The goals of novel therapies in acute myelogenous leukemia (AML) obviously should be to improve the outcome of patients with this disease. The greatest problem we have to deal with despite the toxic therapies we now use is disease resistance. We'd like to get an improved therapeutic index. We'd like to target therapy to the pathophysiologic abnormality and leave the normal cells alone. What would targeted therapy really be defined as? The target should be ubiquitous in all the cells in a given subset of AML. I emphasize the word "subset" because, obviously, AML is a highly heterogeneous disease. The target should be critical to the pathophysiology. CD33, for example, is commonly expressed, but may not be critical to the pathophysiology of AML. Ideally, the normal version of the target, if there is one, should not be required for normal health and happiness of the patient.

To begin in a rudimentary fashion to understand how we might apply the targeted therapy paradigm, I'd like to briefly discuss the pathophysiology of AML. It is obviously a complicated disease. Basically, if you look at myelodysplasia, it's a failure to differentiate the normal hematopoietic compartment into a very gross first approximation that may be due to transcription factor abnormalities that prevent the expression of genes associated with differentiated cells.

On the other hand, chronic myeloid leukemia (CML), in the stable phase, is largely due to 1 particular abnormality, a constituitively activated tyrosine kinase -- BCR-ABL.

Acute myeloid leukemia is at least a combination of those 2 pathophysiological problems. Both transcription factor abnormalities and tyrosine kinase abnormalities would be required together -- experimental studies have suggested this to be the case. Therefore, if you devised a therapy that targeted Problem A, it might not be enough to cause a major response unless you combined the therapy. Please keep that in mind during my discussion.

There are many new approaches in AML. Some people say there are as many new approaches in AML as there are patients with AML, which makes studying these therapies very complicated. Here is a list of some of the possible approaches. I will not dwell on new chemotherapy regimens since I'm not too high on that idea.

I'll speak about drug resistance modulation briefly, focusing my remarks on cell signaling modulation, in other words, actually getting at the primary defects that cause AML.

Drug resistance at large could be due to almost any of the new therapies that I am discussing. But when we use the words, we usually mean something about the chemotherapy pumps that extrude chemotherapy from the cells that really are the direct drug resistance issues. There have been a number of trials with PSC833, which is a nonimmunosuppressive cyclosporin analogue. Trials on PSC833 show that if you use this drug in patients with relapsed refractory disease, you do not improve the outcome compared with chemotherapy alone. In fact, there was a Cancer and Leukemia Group B (CALGB) trial in older adults that was stopped early because of therapy-related deaths in the PSC833 group. Nonetheless, Baer conducted a subgroup analysis in that trial, showing that those who were actually able to extrude chemotherapy based on ancillary laboratory studies actually benefited from PSC833. So it wasn't a completely negative experience.

Moreover, there was a Southwest Oncology Group (SWOG) trial that showed that the use of cyclosporin A in relapsed refractory AML would actually improve the outcome compared with the use of continuous infusion daunorubicin, a high-dose salvage regimen. So there were at least some hints that inhibiting the drug resistance pump might be beneficial. Moreover, the phase 1 part of a large CALGB trial chaired by Kolitz showed that if you actually were assigned to PSC833 -- again, this was not randomized -- you had a slightly higher disease-free survival than the cohorts who weren't assigned to PSC833. So there's still some hope that drug resistance modulation might be useful. The jury is still very much out. Ironically, there is not enough PSC833 available anymore due to a drug company decision not to pursue it further.

Why would antiangiogenic therapy make any sense in AML -- a disease within the blood? Well, actually it has been shown that there is a lot of crosstalk between cytokines that are secreted by endothelial cells and receptors in AML and vice-versa that may provide growth and survival advantages during this microenvironmental issue. In fact, there is increased microvessel density and increased expression of vascular endothelial growth factor (VEGF) and other pro-angiogenic proteins. Some of this expression is associated with a poor prognosis of AML.

There have been some national attempts to target this particular aspect of pathophysiology. One that you may be familiar with is the use of thalidomide in low-risk myelodysplastic syndrome (MDS), which may have nothing to do with AML, but it is an attempt to use that idea in people who have a decreased transfusional requirement. There has been a lot of interest in targeting the VEGF receptor or VEGF itself with drugs like SU5416 or the Novartis compound, PTK747 in AML. And there are some interesting clinical trials that are ongoing or have been recently completed regarding these issues.

Whether thalidomide itself will have a role in AML is doubtful. As you may know, there's a congener of thalidomide called Revimid. There will be some exciting results presented concerning Revimid's role in low-risk MDS. Whether that might have a role in AML in the future is also speculative.

Signaling pathways subsumes a large amount of issues in AML. There are so-called Ras farnesyl transferase inhibitors. Because Ras is often mutated at AML, that was a target. That was really the impetuous for the development of the Ras farnesyl transferase inhibitors, but whether these drugs that are inhibiting the farnesylation of proteins actually inhibit the farnesylation of Ras is the important issue. In AML it is still not clear.

Histone deacetylase inhibitors, like phenylbutyrate, depsipeptide, and suberoylanilide hydroxamic acid (SAHA) or DNA hypermethylating agents like, 5-Azacytidine and gemcitabine, have both been developed largely in MDS, but the idea is that you might be able to promote differentiation -- and that might be useful in getting at the problem in AML.

Bcl-2 overexpresssion, at least in some context, is a bad thing and may prevent the cells from undergoing apoptosis in response to chemotherapy. We might be able to deal with that by directly inhibiting Bcl-2 or even inhibiting proteasomes. I'll discuss tyrosine kinase inhibitors, which try to undo the proliferation issue caused by an overactive tyrosine kinase.

What about targeting Ras? Whether or not Johnson & Johnson's compound Zarnestra (R11577) really targets Ras is unclear. It does work as a single agent in Ras transformed cell lines. This now often cited paper by Karp was a phase 1 trial with this agent in relapse refractory leukemias -- mainly AML but a few acute lympboblastic leukemia (ALL) and CML blast questions were thrown in there. There were remissions seen with this oral agent, 2 complete remissions (CRs) and 8 partial remissions (PRs), and the remissions were seen at all dose levels even though the farnesyl transferase was only inhibited at dose levels above 300 mg twice a day. It's fairly well tolerated.

What's even more interesting about this particular drug is its use in relapsed refractory AML as a single agent more or less confirming prior data. An update of that trial by Harousseau showed a 10% response rate. Even more exciting is its use in people who are not going to get chemotherapy. These are largely older adults who have a really poor prognosis. They get Zarnestra as a single agent. Lancet found that there's a 30% response rate, mostly complete responses, as a single agent in untreated AML. It needs to be corroborated and extended, but it is very interesting and does give more hope that targeted therapy will have a role.

Returning to the differentiation agent issue, we have agents that can promote DNA histone acetylation and remove methyl groups from the DNA, both of which could promote differentiation.

Just to briefly review the pathophysiology, in the favorable subtype of AML, 8;21, the molecular biology behind that is the AML genome, chromosome 21, hooked up to part of the ETO genome, chromosome 8. That novel fusion protein, which only exists in these AML cells, recruits histone deacetylase activity, which removes acetyl groups from the histone, which you need to promote differentiation. Thus, transcription differentiation is prevented.

This concept has already been harnessed therapeutically in acute promyelocytic leukemia (APL), which has a similar pathophysiology in that the fusion protein between retinoic acid receptor-alpha and promyelocytic leukemia (PML) locus on chromosome 17 and chromosome 15, recruits histone deacetylase activity. We are all familiar with the use of retinoic acid to treat APL. Retinoic acid binds to the retinoic acid alpha receptor, and then that gets rid of the histone deacetylase activity, thus allowing transcription. There's an alternative potential explanation for this, but basically that is an example of histone deacetylase inhibitor therapy that's already ongoing.

When the CALGB compared the use of 5-Azacytidine to no therapy in people with all risk groups of MDS, there was a trend toward a survival benefit in those patients who received the 5-Azacytidine. There was a decrease in the P15 methylation associated with response. This drug is currently available for compassionate use patients, and whether or not it might be part of the therapeutic regimen in AML in the future remains an open question.

Bcl-2, as we've mentioned, is an apoptosis inhibitor. It may be involved in chemotherapy resistance of AML in cells with too much Bcl-2. If you get too much Bcl-2 you will not die, but it may be an independent factor for poor prognosis. In version 16, high white cell counts was a bad subgroup within a good subgroup. It may be true that those who overexpressed Bcl-2 with 8;21 have a worse prognosis.

So how can we target Bcl-2? Well, we can give all-trans retinoic acid (ATRA).

We could actually put in a molecule that prevents the translation of Bcl-2. In other words, put antisense mRNA into the Bcl-2 RNA. There have been phase 1 trials with antisense Bcl-2 in AML by Marcuicci and others. That actually is going to be part of a large CALGB randomized trial in which older patients with AML will get standard chemotherapy with or without antisense Bcl-2, and we'll finally understand whether this particular targeted therapy will be useful. We won't know the answer for a few years, though.

Another way to get after disease resistance is proteasome inhibition. Proteasome inhibition is really just in its infancy in AML, but as you may know, the proteasome is a multienzyme complex that degrades proteins that regulate cell cycle progression. One way to think about it may be that this proteasome enhances proteolysis of I-kappa-B, which thereby allows NF-kappa-B, which is a bad thing, and allows the cell to not apoptose and promote survival. So if you inhibit the proteasome, you get more I-kappa-B, less NF-kappa-B, and then perhaps you can allow the cell to undergo apoptosis, etc.

Velcade (bortezomib) is a boronic acid dipeptide that inhibits the proteasome. A phase 2 study in refractory myeloma that was published in The New England Journal of Medicine showed an impressive response rate in refractory myeloma patients. Why am I talking about it with AML? There are fairly good preclinical data that proteasome inhibitors, particularly this one, in conjunction with chemotherapy drugs like Ara-C (cytarabine) and daunorubicin, are synergistic killing AML cells in the test tube.

In fact, right now we are doing a phase 1/2 trial in which we combine this particular proteasome inhibitor with standard chemotherapy in a fashion analogous, developmentally speaking, to the way we developed Mylotarg (gemtuzumab ozogamicin) in AML.

Tyrosine kinases are enzymes that bind to adenosine triphosphate (ATP) thereby transferring a phosphate group onto a tyrosine residue and a substrate. This is the cell's way of signaling and telling the cell to grow more, to divide. If you have a mutation in a tyrosine kinase that leads to overactivity of the signaling communication, you might have too much proliferation -- that's the hallmark of cancer. This slide shows a normal situation.

We have you resting, in this case, a transmembrane tyrosine kinase sitting there waiting for the ligand to bind to it. The ligand binds to it. It gets excited and dimerizes. That dimerization turns on the tyrosine kinase activity thereby promoting the growth.

In the case of mutation in the tyrosine kinase, an acquired one, whether it's BCR-ABL in CML or a FLT3 mutation in AML, you don't require the ligand binding -- you have spontaneous dimerization and constitutive intracellular signaling promoting growth. That's what cancer is all about, at least in part.

You already saw 1 schematic diagram of FLT3; this is the second example. It shows that you can have a mutation in the juxtamembrane region of this tyrosine kinase. You can have a mutation less commonly in the activation loop. The consequence physiologically or biologically of having either one of these 2 mutations is that you have a poor prognosis, and you have constitutive activation of the tyrosine kinase. Experimental systems indicate that that constitutive activation needs ligand-independent growth in cell lines previously ligand-dependent, and it also allows you to get a myeloproliferative disorder in a murine bone marrow transplant model, thereby suggesting that a mutation, like this particular tyrosine kinase, can be leukemogenic. The other consequence of having such a mutation is that you have unbridled signaling down a number of pathways after the dimerization caused by the mutation.

This slide is a watered-down version of a usually very complicated network slide of tyrosine kinase; you might be able to also target some downstream consequences of the tyrosine kinase. In fact, there are Raf inhibitors in development in CML and AML now.

I just wanted to mention the development of direct FLT3 inhibitors, of which there are several in development right now.

And, of course, the big difference between the development of these molecules and the development of imatinib in CML is that there's many, many companies pursuing this and there are different molecules.

This makes things even more complicated.

We already mentioned the role of FLT3 in AML pathophysiology. I just wanted to briefly mention PKC412, which is 1 of the 4 agents being developed that are FLT3 inhibitors. It was actually out there a number of years ago because it was developed as a VEGF and protein kinase C (PKC) inhibitor. There was a phase 1 trial done in the United Kingdom in people with solid tumors. But, it was subsequently shown to be a very potent FLT3 inhibitor with a 50% inhibitory concentration (IC₅₀) in the 10 mnM. In those model systems I alluded to the cause of myeloproliferative disorders in mice due to the mutation in FLT3. If the mice eat the PKC412, their leukemias will go away, by and large. So it had good preclinical rationale.

These are cell line data from Jim Griffin and Allen Weisberg showing that the IC₅₀ in the cell lines transformed by the FLT3 internal tandem duplication (ITD) is about 10 nM, maybe even a bit lower if the cell line is transformed, due to an activating lutein mutation as opposed to an ITD.

We figured out that if this is important in the pathophysiology of AML, and if we inhibit it, then maybe we'll have a therapeutic efficacy. But we had fairly realistic goals. We didn't really expect to see many remissions because we're only dealing with part of the pathophysiological problem in AML. We hope to see results about as good as was seen with Gleevec (imatinib) in CML blast crisis, which is similar to AML in some regards from a biological standpoint.

The key thing here was you could either have relapse refractory AML or you could have untreated AML if you weren't deemed a candidate for chemotherapy because of age or other considerations. Importantly, you had to have 1of the 2 mutations in the FLT3 gene, which was proven by DNA sequencing before the patients went on the trial.

Seventy-five mg given orally 3 times a day was the dose that was determined to be safe in the phase 1 trial in solid tumor patients.

We enrolled 20 patients, which was our accrual goal, obviously this an older age group. Patients had mainly AML, although there were a couple of CMLs and a couple of MDS included in there. A few people had not had chemotherapy. The prior therapies were quite extensive. As you would expect from the epidemiology of FLT3 mutation, the majority had normal cytogenetics and most had the ITD mutation. A couple had the activation of lutein mutation.

The treatment was largely well tolerated. There was some low-grade nausea and vomiting. A couple of patients had severe pulmonary events. We still don't know whether that was due to the drug or to do with their leukemia or complication of an infection we don't normally see.

But we did see fairly exciting clinical results. If you looked at how the number of people had a 50% reduction in bone marrow blast, 5 of 20 -- it was in that category -- 2 actually got down to less than 5%, 1 had a near CR except for hypocellular bone marrow, on day 60. If you looked at just reducing your blast by 50%, that included 14 of 20 patients. Seven of 35 patients had a major clinical benefit if you defined benefit as a loss of 2 logs of leukemic cells in the blood for at least a month. One person, as you can see, went from 110,000 cells to 0. Another went from 16,000 cells to 0, etc. These really are comparable results to that seen with imatinib in blast crisis CML.

Here's an example of the patient who went from blast of about 65,000 down to 0. He eventually relapsed while on the drug but was down to 0 for a few weeks.

Here's an example of a patient who actually remained without any blast for 3 months before he went onto bone marrow transplant.

What have we learned from the phase 2 trial with PKC? We certainly proved the biological concept. In fact, FLT3 autophosphorylation was decreased at 4 to 24 hours. In 5 of 8 patients in whom we could study, 3 were responders, so we could inhibit the target. The response rate was about the same as it had been in the CML blast crisis, and the drug was pretty well tolerated.

Where do we go from here? We're testing additional doses, and, most importantly, we're administering this drug to patients with wild-type FLT3 AML because this drug has other targets beside FLT3 and that might have relevance for the response. Estey is reporting on work with the extension of this particular phase 2 trial in which we've gone onto administer this drug to a bunch of patients with FLT3 wild-type AML. There are some responses, but I would say they are not as dramatic as that seen with the mutant AML. That still remains to be figured out. And, of course, combining this drug with chemotherapy with other agents that target different aspects of the pathophysiology of AML would be a relevant way to go.

There are other drugs, as I said before, that are targeting the FLT3 mutant tyrosine kinase. One is CEP-701, which has been studied largely at Johns Hopkins. In patients with the FLT3 mutant AML, the results are similar to those that have been obtained with PKC412.

SUGEN compound SU5416 has been tested in a phase 2 trial. It was really believed to be most useful as a c-kit inhibitor. There was 1 response and a few minor responses. Interestingly enough, none of the patients with a FLT3 mutation, which were retrospectively looked at in this trial, had a response to this drug. The reason for that is not clear.

Finally, just for completeness, I'll mention immunotherapy in AML, other than antibody-targeted therapy. You could use various compounds to generally upregulate the immune system. You could try to make the AML cells more visible to the immune system by transferring genes, which might allow them to express their antigens in a better way. You might be able to fuse dendritic cells with AML cells, again, hopefully allowing better antigen presentation. The basis for this is that daunorubicin infusions do work in AML, so there are some good clinical data suggesting that immunotherapy could have a role in AML.

In summary, I would reiterate that most AMLs are intrinsically resistant to chemotherapy. We have many targeted approaches currently in development, and I think there are at least 3 challenges in this whole idea. There are multiple hits to go from normal to AML. There are a lot of trials -- do we need phase 3 trials, with every agent to show that it's going to be useful? The answer is probably "Yes." These trials are all coming through the drug companies. We need to buffer our relationship between academia and industry. And, of course, we really need to select patients, if we can, based on pathophysiology rather than just the way the cells look under the microscope.

Hopefully, some day we'll have a therapeutic armamentarium like this. We'll pick 1 drug from column A that maybe targets proliferation of survival mutation, 1 drug from column B with an agent that impairs differentiation, and with this combination we could throw out intensive chemotherapy and allow many of our patients with AML to improve greatly.