we actually knew when we got this compound that we had a specific inhibitor of the BCR-ABL tyrosine kinase,
00:15:53.18 as well as all the other Abelson tyrosine kinases, as well as the PDGF receptor.
00:15:58.28 But my laboratory also showed that it inhibited the KIT tyrosine kinase
00:16:03.08 and when we talk about the gastrointestinal stromal tumors, you'll see exactly why that was important.
00:16:09.04 We also tested as many different kinases that we could get our hands on,
00:16:13.12 and showed that we had a very, very selective inhibitor, inhibiting only those three, but no others.
00:16:19.12 For example, EGF receptor, insulin receptor family, as well as a variety of other transmembrane receptors,
00:16:26.08 like FLT-3 and the CSF-1 receptor, and none of the intracellular tyrosine kinases, like the JAK family or the SRC family.
00:16:34.16 So, for those days and even now, even more so, this was a very selective inhibitor of tyrosine kinases.
00:16:41.26 Just to show you some of our pre-clinical data, this is an anti-phosphotyrosine immunoblot
00:16:49.25 that shows increasing doses of imatinib. So, we start with very low doses
00:16:55.03 and then move on to very high doses.
00:16:57.13 And you can see at the low doses, there's very little inhibition of tyrosine phosphorylation,
00:17:02.02 but you can see a very nice dose-dependent inhibition as we increase the dose of imatinib.
00:17:08.00 And this band up here is the BCR-ABL tyrosine kinase itself.
00:17:11.29 Now, just to show that we're not inhibiting the protein,
00:17:15.15 this is an ABL immunoblot that shows you that there's no inhibition of the protein expression.
00:17:21.08 So, we're seeing a specific effect on the kinase activity.
00:17:26.06 When we now go to cells, these are actually cells, either parental cells or cells that are engineered to express BCR-ABL,
00:17:34.07 and you can see that the parental cells and BCR-ABL cells, in the absence of imatinib, grow very, very nicely.
00:17:44.00 But if we now we incubate the BCR-ABL expressing cells in the presence of as little of 1 micromolar imatinib,
00:17:50.26 these cells are dead within two to three days.
00:17:53.23 And it's a specific effect on cells that express BCR-ABL.
00:17:57.18 If we now go to animal models, this is just a dose-finding study,
00:18:03.07 where we start with either placebo or increasing doses of imatinib injected as up to 50 milligrams per kilogram, once per day.
00:18:11.24 These are established tumors that we start with over here,
00:18:15.10 and we're just looking at whether we inhibit the growth of the tumors or not.
00:18:19.29 And you can see a very nice dose-dependent decrease in the growth of these tumors.
00:18:26.12 This is specific to BRC-ABL expressing tumors, we don't see this activity against other types of tumors,
00:18:32.15 for example v-SRC expressing tumors.
00:18:36.06 So, to summarize our pre-clinical data, imatinib is a potent selective inhibitor of the ABL, PDGF and KIT tyrosine kinases.
00:18:45.19 We can show that imatinib selectively kills BCR-ABL expressing cells,
00:18:50.08 both in vitro, as well in vivo in our animal models.
00:18:54.09 Now, as it turns out, the chemists at Novartis were able to make this into a highly bio-available oral formulation.
00:19:01.16 What that meant was that when we went to clinical trials, we actually had pills that we could use,
00:19:06.21 we didn't have to hook patients up to IV's.
00:19:09.05 Now, I wish I could tell you that everything went very smoothly,
00:19:12.21 and relatively quickly, but the reality is there were lots of hurdles to getting this drug into the clinic.
00:19:18.24 And there were many, many reasons that came up why imatinib should never be developed.
00:19:24.07 The first was lots of skeptics thought that kinase inhibitors would never work.
00:19:30.09 A kinase inhibitor had never been used in man, or humans, before,
00:19:34.14 so the view was that if we shut down some of these kinases, they probably aren't that critical of a target
00:19:40.23 and they probably won't work.
00:19:42.27 So, there was still a lot of skepticism about whether BCR-ABL was the right target.
00:19:47.20 The next concern was that these kinase inhibitors will be terribly toxic.
00:19:52.02 Although we had done a lot of profiling, a lot of preclinical toxicology,
00:19:56.14 there were some hints that we might see some toxicity in our human studies,
00:20:00.20 but because this type of agent had never been given to people,
00:20:04.16 there was a lot of concern about whether it might see a lot of toxicity.
00:20:07.21 But the biggest concern from a drug company was that these kinase inhibitors, at least for CML,
00:20:14.18 would never make enough money to justify their development.
00:20:18.05 So, if we think a drug development, it can cost as much as 500 million to 1 billion dollars
00:20:23.19 to make back, to develop a drug. And if you think about chronic myeloid leukemia, I said 5,000 cases per year in the United states,
00:20:33.19 that might never be enough to justify its development.
00:20:37.21 Now, as it turns out, all three of these things were absolutely wrong.
00:20:41.08 We'll talk about how well this drug worked, how non-toxic it was,
00:20:45.24 and in fact, because patients continue on this drug potentially life-long,
00:20:51.20 in fact, this drug is making quite a bit of money for Novartis,
00:20:55.08 as it turns out because the drug was patented when it came to my lab,
00:20:58.13 I don't see any revenue, either to my laboratory or to my university, but that's actually a separate story.
00:21:04.26 Once we were able to convince the drug company, though, and this took a couple of years,
00:21:10.11 we ultimately did begin phase I clinical trials in June of 1998.
00:21:15.00 Now, these were very typical phase I clinical trials.
00:21:19.18 It was a dose escalation starting with very, very low doses and then as we saw that these lower doses were safe,
00:21:26.27 we began to increase the dose stepwise and ultimately went up to doses of a hundred milligrams.
00:21:32.25 Now, as it turned out, the biggest capsule size we had was 100 milligrams,
00:21:37.09 so at 1000 mg, patients were taking 10 pills once a day,
00:21:41.17 and we almost got to sort of maximal pill tolerance per day.
00:21:45.15 But in terms of side effects, even at a thousand milligrams, we didn't reach a maximally tolerated dose,
00:21:52.11 which was actually the intent of the phase I trial, was to define a maximally tolerated dose.
00:21:58.14 But the reality is is that long before we reached a thousand milligrams,
00:22:03.24 we were seeing significant therapeutic benefits.
00:22:07.13 So, once we got to 300 milligrams per day and above,
00:22:10.15 we were seeing significant therapeutic benefits with minimal side effects.
00:22:15.19 And just to give you an example of what we were seeing,
00:22:19.07 in chronic phase patients who had failed standard therapy with interferon,
00:22:23.13 98 percent, in fact, that was 53 out of 54 patients, had their blood counts return to completely normal,
00:22:30.04 a so-called complete hematologic response, or CHR.
00:22:34.17 And with one year of follow-up, only one of those patients relapsed.
00:22:39.08 And in blast crisis patients, we've historically had a response rate of less than 5 to 10 percent,
00:22:45.20 to high-dose chemotherapy, we saw close to a 60 percent response rate.
00:22:51.06 Eighteen percent of those were durable at one year duration.
00:22:55.22 So, just to give you an example of our five hundred milligram dose cohort,
00:23:00.16 this is our chronic phase cohort. The line at 10 represents a normal white blood count.
00:23:06.17 And what you can see is that all of our patients come in with high white blood counts,
00:23:10.13 anywhere from 20 to 30, even up as high as a hundred thousand.
00:23:14.15 Within two to three weeks, their white blood count begins to fall, within four to six weeks, it's normal,
00:23:20.16 and it remains normal throughout the duration of therapy.
00:23:23.17 And this was typical and shows just how rapidly and reliable our responses were with this drug.
00:23:29.11 Now, it's said that when you develop a successful drug,
00:23:32.25 you really only need on patient to tell you that you have a successful drug.
00:23:37.18 And for me, this was my one patient. And this essentially is a before and after picture.
00:23:44.08 Now, this is a patient I had been following in my clinic for close to two years.
00:23:48.24 And I'd been adjusting his dose of a medication called hydrea for his CML
00:23:53.19 to try to dampen the amplitude of these waves in his white blood count.
00:23:59.21 Now, the reason I have these blood counts is the patient's wife is an accountant.
00:24:04.05 And she used to keep track of how well I was doing and she'd come in at this point
00:24:08.13 and she'd say to me "Dr. Druker, you're really not doing all that well at controlling my husband's blood counts."
00:24:15.05 And I'd look back at her and I'd say, well, I'm doing the best I can.
00:24:18.27 And in fact, it had taken me a couple of years to figure out how to control her husband's blood counts.
00:24:24.15 In April of 1999, we started this patient on 300 milligrams per day of imatinib.
00:24:30.17 Within three weeks, his blood count returned to normal and by December of that year, it remained normal.
00:24:36.18 And the reason I don't have blood counts out beyond here is the patient's wife finally conceded
00:24:41.11 that yes, Dr. Druker, you're doing ok at controlling my husband's leukemia,
00:24:45.08 and you can continue to manage him.
00:24:49.07 Now, as it turns out, I generally don't like anecdotes, and neither does the FDA.
00:24:54.07 So, we had to go to much larger phase II studies.
00:24:57.04 And in our phase II studies, these included chronic phase patients who had failed standard therapy of interferon,
00:25:03.05 as well as accelerated phase and blast crisis.
00:25:06.23 And pretty much the results that we saw in the phase II studies were the same was we saw in the phase I study.
00:25:13.10 So, in terms of return of blood counts to normal, in chronic phase patients who had failed interferon,
00:25:18.28 95 percent of patients had their blood counts return to normal.
00:25:23.01 In blast crisis, about half of the patients had their blood counts return to normal
00:25:27.26 and accelerated phase was somewhere in between.
00:25:31.08 So, now, these are obviously much larger numbers. We enrolled 450 patients in chronic phase,
00:25:37.28 and about another five hundred patients in accelerated phase and blast crisis combined.
00:25:43.03 When we also, if we looked at the disappearance of the Philadelphic chromosome,
00:25:48.10 a so-called major cytogenetic response where we can't detect the Philadelphic chromosome
00:25:53.07 or it's present in less than 35 percent of metaphases,
00:25:57.00 again, we saw very high responses. Sixty percent in chronic phase patients who had failed interferon,
00:26:03.27 and even 16 percent in blast crisis. And historically, it was incredibly rare to see a cytogenetic response,
00:26:11.23 or lowering of the number of metaphases that had the Philadelphic chromosome in a blast crisis patient.
00:26:17.06 So, it was pretty clear from these clinical trials that we had something that was unique.
00:26:21.18 Now, these phase II trials are what ultimately led to FDA approval of this drug,
00:26:27.16 in less than three years from our initiation of clinical trial and with a six-week review period at the FDA.
00:26:34.18 From there, we then moved into randomized studies in newly diagnosed patients.
00:26:41.22 So, the view in oncology is you start with the more advanced phases
00:26:45.10 and then move therapy up into newly diagnosed patients.
00:26:48.16 And this was a head to head comparison to standard therapy and it was an enormous trial.
00:26:54.06 A hundred and seventy-seven centers in 16 countries, over a thousand patients enrolled in about a six month period.
00:27:01.28 And when you think about it, 5,000 patients per year diagnosed in the United States with CML,
00:27:07.23 over a thousand patients enrolled in a 6 month period, that's close to 20 percent of all newly diagnosed patients,
00:27:15.05 and you can immediately tell why it had to go to so many centers in so many countries.
00:27:19.05 But this is still the most rapidly accrued clinical trial in leukemia in perhaps even any oncologic disease.
00:27:27.19 Half the patients got imatinib in 400 milligrams per day as a pill
00:27:31.27 and half the patients got interferon as a subcutaneous injection,
00:27:36.04 plus another drug called Ara-C.
00:27:38.27 Now, I'll spare you all the typical clinical trials detail,
00:27:42.05 and just tell you that these 553 patients assigned to either arm were well matched
00:27:48.14 for any possible prognostic factor you could imagine.
00:27:52.10 And I'll just give you the bottom line results.
00:27:54.16 So, if we just look at the return of blood counts to normal,
00:27:57.29 imatinib, almost everybody, 97 percent, interferon plus Ara-C, about 69 percent.
00:28:05.13 If we next look at the disappearance of the Philadelphic chromosome,
00:28:10.12 this is known as a complete cytogenetic response, no Philadelphic chromosome detected in any metaphases,
00:28:16.27 three-quarters of the patients randomized to imatinib achieved that level of response
00:28:22.05 and a pretty typical 14 percent of patients randomized to interferon.
00:28:26.24 So, imatinib, five times better.
00:28:30.05 If we now look at tolerance, how well was this drug tolerated,
00:28:34.08 one-third of patients randomized to interferon had to discontinue therapy.
00:28:38.29 And if you think about why, think about the worst case of a cold or flu you've ever had.
00:28:44.24 Fevers, chills, muscle aches, wanting to stay in bed all day, no energy, depression,
00:28:51.01 that's what happens to our patients on interferon and the reason is, is that when you get a cold or the flu,
00:28:58.16 your body produces interferon. And for patients with CML that were treated with interferon,
00:29:03.20 we're giving them a massive dose of interferon every single day for possibly the rest of their life.
00:29:10.14 But in contrast, imatinib, only three percent of patients discontinued therapy for side effects.
00:29:17.02 I should also mention that if you use statistics on all of these three findings,
00:29:21.02 all the p-values are less than 0.001.
00:29:25.01 Now, we have five years of follow-up on these patients, who initially received imatinib.
00:29:30.03 And if you look at their overall survival, the overall survival is 90 percent at five years.
00:29:36.02 If you just look at deaths specifically related to CML, the CML-specific survival is actually 95 percent.
00:29:45.08 So, earlier on, I told you that the median survival was somewhere around 5 years,
00:29:51.27 now we've completed changed the prognosis for this disease with a 90 percent or better five year survival.
00:29:59.10 If we now look at relapses or disease progression,
00:30:03.12 only six to seven percent of patients have progressed to accelerated phase or blast crisis.
00:30:10.00 Again, historically, the average duration of the chronic phase was around four to five years,
00:30:16.10 and now, at five years, 93 percent of patients have not gone into accelerated phase or blast crisis.
00:30:24.17 Overall, about 17 percent of patients have had some indication of disease progression.
00:30:30.19 This includes the seven percent of patients who progressed to accelerated phase or blast crisis,
00:30:36.01 but also includes patients who've had their blood counts become abnormal again,
00:30:40.23 and patients who've lost their complete cytogenetic response.
00:30:44.29 But, one of the things you really can't glean from the yearly relapse curves
00:30:50.13 is the fact that the yearly risk of relapse is actually decreasing over time.
00:30:55.22 So, if you look, there's actually a peak of relapses in the first and second year
00:31:00.19 for either progression to accelerated phase or blast crisis, or any event, including loss of a complete cytogenetic response,
00:31:10.05 or loss of complete hematologic response.
00:31:12.23 But in year three, the risk of relapse goes down, year four and five it continues down,
00:31:18.17 and so that in year five, it's less than one percent overall.
00:31:23.21 And what that means, and what we hope if this trend continues that means,
00:31:28.27 that the duration of this disease will be significantly prolonged if patients make it beyond the, say the second or third year,
00:31:36.04 the risk of relapse really becomes quite negligible,
00:31:39.09 and it may well mean that our relapse curves at five or ten years may actually begin to plateau,
00:31:44.24 and that would certainly be very good news for patients with this disease who are on therapy currently.
00:31:50.11 So, to summarize our chronic myeloid leukemia trials,
00:31:54.17 we have a drug that yields a very high response rate with minimal toxicity in all phases of CML.
00:32:01.17 Our chronic phase patients are achieving very durable responses,
00:32:05.10 with a close to 90 percent five year survival.
00:32:08.28 But as you saw in advanced phase, resistance is relatively common
00:32:13.03 and that is certainly one of the problems that needs to be addressed.

Part 2: Imatinib Resistance and Other Diseases Targeted by Imatinib

00:00:02.13 I'd now like to talk about why imatinib resistance occurs and then also talk a little bit about some other diseases
00:00:08.23 that are targeted by imatinib.
00:00:10.16 So, to summarize from my first part of the talk, I noted that imatinib yields very high response rates with minimal toxicity
00:00:19.09 in all phases of CML. In chronic phase patients, which comprise the majority of patients with CML,
00:00:25.28 responses are quite durable and relapses only occur in somewhere between 10 to 15 percent of patients at five years.
00:00:33.14 But in advanced phase, resistance is quite common and, in fact, a major problem.
00:00:39.18 So, just to remind you, these are relapses and disease progression in our chronic phase patients.
00:00:45.03 So, in our chronic phase patients, about 7 percent of patients have progressed to accelerated phase or blast crisis,
00:00:51.01 and overall about 17 percent of patients, including those 17 percent of patients who progress to accelerated phase or blast crisis
00:00:59.23 have relapsed at five years.
00:01:01.26 But it's a very different story in patients with more advanced disease.
00:01:06.03 So, at four years, if patients have failed interferon therapy, about a quarter have relapsed,
00:01:11.09 and at four years, about close to 75, or three quarters of patients with accelerated phase of relapse,
00:01:17.12 and in blast crisis, it's 95 percent,
00:01:20.06 and in fact, the reality is in blast crisis, 80 percent will relapse within the first year,
00:01:25.13 and about 85 to 90 percent by the end of the second year.
00:01:29.13 So, clearly in the advanced phase of this disease, relapses are quite common.
00:01:35.06 Now, as I mentioned, if we're going to understand what to do about relapses,
00:01:40.19 much like understanding the cause of chronic myeloid leukemia,
00:01:44.12 we have to understand the cause of relapse.
00:01:47.17 So, we have to start by asking the question: why do some patients relapse?
00:01:52.11 So, we've developed imatinib, a specific inhibitor of the BCR-ABL kinase.
00:01:58.04 So, the first question I would ask is, in a patient who relapses, is the kinase still inhibited or not?
00:02:05.12 Because if the kinase is inhibited and the patient relapses, it has to mean that now there's additional mutations besides BCR-ABL
00:02:14.13 that are driving the growth of the malignant cell.
00:02:17.09 If, however, the kinase is no longer inhibited,
00:02:21.23 it has to mean that somehow the drug is either pumped out of the cells by a drug efflux pump,
00:02:27.14 maybe the target's been amplified or mutated
00:02:30.22 or perhaps the drug is metabolized, bound to proteins, or maybe some other mechanism.
00:02:36.06 So, how are we going to distinguish between these two?
00:02:39.14 Well, we know that BCR-ABL functions as a tyrosine kinase
00:02:44.16 and what does a tyrosine kinase do?
00:02:46.06 It puts phosphate onto tyrosine residues of specific substrate proteins.
00:02:51.04 As it turns out, the most heavily tyrosine phosphorylated protein in a CML patient is BCR-ABL itself.
00:02:57.26 But BCR-ABL is an extremely unstable protein and in fact, if we try to isolate it from the CML patient,
00:03:04.13 we fail. But this other protein, called CrkL, it turns out to be a much better marker.
00:03:10.13 So, what is CrkL? CrkL is an SH2/SH3 domain adapter protein.
00:03:16.13 Actually, in 1994, our laboratory showed it was the most heavily tyrosine phosphorylated protein in CML patient samples.
00:03:25.08 We then went on to show that it's a direct substrate of BCR-ABL,
00:03:28.26 in fact, one of the few times in which CrkL is tyrosine phosphorylated is in cells that express BCR-ABL.
00:03:37.26 We've also gone onto show that in CrkL null animals, BCR-ABL does not transform those cells quite as well.
00:03:45.07 So, in fact, it's a required substrate for BCR-ABL transformation.
00:03:49.29 But we use the fact that it's so heavily tyrosine phosphorylated in CML patients
00:03:55.04 to turn this into an assay. So, in this assay we're looking at CrkL phosphorylation.
00:04:01.11 And like most phosphoproteins, CrkL migrates as a doublet.
00:04:06.04 So, the phosphorylated form migrates here, slightly slower, because it's phosphorylated, more heavily charged,
00:04:12.22 and has a higher molecular weight, so it migrates a little bit more slowly.
00:04:17.11 When CrkL is unphosphorylated, because of less charge, lower molecular weight, it migrates a little bit faster, right there.
00:04:25.17 So, in our clinical trials, we assessed the status of CrkL phosphorylation, either prior to therapy, meaning here,
00:04:33.23 where's it's mostly tyrosine phosphorylated,
00:04:36.06 or one day into therapy, and we can see that even one day into therapy,
00:04:40.26 most of the CrkL is unphosphorylated.
00:04:44.08 Now, we can then use this assay to see, is CrkL phosphorylated when a patient relapses?
00:04:51.11 And this is some work from Charles Sawyers's laboratory, showing a typical patient who responds and then relapses.
00:04:58.23 So, at diagnosis, this is in fact a blast crisis patient,
00:05:02.27 most of the CrkL migrates as this higher form, the phosphorylated form.
00:05:08.19 Five days into therapy, most of the CrkL is now unphosphorylated,
00:05:13.10 this patient in fact had a very transient response, and at day 22, most of the CrkL is again tyrosine phosphorylated.
00:05:21.08 And this has been the typical pattern of patients who respond and then relapse.
00:05:25.15 CrkL is unphosphorylated when patients respond,
00:05:29.02 and then re-phosphorylated, consistent with re-activation of the BCR-ABL tyrosine kinase at relapse.
00:05:36.03 So, that suggests that it's elements in this category.
00:05:39.25 The BCR-ABL kinase is no longer inhibited relapse, so it must mean amplification, mutation or some of these other categories.
00:05:48.03 Now, as it turns out, this category, BCR-ABL kinase mutations,
00:05:52.27 the target of imatinib is mutated, turns out to be the major cause of relapse in patients who respond and then relapse.
00:06:01.11 So, this slide, although it's a little bit complicated, shows you the ABL kinase domain,
00:06:07.11 from the N-terminal of the kinase domain,
00:06:09.10 to the C-terminus of the kinase domain, with the residues that have been mutated.
00:06:15.01 And what you can see, there's a cluster of mutations here,
00:06:18.14 in what's called the P-loop, or ATP-binding loop, and this just gives you an idea of their frequency
00:06:23.26 at which they're identified. There are other mutations scattered throughout the kinase domain.
00:06:28.29 There's one very common one here that you'll hear a fair bit about, this called threonine 315-I.
00:06:35.20 And then others here down in the catalytic inactivation loop.
00:06:39.13 So, now the first question you'd ask is ok, so we see kinase domain mutation.
00:06:44.11 Are they really resistant to imatinib?
00:06:47.08 So, what we've done is we've taken all of these different mutations that have been observed in patients
00:06:52.06 and expressed them in cells and looked at their sensitivity to imatinib.
00:06:56.04 So, here, first of all, just to orient you, this is imatinib and what you can see is, imatinib with unmutated, or wild type, BCR-ABL,
00:07:06.26 very sensitive to cell killing, IC 50, meaning the concentration that inhibits 50 percent of the cell proliferation,
00:07:15.03 of about 250 nanomolar, by one micromolar, all these cells are dead.
00:07:21.05 But, now you can see all these mutations and the curves have shifted to the right.
00:07:25.08 And you can see that there's variable sensitivity.
00:07:28.17 There's this mutation here, this is M351T, which is only slightly insensitive,
00:07:35.17 and it suggested that if you increase the dose of imatinib, you may see responses if you had that mutation.
00:07:41.29 Here are all the P-loop mutations, and you can see they're relatively insensitive
00:07:47.19 and then up here is this T315I, which is completely insensitive.
00:07:53.17 So, clearly these mutations are resistant to imatinib and they are driving the resistance that's observed in patients.
00:08:01.21 Now we need to know, is, ok, how does this resistance, what does this resistance look like?
00:08:09.09 And the reality is, we've been aided by the crystal structure of imatinib,
00:08:13.10 this was done by John Kuriyan's lab at UC-Berkeley.
00:08:16.24 And here what you can see in green is the imatinib structure.
00:08:20.18 So, there's imatinib, and in orange are amino acids that contact.
00:08:25.28 And here is one in particular, 315. Threonine 315, you can see forms a hydrogen bond with imatinib.
00:08:34.23 So, you can imagine that if you mutated any of these direct contact sites,
00:08:39.18 you would have imatinib resistance.
00:08:41.29 And let's look a little bit more carefully at how this threonine 315 to isoleucine substitution works.
00:08:48.19 So, here is threonine with imatinib. So, here's imatinib, and here's threonine,
00:08:57.10 at 315 and you can see it stacks up, makes this hydrogen bond.
00:09:01.24 Now, imagine that we mutate this to a bulkier isoleucine residue,
00:09:06.27 and you can see that there's a steric clash, it's too bulky, it goes right into the binding pocket and imatinib can't bind.
00:09:14.28 What about other residues?
00:09:17.23 And it turns out the mechanism of some of the other residues are a little bit more complicated.
00:09:22.21 So, what I want to talk about next are the P-loop mutants.
00:09:26.05 And here I've just shown you a crystal structure of the P-loop,
00:09:29.12 and here you can see the P-loop with an ATP analogue.
00:09:33.04 And notice how vertical these two structures stack.
00:09:40.04 So, the P-loop, the ATP analogue, and they line up just perfectly and they're vertical.
00:09:46.21 Now, let's look at what happens when imatinib binds at this P-loop.
00:09:50.29 So, remember, here's our P-loop here, this is what it's supposed to look like.
00:09:55.10 When imatinib binds, it kind of looks like a chair.
00:09:59.05 And imatinib, which is here, binds at this P-loop and causes this massive conformational change.
00:10:06.21 And so, the reality is that this P-loop has a lot of flexibility and accommodates imatinib binding.
00:10:13.23 But, look at the residues, right at the entrance,
00:10:16.01 253, 255. If those two residues are mutated,
00:10:20.24 this P-loop presumably doesn't have the flexibility to adopt a conformation to allow imatinib binding
00:10:28.23 and therefore, that's why imatinib resistance occurs with some of these P-loop mutants.
00:10:34.17 Now, the last set of mutants requires slightly more complicated explanation
00:10:39.16 and I have an animation here that will show you the conformation.
00:10:52.22 So, if you look at the activation loop, it has two conformations:
00:10:55.18 closed, and open. We'll do that again.
00:10:59.13 Closed. Imatinib binds now.
00:11:03.09 And open. So, you can see that this activation loop is quite flexible.
00:11:09.10 It can go back and forth, from open to closed, and as we'll see, this has a huge impact on the ability of imatinib to bind.
00:11:19.09 So, I'm going to go to a static set of diagrams.
00:11:22.07 So, here's imatinib. And remember, here's our activation loop: open and closed,
00:11:27.28 just like you saw in that animation. Now, here's imatinib.
00:11:31.05 And you can see what imatinib does is it hangs down over this activation loop
00:11:36.15 and actually locks it in this off conformation.
00:11:39.27 Now, if you could imagine, if a mutation favored this open conformation,
00:11:47.23 where the activation loop is here in this open, active conformation,
00:11:52.02 there's this part of imatinib here that would clash and couldn't bind.
00:11:57.19 And that's presumably why mutations in the activation loop or other mutations that favor this open or active conformation,
00:12:05.27 cause imatinib resistance. So, now we understand imatinib resistance,
00:12:10.25 we understand mutations, we understand the structural basis of these mutations.
00:12:15.21 So, now it's time to do something about it.
00:12:18.11 And the reality is, looking at these crystal structures, there're two things we can do about it.
00:12:23.09 First of all, we can look at the structure of imatinib and we can actually modify the imatinib structure,
00:12:30.03 and make it bind even more tightly than it does.
00:12:33.10 Or, we could think about a completely different structural class.
00:12:38.00 Maybe a class that's a little bit smaller, maybe that doesn't care whether this activation loop is closed or open.
00:12:47.03 And maybe that kind of a compound might inhibit some of the common mutations.
00:12:52.01 Now, as it turns out, two major drug companies have taken both of these approaches.
00:12:56.10 Modify imatinib, come up with a distinct structural class.
00:13:00.20 So, one of the companies, Novartis, which owns imatinib, modified the imatinib structure to allow tighter binding,
00:13:08.19 and the drug they developed is called Nilotinib, or tasigna or historically AMN107.
00:13:14.13 The second company that got into this is Bristol-Meyers-Squib,
00:13:18.21 which developed Dasatinib, or sprycel, historically was called BMS-354825.
00:13:25.14 This comes from a structurally distinct category of inhibitor
00:13:29.08 that also inhibited the SRC tyrosine kinase.
00:13:32.07 As it turns out, there are fewer structural constraints to binding,
00:13:35.28 but because of that, it inhibits more kinases than imatinib or even nilotinib.
00:13:41.28 So, let me take you through some of the data for these compounds.
00:13:45.20 Or, actually, turn out to both to be more potent ABL inhibitors than imatinib.
00:13:50.00 So, here's nilotinib as compared to imatinib.
00:13:53.16 And you can see there's the core structure shared between imatinib and nilotinib.
00:13:58.21 They share this part of the molecule, but sort of this back half is structurally distinct,
00:14:05.02 whereas the front half is identical.
00:14:07.10 So, modification of the imatinib structure allows tighter binding.
00:14:11.13 So, now we can look at, how does this compound do against cells that express the imatinib resistant mutations?
00:14:18.21 So, here in this dotted line, this is imatinib against unmutated BCR-ABL.
00:14:26.02 And you can see nilotinib against unmutated is way over here,
00:14:30.18 way over here to the left, with nanomolar potency.
00:14:33.11 And virtually all the mutants fall on this line, or below, with one exception: T315I.
00:14:41.03 It still makes that direct contact, so it's still not inhibited by nilotinib.
00:14:47.11 What about Dasatinib? As, I mentioned, it's a structurally distinct category and this just shows you the structure of Dasatinib.
00:14:55.29 And here is the cell proliferation data. As you can see, very low, nanomolar potency
00:15:03.24 against all the imatinib resistant mutations, again, one exception, T315I,
00:15:09.28 and again, because it's at the ATP binding loop and because it also makes a direct contact,
00:15:16.05 you could imagine why T315I would also be resistant to Dasatinib.
00:15:23.08 Both of these drugs have gone into phase I and phase II clinical trials in patients with imatinib resistance,
00:15:29.29 and both have achieved very high levels of response in patients with imatinib resistance.
00:15:35.03 So, here, return of blood counts to normal in imatinib-resistant, or intolerant, patients,
00:15:40.07 between 84 to 100 percent. Very much like imatinib achieved in chronic phase patients,
00:15:46.00 where we saw close to 90 to 98 percent of patients achieving return of their blood counts to normal
00:15:51.25 and as far as loss of the Philadelphic chromosome,
00:15:55.01 or decrease in the number of metaphases with the Philadelphic chromosome,
00:15:58.11 again, between 50 to 80 percent of patients with imatinib resistance treated with Dasatinib.
00:16:04.16 Very similar results have also been observed with nilotinib.
00:16:09.24 And in fact, Dasatinib is already FDA approved for patients with imatinib resistance.
00:16:15.16 So, to summarize the Dasatinib and nilotinib trials,
00:16:20.02 we've seen significant activity in patients with imatinib resistance,
00:16:24.11 activity is observed in all imatinib resistant mutations except T315I
00:16:31.00 and much like the imatinib clinical trials, where relapses were common in advanced phase patients,
00:16:36.04 again with Dasatinib and nilotinib, the best responses are observed in the chronic phase patients,
00:16:42.00 good responses are also observed in advanced phase, but relapses are quite common.
00:16:47.03 T315I turns out to be the major cause of resistance or relapse in both of these, with both of these agents.
00:16:54.03 However, there are some other causes.
00:16:57.09 So, to summarize imatinib resistance, relapses are mostly due to ABL kinase domain mutations.
00:17:05.19 There are now two novel ABL inhibitors and several others in earlier clinical development
00:17:10.29 that have significant activity against most of the imatinib resistant mutations.
00:17:15.23 The one exception to that is T315I remains insensitive to all current inhibitors,
00:17:22.09 but fortunately for patients, about the 10 percent or so of patients or less that harbor that mutation,
00:17:27.15 there are already numerous pre-clinical compounds
00:17:30.10 and it's likely sometime in 2007 that one or more of these compounds will enter clinical trials.
00:17:36.28 I'd now like to talk about some other diseases that are targeted by imatinib.
00:17:43.03 Two examples are gastrointestinal stromal tumors, which are driven by KIT mutations,
00:17:50.01 and hypercosinophilic syndrome, which is driven by PDGF receptor rearrangements
00:17:55.11 and in fact, there are a few other rare diseases that are targeted by imatinib that I won't cover.
00:18:00.16 I want to start, however, with imatinib and gastrointestinal stromal tumor.
00:18:05.22 Gastrointestinal stromal tumors are intestinal sarcomas.
00:18:10.28 They actually used to be called intestinal leiomyosarcomas.
00:18:15.12 As it turned out, when patients with leiomyosarcomas were lumped into clinical trials,
00:18:21.22 the patients with uterine leiomyosarcomas, which are the common ones,
00:18:25.20 responded very well to chemotherapy, but patients with these intestinal leiomyosarcomas didn't.
00:18:31.25 As it turned out, intestinal leiomyosarcomas are KIT positive, whereas uterine leiomyosarcomas aren't.
00:18:39.23 So, it was very clear that these were clinically and pathologically distinct entities,
00:18:45.03 so they were reclassified as gastrointestinal stromal tumors.
00:18:49.21 The annual incidence of gastrointestinal stromal tumors are just, is estimated at about 5,000 cases per year.
00:18:56.17 As I've mentioned, the response to chemotherapy with this gastrointestinal stromal tumors with very high dose chemotherapy,
00:19:06.05 or multi-agent chemotherapy is less than five percent,
00:19:09.08 and in fact, may be closer to 1 in 200 and historically,
00:19:13.17 if you saw a response to chemotherapy with a gastrointestinal stromal tumor, it was almost worthy of a case report.
00:19:20.10 Now, in 1998, a Japanese group reported that activating KIT mutations
00:19:26.06 were present in the majority of patients with GIST.
00:19:29.08 Now, remember, that paper was published January 1998, I was very excited about this
00:19:33.27 because we knew from our pre-clinical profiling that imatinib inhibited KIT.
00:19:38.28 And it seemed to me that this would be a very useful tumor in which also to test imatinib,
00:19:45.01 besides chronic myeloid leukemia. So, I called up the drug company and told them that what I thought was very good news.
00:19:51.16 The first question they asked me was, well, how many cases of gastrointestinal stromal tumor are there?
00:19:56.20 And in those days, it was probably close to maybe 250 to a thousand at the most,
00:20:02.14 and no big surprise, there was dead silence on the other side of the telephone.
00:20:07.16 I just identified another, from 5000 to maybe 6000, at best, patients,
00:20:12.12 and it still wasn't going to make it over the hurdle of a profitable drug.
00:20:16.16 But fortunately, a number of investigators, including George Demetri and myself,
00:20:21.00 were able to convince the drug company to expand their clinical trials to gastrointestinal stromal tumor
00:20:26.20 and once again, rapid, dramatic responses were seen.
00:20:31.02 So, a disease that historically had a response rate of less than five percent,
00:20:35.14 with imatinib, 54 percent. And another 28 percent of patients had stable disease.
00:20:41.15 And these patients also derived a survival benefit.
00:20:44.22 So, we're completely reversing the prognosis for patients with this disease with a simple, oral therapeutic that they had to take once a day.
00:20:52.20 Again, response is rapid and dramatic.
00:20:57.10 This is the pre-therapy PET scan, and you can see massive uptake in this liver tumor, pre-therapy.
00:21:05.14 One month into therapy, the PET scan is completely cold.
00:21:09.11 That mass you see over here, no longer visible on this PET scan.
00:21:13.12 The only uptake you see is in heart, kidneys and bladder, which is just physiologic uptake.
00:21:18.17 Again, rapid, dramatic responses to a targeted agent.
00:21:23.02 Another disease where activity has been seen is a disease called idiopathic hypereosinophilic syndrome.
00:21:31.04 This disease is characterized by prolonged eosinophilia
00:21:35.09 defined as greater than fifteen hundred eosinophils per mil, per microliter, in the blood.
00:21:42.26 You also have to exclude other etiologies of an elevated eosinophil count, like parasitic diseases and allergies,
00:21:49.16 and a variety of other diseases that can cause an elevated eosinophil count,
00:21:55.11 plus, you have to see some evidence of end-organ damage.
00:21:58.14 So, for example, heart damage, lung damage, brain or infiltration of the skin.
00:22:04.07 And some of these patients with the infiltration of the skin and other organs can be quite debilitated.
00:22:10.25 Now, as it turned out, there were some anecdotal reports of incredible, dramatic responses to actually quite low doses,
00:22:20.21 one hundred milligrams per day of imatinib.
00:22:23.15 Now, the reality is, that imatinib had been tried empirically.
00:22:27.08 The reality is the first investigator who tried imatinib in a patient with hypereosinophilic syndrome
00:22:33.17 did it because historically interferon had had some activity,
00:22:37.01 and because historically interferon had some activity in chronic myeloid leukemia,
00:22:40.17 this investigator thought that he'd also try imatinib, and it worked.
00:22:44.26 But they didn't, nobody knew what the cause of hypereosinophilic was.
00:22:48.28 But it suggested that because we knew how imatinib worked,
00:22:52.08 it targets ABL, KIT, PDGF receptor, that hypereosinophilic syndrome had to be caused by a kinase that was inhibited by imatinib.
00:23:02.23 So, from some work with Jan Cools and Gary Gilliland's lab, they showed that on chromosome four,
00:23:10.19 where the PDGF receptor resides, there's an interchromosomal deletion
00:23:16.00 that truncates the PDGF receptor and fuses it to this gene of unknown function called FIP1L1.
00:23:24.18 And what you end up with is this fusion gene that truncates and activates the PDGF receptor.
00:23:32.13 And we know that the PDGF receptor is a good target for imatinib and by shutting down the PDGF receptor,
00:23:39.19 that drives the growth of cells in hypereosinophilic syndrome,
00:23:43.09 rapid, dramatic responses were seen. And in fact, the PDGF receptor in this fusion protein
00:23:50.20 is a little bit more sensitive to imatinib than BCR-ABL,
00:23:54.15 which explains why slightly lower dose therapy actually works in patients with this disorder.
00:23:59.23 So, what this example tells us is that an understanding of the molecular targets of a drug can lead to insights into molecular pathogenesis
00:24:08.09 and this example I just gave you is actually a reverse of how imatinib was developed.
00:24:13.13 So, with imatinib, we understood the cause of a disease, develop a drug and took it into clinical trials.
00:24:18.27 Here, we had an empiric clinical trial that led us to the insights into what caused the disease.
00:24:27.07 And what that also tells you is that if you have a responding patient
00:24:30.18 with a rapid, dramatic response, that patient's trying to tell you something.
00:24:34.22 And what you need to do is you need to study that patient so you can understand why the response occurs,
00:24:40.06 so that you can extend that discovery to many, many more patients.
00:24:46.00 So, with this part of my talk, I hope I've given you some insights into why resistance occurs to imatinib,
00:24:51.25 but also some of the other diseases where imatinib has worked,
00:24:55.20 and how we can extend the paradigm by careful studies of responding patients
00:25:00.08 and extending this paradigm, and examining responding patients, will be the topic of the third part of my talk. Thank you.

Part 3: Extending the Imatinib Paradigm

00:00:02.22 I'd like now, in the third part of my talk, to discuss how we can extend the imatinib paradigm to other diseases.
00:00:10.19 We've heard about how imatinib results in remarkable and durable responses in patients with chronic myeloid leukemia,
00:00:17.26 as well as gastrointestinal stromal tumor and a of couple of other diseases,
00:00:21.28 but I'd really like to delve into in this part of my talk is what lessons have we learned from the development of imatinib,
00:00:28.11 and how can that be applied to other diseases, and particularly other cancers.
00:00:32.12 So, let's start with where has imatinib worked?
00:00:35.10 It's worked in tumors in which ABL, KIT or PDGF receptor have a critical role in the growth and survival of the cancer.
00:00:42.18 So, where is that? In chronic myeloid leukemia, we're targeting the BCR-ABL oncogene,
00:00:47.13 in gastrointestinal stromal tumor, we're targeting a mutated KIT,
00:00:52.02 and in hypereosinophilic syndrome, we're targeting again a fusion to the PDGF receptor alpha, that activates it.
00:00:58.24 So, in each of these tumors, we have an example of one of the targets of imatinib being critical for the growth and survival of the cancer
00:01:06.12 and if you target it, you'll see rapid, dramatic and durable responses.
00:01:11.21 But imatinib actually has been tried in lots of other tumors.
00:01:14.17 Brain tumors, breast cancer, prostate, melanoma, and numerous others.
00:01:19.03 And in these clinical trials, which have been a little bit more empiric, imatinib has not worked very well, if at all.
00:01:26.06 And the reason presumably that it hasn't worked is that despite the fact that many of these tumors express targets of imatinib,
00:01:34.02 they're not critically dependent on one of those targets.
00:01:37.26 So, what lessons can we learn from the clinical trials of imatinib?
00:01:43.06 And we need to start first with target expression and how does that relate to response.
00:01:50.07 Now, I want to start with a hypothetical clinical trial.
00:01:53.27 And in our hypothetical clinical trial, we're going to enroll one hundred patients.
00:01:58.10 We're going to put one caveat on our clinical trial. We're going to vary the target frequency.
00:02:05.01 So, in our first group of patients, all of our patients will express the target, and in our last group of patients,
00:02:11.26 only ten percent. So, we'll vary the frequency at which our target of our drug is expressed.
00:02:17.25 We'll set the target response rate at sixty percent.
00:02:22.19 So what you can see is that in the patients, this group of patients, our response rate's sixty percent, and
00:02:29.15 and in this group of patients, our response rate is only 6 percent.
00:02:33.11 Now, in this clinical trial, we'd congratulate ourselves for developing a very successful agent,
00:02:38.17 and in this group of patients, we'd say that this agent wasn't very effective.
00:02:43.11 But the reality is is that in this clinical trial, we were very smart. We selected the right patients.
00:02:50.23 In this clinical trial, we weren't very smart, and we didn't select the right patients.
00:02:55.26 But the reality is that our response rate was always sixty percent. It's just whether you selected the right patients,
00:03:02.02 or the wrong patients. Now, does that have any bearing on reality?
00:03:07.28 And if we look at imatinib responses in advanced malignancies,
00:03:12.08 we have a situation with BCR-ABL in CML and gastrointestinal stromal tumor with KIT,
00:03:20.03 where all the patients enrolled in our clinical trial express a target of imatinib.
00:03:25.06 And our response rate is between fifty to sixty percent,
00:03:28.14 in advanced cancers, so blast crisis or metastatic gastrointestinal stromal tumor, a fifty to sixty percent response rate.
00:03:36.26 So, in this example, it looks like expression of a molecular target correlates with a response to an agent directed against that target.
00:03:46.29 But we really need to delve a little bit more deeply and ask a more pertinent question:
00:03:53.05 is expression really sufficient to predict responses?
00:03:57.15 So, we're going to go back to our hypothetical clinical trial.
00:04:00.25 But here, we're going to say, take our hundred patients and we're going to have all of them expressing our target.
00:04:07.03 So, a hundred percent of our patients express the target.
00:04:09.21 We're actually going to even bump our response rate a little bit.
00:04:13.12 So, we'll set our response rate at 80 percent.
00:04:16.06 But we're going to put one condition on whether patients respond or not.
00:04:20.29 They have to have some evidence that the target is activated,
00:04:24.13 in other words, that it's critical to the growth and survival of the tumor.
00:04:28.00 And we'll vary that from 90 percent down to 10 percent.
00:04:32.00 Now, let's look at what effect that has on our response rate.
00:04:35.06 In this group of patients, the response rate is 72 percent, and in this group of patients the response rate is 8 percent.
00:04:43.03 Now, in this clinical trial, we'd say that, first of all, we'd say this is a very successful agent, of course,
00:04:50.02 but we'd say that target expression correlates with response.
00:04:55.04 And in fact, we're wrong. We're wrong because it's actually target activation.
00:05:01.10 It looks like target expression because almost all the patients that express the target express an active target.
00:05:08.09 Now, in this example, there are a couple of things that we'd conclude.
00:05:12.08 First of all, we'd conclude that target expression doesn't correlate with response.
00:05:17.06 And in fact, we're right. Target expression, all have it, only eight percent respond.
00:05:22.05 So, target expression doesn't correlate with response.
00:05:25.11 But we'd also conclude that this agent has no activity, or virtually no activity, but we're wrong.
00:05:31.28 It has 80 percent response rate, we just selected our patients wrong.
00:05:36.24 So, I'm actually going to show you examples of both this first example and this last example,
00:05:42.19 of high response rate with all the patients expressing, target having, target activation.
00:05:49.13 And examples here, where all the patients expressed the target, response rates are low,
00:05:54.18 but we just need to understand what this ten percent of patients are.
00:05:58.22 So, let's start with gastrointestinal stromal tumors.
00:06:02.10 The most common KIT activating mutation is this exon 11 mutation.
00:06:08.00 If you have it, your response rate is 80 percent. In contrast, if you express wild-type KIT, no mutations, and
00:06:17.03 you express the target, but it doesn't have any mutations, your response rate is under 20 percent.
00:06:23.01 So, this was work done by Mike Heinrick in Oregon and Jonathan Fletcher in Boston.
00:06:27.01 So, here it's absolutely clear that mutational status predicts response.
00:06:33.07 You have mutated KIT, 80 percent response. No mutations in KIT, very low response.
00:06:40.19 So, here what this example tells us is that expression of a target doesn't guarantee response to an agent that modulates that target.
00:06:49.08 You need something else, you need some evidence of activation,
00:06:53.02 some evidence that the target is actually critical to the growth and survival of the tumor.
00:06:58.04 But for those of you that are listening carefully, when you looked at this example,
00:07:05.04 it wasn't in fact a zero percent response in these patients that expressed the wild-type KIT.
00:07:11.00 And in fact, some of these patients in this sixteen to eighteen percent that are responding,
00:07:17.26 no mutations in KIT, but rapid, dramatic responses compared to all these other patients that didn't respond.
00:07:24.20 So, what Mike and Jonathan did is they went on to look at these patients a little bit more carefully.
00:07:29.26 And what they found was that in fact, even though these patients had wild-type KIT,
00:07:35.25 about one-third of them had PDGF receptor activating mutations.
00:07:41.13 Now, we know that imatinib is capable of inhibiting the PDGF receptor
00:07:46.07 and in fact, one of these sets of mutations was imatinib sensitive.
00:07:50.21 And two-thirds of these patients had good responses.
00:07:53.24 So, now what we've learned is that in some patients who have wild-type KIT,
00:07:58.28 they have PDGF receptor activating mutations that are sensitive to imatinib that are driving the growth of these gastrointestinal stromal tumors
00:08:08.27 and that's why they respond.
00:08:10.24 So, what about another example?
00:08:13.25 Gefitinib and Erlotinib, also known as Iressa and Tarceva.
00:08:17.20 These two inhibitors inhibit the EGF receptor kinase.
00:08:21.19 Now, as it turns out, the EGF receptor kinase is broadly expressed in cancer.
00:08:25.22 These agents were tried in advanced non-small cell lung cancer,
00:08:30.11 all of which expressed the epidermal growth factor receptor.
00:08:33.26 But the response rate was disappointingly low, only ten to twenty percent.
00:08:39.05 But, as it turns out, there's no correlation, I said all, virtually all, the patients with non-small cell lung cancer express EGF receptor,
00:08:47.14 but only ten to twenty percent respond.
00:08:50.05 But hidden in this clinical trial was a clinical finding.
00:08:53.27 These ten to twenty percent of patients who were responding were having very rapid, very dramatic and relatively durable responses.
00:09:01.06 And the patients who had these responses were females that never smoke,
00:09:06.01 and had this slightly unusual histology, which was known as a bronchoavelolar histology.
00:09:11.11 Well, as it turns out, a number of laboratories took that clinical finding, some patients were having rapid, dramatic response,
00:09:19.02 and asked, why does that occur?
00:09:21.02 And what they found was that responding patients had EGF receptor mutations.
00:09:26.29 As it turns out, these EGF receptor mutations are more sensitive to inhibitors than the wild-type receptor.
00:09:34.17 So, both of these examples, the gastrointestinal stromal tumor example and EGF receptor inhibitor example
00:09:41.08 tell you that careful study of subsets of patients reveal important insights.
00:09:46.24 So, again, going back to one of the early points of my talk,
00:09:51.06 you see an effect in patients, you need to go back to the lab to study why that occurs
00:09:57.20 so you can learn how to design your trials even better, identify responding patients, enroll those in your clinical trials,
00:10:04.15 and now you can identify the right 100 patients to enroll in your clinical trial and get a 68 percent response rate.
00:10:12.06 So now we're in the position to ask a relatively broader question:
00:10:17.25 what if the response to a molecularly targeted agent is low?
00:10:21.14 The first question I want to ask: is the target even expressed?
00:10:26.00 The next question I want to ask: is the target modulated by your agent?
00:10:30.25 If you have an EGF receptor or an ABL inhibitor and you're not shutting down your target, you'd never expect to see a response.
00:10:38.01 But if your target's expressed and your target's modulated by your agent,
00:10:42.20 then you have to ask a more critical question: is this target really as good as you think it is?
00:10:48.07 Is the target critical to the growth and survival of the tumor?
00:10:52.05 Or, is there a subset of patients who are responding well? And as we saw in the gastrointestinal stromal tumor trials,
00:10:59.24 and in the Iressa and Tarceva trials, the EGF receptor inhibitors,
00:11:04.00 there in fact was a subset of patients who were responding extremely well and those patients deserve more careful study.
00:11:10.28 So, what lessons have we learned from the clinical trials?
00:11:14.27 And if nothing else, what we've learned is, it's the target!
00:11:18.23 The target turns out to be the most critical factor that drives whether we have a good agent or not.
00:11:26.00 And if we have a good target, and a good drug, we really can get good results, and it really can be that simple.
00:11:34.06 But let me remind you what we had, why we had such a good target, what makes BCR-ABL such a good target.
00:11:41.08 First of all, we have a causative molecular abnormality of the leukemia.
00:11:46.23 And in the early stages of the disease, it's likely it's the sole oncogenic event,
00:11:51.26 so we have a causative, early molecular pathogenic event that's driving the growth and survival of the tumor.
00:11:58.14 But in addition, we actually ran a pretty smart clinical trial.
00:12:02.11 And it wasn't because we were so smart, it was actually because we had it pretty easy.
00:12:06.11 We actually could select patients that had a cancer driven by activated ABL,
00:12:12.05 and enroll them in a clinical trial of an ABL inhibitor.
00:12:15.01 And we could select those patients because they had the Philadelphic chromosome, or BCR-ABL.
00:12:20.07 So, we only enrolled patients in our ABL inhibitor trial that had a disease driven by activated ABL.
00:12:26.26 And so, because of that, we actually, as I said, ran a pretty smart clinical trial.
00:12:32.09 What about KIT, why is that such a good target?
00:12:35.10 Well, as it turns out, KIT mutations are again an early, pathogenic event.
00:12:40.27 If patients undergo surgery for other abdominal disorders and less than one millimeter gastrointestinal tumors are found,
00:12:50.05 they already have KIT mutation. And these KIT mutations were acquired before many of the other cytogenetic abnormalities
00:12:58.10 that are observed in metastatic gastrointestinal stromal tumor.
00:13:02.21 And lastly, there's a familial syndrome of gastrointestinal stromal tumors and these families have germ line KIT mutations,
00:13:10.27 again, an early molecular pathogenic event.
00:13:14.28 We got some other news, or lesson, from the clinical trials of imatinib and this is some pretty old news.
00:13:22.11 Anybody who treats any disease knows the earlier you treat in the course of a disease, the better your responses are going to be.
00:13:29.17 So, if we look at our imatinib clinical trials in chronic myeloid leukemia, you can see newly diagnosed chronic phase patients,
00:13:37.01 very high response rate, you get to blast crisis, only 8 percent of patients have their blood counts return to completely normal.
00:13:45.14 So, the earlier you treat in the course of the disease, the better your responses are going to be.
00:13:50.01 So, if we're going to translate the success of imatinib to other malignancies,
00:13:55.00 it's my opinion that the first task is to identify the right targets.
00:13:59.12 And in my view, the right targets are going to be the early molecular pathogenic events.
00:14:04.23 And these events are carried on into the advanced malignancies, so even in blast crisis of CML,
00:14:11.13 we see a fifty to sixty percent response rate, even in metastatic gastrointestinal stromal tumor, fifty to sixty percent response rate,
00:14:18.18 and even in metastatic lung cancer, between a fifty to seventy percent response rate to targeted agents against the early, molecular pathogenic events.
00:14:29.00 But if we're really going to make major impacts, we have to treat early in the course of the disease.
00:14:35.14 And to treat early in the course of the disease, we have to identify the patients early.
00:14:39.12 And to do that, we need very reliable techniques. I wouldn't want a technique that's overly sensitive.
00:14:45.22 By sensitive I mean it would detect a thousand patients that might have disease,
00:14:50.01 and only a hundred of them turn out to have the disease. We need diagnostic techniques for early detection
00:14:56.13 where if you have a positive result, you have the disease or will develop the disease.
00:15:01.05 Those are the types we need if we're going to actually make an impact and start a treatment.
00:15:06.10 Because we wouldn't want to be treating patients that don't actually have the disease.
00:15:09.23 But the third thing we have to do is we have to run smarter clinical trials.
00:15:14.00 And those smarter clinical trials will match the right patients with the right drug.
00:15:18.26 If you have activated ABL driving your disease, you need an ABL inhibitor.
00:15:22.24 If you have activated PDGF driving your disease, you need a PDGF receptor inhibitor, and so on and so forth.
00:15:28.25 So out of that understanding comes the ability to match the right patient with the right drug.
00:15:34.06 So, in my view, in the twenty-first century, we have several goals.
00:15:40.20 First of all, we have to identify all the molecular pathogenic events in all cancers.
00:15:46.20 We have to develop improved diagnostic and imaging techniques so we can identify patients as early as possible
00:15:54.27 in the course of their disease. We also have to get faster drug discovery;
00:16:00.08 we want this drug discovery, instead of taking fifteen years, to take three to five years.
00:16:05.19 And lastly, a major goal of the 21st century will be to actually understand our own individual risk for cancer development,
00:16:15.14 based on genetic analyses. Because ultimately what that will mean is that we can actually move to preventative therapy,
00:16:22.01 based on molecularly targeted therapies based on our own individual genetic analyses and individual risks.
00:16:29.11 Now, that might seems like a daunting task, to do all of this in a hundred years or less,
00:16:34.29 and hopefully it could be significantly less, with the kind of technologies we currently have in hand today.
00:16:42.13 But, what I'd like to do is have you think about the last hundred years.
00:16:47.13 So, if we go back to 1900, the leading three causes of death in most industrialized countries were infectious diseases.
00:16:57.02 So, in the United States, in 1900, the three leading causes of death: pneumonia, tuberculosis and diarrheal diseases,
00:17:05.13 infectious diseases. Cancer isn't even on this list, it was actually down about number 8.
00:17:11.04 The reality is in 1900, life expectancy in the United States was about 47 years, people didn't live long enough to get cancer.
00:17:19.08 But where are we in the year 2000?
00:17:22.02 Cancer is now number two and in the next ten to fifteen years,
00:17:26.00 it's projected that it will have the indubitable, or enviable, unenviable view of being number one as the leading cause of death in this country.
00:17:38.27 So, what happened between 1900 and 2000 to take infectious diseases from numbers one, two and three
00:17:47.02 down to number six or less?
00:17:49.15 And there were many factors. The main factors, however, were things like improved sanitation
00:17:56.17 and refrigeration, and if you think about chlorination of water, that was introduced in 1903,
00:18:03.07 pasteurization of milk, 1908, having refrigerators in most of our houses was around 1940, 1945.
00:18:11.23 These are the things that made our food and water supply safe.
00:18:15.13 In the 1940's, antibiotics, not even conceived of in the 1900's, began to come into common use.
00:18:24.00 Penicillin, 1942. And antibiotics truly opened people's eyes in that era to what specific, target therapies could do for infectious diseases.
00:18:35.12 And, in the 1950's, vaccinations, like the polio vaccine, 1955. Before 1955, there were massive epidemics of polio,
00:18:45.25 and now polio is on the verge of being eradicated.
00:18:49.19 So, if we re-cast this, this is prevention, this is specific treatments and this immune modulation.
00:18:59.06 So, if we think about moving forward into the twenty-first century,
00:19:03.01 and impacting on cancer, we need to take a very similar, broad-based approach.
00:19:08.23 We need specific therapies directed at critical targets.
00:19:12.16 We'd like to have a Gleevec for every single cancer,
00:19:15.28 but it's only going to come about through a precise understanding of molecular pathogenic events.
00:19:21.12 We need to do better prevention and early diagnosis.
00:19:25.19 And also, we need to think about modulating the immune system to harness the power of the immune system to target and eradicate cancers,
00:19:33.06 particularly in patients with minimal residual diseases.
00:19:36.21 So, my hope is that for the twenty-first century, we can take a very similar approach to the approach we took in the last century
00:19:43.23 for infectious diseases and sometime in the next ten to twenty to even fifty years,
00:19:48.22 we'll see significant impacts on cancer and that by the time my children, your children, grandchildren,
00:19:55.18 get to be our ages, cancer will no longer be the feared disease that it is today.
00:20:00.19 So, in closing, I'd like to thank all the people that work at the Oregon Health and Science University Cancer Institute,
00:20:07.00 in my lab, as well as in our clinics that have helped me on these clinical studies.
00:20:11.07 Novartis, who organized many of the clinical studies and who are essential to the pre-clinical studies.
00:20:16.27 All the people who have helped put people on our imatinib studies,
00:20:21.00 as well as the funding agencies who have funded me over the long haul,
00:20:24.16 particularly the Leukemia and Lymphoma Society, the National Institutes of Health and most recently,
00:20:29.12 the Howard Hughes Medical Institute.
00:20:31.28 But in closing, the people I'd really like to thank are the patients.
00:20:36.16 And the patients are the people who really spurred me to become an advocate for a drug like imatinib.
00:20:42.12 These are also the pioneers who went on a phase I trial of a completely unknown agent
00:20:48.12 and these patients are now, who have gone on this journey with me,
00:20:52.02 are now here today doing the things they truly enjoy doing.
00:20:55.11 Like this patient, gardening, like this patient, dancing, this patient spending time with her children and grandchildren,
00:21:03.02 that she thought she'd never live to see. This patient, like many others, traveling to Oregon, for our clinical trials.
00:21:10.12 Or, some patients doing truly remarkable things, like this patient,
00:21:15.18 who was actually the first patient who was from Australia treated with imatinib,
00:21:20.11 she came to Oregon, received imatinib, returned to Australia and was selected as one of the torch-bearers
00:21:27.00 for the Sydney Olympics in the year 2000.
00:21:30.22 But, much like I've take each one of these patients one at a time, my hope is we'll take each cancer one at a time,
00:21:37.23 and as we take each cancer one at a time, it all begins to add up.
00:21:42.13 And these are my phase I patients now who are a large group of patients,
00:21:46.14 many patients throughout the world have benefitted from understanding leading to better treatments.
00:21:52.19 And my hope is that this is exactly what weâ€™ll see in cancer therapy over the next twenty to fifty years.
00:21:58.21 More and more patients alive, surviving and thriving, despite a diagnosis of cancer.
00:22:04.03 Thank you very much.