¹Adult Leukemia Program, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

Correspondence: RM Stone, Harvard Medical School, Adult Leukemia Program, Dana-Farber Cancer Institute, 44 Binney Street, Room D-840, Boston, MA 02115-6084, USA. Fax: 1 617 632 2933

Received 19 September 2002; Accepted 15 November 2002.

Treatment of acute myeloid leukemia (AML) has not changed in a number of years; so there is a definite need for new therapies. Anthracyclines and cytarabine are still the mainstays of first-line therapy. Novel chemotherapeutic agents, such as temozolomide, an oral alkylating agent, and troxacitabine, a novel nucleotide analog, have been studied, but it is unclear whether they are better than conventional therapy. Gemtuzumab ozogamicin, a novel targeted therapy, may provide benefit in highly selected patients. Arsenic trioxide is efficacious in acute promyelocytic leukemia (APL), but not in other subtypes of leukemia. The successes of all-trans retinoic acid (ATRA) and arsenic trioxide in APL suggest that our goal should be to develop novel therapies to target the specific pathophysiologic processes in AML.

The pathophysiology of AML deals with two issues from the clinician's standpoint. If myelodysplastic syndrome (MDS) is a failure to differentiate because of transcription factor abnormalities, and if chronic myelogenous leukemia (CML), at least in chronic phase, is due to overabundant proliferation resulting from constitutive tyrosine kinase activity, AML involves both problems.¹ It would be naïve to think that targeting either one of those two mechanisms by itself will cure the disease, but perhaps aiming at each will begin to approach effective therapy in AML. At least for APL, we currently use chemotherapy to target proliferation and retinoic acid for differentiation, so a multifaceted approach could be feasible in other subtypes of AML. Proliferation can be decreased by inhibiting tyrosine kinases² and farnesyl transferases,³ as well as by promoting apoptosis. Differentiation can be promoted by hypomethylation of DNA by using DNA methyltransferase inhibitors,⁴ or histone deacetylase inhibitors,⁵ and/or manipulating associated phenomena such as drug resistance⁶ and angiogenesis.⁷ In addition, immunotherapeutic strategies that do not require as much intrinsic knowledge about the way a cell transforms to a malignant phenotype are also under development.

In CML, imatinib mesylate (Gleevec) has been shown to target successfully the constitutively active tyrosine kinase resulting from the mutant BCR/ABL gene that is responsible for uncontrolled proliferation. The tyrosine kinases c-kit and FLT-3 are overexpressed in AML myeloblasts. Published reports indicate that FLT-3 inhibitors kill cell lines and primary AML cells that are dependent upon expression of mutant FLT-3 for growth.^8,9,10,11 Some 20–30% of AML patients have leukemic cells that have an FLT-3 mutation; these proteins may be viable targets in AML, unlike c-kit, which is rarely mutated. FLT-3 is necessary for the development of hematopoietic and immune cells (natural killer cells and dendritic cells). FLT-3 is overexpressed in >80% of AML and acute lymphocytic leukemia (ALL). FLT-3 mutations, both the internal tandem repeat (ITD) in the juxtamembrane region and the point mutation at D835 in the activating loop, promote proliferation.¹² FLT-3 ITD mutations also induce myeloproliferative disorders (MPD) rather than leukemia in a murine bone marrow transplant model.¹ Thus, these ITD and D835 mutations cause the cell proliferation to be growth factor independent and cause MPD in mice.¹³

There are several oral FLT-3 inhibitors in development. CEP-701 is being tested at John Hopkins.⁸ PKC-412 and CT53518 are also in clinical trials in AML and MPD at several US sites.¹² One might expect that FLT-3 inhibitors will lead to hematological response rates of at best in the 25% range and complete responses, if at all, under 10%, because only part of the pathophysiology is being attacked. This is analogous to the situation with imatinib mesylate in blast crisis CML.²

Investigators at Dana-Farber Cancer Institute, along with collaborators at Memorial Sloan-Kettering Cancer Center, and MD Anderson Cancer Center, are involved in a phase II trial of the PKC-412 inhibitor, a very effective inhibitor of vascular endothelial cell growth factor (VEGF)-mediated cellular signaling.¹⁴ This drug inhibits the VEGF cellular receptor KDR and protein kinase C (PKC) in vitro and VEGF-dependent angiogenesis in vivo in a growth factor implant model. Phase I studies were carried out in patients with solid tumors. Treatment-related toxicities were nausea, vomiting, fatigue, and diarrhea.¹⁵ The patients enrolled in the current study are all required to have an FLT-3 mutation, have relapsed AML, and not be appropriate candidates for chemotherapy.

Farnesyl transferase inhibitors (FTI) prevent post-translational modification of Ras by blocking farnesylation, thus preventing its translocation to the membrane and achieving functionality. R115777, a potent, orally active FTI, induced complete remissions in patients with AML and CML with blast crisis.¹⁶ However, there was no good correlation between the in vitro assay for farnesyl transferase inhibition and response. Although the farnesylation of the Ras proto-oncogene is one presumed target, many proteins in the cell are farnesylated, and this drug could affect all of these cellular proteins. Phase I data suggested that R115777 could be administered with acceptable toxicity at a dose of 300 mg b.i.d.³ Phase II data on R115777 in myelodysplastic syndrome were reported at ASH 2001 on 23 patients. Only one patient had a Ras mutation. In this study, 600 mg was given orally for 4 weeks; 2/16 evaluable patients achieved a complete remission, indicating activity.¹⁷ Phase II trials of R115777¹⁸ and phase I trials with Sch-66336, another oral FTI,^19,20 and BMS-214662²¹ are ongoing.

Overexpression of BCL-2, an antiapoptotic molecule, is linked to poor prognosis in the pathophysiology of AML.^22,23 BCL-2 is also associated with survival and chemoresistance in AML. Genasense is an 18-mer anti-bcl-2 antisense oligomer that enhances chemosensitivity in AML cell lines and primary cells.²⁴ Genasense monotherapy has been reported to be tolerable in chronic lymphocytic leukemia.²⁵ A trial of combination therapy with Genasense and gemtuzumab ozogamicin is ongoing. The patients eligible for this trial are older adults who have CD33 positive AML and at least a 3-month disease-free interval. Genasense has been given in combination in relapsed AML with fludarabine, cytarabine, and GCSF (FLAG).²⁶ Genasense will likely not be effective as a single agent, however, and must be used in combination with other agents.

Another possible target is histone deacetylase (HDAC). Chromosomal rearrangements in AML result in fusion proteins (AML1-ETO), which can impose an altered interaction with transcriptional coregulators (eg. NCoR/SMRT). Transcriptional coregulators are responsible for recruitment of HDAC, which is required for repression of target genes (ie, promyelocytic leukemia–retinoic acid receptor (PML–RAR)), and for the transforming potential of the fusion protein. In APL, ATRA relieves NcoR/histone deacetylase corepressor complex inhibition of transcription.²⁷

DNA hypomethylating agents have also shown activity in leukemias. In a CALGB randomized trial comparing 5-azacytidine with observation in MDS patients, treatment with 5-azacytidine resulted in a better response, quality of life, and survival.⁴ In AML, DNA-modifying agents can also be effective, since in order to transcribe genes, which encode proteins involved in the differentiated phenotype, both acetylated histones and demethylated DNA are needed. Decitabine, a new DNA-hypomethylating agent with activity in MDS,²⁸ and HDAC inhibitors such as phenylbutyrate are already in clinical trials.²⁹ Preliminary information on trials combining the two agents, 5-azacytidine and phenylbutyrate, were presented at ASH 2001.³⁰ A phase III trial of decitabine vs observation in high-grade MDS is ongoing, as are trials of an HDAC inhibitor depsipeptide.³¹

Poor prognosis in AML is also associated with drug resistance, which is controlled by the expression of proteins such as Pgp170 (MDR-1 gene product).^32,33 These proteins efflux naturally occurring cytotoxic agents and chemotherapeutic drugs. Agents that block these proteins³⁴ may play a role in patients whose leukemia cells express high levels of MDR-1 protein and therefore, efflux drugs. Other drug resistance proteins that may be implicated are LRP, MRP-1,³⁵ and BCRP.³⁶

Angiogenesis is another potential target in AML. Data from MD Anderson Cancer Center suggest that VEGF and other proangiogenic proteins are operative in the pathophysiology of AML, indicating that antiangiogenic agents may be useful for AML. Thalidomide has demonstrated efficacy in MDS,³⁷ but has had minimal activity in AML.³⁸ Another avenue would be to interfere with VEGF and its receptors. VEGF-related tyrosine kinase inhibitors are in development. Sugen's SU5416 ^39,40and PTK787,⁴¹ a Novartis drug, may be tested by CALGB in the future.

Immunotherapy may be targeted against specific antigens expressed on leukemia cells. One antigen, CD33, has been chosen because it is expressed in 90% of AML cases. Another targeted therapy is a fusion protein containing the catalytic and transmembrane portions of diphtheria toxin linked to granulocyte macrophage colony-stimulating factor (GM-CSF). This fusion toxin (DT-GM) selectively binds and kills cells that express the GM-CSF receptors. Almost 90% of AML cells express GM-CSF receptors. DT-GM selectively kills AML cells, but not normal stem cells.⁴² In a phase I trial with refractory or relapsed AML patients, DT-GM resulted in complete and partial remissions, but produced severe hepatic toxicity.⁴³ AntiCD33 (HUM195) has been administered in a randomized phase III trial after mitoxantrone, etoposide, and cytarabine for relapsed AML, without any clearcut benefit.⁴⁴

Perhaps the most effective therapy would be to make the AML cells visible to the patient's immune system. There are a number of ways one could change the immune response toward AML cells. One approach involves transducing GM-CSF into myeloblasts, thereby making them more effective as antigen-presenting cells. This therapy was used in a small phase I trial for which we have enrolled five relapsed patients with AML or MDS. Their myeloblasts were isolated, transduced with adenovirus vector containing GM-CSF, which then begin to express GM-CSF. The cells were irradiated and administered as a vaccine. Mean secretion of GM-CSF was 11 ng/10⁶ cells/24 h. Four patients received 1x10⁷cells/dose and one patient received 3x10⁶ cells/dose. No significant toxicities were observed. Based on delayed-type hypersensitivity testing at a different site with nontransduced myeloblasts, an immune infiltrate has been noted. Such a response was seen in all patients, and one patient had a stabilization of disease. Alternatively, dendritic cells can be pulsed with killed myeloblasts, which will express antigens in such a way that they can be recognized by the immune system.⁴⁵

Most AML patients, especially the ones at the median age (65 years) and above, are intrinsically resistant to chemotherapy. There are many targeted approaches currently under development, but the major challenge remains to identify targets of critical pathophysiologic importance. Questions exist as to how we will design strategies to incorporate these agents into current therapy. Will we need to do a phase III trial for every biological agent? Ultimately, the patients should be defined based on their genotype (mutations in their blasts) and their phenotype (protein overexpression) to design the ultimate risk-adapted therapy.