Peripheral Blood

Most patients with AML present with anemia (median hemoglobin 8 g%), thrombocytopenia (median platelet count 40,000–50,000/μl), and leukocytosis (median white blood cell count 10,1000–20,000/μL). The red blood cell morphology is usually relatively normal. Large, sometimes hypogranular platelets can be seen, and functional defects can contribute to hemorrhagic manifestations. Most patients are neutropenic, and morphologic abnormalities (hypogranulation, nuclear hyperlobulation, Pelger-Huët anomaly) are often noted in the remaining neutrophils. Careful examination will detect blasts in most patients, although it can be difficult to distinguish among leukemia subtypes (or occasionally even to be confident of the diagnosis of acute leukemia) in patients with a low number of circulating blasts. In occasional patients, marked leukopenia at presentation (aleukemic leukemia) may obscure the diagnosis until a marrow examination is performed.

MO (Minimally Differentiated AML)

In most patients, it is relatively simple to distinguish between AML and ALL on morphologic grounds. In general, the blasts from patients with AML are larger, with more abundant cytoplasm and more prominent, often multiple nucleoli. The definitive diagnosis depends, however, on the presence of Auer rods, concentrations of myeloid-containing granules (present in 50% of those with AML), or demonstration of at least 3% granulated precursors in AML,³⁵ usually visible on Wright stain and confirmed by staining with either peroxidase or Sudan black B. It should be noted that it can occasionally be difficult to distinguish between positivity in blasts versus staining in residual myeloid elements in patients with ALL; careful correlation with the Wright stain is mandatory.

As shown in Fig. 124.2, some patients have blasts that resemble myeloid blasts but are negative at the light microscopic level when examined with peroxidase, Sudan black, or other histochemical stains. The myeloid nature of these leukemias can be detected, however, by immunologic means or by election microscopy of peroxidase-stained preparations. Electron micrographs reveal ultrastructural peroxidase-positive granules, whereas immunologic phenotyping demonstrates reactivity with antibodies directed against myeloid antigens and nonreactivity with antibodies that characterize lymphoid differentiation.³⁹ The cells are often reactive with antibodies directed against CD34. Terminal deoxynucleotidal transferase (TdT) is generally absent but can sometimes be detected in a minority of blasts.

Approximately 7% of patients with untreated AML reportedly have minimally differentiated AML (so-called MO AML),^53,54 which is relatively treatment resistant. This undifferentiated leukemia can easily be confused with ALL, and the Cancer and Leukemia Group B (CALGB) has noted that occasional patients entered on ALL chemotherapy trials actually have minimally differentiated AML.⁵⁵ This has undoubtedly occurred in other clinical trial settings and potentially influences interpretation of the clinical outcome. It is therefore critically important to obtain immunophenotyping of blasts from patients with morphologically undifferentiated leukemias.

Other than an apparent resistance to chemotherapy, this MO variant does not appear to be associated with other specific clinical findings. Although a number of patients have had complex karyotypic abnormalities, no distinctive cytogenetic pattern has been noted, with the possible exception of trisomy 13, which has been reported to occur in some patients with morphologically less differentiated leukemias.⁵⁶

FAB M1 (Myeloid Leukemia without Maturation)

The blasts from patients with FAB M1 have round nuclei with moderate amounts of sometimes lightly granulated cytoplasm, which can contain Auer rods (Fig. 124.3). In contrast to FAB M2, there is little evidence of myeloid maturation, with <10% of cells beyond the level of the promyelocyte. There is no particular age, gender, clinical features, or characteristic cytogenetic abnormality associated with this morphologic variant. A retrospective analysis has suggested that the overall prognosis is poorer in patients in whom there is a lower fraction (50%) of blasts that are peroxidase-positive.⁵⁷ Although this observation is compatible with the more refractory nature of MO AML, these results require confirmation.

Figure 124.3

One of the blasts from this patient with FAB M1 AML contains a prominent Aver rod.

FAB M2 (Myeloid Leukemia with Maturation)

In contrast to FAB M1, there is obvious continued maturation in the myeloid series with the presence of promyelocytes, myelocytes, and often more mature myeloid elements. Granulation is generally more obvious, Auer rods are often prominent, and there is virtually never any difficulty in distinguishing this variant from ALL (Fig. 124.4). Approximately 20 to 25% of patients with FAB M2 AML have a characteristic translocation between chromosomes 8 and 21 [t(8; 21)(q22; 22)]; this translocation is seen almost exclusively in patients with FAB M2 and Auer rods. Such patients have a lower median age (approximately 30 years), a very high initial response rate to chemotherapy (>85% in most series), a lower relapse rate, and improved long-term disease-free survival compared with most other AML variants and with other patients with FAB M2.^58–60 It has recently been suggested that the incidence of extramedullary granulocytic sarcomas, often in unusual sites, may be increased in patients with t(8;21).⁶¹ This extramedullary manifestation often presents as discrete tumor masses, often in paraspinous locations, and is distinct from the gingival and cutaneous involvement found in monocytic leukemia.

Figure 124.4

FAB M2. Leukemia is characterized evidence of continued myeloid differentiation with myelocytes and more mature myeloid elements present.

The t(8;21) has recently been cloned and involves what has been termed the AML 1 gene on chromosome 21 and the ETO gene on chromosome 8.⁶² The AML 1 gene has homology to the Drosophila runt gene and encodes the alpha chain of the heterodimeric transcriptional apparatus, core binding factor (CBF).⁶³ Normal CBF function is critical for mammalian hematopoietic development.⁶⁴ The AML-ETO fusion protein recruits nuclear corepressor molecules including histone deacetylase,⁶⁵ which prevents transcription of genes required for myeloid differentiation. The AML-ETO fusion product can be detected by reverse transcriptase-polymerase chain reaction (RT-PCR) and used for both diagnostic testing and monitoring for minimal residual disease. Of note is that the t(8;21) has been detectable in the peripheral blood cells of some patients in long-term CR after treatment, suggesting that molecular detection by highly sensitive RT-PCR may not necessarily presage relapse.⁶⁶

FAB M3 (Progranulocytic Leukemia)

APL is one of the most distinctive subtypes of AML with regard to morphologic, clinical, and cytogenetic features.^67–70 In most patients, the morphologic diagnosis is straightforward, with the marrow being replaced by blasts that resemble abnormal, unusually heavily granulated progranulocytes. The nuclei are round, usually with obvious nucleoli, and the cytoplasm is filled with multiple, large, and often coalesced, azurophilic granules (Fig. 124.5). Auer rods are usually seen, and multiple Auer rods seen in sheaves (so-called faggot cells) are frequently noted. In a minority of patients, the blasts are hypogranular and granules sometimes can only be seen with electron microscopy. This M3 variant often has cells with bilobed or lobulated nuclei, which at first glance can sometimes be confused with monocytic variants of AML. In contrast to the typical leukopenic presentation, patients with the hypogranular variant tend to have a high white count. In both types of APL, staining with Sudan black or peroxidase is strongly positive.⁷¹ Class II HLA antigens (HLA DR), which are found on all hematopoietic precursors, are usually not detected or expressed on the surface of the malignant progranulocyte.^72,73 The explanation and biologic implications of this finding are not known.

Figure 124.5

FAB M3. Progranulocytic leukemic cells usually have round nuclei with heavily granulated cytoplasm. Extracellular granules are often noted, and blasts with multiple Auer rods (not shown) are common. This leukemia has typical 15; 17 translocation and a (more...)

Patients with APL tend to be somewhat younger, with a median age of 30 to 40 years, although this variant is seen in patients of all ages. APL accounts for approximately 10% of AML and is somewhat more prevalent in Latinos⁷⁴ and in obese people.⁷⁵ APL is almost uniformly characterized by hypofibrinogenemia, variable depletion of other coagulation factors, elevated levels of fibrin degradation products, and accelerated consumption of endogenous and transfused platelets.⁷⁶ The granular contents have been shown to contain potent procoagulants, and DIC is generally accelerated following lysis of blasts by chemotherapy, often with increased bleeding.⁷⁷ In some patients, there is evidence that accelerated fibrinolysis may be the primary event triggering the coagulopathy.⁷⁸ This type of AML is associated with the highest frequency of hemorrhagic morbidity and mortality, the latter usually related to intracranial hemorrhage. During chemotherapeutic treatment with accelerated cell lysis, these patients required aggressive support with platelet transfusions and often require transfusions 2 to 3 times per day during the first few days of treatment.⁷⁹ The plasma accompanying the platelet transfusion is usually sufficient for coagulation factor replacement. Except for patients with very severe hypofibrinogenemia who may require supplementation with cryoprecipitate, most patients do not require supplemental support with clotting factors. Aggressive blood product support is probably as good as heparin, once frequently used to interrupt the clotting cascade during drug therapy.^79–81 However, the heparin versus transfusion debate has been largely silenced because treatment with all-trans retinoic acid (ATRA) ameliorates DIC within 48 hours of its administration.⁸²

Cytogenetic studies have demonstrated that almost all patients with APL have a characteristic translocation involving chromosomes 15 and 17 [t(15;17)(q22;q12)].^68,83 This abnormality persists at relapse and may be associated with additional cytogenetic abnormalities, such as trisomy 8.⁸⁴ The t(15;17) breakpoint has been cloned,^85–87 thereby allowing detection of the fusion transcript as an assay for minimal residual disease as well as permitting the proper classification of the occasional patient with clinically and morphologically typical APL but with an apparently normal karyotype.⁸⁸ The breakpoint on chromosome 17 is in the intron of the retinoic acid receptor alpha gene.^85,86 A gene that has been termed PML, also with DNA-binding capability, is translocated from chromosome 15, resulting in the formation of a fusion protein that functions in a dominant fashion to block transcription of genes controlled by RAR-α, probably by recruiting nuclear co-repressor activity.^89,90 It is likely that retinoic acid treatment relieves the co-repression activity, allowing transcription of genes involved in differentiation.⁹⁰

Recently, a group of patients with a leukemia similar in morphology to APL but with a t(11;17)(q23;q21) have been described.⁹¹ Although there was a rearrangement of RAR-α, these patients failed to respond to ATRA. A novel zinc finger gene termed PZLF from chromosome 11 was translocated to the RAR-α, rather than the PML gene from chromosome 15, creating a fusion protein that does not allow the ATRA-mediated release of co-repressing transcriptional co-repression activity.⁹²

Historically, the initial remission rate in patients with APL treated with chemotherapy was quite high. Initial drug resistance is very unusual, and most failures of therapy are related to hemorrhagic or infectious deaths. APL seems to be uniquely sensitive to anthracycline therapy, and CR rates in excess of 80% can be seen with anthracycline therapy alone.⁹³ It is critical to recognize that in contrast to other types of AML, remission can be attained in APL with chemotherapy without producing bone marrow aplasia.^93,94 Post-treatment bone marrows frequently remain cellular with abnormal progranulocytes, with follow-up marrows demonstrating disappearance of these cells and return of normal hematopoiesis; DIC does not reappear despite the apparent persistence of abnormal cells. Stable or improving blood counts and the return of erythroid and megakaryocytic precursors are other signs that further chemotherapy is not necessary. Undoubtedly, this unique feature of APL is in some way related to the equally unique sensitivity of APL to agents that have a differentiating and noncytotoxic mechanism of action (see “Differentiating Agent Therapy” below).

FAB M4 (Myelomonocytic Leukemia)

Myelomonocytic leukemia is characterized morphologically by a mixture of both myeloid and monocytic elements and represents about 15 to 20% of newly diagnosed patients with AML. According to the FAB criteria, greater than 20% of the leukemic cells must be monocytic in morphology to distinguish this variant from FAB M1 and particularly from FAB M2.³⁸ The monocytic elements can be defined morphologically by the presence of partially differentiated monocytes with lightly granulated, grayish cytoplasm and folded nuclei, which are frequently seen in the peripheral blood (Fig. 124.6). Monocytic derivation can be confirmed by staining with nonspecific esterases such as alpha naphthyl acetate or alpha naphthyl butyrate.

Figure 124.6

FAB M4. Myelomonocytic leukemia has blasts with both myeloid and monocytoid appearance.

There is no distinct clinical picture associated with this morphologic variant, perhaps because this classification encompasses a wide spectrum of patients because of the generous morphologic criteria for inclusion. The median age tends to be somewhat higher, and there may be an increased incidence of hyperleukocytosis and extramedullary leukemic involvement, as can be seen with monocytic leukemia (see below). There is no particular cytogenetic clustering, and it is impossible to predict reliably short- or long-term outcome in patients with FAB M4.

FAB M4EO (Myelomonocytic Leukemia with Eosinophilia)

Approximately 5% of patients with de novo AML present with typical morphologic features of myelomonocytic leukemia with the presence of variable numbers of dysplastic eosinophils at various stages of maturation. The distinctive eosinophils usually represent only 5 to 10% of the cells of the marrow.^95,96 These cells generally contain large basophilic granules in addition to typical eosinophilic granules and are easily detected by experienced morphologists. It is unclear whether these cells are part of the leukemic process or are reactive in nature. Occasional patients with apparent FAB M2 morphology with eosinophilia have also been described.

FAB M4E0 tends to occur in patients of younger age (median age 35–40 years) and is associated with an excellent prognosis. CR rates are high (generally >85%), and failure due to initial drug resistance is unusual. In some series, this variant represents the subtype with the most favorable long-term prognosis. In addition to long initial CRs, second remissions, which are often quite sustained, are generally easier to accomplish in patients with FAB M4E0.⁹⁷ Older series have suggested that there may be a high rate of central nervous system (CNS) relapse in patients with bone marrow eosinophilia.⁹⁸ With more intensive regimens using higher doses of cytosine arabinoside (ara-C), CNS relapse in AML is now an unusual occurrence, and this possible predilection in patients with FAB M4E0 does not require prophylactic CNS therapy.

Essentially all patients with FAB M4E0s have a cytogenetic abnormality involving chromosome 16 at band q22. In most patients, the cytogenetic changes involve a pericentric inversion [inv¹⁶], although translocations between the two chromosomes 16 with homologous deletions at 16q22 have also been noted.^94,95 This can be a very subtle cytogenetic finding, and cytogeneticists should be prompted to look for these changes by the characteristic morphologic changes. This breakpoint has been cloned and involves a fusion between the CBF-β chain and the gene encoding the smooth muscle myosin heavy chain.⁹⁹ The fusion protein thus generated may recruit nuclear co-repressor activity (in the form of histone deacetylase), which prevents transcription of genes required for myeloid differentiation in a fashion analogous to the CBF-α ETO fusion in t(8; 21) M2 AML. Although it is clear that patients with inv 16 leukemia respond very well to intensive chemotherapy (> 60–70% 3-yr DFS rate in those receiving high-dose ara-C [so-called HIDAC])¹⁰⁰, the reason for this is uncertain. One group has suggested that the gene coding for the multidrug resistance protein (MRP) is mutated as a consequence of this translocation, thereby hypothesizing that the enhanced chemosensitivity may be related to a decrease in the normal amounts of this resistance factor.¹⁰¹

FAB M5 (Monocytic Leukemia)

Two variants of monocytic leukemia have been described; in both, >80% of the blasts are of monocytic derivation. Less common is so-called FAB M5a, in which the monocytic blasts usually have round nuclei and small amounts of sometimes deeply basophilic cytoplasm without evidence of morphologic differentiation. In monocytic leukemia with differentiation (FAB M5b), at least 20% of the blasts resemble promonocytes with folding of the nuclei and abundant, lightly granulated cytoplasm, generally without Auer rods. The nuclear folding can often be quite marked with rarification of the nuclear chromatin (Fig. 124.7). Phagocytosis of other hematopoietic elements by these cells is frequently noted in bone marrow preparations. These monocytic elements stain prominently with nonspecific esterase that is inhibited by fluoride.

Figure 124.7

FAB M5. Monocytic leukemia. Prominent nuclei filled with nucleoli in some cells, light granulation, and large amounts of lightly basophilic cytoplasm give these cells the appearance of promonocytes.

Although seen in patients of all ages, monocytic leukemias are somewhat more common in older adults. Patients with FAB M5 have higher blast counts at diagnosis, and problems with hyperleukocytosis are most common in this morphologic variant (see “Complications” below).^102,103 In addition, the incidence of extramedullary leukemia is highest in FAB M5, particularly those with evidence of morphologic differentiation.^104,105 For example, it is common for patients to present to the dentist with gingival hypertrophy, an example of which is seen in Fig. 124.8. Skin infiltration is common both at diagnosis and relapse and frequently represents the initial site of recurrence, often while the bone marrow is still morphologically normal. Other less common areas of extramedullary involvement include the gastrointestinal tract, conjunctivae, and the CNS.^106,107 It is likely that the extramedullary infiltration is related to active migration of the leukemic promonocytes to these sites. It has been shown that these partially differentiated cells are capable of migration to skin windows in vivo as well as phagocytosis of microorganisms and adherence to nylon fibers in vitro.¹⁰⁸

Figure 124.8

FAB M5. Gingival hypertrophy due to infiltration by leukemic cells in acute monocytic leukemia.

Serum levels of lysozyme are elevated in most patients with AML but are generally much higher in patients with monocytic leukemia.¹⁰⁹ Lysozyme can affect renal tubular function, and severe, symptomatic hypokalemia can occur in patients with FAB M4 and M5 leukemia. This problem generally resolves with cytoreduction but can also be additive to the hypokalemic side effects of antibiotics and amphotericin B, vomiting, and diarrhea.

In addition to the initial problems presented by complications of hyperleukocytosis, patients with monocytic leukemia have lower complete response rates related to initially drug-resistant disease.^105,110,111 In parallel, CR duration tends to be shorter with consequent very low rates of long-term disease-free survival. A variety of cytogenetic abnormalities can be detected, although the most common findings involve abnormalities of chromosome 11 at band q23.^111,112 This breakpoint, at what has been termed the MLL gene, has recently been cloned and can be involved in leukemias of myeloid or lymphoid origin as well as in leukemias associated with therapy with epidophyllotoxins and other drugs directed at topoisomerase II.¹¹³ The MLL gene, also called All-1 or HRX, may partner with at least 16 different genes in balanced translocation.¹¹⁴ MLL is homologous to a gene important in drosophila development and includes DNA binding elements.¹¹⁵ A relatively common translocation involving the MLL gene, the t(9;11) associated with monocytic leukemia may actually confer a better prognosis than formerly thought,¹¹⁶ with higher initial CR rates. There is an association between M5b and extensive erythrophagocytosis, and the t(8;16) (p11;p13)^117,118 a translocation involving that CBP class of translocation factors that are positive regulators of myeloid differentiation.¹¹⁹

FAB M6 (Erythroleukemia)

Erythroleukemia, often termed de Guglielmo’s syndrome in earlier literature, is the variant of AML in which morphologic abnormalities of erythropoiesis are most prominent. Cases of pure erythroleukemia in which the predominant malignant cell is clearly identified as a pronormoblast are extraordinarily rare. Rather, this is almost certainly a disease of the myeloid stem cell with marked dysplastic changes in all three hematopoietic lines with eventual increases in the number of myeloid blasts. Along with the increase in myeloid-appearing blasts, there is persistence of morphologic abnormalities in the erythroid series with profound megablastosis, multinuclearity, karyorrhexis, increased numbers of mitoses, and staining with periodic acid–Schiff, often in a block pattern (Fig. 124.9). Increased iron stores are usually seen, often with ringed sideroblasts.

Figure 124.9

FAB M6. Erythroleukemia is characterized by the presence of bizarre megaloblastic and often multinucleated erythroid precursors. Karyorrhexis is seen in some cells. The somewhat arbitrary distinction between FAB M6 and myelodysplastic syndrome with excess (more...)

These changes are morphologically identical to those seen in patients with myelodysplasia, and many observers feel that it is likely that most cases of erythroleukemia are biologically similar, if not identical, to patients with refractory anemia with excess blasts, or refractory anemia with excess blasts in transformation (RAEB or RAEB-T). This contention is supported by the very poor response to therapy in both groups, the tendency for the disease to occur in patients of older age, and the presence of similar cytogenetic abnormalities (complex karyotypic abnormalities with loss of part of all of chromosomes 5 and/or 7 and marker chromosomes).¹²⁰ Nonetheless, in order to provide some consistency in terms of protocol entry and reports of clinical trials, the FAB group, with a minor recent revision, has somewhat arbitrarily distinguished among FAB M6, myelodysplasia (RAEB or RAEB-T), and other FAB subtypes with significant numbers of erythroblasts by quantification of the number of erythroblasts and myeloblasts.^1,38 FAB M6 is defined by the presence of > 30% blasts among nonerythroid cells when > 50% of the marrow nucleated elements are erythroid. Antibodies against glycophorin A are lineage-specific for erythroblasts, but immunologic phenotyping is rarely needed to distinguish among these various morphologic subtypes.

Except for the more advanced age and higher frequency of primary resistance to therapy, there are no distinct clinical or diagnostic karyotypic features associated with FAB M6.¹²¹ In occasional patients, transfusion of red blood cells can result in a marked decrease, if not elimination, of the marrow erythroid elements, suggesting the continued response to normal feedback mechanisms.

FAB M7 (Megakaryocytic Leukemia)

Morphologic abnormalities of megakaryocytopoiesis, usually characterized by the presence of mono- or binucleated micromegakaryocytes, are common in many variants of AML and can be particularly prominent in patients with FAB M6 or myelodysplasia. A minority of these patients have thromboyctosis and abnormalities of chromosome 3[inv(3)(q21;q26)], usually in association with other chromosomal deletions, with a variety of primary FAB morphologies (M1, M2, M4).^122,123 These patients have a poor response to initial treatment and poor overall survival. Thrombocytosis is not a unique feature of the inv(3) karyotype but can also be present in other patients with AML at the time of diagnosis, particularly those in whom normal numbers of megakaryocytes are present.¹²⁴ The gene at the chromosome 3q21 breakpoint is associated with the activation of EV11 transcription factor.¹²⁵

The diagnosis of FAB M7 is reserved for those patients in whom the predominant leukemic cell is of megakaryocytic lineage.³⁷ In some patients, there is obvious evidence of megakaryocytic dysplasia or multinucleated cells that strongly points morphologically toward principal involvement of the megakaryocyte (Fig. 124.10). In many others, however, the leukemia is very undifferentiated morphologically with variable amounts of generally agranular cytoplasm and can sometimes be confused with FAB M1 or even ALL. Sudan black B (Fig. 124.11), peroxidase, and alpha naphthyl butyrate stains are negative, whereas PAS and acid phosphatase may be positive, usually in a diffuse pattern. Histochemical staining is nondiagnostic, however, and the definitive diagnosis depends on either the demonstration of platelet-specific peroxidase by ultrastructural techniques¹²⁶ or, more commonly, by the demonstration of a variety of platelet antigens (usually glycoprotein IIb/IIIa [CD41] or von Willebrand’s factor) on the surface of the blasts.¹²⁷ At times, the diagnosis can be quite difficult to confirm, particularly because in most patients there is increased marrow reticulin, rendering the marrow fibrotic and inaspirable. Careful evaluation of peripheral blood blasts is necessary in such patients. It is likely that most patients with what has been termed acute myelosclerosis actually have acute megakaryocytic leukemia.^128,129 Acute megakaryocytic leukemia should not be confused with the late stages of myloid metaplasia, and, indeed, prominent splenomegaly is not a clinical feature of FAB M7.

Figure 124.10

FAB M7. Megakaryocytic leukemia. Blasts in this category are often morphologically undifferentiated. The presence of multinucleated cells, dysplastic micromegakaryocytes, and cytoplasmic budding can be helpful diagnostic clues. The diagnosis is confirmed (more...)

Figure 124.11

Typical granular staining with Sudan black B of a blast and a neutrophil from a patient with FAB M1 AML.

Although an uncommon variant of AML, most series suggest that this subtype is associated with a very poor prognosis. Prolonged aplasia is common following induction chemotherapy, and, because of the marrow fibrosis, it is often difficult to follow the results of therapy with repeated marrow aspirations. There have been relatively few cytogenetic evaluations of this variant, and except for the inv(3) and a few cases of t(1;22) (p13;q13) found in infants,¹³⁰ no consistent abnormality has been identified to date.

Hybrid Leukemias

Morphologists have long been perplexed by cases that defy easy categorization or seem to have features of both lymphoid and myeloid histology. A variety of terms have been used to describe such cases including hybrid leukemia, mixed lineage leukemia, biclonal leukemia, lineage infidelity or promiscuity, and biphenotypic leukemias.^131–133 With the advent of new immunologic and molecular technology, it was hoped that most, e.g., if not all, leukemias could be assigned to a specific lineage. Paradoxically, the serial application of biochemical, immunologic, and molecular biologic techniques has demonstrated that sharing of myeloid and lymphoid characteristics may be a relatively common feature of both myeloid and lymphoid leukemia. The clinical and therapeutic implications of these findings are as yet unclear. There are no accepted diagnostic criteria for hybrid leukemia, and, given the rapid pace of investigation, it may be premature to establish rigorous definitions. Most reports suggest that individual cells co-express myeloid and lymphoid features; it is likely that this pattern is much more common than the simultaneous occurrence of two separate leukemias. The frequency and importance of subpopulation of myeloid only, lymphoid only, null, or mixed Leukemia cells in an individual patient.

Leukemias that co-express features of both lymphoid and myeloid development are generally believed to arise from a relatively less committed stem cell. Such a concept is supported by the finding that the less differentiated myeloid leukemias (e.g., MO AML) have a higher frequency of expression of markers traditionally associated with lymphoid differentiation. TdT, an enzyme found in primitive lymphoid precursors and thymocytes, may be expressed in 20 to 40% of blasts from those with FAB MO AML.^53,54

The significance of expression of TdT or other lymphoid markers in those with typical peroxidase positive AML is unclear. Although up to 25% of AML patients may express TdT,^134,135 there are conflicting data about whether such a feature confers an adverse prognosis. The importance of T-cell antigen expression or T-cell gene rearrangement in children with AML may¹³⁶ or may not¹³⁷ correlate with a higher level of resistance to regimens designed for AML patients.¹³⁶ In adults with AML treated on CALGB clinical trials, lymphoid antigen expression had no effect on outcome.¹³⁸ The significance of a T-cell receptor rearrangement, a finding not uncommon in the AML patient,¹³⁹ is even less clear. In fact, CD7 expression is not uncommon in t(8;21) AML which clearly responds well to chemotherapy.^140–142 Patients, generally infants, with a t(4; 11) abnormality frequently display biphenotypia.¹⁴³

Although the phenotypic features of the leukemic blast are generally constant throughout the patient’s course (e.g., at diagnosis and relapse), examples of lineage switch, usually AML followed by ALL, have been noted.^144,145 The optimal therapeutic approach to these hybrid leukemias remains unclear. For patients whose blasts are clearly peroxidase positive and/or nonspecific esterase positive, who express mainly myeloid antigens but who display TdT positivity or express one lymphoid antigen, an AML regimen is appropriate. For those ALL patients whose blasts express one or two myeloid antigens, enrolment onto an ALL protocol is reasonable.^55,146,147 For patients who truly have equal myeloid and lymphoid features, an obvious initial approach would be to combine AML and ALL regimens, but there are few data concerning the outcome of this strategy in such patients.

Induction Therapy

General Principles

Induction therapy is designed to produce rapid clearing of leukemic cells from the peripheral blood with subsequent marrow aplasia. Perhaps the only exception to this principle occurs in patients with acute progranulocytic leukemia (FAB M3), in whom, as described earlier, remission can be achieved despite the persistence 2 to 3 weeks later of what appear morphologically to be viable leukemia cells.^93,94,156 Approximately 1 week after standard induction therapy is completed (generally 2 weeks after the initiation of treatment), bone marrow aspirates and biopsies are done to evaluate the magnitude of cytoreduction. If indeed the marrow is profoundly hypoplastic, the marrow is repeated at approximately weekly intervals to assess whether there is return of normal hematopoiesis, leukemia cells, or an apparent mixture of both. If the marrow is not hypoplastic and only leukemia cells are noted on the day 14 marrow, then a second course of therapy is generally administered. At times, particularly if the marrow is hypocellular, it can be difficult to distinguish morphologically between residual leukemia cells and normal undifferentiated hematopoietic progenitors. In this instance, it is advisable to delay a few days and perform another marrow aspirate. If there is no evidence of further maturation, then the second course of treatment is indicated. The presence of erythroid precursors, juvenile megakaryocytes, or increases in the peripheral blood platelet or neutrophil counts serve as clues that normal regeneration is occurring and that a second course of treatment should be delayed. With standard regimens, approximately 30% of patients with AML receive two courses of treatment to enter remisssion. It should be noted, however, that despite these guidelines, there remains considerable variability among clinicians in terms of the decision as to when a second course of therapy is indicated. Some clinicians also have advocated performing bone marrow aspirates and biopsies on the day after chemotherapy is completed because of a suggestion that evidence of residual leukemia at this time may be a signal of inadequate cytoreduction and a decreased probability of CR.^157,158 Although it has been advocated that further therapy be administered immediately to such patients, there is no conclusive evidence that this approach increases the rate of CR or improves survival; such bone marrows that have been so recently exposed to intensive chemotherapy are also often difficult to interpret.

Patients can fail to achieve remission for a number of reasons. Until recently, most papers reported results as complete response or no response. It is, however, more helpful in investigations of prognostic factors or assessment of the cytotoxic activity of different regimens to classify more rigorously the causes of failure to achieve remission. One such classification divides nonresponders into those with apparent chemotherapy-resistant leukemia, those who die with aplastic bone marrows in whom the response to chemotherapy cannot be determined, and those in whom either early death or failure to obtain adequate bone marrow studies prior to death preclude determination of whether persistent leukemia was present.¹⁵⁹ Patients with drug-resistant leukemia include those patients who survive treatment and those who die but have morphologic evidence of leukemia in the bone marrow or blood antemortem or at autopsy. This categorization is relatively simple to perform in most patients and can be helpful in distinguishing between failure due to ineffective chemotherapy and failure related to inadequacies of supportive care.

The overall rate of CR in large cooperative group studies is approximately 65%. This CR rate is related to a number of prognostic factors, many of which are discussed below. Patient age is the most critical clinical variable. Comparisons among studies can be difficult because of imbalances in prognostic factors. In addition, many reports do not stipulate whether patients with secondary leukemias or prior myelodysplasia were included in the trial. Exclusion of such patients, who represent approximately 5 to 10% of adult patients diagnosed with AML, produces apparently improved results. It is preferable to present results in these patients and in patients with de novo leukemia separately.¹⁶⁰ The achievement of CR is highly dependent on the ability to administer appropriate supportive care to prevent hemorrhagic and infectious deaths during the approximately 3 weeks of aplasia. The criteria for CR are generally reached a median of 30 to 35 days after treatment has begun. Perhaps more importantly, patients develop adequate levels of circulating neutrophils (> 500/μl) and no longer require platelet transfusion (at counts of approximately 10–20,000/μl), at least 7 to 10 days earlier.

Although there have been gradual overall improvements in the CR rates worldwide during the past 10 to 15 years, much of this can be attributed to more widespread availability of sophisticated supportive care and not to changes in therapy. With the exception of the use of ATRA for patients with APL (see below), relatively few changes in therapy have been made since the introduction of combined therapy with daunorubicin and ara-C, the so-called “7 and 3” regimen.^161–163 This two-drug combination was derived from observations of single agent activity with either compound.¹⁶⁴ Daunorubicin is generally administered by intravenous push at doses of 30 to 50 mg/m²/d for 3 days; ara-C is now generally administered at doses of 100 to 200 mg/m²/d by continuous infusion for 7 days (Table 124.3). A series of randomized studies^{159,162,165–168} by the Cancer and Leukemia Group B (CALGB) demonstrated that (a) results were superior using the 7 and 3 regimen compared with 5 days of ara-C and two doses of daunorubicin; (b) the addition of oral 6-thioguanine (DAT regimen) to the 7 and 3 program did not increase CR rate; (c) continuous infusions of ara-C produced slightly better outcome than twice-daily short intravenous infusions of ara-C when combined with daunorubicin; (d) results were not improved when 10 days of ara-C by continuous infusion were administered compared with 7 days; (e) substitution of Adriamycin for daunorubicin produced almost identical CR rates, although mucosal toxicity was greater with Adriamycin; and (f) there was no overall benefit from doubling the dose of ara-C from 100 mg/m² to 200 mg/m²/d.

Table 124.3

Representative Regimens for AML.

Other attempts to improve the 7 and 3 induction regimen have also been relatively disappointing. Although there was a suggestion that older patients more frequently achieved CR with a single course of treatment using mitoxantrone in place of daunorubicin, this did not result in enhanced overall patient survival. Alternative anthracyclines or other agents such as rubidizone, aclacinomycin, amsacrine, mitoxantrone, and idarubicin have been used in several trials.^5,169–172 The largest reported experience has been with idarubicin where three trials have demonstrated that induction results are at least equivalent to results achieved with daunorubicin and ara-C.^169,171,172 None of these studies demonstrated a survival or disease-free survival advantage with these different agents, perhaps because most are relatively similar in structure and mechanism of action and hence become susceptible to the same mechanisms of resistance. Moreover, the duration of myelosuppression was greater in the idarubicin cohorts, calling into question the equitoxicity of the arms. Therefore, it is unlikely that the introduction of these compounds will have a major impact on the long-term disease-free survival rate of patients with AML.

Another recent trial from Australia added etoposide (VP-16) to the 7 and 3 regimen.¹⁷³ There were similar complete response rates in older patients when compared with daunorubicin and ara-C alone and a possible slight increase in CR rate and modest prolongation of CR duration in younger patients receiving the etoposide. This was a relatively small study and requires confirmation, since the overall disease-free survival in the group receiving the etoposide during both remission induction and postremission therapy was similar to other larger studies using daunorubicin and ara-C alone. Insofar as HIDAC is a beneficial postremission therapy (see below), several groups have tested the concept of using this approach during induction. Studies by the Southwest Oncology Group,¹⁷⁴ the Australia Leukemia Group,¹⁷⁵ and UCLA¹⁷⁶ compared standard 7 and 3 to daunorubicin plus intermediate- or high-dose ara-C (2 g/m² for 8–12 doses). These studies failed to show an improved CR rate for the recipients of higher-dose ara-C, although the Australian study documented a more prolonged duration of CR (but no change in overall survival) in the patients randomized to HIDAC. The addition of HIDAC to standard daunorubicin/ara-C during induction has also been studied. Although a limited access trial demonstrated an 87% remission rate in patients under 60 years old,¹⁷⁷ a cooperative group trial,¹⁷⁸ failed to confirm those positive results. Topotecan-based regimens are now being evaluated¹⁷⁹; unfortunately, there are few convincing data to suggest that new drugs or alternative induction regimens are likely to make a significant difference in either the CR or cure rates.

The reasons for treatment failure vary according to patient age. With the availability of sophisticated supportive care, it is relatively uncommon for patients less than 50 years of age to die from complications of treatment; most of the approximately 25% induction failure rate is a consequence of drug-resistant leukemia, and most patients survive initial therapy. In contrast, the CR rate in patients >60 years of age is approximately 50%,¹⁸⁰ with failures divided equally between drug-resistant leukemia and deaths occurring during marrow aplasia as a consequence of reduced end organ tolerance. Thus, whereas younger patients with poor prognostic factors would benefit from entirely new approaches to achieve CR, including the possible use of new chemotherapeutic agents, older individuals would also benefit from methods by which the period of aplasia can be shortened.

Some observers debate whether older patients or patients with secondary leukemias or prior myelodysplasia benefit from intensive induction approaches at all. In particular, there has been interest in using very low-dose ara-C (10–20 mg/m²/d) in these patients to avoid the acute side effects associated with daunorubicin and even conventional-dose ara-C. Although responses can be seen with low-dose ara-C, many weeks of therapy can be required with consequent pancytopenic complications equivalent to those seen with higher-dose treatment.¹⁸¹ In addition, the CR rate is lower, and although median survivals may be similar using the two different dosage approaches, the prospect for long-term disease-free survival is greater, when, the intent of treatment is to achieve CR.¹⁸² Another study documented a survival benefit for the prompt initiation of induction therapy at the time of diagnosis compared with waiting and administering chemotherapy only on development of symptoms.¹⁸³ However, the degree of benefit was small. To answer the question of whether a 1- to 2-month aggregate survival advantage justifies the side effects associated with induction therapy, a quality of life component in those types of trials is critical.

CRs can be achieved in approximately 20 to 30% of patients whose leukemia followed treatment for another cancer. Responses may be better in patients who do not have an extended pancytopenic or myelodysplastic period prior to developing frank leukemia. Although responses tend to be short in such patients, intensive induction should be offered, particularly to younger patients in whom subsequent bone marrow transplantation (BMT) may be an option.^160,183,184 Some younger patients with apparent secondary leukemia have been found to have cytogenetic findings associated with de novo AML of good prognosis such as t(8;21) or t(15;17) and may have a more favorable clinical course than expected.^185,186 These patients should be treated in the same fashion as those with de novo disease. It can also sometimes be difficult to distinguish morphologically between FAB M2 with t(8;21) and patients whose marrows contain < 30% blasts who would be classified as RAEB-T according to strict FAB criteria. Cytogenetic studies should therefore be done in all such patients.

Whether to treat patients with well-documented prior MDS and leukemia evolving from RAEB or RAEB-T can be more problematic. Complete response rates are lower, sustained responses are unusual, and prolonged periods of aplasia are common. In addition, these patients are usually older and less able to tolerate pancytopenic complications. However, data from M.D. Anderson Cancer Center suggest that the absolute number of marrow blasts (e.g., whether a patient has RAEB-T or AML) is not of prognostic significance.¹⁸⁷ Instead, cytogenetic status and a truly prolonged history of MDS are critical factors. It is possible, however, to define patients with AML who have such poor prognostic factors that enrollment on phase I/II studies using new agents would be appropriate. If a histocompatible donor is available, allogeneic BMT should be considered for younger patients.^188–190

Postremission Therapy

Although CR is defined as the absence of morphologically detectable leukemia on bone marrow aspirate, morphologic assessments are subjective and relatively insensitive. It is estimated that as many as 10⁹ leukemia cells may still be present in patients in apparent CR. Rare patients can remain in CR for 1 to 2 years without further treatment, but it has been generally accepted that some form of therapy after CR is required to achieve long-term disease-free survival. Data to support this contention are based on two randomized trials conducted well over a decade ago. In a trial conducted by the German Cooperative Leukemia Group, a subset of 37 patients did not receive postremission therapy for a variety of protocol and medical reasons; all of these individuals relapsed.¹⁹¹ The Eastern Cooperative Oncology Groups (ECOG) reported a randomized study in which patients achieving CR were randomized to either no therapy, lower-dose maintenance therapy, or an intensive postremission program.¹⁹² All of the patients in the no-treatment arm relapsed rapidly, with a median CR duration of 4 months, resulting in early termination of this arm of the trial. Similar results were noted in a smaller randomized study reported by Embury and colleagues.¹⁹³ Although timed sequential therapy in which patients receive additional chemotherapy during the early postinduction recovery phase has been associated with prolonged remissions in the absence of postremission chemotherapy,¹⁹⁴ it remains standard practice to administer chemotherapy after remission is achieved.

A variety of different approaches have been evaluated as postremission therapy. Considerable uncertainty and controversy remain about which approach is preferable. These different approaches can be generally categorized according to the intensity of the therapy administered. The terms consolidation and intensification have generally been used to describe therapies that are of equal and greater intensity, respectively, than the regimens used in initial induction therapy.¹⁹⁵ It should be noted, however, that these phrases have been used rather loosely in the literature, and readers are advised to assess carefully the doses stipulated in various regimens, the doses actually delivered, and the periods of aplasia experienced by the recipients. The term maintenance has generally referred to lower-dose therapy administered on an intermittent basis for months to years, often on an outpatient schedule, and frequently at doses that do not produce significant myelosuppression. Again, however, the ability to deliver such regimens as scheduled varies considerably. In addition, there have been combinations of consolidation and maintenance approaches, including the use of late intensification treatment for patients who have already had sustained remissions.¹⁹⁶ A wide variety of antineoplastic agents have been used including the agents successfully administered in initial induction, sometimes in different doses or schedules, as well as different classes of compounds, some of which have proven activity in AML, others of which may have had only limited activity.

From a historical perspective, earlier postremission approaches tended to focus on maintenance-type therapy administered on a monthly or a bimonthly basis for up to 2 to 3 years.^197–199 This is largely derived from the model of maintenance treatment successfully used in childhood ALL. In more recent years, in part based on some of the positive results reported following BMT,^200,201 the emphasis has shifted toward the use of intensive postremission consolidation with a particular focus on HIDAC.^202–208 There has been considerable variation in the doses and numbers of courses of HIDAC administered, and newer programs are being initiated assessing high-dose therapy with agents shown to be active in phase II trials but that have not previously been used extensively in nonrelapsed patients. Lastly, both allogeneic and autologous BMT have been advocated for selected patients in first remission.²⁰⁰

Although many nonrandomized studies purport to demonstrate better results, using different postremission approaches, these are invariably small studies with potentially selected, generally younger patient populations. Since a variety of prognostic factors significantly affect ultimate outcome, independent of the type of therapy administered, small studies that may reflect inadvertent patient selection, single or oligoinstitutional superiority, or even random luck must ultimately be confirmed by multi-institutional applicability.

Overall, in adults, it can be expected that the administration of some sort of postremission therapy will result in a median CR duration of 12 to 18 months with approximately 20 to 25% of complete responders remaining as long-term disease-free survivors. A review of approximately 1,700 CALGB patients who achieved CR demonstrated a relatively constant relapse rate of 4.7% per month during the first 6 months following CR.²⁰⁹ The overall failure rate decreased in subsequent 6-month intervals (3.5% per month in months 7–12 and 2.4% per month in months 13–18), with a further flattening of the curves after this point. It is likely that patients relapsing earlier have leukemia that is more drug-resistant than those who relapse late.¹⁶⁰ Many clinical/laboratory correlative studies that evaluate prognostic factors and mechanisms of drug resistance now combine individuals with very short CR durations together with individuals with primarily resistant leukemia because of probable similarities in the underlying biology of their disease.

Earlier randomized trials of postremission therapy conducted by the CALGB^{162,165,166,199} have failed to demonstrate long-term benefits from (a) an alternate-month compared with a monthly schedule of maintenance therapy; (b) 3 years of relatively low-dose maintenance therapy compared with 8 months of a similar program (indeed, there was a modest survival advantage for patients randomized to stop therapy after 8 months); (c) doubling of the dose of ara-C from 100 to 200 mg/m² during maintenance therapy; and (d) addition of nonspecific immunotherapy in the form of methanol-extractable residue of BCG (MER) to maintenance therapy. An earlier ECOG study demonstrated no benefit from the addition of 2 years of maintenance therapy following two courses of postremission treatment with DAT.²⁰³ In contrast, results from Germany reported by Buchner and colleagues¹⁹¹ suggest a modest prolongation of CR duration, although with a questionable effect on long-term survival, when long-term maintenance therapy was administered after a single course of DAT. It is unclear whether the same results would have been achieved if more intensive postremission consolidation had been used before the maintenance.

Two important randomized trials have shown that HIDAC in the post-remission setting is better than lower doses of the drug. In an ECOG study, patients randomized to receive one course of very intensive postremission consolidation with a HIDAC-type regimen had a longer median duration of remission than patients receiving 2 years of lower-dose maintenance therapy.²¹⁰ A CALGB study provides additional evidence of the value of more dose-intensive postremission treatment.²¹¹ Five hundred ninety-six AML patients achieving CR were randomized to receive four courses of ara-C administered at three different dose levels (100 mg/m² by continuous IV infusion [CIV] for 5 days, 400 mg/m² CIV for 5 days, 3 g/m² IV for 3 hours q12h on days 1, 3, and 5 [total 6 doses/course]). There was no benefit from the higher-dose arms in patients > 60 years of age (see Fig. 124.1, bottom) with a median duration of CR of approximately 13 months and with only 10 to 12% long-term DFS. In addition, there was a substantial incidence of CNS neurotoxicity in the older patients manifested primarily as cerebellar dysfunction. In contrast, patients younger than 60 years of age benefited substantially from the HIDAC regimen in terms of both relapse-free and overall survival (see Fig. 124.1, top). The long-term results in patients < 40 years of age were similar to those reported with autologous or allogeneic BMT.

The CALGB and ECOG studies strongly support the use of HIDAC-based consolidation programs in younger patients with AML. It is not known how many courses of such therapy are needed, how to most safely and effectively administer HIDAC, and whether the addition of other active agents will improve on these results. Analysis of the CALGB data suggests that the benefit from HIDAC was most pronounced in patients with favorable cytogenetic findings [t(8;21); and inv16] with much less effect in patients with unfavorable karyotypes typically associated with drug resistance.²¹² When medically feasible, most clinicians attempt to administer at least two to three courses of reasonably intensive HIDAC-based therapy postremission regimens. Programs that use HIDAC in the range of 2 to 3 g/m² for 8 to 12 doses often are only able to administer one to two courses of such treatment. Obviously, the decision about the type of postremission therapy must take into consideration the patient’s medical condition and the possible persistence of infection (particularly with fungal organisms) acquired during induction, as well as the ability to provide adequate platelet transfusion therapy, in an attempt to balance the risk of intensive postremission approaches with the potential benefit. Depending on the intensity of the consolidation program, a 5% mortality rate in CR is to be expected and must be carefully explained to the patient (see Table 124.3).

In addition, a number of clinical trials suggest that between 5 and 15% of patients achieving CR do not go on to receive intensive postremission programs because of physicians’ concerns about the patient’s ability to tolerate and survive such therapy. Such patients should receive some sort of lower-dose treatment if possible, and we frequently use ara-C (100–200 mg/m²/d by continuous infusion for 5 days or sometimes on a b.i.d. subcutaneous schedule), with or without 1 to 2 doses of an anthracycline for 2 to 4 courses, depending on patient tolerance. It is difficult to recommend a particular postremission approach in older patients not participating in clinical trials. Intensive chemotherapy has not proved beneficial compared with standard doses^211,213; some form of therapy, even low-dose ara-C,²¹⁴ is required, however, to yield any chance at long-term disease-free survival. Since there is no proven role for HIDAC-based treatment in the older adult in remission, we generally use lower-dose regimens as described above. In view of the lack of proven benefit and the expense and morbidity associated with long-term maintenance therapy, prolonged treatment cannot be recommended.

Bone Marrow Transplantation

BMT represents the other major therapeutic option available as postremission therapy. Syngeneic BMT, using identical twins as donors, and allogeneic BMT, using HLA identical siblings, were first evaluated in patients with advanced, refractory AML.^215,216 After demonstrating an approximately 10% rate of long-term disease-free survival in this refractory group of patients and a 40 to 50% disease-free survival in patients in first remission receiving syngeneic transplants, studies of allogeneic transplantation in first remission AML were begun.^215,217 Early small series were promising, demonstrating a low rate of relapse with most of the approximately 50% mortality being related to complications of acute and chronic graft-versus-host disease (GVHD). Following publication of these initial results, the suggestion was made that all suitable patients with available donors should be considered for allogeneic transplant as the treatment of choice. For years, contentious debate ensued and still continues, despite several large prospective trials designed to address the role of allogeneic as well as the appropriateness of autologous BMT relative to chemotherapy in the postremission setting.^218–221

At least in part because of difficulties in identifying suitable donors for many patients as well as the complications of GVHD. considerable attention has been focused on the use of autologous BMT. It has been demonstrated that administration of cryopreserved autologous bone marrow can rapidly reconstitute heavily treated recipients, although marrow recovery, and particularly recovery of platelet count, may be delayed compared with allogeneic BMT recipients.²²² Recent experience using cytokine-mobilized peripheral blood stem cells (PBSC) suggests that the duration of neutropenia and particularly thrombocytopenia can be shortened further compared with the use of bone marrow.²²³ Because of the absence of GVHD, the procedure is much better tolerated and can be readily applied to patients up to 65 years of age and perhaps older. The major theoretical disadvantages of this approach include the absence of a graft-versus-leukemia effect, as well as concern that viable leukemic progenitors will be administered with the autologous stem cells. A number of purging techniques have been used including incubation with monoclonal antibodies directed against myeloid blasts,^224,225 as well as incubation with very high concentrations of cytotoxic agents that rather remarkably spare hematopoietic progenitors, although delayed marrow reconstitution is not uncommon.²²² It is unknown whether these in vitro manipulations are necessary, and randomized trials have not been reported. When autologous transplantation is performed early in CR, many centers use some type of purging technique. In studies in which autologous BMT was harvested after many months of CR, or after the administration of high-dose consolidation therapy, retrospective comparisons do not suggest an increased rate of relapse after the unpurged transplant, perhaps as a consequence of the therapy that had been administered during these intervening months.^220,226,227

A number of centers and groups have reported relapse-free survival rates in excess of 40% in recipients of autologous BMT in first CR.^{222,224,227–230} Historically, syngeneic transplant experience in this situation demonstrated relapse rates of approximately 50%, which is probably the best that can be expected with the autologous approach. Perhaps the strongest evidence of the potential of this approach consists of reports of apparent cure rates of 20% in patients transplanted in second and third remissions of AML.²²² The potential criticisms with regard to patient selection noted above are also applicable to most of the autologous BMT studies.

The controversy about whether a patient in first remission should receive chemotherapy, an autologous BMT, or allogeneic BMT has prompted prospective comparative trials dating back to the early 1980s. Several of the initial trials,^231–233 which compared the outcome in patients assigned to allogeneic BMT on the basis of having a histocompatible sibling to those receiving chemotherapy alone, documented a benefit for the transplant. However, the situation in recent times is more complex. Four recently available prospective trials each compared results in first remission patients genetically assigned to allogeneic BMT (usually only in those 45 years old or less) to autologous BMT versus chemotherapy. The latter comparison was based on randomized assignment. In the first of these trials, conducted in Europe,²¹⁸ 422 patients < 45 years of age (median 33 years) were either allocated to allogeneic BMT if an HLA-identical sibling was available or randomized to autologous BMT or a second course of consolidation chemotherapy with intermediate-dose ara-C and m-amsacrine. Disease-free survival was similar in patients undergoing allogeneic BMT (55% projected at 4 years) and autologous BMT (48%) and superior to the chemotherapy group (30%). Overall survival was similar, however, because many patients relapsing after chemotherapy could be successfully reinduced and undergo BMT in second CR. Many patients did not undergo treatment as randomized, and no information was provided about responses to the three treatments in different cytogenetic or FAB risk groups.

Another similarly designed trial performed in Europe by a French group failed to show a benefit for either type of BMT compared to patients receiving intensive postremission chemotherapy.²¹⁹ The MRC 10 trial conducted in Great Britain²²⁰ appeared to show that autologous BMT was beneficial. All patients not assigned to allogeneic BMT received three cycles of post-CR chemotherapy and were then randomized to autologous BMT (nonpurged) or observation. However, there was no significant difference in overall survival among the two groups. Allogeneic BMT was not significantly better than autologous BMT; in no cytogenetic risk group was one modality clearly better than another. The most recently published trial conducted by the North American Intergroup²²¹ actually found that chemotherapy was at least as good as autologous or allogeneic BMT in leading to cure, at least based on the intent-to-treat analysis. Many patients assigned to autologous BMT did not receive this therapy for a variety of reasons. Although the study was not powered to do so, an attempt was made to review the result according to cytogenetic status at diagnosis.²³⁴ In this preliminary analysis, both the favorable and unfavorable risk groups benefited from BMT.

Based on the available evidence, no firm conclusion can be drawn. However, a few general points can be made concerning the role of allogeneic BMT in the management of patients with AML in first CR. (a) The applicability of the technique is severely limited by patient age. Mortality from GVHD increases in a step-wise fashion from decade to decade, and most centers do not perform allogeneic transplantation in patients older than 60 years, thereby excluding most patients with AML. Whether so-called mini-allogeneic transplants (nonmyeloablative conditioning regimens)²³⁵ will change this is unclear. (b) Less than one-third of potential recipients have suitable HLA-matched family donors. Although alternatives include the use of partially mismatched donors or matched unrelated donors,^236–238 there is a considerable increase in the incidence of severe GVHD with both approaches and appreciable administrative delays in rapidly identifying suitable donors despite the available thousands of HLA-typed donors throughout the world. This is particularly true for patients with less common HLA types.²³⁹ (c) Although the relapse rate is lower, and allogeneic BMT can cure some patients with chemotherapy-resistant disease, there remains an appreciable relapse rate post-BMT, with at least some of the factors predictive of drug resistance and relapse after chemotherapy also applying to BMT recipients.^240,241 (d) A substantial component of the antileukemic effect following BMT correlates with the occurrence and severity of GVHD, presumably as a result of a graft-versus-leukemia effect. Attempts to attenuate GVHD with a variety of immunosuppressive approaches (T-cell depletion of the infused marrow, cyclosporin) are associated with a decrease in GVHD but also an increase in the relapse rate,^242,243 although selective T-cell depletion approaches hold promise in making BMT safer without an increased relapse risk.^244,245 (e) Although important advances in supportive care have been made, the consequences of the immunosuppression and direct organ sequelae of GVHD still represent formidable barriers, particularly with regard to infections with cytomegalovirus and interstitial pneumonia. (f) As many as 10 to 20% of surviving patients may have significant symptoms and impairment of performance status because of chronic GVHD. There is also a small increase in the development of secondary tumors in long-term survivors.²⁴⁶

It is hoped that additional ongoing studies, perhaps permitting (or eventually requiring) meta-analysis of results, will clarify many of these issues, particularly with regard to optimal treatments for different risk subgroups of patients. It would appear, however, that all approaches are in need of considerable improvement. Although the causes of failure are different (high rate of relapse with chemotherapy, lower rate of relapse but high mortality due to GVHD for BMT), historic results suggest that the outcomes may be similar with the inevitable conclusion that improved cytoreductive approaches are needed in both arenas. The greater the improvement in chemotherapy, the less will be the need for transplantation.

Therapy of Relapsed and Refractory AML

There are a number of agents available that have clear-cut activity when used alone or in combination in patients with relapsed or refractory AML, including amsacrine, mitoxantrone, diaziquone, idarubicin, fludarabine, 2-clorodeoxyadenosine, etoposide, homoharringtonine, topotecan, and carboplatin.^247–258 The pros and cons of these and other individual drugs or therapeutic regimens have been reviewed.^251,259,260 There is considerable heterogeneity among this group of patients, and a number of issues, in addition to the specific drugs used, influence the outcome of treatment. Most phase I/II trials enlisting new agents contain relatively small numbers of patients and frequently do not adequately describe their clinical characteristics. Some consistent trends are found in most such studies; however, (a) response rates are uniformly very low in patients with primarily refractory leukemia; (b) patients with short initial CR durations or who have relapsed while receiving postremission chemotherapy also tend to fare poorly with reinduction therapy; (c) secondary leukemias, or leukemias that evolved from a prior hematologic disorder, are particularly resistant to further therapy; and (d) patients in second and subsequent relapse tend to have a poorer prognosis than patients in first relapse, and the durations of subsequent remissions tend to decrease progressively.

Results from the Medical Research Council (MRC) in Britain, which prospectively followed all patients entered on an induction AML trial, are supportive of these conclusions.²⁶⁰ Overall, 155 of 531 relapsed patients (29%) achieved second CRs. The response rate with a variety of different reinduction regimens was directly related to the duration of the first remission (Table 124.4) and patient age, emphasizing the enormous influence of patient selection on the results of phase II trials. In the absence of comparative trials, differing results are more likely the consequence of patient selection than the superiority of a particular regimen. Others^249,261 have also reported results essentially identical to the MRC. In one study, patients whose initial remission was greater than 18 months had a reinduction rate of 64% (37/58) compared with a 29% CR rate in 278 patients with shorter initial remissions. The duration of second CR was also longer in the former group (8 months vs. 3 months), with three patients remaining in remission for more than 4 years. In addition to the duration of first CR, favorable karyotypes [inv16; t(8;21), or t(15;17)] were more common in patients with longer first remissions. Such patients tend to be younger, perhaps accounting for the improved results in younger patients noted by the MRC.

Table 124.4

Treatment of AML in Relapse*.

The type and timing of therapy for relapsed patients should therefore be individualized. Patients with other medical problems who have had poor responses to initial therapy have a small likelihood of sustained benefit from reinduction therapy and, indeed, may have life shortened by intensive therapy. Some such patients can be supported for many months with maintenance of a reasonable quality of life, with a more conservative approach using red blood cell and platelet transfusions and oral hydroxyurea to control elevated white blood cell counts or symptoms such as bone pain. However, there is no potential for long-term benefit. Conversely, patients with longer initial responses may derive prolonged benefit from intensive reinduction therapy.

Recurrence of leukemia can be detected in a variety of clinical circumstances. It is common to identify relapse at a time when the patient is asymptomatic, blood counts are normal, and there is modest marrow infiltration by blasts. Although it is logical to begin reinduction therapy at the earliest sign of relapse, when tumor burden is presumably the lowest, there are no data to demonstrate the validity of this approach except possibly for allogeneic BMT.²⁶² The drawback to early treatment is that reinduction therapy is often unsuccessful and may result in excessive early morbidity and premature patient death. Our general policy is to observe such patients until there has been a pattern of deterioration in blood counts but to initiate therapy while adequate numbers of circulating granulocytes and platelets are still present, prior to the development of symptoms or infection. Some patients have relapse diagnosed when bone marrows are done because of falls in blood counts. Such patients usually require reinduction therapy more immediately. In occasional patients, relapse develops in an explosive fashion with high circulating blast counts, fever, and infection. Although there are few prospective data available, it is likely that the outcome in this group of patients is poorer.

There are essentially no comparative trials to provide therapeutic guidelines. The initial decision can generally be outlined as a choice between reuse of drugs that have previously been effective in a given patient, compared with the use of new drugs, some of which may be investigational. If a patient has relapsed while receiving chemotherapy, it makes little sense to use these same agents for reinduction. The oft-quoted exception to this guideline is the use of HIDAC (2–3 g/m² q12h for 8–12 doses), even in patients who had recently received more conventional doses of this agent.²⁰⁵ Although CR can be achieved in this setting, the duration of CR appears to be short, and it is not clear that the HIDAC approach is of long-term value in such patients.

In contrast, in patients who have longer first remissions, there is logic to reusing initial treatment. There are no clear-cut guidelines as to what constitutes a satisfactory initial first remission. Many centers have arbitrarily felt that patients relapsing more than a year after completion of therapy should again receive the standard treatment. Although the CR rate in such patients is probably in excess of 50% when the original drugs are reused, it is not clear whether the same results could have been achieved in this favorable group using a variety of different tactics, including new investigational agents.

There is no proof of the benefit of postremission therapy in patients achieving second or third remissions. The obvious drawback of such an approach is that one is potentially costing patients time when they would be asymptomatic and out of hospital by the administration of therapy of variable toxicity but no proven long-term efficacy. At the moment, most reinduction studies do not specify that patients in second or later remission receive either maintenance or consolidation therapy.

Although some relapsed patients may have long-term benefit from chemotherapy, it is likely that BMT offers a greater chance of long-term survival than does chemotherapy alone. What is unclear is whether transplantation should be offered at the time of initial relapse or whether patients should be reinduced and transplanted when a second remission is achieved. Unfortunately, a large percentage of patients do not achieve second remission, and a significant fraction of patients develop medical problems as a consequence of reinduction therapy that preclude subsequent transplantation. Enthusiasm for transplantation in relapse derives largely from reports from Seattle,²⁶² indicating that the results of allogeneic transplantation in early relapse were equivalent to or better than those achieved in patients transplanted in second or subsequent remission. It is often a practical problem both to identify patients during early relapse and to be able to refer such patients for transplantation rapidly; the waiting lists at most transplant centers can preclude this approach. Nonetheless, it would probably be best to consider allogeneic transplantation for patients in early relapse if a suitable donor is available and the patient is under the age of 55 to 60 years. Similarly, allogeneic BMT can produce long-term survival in some patients with primary refractory leukemia, and it is advisable to HLA-type patients and their families at diagnosis to allow for this possibility. It is more difficult to use nonrelated donors for these purposes because of delays caused by tissue typing, identification of donors, and marrow procurement.

Patients who achieve second remission probably have a greater chance of long-term survival if transplanted in second remission. Data using both allogeneic and autologous approaches are promising in this regard, although it should be kept in mind that all such reports include highly selected patient populations, most of whom have had second remissions lasting a few months prior to transplantation. These patients have thereby identified themselves as having more chemotherapy-sensitive disease. Some centers now obtain and cryopreserve bone marrow from patients in first CR for use in either early relapse or as consolidation in second CR. Although this is an expensive, administratively cumbersome approach, it merits study in investigational settings, particularly in high-risk patients, prior to its more widespread application.²²⁴ Lastly, some selected patients who relapse following BMT can derive benefit from a second BMT and, in general, can receive and tolerate further chemotherapy.²⁶³

Hematopoietic Growth Factors

Because the rate of myelosuppression-associated mortality is so high, particularly after induction therapy is administered to older adults with AML, the cloning of a number of hematopoietic growth factors (HGFs) held promise to aid in amelioration of such side effects. Granulocyte-macrophage colony-stimulating factor (GM-CSF), granulocyte colony-stimulating factor (G-CSF),²⁶⁴ and interleukin-3 have been evaluated extensively; the potentially thrombopoetic factors MGDF and IL-11 are also in clinical development. The use of these agents in AML lagged behind that in solid tumors because of the concern that pharmacologic administration would stimulate the endogenous HGF receptors on the blasts, leading to proliferation (a phenomenon well recognized in the laboratory²⁶⁵) and a poor clinical outcome. Although fears regarding this clinical problem have been unfounded, the HGFs have not lived up to their promise for other reasons.

The first two available studies suggested that G-CSF²⁶⁵ and GM-CSF²⁶⁶ were safe and possibly effective in patients with poor-risk AML. In a pioneering study conducted in Germany, GM-CSF was administered to patients with AML shortly after completion of standard induction chemotherapy. Clinically overt stimulation of leukemia cell growth occurred very infrequently, and duration of neutropenia was slightly shortened compared with historical controls.

The growth factors were therefore considered for use in two circumstances: as hematopoietic support (to shorten neutropenic duration and decrease infectious deaths) and as “priming” (to stimulate leukemic cells into S-phase, thereby making them more sensitive to chemotherapeutic agents such as ara-C). In older patients, complications related to prolonged neutropenia are the major causes of failure to achieve remission. Randomized studies have recently been completed in which older patients received an HGF or placebo beginning on the day following the completion of initial induction therapy with the goals of assessing the duration of severe myelosuppression and the influence on complete response rate.^267–273 In addition, it was hoped that if leukemia proliferation was not a significant problem with growth factor given after cytoreductive therapy to patients with known leukemia, then the growth factor should also be usable with safety to attenuate neutropenia in patients in remission receiving intensive consolidation therapy.

Two large placebo-controlled randomized trials using different preparations of GM-CSF given after the completion of standard induction therapy in older patients with AML have been reported. Both trials noted a small decrease in the number of days of neutropenia. In a study of 121 patients reported by the ECOG, there was a significant reduction in infectious deaths and a trend toward a higher CR rate in the patients randomized to GM-CSF.²⁶⁷ There was no difference in the duration of neutropenia in the two groups following intensive consolidation therapy. A larger CALGB study evaluated 365 patients >60 years of age and failed to show a difference in the incidence of severe infection, deaths from infections, CR rate, duration of hospitalization, or survival.²⁶⁸ Other smaller studies suggest a modest shortening of the duration of severe neutropenia, but with variable or no effect on CR rate and infectious morbidity.^272,273 One published randomized trial using G-CSF after induction therapy for older patients with AML showed an improved CR rate for the G-CSF recipients.²⁷¹ Somewhat unexpectedly, the infectious mortality was similar in the two groups of patients, and the reason for the increased CR rate is unclear. There was no difference in overall survival. Other trials^269,270 evaluating G-CSF as a supportive agent showed a consistent effect on decreasing the duration of neutropenia by 5 d. Although one of these studies documented a reduction in the duration of hospitalization, there was no effect on CR rate, DFS, or infectious deaths. In aggregate, it would appear that the benefits of growth factors administered after the completion of induction therapy are likely to be modest at best. In contrast, those of G-CSF following intensive consolidation therapy have produced more marked decreases in the duration of neutropenia, albeit without effects on CR duration or survival. Because of the potential for eliminating the need for hospitalization,²⁶⁹ the use of HGF can be recommended following consolidation therapy. Preliminary studies with pegylated thrombopoietin^274,275 failed to demonstrate its ability to reduce the need for platelet transfusions during the induction treatment of AML.

Although there is in vitro evidence suggesting that the HGFs can increase the cytotoxic effects of ara-C, at least in part by increasing the fraction of cells in S-phase,^276–280 the clinical results with the priming strategy have been disappointing. An early study from the M.D. Anderson group suggested that GM-CSF administered before and during chemotherapy produced lower response rates and survival than seen in a historical control group treated with chemotherapy alone.²⁸¹ Studies in which patients have been randomized to receive GM-CSF or placebo before/during/after chemotherapy^272,273 also generally fail to show an advantage for patients primed with growth factor before initial induction therapy. Some of these studies used HIDAC, whereas others evaluated more conventional continuous-infusion schedules. Two recently presented abstracts suggest that GM-CSF priming does not have a role in older patients with newly diagnosed AML²⁸² or those with relapsed disease.²⁸³ Despite all of the relatively uninspiring results with HGFs, they are still widely used, especially during consolidation therapy,²⁸⁴ are safe, and may have a more important future role.

Minimal Residual Disease

A variety of techniques of varying sensitivity are being evaluated to detect residual leukemia cells in patients in morphologic CR, including (in approximate descending order of sensitivity) conventional cytogenetics, Southern blotting for known gene rearrangements, fluorescent in situ hybridization (FISH), multiparameter flow cytometry, and RT-PCR.^331,332 Technical problems abound, including the requirement for an already cloned abnormality when applying molecular techniques and the potential for changes in antigen expression when using immunologic monitoring. There are few large prospective studies available. Although preliminary data tend to be compatible with the logical premise that persistence of detectable disease presages eventual relapse, false-positive [see comments on t(8;21) above]⁶⁶ and false-negative rates may be appreciable and could result in incorrect decisions about further treatment. Serial monitoring is cumbersome, and it may be that detection of MRD at a particular time point after completion of therapy will be most convenient and prognostically significant. This appears to be true of persistence of the t(15;17) in APL after ATRA therapy.^323,333 A recent report from Italy suggests that a positive PCR test after all chemotherapy is completed is an ominous sign.³³⁴ The key question, of course, is whether earlier intervention with further therapy or allogeneic BMT will be of value if applied earlier, prior to gross relapse. There are no data addressing this issue. Because of the presence of residual disease, autologous collection and high-dose therapy may be predicted to be of less value in this circumstance.

Prognostic Factors

A number of clinical and laboratory factors have been described that can help predict the probability of attaining complete remission, the duration of CR, and, by extension, the probability of disease-free survival. It should be emphasized, however, that, with few exceptions, prognostic factors provide only relative guidelines to be used when discussing the probabilities of treatment outcome with individual patients. Conversely, they are of critical importance in focusing clinical and laboratory research efforts directed toward improving outcome. Most clinical factors are associated with decreased response to initial chemotherapy because they result in poor tolerance of the complications of prolonged pancytopenia. Advanced age (and particularly age > 70 years), renal insufficiency, congestive heart failure, concurrent infection, decreased serum albumin, and history of alloimmunization to blood products are examples.³³⁵ Many can be summed in the assessment of the patient’s performance status at the time of presentation.

As supportive care improves, some of these issues become less critical, and the intrinsic biologic characteristics of the patient’s leukemia become the dominant factors. Because the major cause of treatment failure tends to be associated with drug resistance, these biologic predictive factors usually are associated with both decreased complete response rates as well as short durations of remission should CR be achieved. A number of large multivariate analyses have consistently demonstrated that patients with hyperleukocytosis and/or FAB M5 morphology, increases in serum lactate dehydrogenase (which is probably simply a reflection of the two previously mentioned factors), prior history of myelodysplasia or other hematologic disorder, and FAB M6 or M7 and probably M0 morphology are associated with poor initial response rates.³³⁵ More advanced age is associated with decreased CR rates because of an increased rate of primarily resistant leukemia as well as difficulties with supportive care. There is evidence suggesting that age greater than 60 years may also presage shorter CR durations, probably because of association with less favorable cytogenetic findings.³³⁶ In contrast to patients with ALL, immunologic phenotyping using currently available reagents does not appear to identify distinct prognostic groups, except for the negative influence of CD34 positivity,^41,43,44 and the conflicting data about whether patients with AML and associated lymphoid antigens may have an improved or impaired outlook.¹³⁹ Studies have suggested that overexpression of bcl-2, with a presumed inhibitory effect on apoptosis³⁴¹ or underexpression of Rb, a tumor suppressor gene,³⁴² may be negative prognostic factors.

Major attention has focused on the prognostic impact of karyotypic abnormalities. A large number of studies demonstrate a high frequency of nonrandom cytogenetic abnormalities in the blasts from patients with AML.^339–342 The overall conclusions from these studies are quite consistent. In brief, clonal abnormalities can be identified in 60 to 75% of patients, although some studies maintain that abnormalities can be detected in greater than 95% and, indeed, perhaps in all patients with AML.³³⁹ Certainly, it is not unreasonable to assume that genetic abnormalities are present in all patients and that morphologic evaluation of chromosomal changes represents a relatively crude technique compared with molecular approaches. It is also clear that certain recurring abnormalities have a very tight relationship with both morphology and treatment outcome (see Table 124.1). The balanced translocations t(8;21), inv 16, and t(15;17) tend to occur in younger patients and area associated with a high incidence of initial complete response to treatment with a very low incidence of initial failure related to drug-resistant leukemia³⁴² and a better rate of overall disease-free survival. In contrast, multiple abnormalities or deletions of chromosomes 5 and/or 7 are rarely associated with cure. Those with normal chromosomes, trisomy 8, and abnormalities of chromosome 11q23, and inv3 are associated with an intermediate prognosis^342,343 (Fig. 124.12). In almost all studies, cytogenetic findings are an independent variable predicting outcome, and in some studies, they represent the major prognostic factor other than patient age. These findings have many intriguing biologic implications and have produced considerable interest in the possibility of specific therapy directed according to different risk groups. Although it is tempting to apply the knowledge of karyotypes to therapeutic decisions, a number of important and unanswered biologic and clinical questions remain.

Figure 124.12

Overall survival according to cytogenic findings in patients with de novo AML treated with intensive induction and postremission chemotherapy treated on CALGB studies.

Because of the close association between certain cytogenetic changes and morphology, the same prognostic information can frequently be gleaned by a careful assessment of the FAB classification or, as has been done in the past, by evaluation of patient age, performance status, and initial blast count. In order for the cytogenetic findings to represent more than a simple addition to this descriptive list, basic investigations into the mechanisms by which the specific cytogenetic changes are associated with these clinically recognizable effects are required. With the extraordinary pace of development of molecular biologic and immunologic techniques in the past few years, it is this aspect of the impact of cytogenetics that is potentially most exciting. These chromosomal alterations result in changes in the level or function of normal or mutated protein products, which result in the halt in differentiation that is morphologically recognized as acute leukemia as well as the associated cell kinetic and/or biochemical changes that may be correlated with resistance to therapy. Although it may be naive to assume that single DNA abnormalities are at the root of the biologically complex leukemic process, pharmacologic or molecular therapies, such as antisense fragments, might eventually become available for rational therapeutic attacks in specific chromosomal abnormalities.

Although cytogenetic analyses are not as yet routinely used to guide induction therapy, such data are used, perhaps with insufficient support, to guide the BMT versus chemotherapy postremission decisions. With regard to influencing decisions about induction, for example, it is possible to screen for gross deletions or additions of whole chromosomes or parts of chromosomes by performing Southern blots using appropriate molecular probes with densitometry to quantify the signal and estimate the number of alleles expressed.¹⁵¹ This technique depends on a relatively pure preparation of blasts as well as the assumption that the same finding is present in all of the leukemia cells. Another approach uses in situ hybridization with probes coupled to compounds that are easily visible on cell smears when viewed under fluorescence microscopy. This technique has been used to identify additions or deletions of individual chromosomes and translocation breakpoints for which probes have been adequately cloned.³⁴⁴ This technique permits correlation with cell morphology, can assess the frequency with which abnormalities are found within a given cell population, and can provide information within 48 to 72 hours of patients referral. Either approach requires the serial application of different probes directed against the most common and clinically relevant cytogenetic abnormalities, a tedious task potentially applicable to at most one half of patients, even as new probes are developed. In addition, these techniques are unlikely to replace standard cytogenetics because they do not easily detect other changes that are likely to be biologically and clinically relevant. Nonetheless, molecular procedures can be more sensitive both at the time of diagnosis and for the detection of minimal residual disease and are more readily adaptable to serial evaluations once the cytogenetic abnormality has been defined. Even if the technologies described above become readily available, it is likely that cytogenetic findings could correctly classify fewer than 50% of patients into prognostic groups. A significant fraction of patients have pseudodiploid karyotypes of uncertain prognostic impact, with others having miscellaneous hyper- or hypodiploid karyotypes. In addition, about one-third of patients have apparently normal karyotypes using current cytogenetic techniques, although it is probable that improved cell culture or molecular techniques will somewhat decrease this fraction of normals.

The real utility of adequate, complete, and rapidly available cytogenetic analysis will be in (1) further definition of genes at cytogenetic breakpoints to define pathophysiology and provide therapeutic clues, (2) assign prognosis, if in fact other cytogenetic abnormalities are shown to have independent importance, and (3) determine therapy if prospective evaluations demonstrate a benefit of one form of therapy relative to another for specific cytogenetic subgroup.

Hyperleukocytosis

Leukemic blasts are considerably less deformable than mature myeloid cells^345,346 and are “stickier” than lymphoblasts due to expression of cell surface adhesion molecules. With increasing blast counts, usually at levels greater than 100,000/μl in the myeloid leukemias, blood flow in the microcirculation can be impeded by plugs of these more rigid cells. Local hypoxemia may be exacerbated by the high metabolic activity of the dividing blasts, with endothelial damage and hemorrhage. The situation can be worsened by red blood cell transfusions that rapidly increase whole-blood viscosity in the presence of hyperleukocytosis, thereby further compromising local blood flow. Red blood cell transfusions should therefore be administered very slowly or, if possible, withheld until the blast count is reduced. Lastly, coagulation abnormalities, including disseminated intravascular coagulation, further increase the risk of local hemorrhage. Liberal use of platelet transfusions is recommended, particularly since the platelet count is frequently overestimated because of the presence of fragments of blasts on blood smears,³⁴⁷ which can be mistakenly counted as platelets by automated blood cell counters.

Although pathologic evidence of leukostasis can be found in most organs in patients with extremely high white cell counts, clinical symptomatology is usually related to CNS and pulmonary involvement.³⁴⁸ Occasionally, pulmonary symptomatology with worsening hypoxemia can occur with therapy and lysis of trapped leukemic cells. Spurious elevation of serum potassium can occur because of release from white blood cells during clotting, and it is sometimes necessary to measure potassium levels on heparinized plasma. Similarly, pO2 can appear falsely decreased because of the enhanced metabolic activity of the high level of white blood cells, even when the specimen is appropriately placed on ice during transport to the laboratory. Pulse oximetry provides an accurate assessment of O2 saturation in such circumstances. Hyperleukocytosis is more common in patients with myelomonocytic or monocytic leukemia,¹⁰² and it is possible that the clinical manifestations are exacerbated by the migration of leukemic promonocytes into tissue where further proliferation occurs.¹⁰⁴

The initial mortality rate for patients with AML and symptomatic hyperleukocytosis is high.^103,349 If patients survive the initial period, they tend to have somewhat lower remission rates. Remission durations are also shorter, possibly because of larger initial tumor mass, but more likely related to the biology and intrinsic chemoresistance of their leukemia.^103,349 Thus, symptomatic hyperleukocytosis in AML (and rarely in ALL) constitutes a medical emergency, and efforts should be made to lower the white blood cell count rapidly. In most patients, rapid cytoreduction can be achieved by chemotherapy, with either standard induction agents or with high doses of hydroxyurea (3 g/m²/d).³⁵⁰ Some centers also advocate low-dose cranial irradiation, including the retina, in order to prevent further proliferation of leukemic cells in CNS sites where drug delivery may theoretically be compromised. This treatment is well tolerated, although there are no comparative studies to determine whether the results are superior to chemotherapy alone. In some patients, however, the response to chemotherapy can be somewhat delayed, whereas in others, it is impossible to initiate chemotherapy immediately because of renal insufficiency, metabolic problems, delays in initiating allopurinol therapy so as to prevent hyperuricemia, or similar considerations. In such patients, emergency leukapheresis has been used to lower or stabilize the white count.^351,352 Although intensive leukapheresis, with procedure times often lasting many hours, can produce rapid improvement of pulmonary and CNS symptomatology, there are theoretic and practical limitations to its benefits. It is difficult, for example, for leukapheresis to affect already established vascular plugs, particularly if vascular invasion has taken place. In such cases, chemotherapy and possibly radiotherapy remain the primary modalities, although, theoretically, leukapheresis could decrease further accumulation of leukocytes at these sites. Furthermore, it is precisely the patient in whom leukostasis is most likely to occur, that is, the patient with high and rapidly rising blasts counts, in whom the technical limitations of leukapheresis are relevant. It is often difficult, even with highly efficient cell separators, to reduce the rapidly rising count. In such patients with a high proliferative thrust, cycle-specific chemotherapeutic agents are more likely to be most rapidly effective, however. Leukapheresis is often of modest benefit to patients who develop pulmonary problems during cytotoxic treatment, since in some such patients the symptoms are related at least in part to a local inflammatory response following leukocyte lysis. Leukapheresis can clearly be of value to the occasional patient to whom chemotherapy cannot be administered immediately.

Central Nervous System Leukemia

Involvement of the CNS is considerably less common in patients with AML than in both adults and children with ALL.^353–356 Most clinicians have the impression that the incidence has decreased even further in recent years, perhaps related to the use of higher doses of ara-C, which can penetrate into the CNS, during both initial induction and postremission therapies. It is therefore somewhat difficult to provide a precise estimate of the incidence of CNS leukemia in AML, although it has occurred in less than 5% of large clinical trials in recent years. Because of this low frequency, diagnostic lumbar punctures are not routinely indicated in the absence of CNS symptoms, and chemotherapeutic approaches do not include CNS prophylaxis. The most common presenting symptoms are consequences of increased intracranial pressure and usually consist of a constant headache, sometimes associated with lethargy or other mental changes. Cranial nerve signs (most commonly cranial nerves III or VI) and occasionally peripheral nerve manifestations are secondary to nerve root involvement and can be accompanied by headaches or occur alone.

The diagnosis is usually easily confirmed by examination of cytocentrifuge preparations of cerebrospinal fluid (CSF) after lumbar puncture. Cell counts can vary from as few as 5/μl to > 1,000. Most patients have moderate elevations in CSF protein with a moderate decrease in glucose. After the administration of intrathecal therapy, cytocentrifuge preparations can demonstrate reactive ependymal cells that, particularly if the leukocyte count is low, can be difficult to distinguish from leukemia cells. Treatment consists of the administration of intrathecal chemotherapy, with the addition of cranial radiation (usually 2,400 cGy), to patients who do not respond fully to chemotherapy or in whom cranial nerve involvement is present.³⁵⁷ Either methotrexate (12 mg/dose) or ara-C (50 mg/dose) can be used as initial therapy, with a cross-over to the other agents in the event of refractoriness or relapse. A typical schedule includes treatment 2 to 3 times a week until the CSF has cleared, generally occurring after a few injections. Treatment is then given at weekly intervals for two more doses to be followed by monthly administration, for a total of a year. An Omaya reservoir to permit intraventricular drug administration is frequently needed either because of difficulties in performing repeated lumbar punctures or because of concern that in some individuals, the CSF flow does not deliver sufficient amounts of the drug from the lumbar space to the entire CNS. Successful therapy with systemically administered agents that penetrate the CNS such as diaziquone (AZQ), high-dose methotrexate, or ara-C have also been reported.³⁵⁷ Unfortunately, the relapse rate, even after initially rapidly successful therapy, is high, either in association with bone marrow relapse or independently.

It has been suggested that the incidence of CNS leukemia is higher in patients with FAB M4 E0 morphology and possibly in patients with monocytic leukemia and high circulating blast counts.³⁵⁸ This may no longer be the case using more contemporary treatment regimens, and prophylactic therapy is not indicated in such individuals.

Ophthalmic Complications

Essentially every ocular structure has been described to be involved in the leukemias, sometimes dominating the clinical picture in the prechemotherapy era.³⁵⁹ Leukemia cells can infiltrate the conjunctiva and lacrimal glands, producing obvious masses that may require treatment with radiation therapy. Involvement of the choroid and retina is most common, however. A comprehensive, prospective study of 53 newly diagnosed adults with AML documented retinal or optic nerve (3 patients) abnormalities in 64% of patients.³⁶⁰ Hemorrhage and cotton wool spots (a consequence of nerve fiber ischemia) were most frequent, and the occurrence of these findings was unrelated to patient age, FAB type, white blood cell count, or hematocrit. Initial platelet counts were lower in patients with retinopathy. Ten patients had decreased visual acuity, including five with macular hemorrhages. It was felt that many of the cotton wool spots were either a consequence of or exacerbated by ischemia due to anemia. Definite leukemic infiltrate of the retina could not be confirmed. All patients received aggressive chemotherapy and platelet transfusion support; no patient received cranial or ocular irradiation. All ocular findings resolved in patients achieving CR, and there was no residual visual deficit in any patient. Infectious ocular problems were not noted and seem to be uncommon, perhaps because the rapid, empiric use of antibacterial antibiotics (and more recently of amphotericin B) has decreased the possibility of hematogenous spread of infections to the eye.

Pregnancy

AML is occasionally diagnosed during pregnancy, either because of clinical manifestations or as an incidental finding during blood count checks. If detected during the first trimester, termination of pregnancy followed by treatment of the leukemia is advisable. The management of patients diagnosed later in pregnancy, such as late in the second trimester or during the third trimester, is more problematic. In the latter group of patients, if the leukemia is relatively indolent, it is sometimes possible to manage patients conservatively with leukapheresis and/or transfusion with induction of labor and delivery of the fetus as soon as possible.³⁶¹ There have also been many reports of patients treated with chemotherapy later in their pregnancy.^362,363 The majority of these women have not aborted, and there have been no reports of leukemia occuring in the children, nor an increased incidence of fetal abnormalities in infants who have been exposed to AML chemotherapy during the later stages of gestation.

Metabolic Abnormalities

Patients receiving treatment for AML can experience a wide range of metabolic problems as a consequence of vomiting, diarrhea, impaired nutrition, or renal dysfunction, usually due to side effects from antibiotics or amphotericin B. Some metabolic disorders are related to the leukemic process itself. Hyperuricemia, occasionally accompanied by urate nephropathy with renal insufficiency, is the most frequent metabolic accompaniment of AML. All patients should receive allopurinol, 300 mg/d or more, as soon as the diagnosis of acute leukemia is established, so that the chemotherapy can be administered as soon as is medically appropriate. In most patients, the urate nephropathy can be avoided or ameliorated with vigorous hydration and urinary alkalization with systemic or oral administration of sodium bicarbonate. Allopurinol can usually be discontinued within a day or two after chemotherapy is completed.

Tumor lysis syndrome occurs more frequently in patients with ALL, although some AML patients experience hyperphosphatemia, hypocalcemia, hyperkalemia, and renal insufficiency as a consequence of massive leukemic cell death.³⁶⁴ The inciting cause seems to be release of large amounts of phosphate from lysed blasts, which co-precipitates with calcium in the kidneys, leading to hypocalcemia and sometimes to oliguric renal failure. Hyperuricemia further contributes to this problem, which is usually self-limited and responds to judicious hydration. Despite markedly hypercellular marrows, hypercalcemia is extremely unusual in patients with AML. Severe, occasionally symptomatic, hypokalemia is not infrequent, particularly in patients with monocytic leukemias. The mechanism appears to be renal potassium loss because of tubular damage induced by the high levels of lysozyme often noted in these patients.¹⁰⁸ Aggressive replacement with parenteral potassium is required; the syndrome usually abates after cytoreduction by chemotherapy. Lastly, rare patients have been described in whom lactic acidosis has been a constant metabolic accompaniment of the leukemia both at the time of presentation and relapse.³⁶⁵ The mechanism is unclear, although anaerobic metabolism by the leukemic cells at sites of leukostasis has been postulated.