CHAPTER 21: Acute Leukemias

Acute leukemias and related disorders are aggressive neoplasms caused by acquired somatic mutations in early hematopoietic progenitors. The most obvious pathologic feature in the acute leukemias is the accumulation of undifferentiated blasts in the marrow and other tissues, indicating that, unlike the myeloproliferative disorders, acute leukemias have defects that block or significantly retard differentiation. We now know that specific subtypes of acute leukemia are often associated with mutations that alter the function of transcription factors that are required for normal differentiation of hematopoietic progenitors (Table 21-1). Sometimes these mutations consist of chromosomal rearrangements that create chimeric fusion genes, in which one or both partners encode a transcription factor; in other cases, the pathogenic mutations are more subtle point mutations or deletions. In most instances, the net result of the mutations is to decrease the function of a transcription factor that is required for the differentiation of cells of one or another of the hematopoietic lineages. An exception is mutations in the NOTCH1 gene, which increase the transcriptional activity of the NOTCH1 protein.

The expression of NOTCH1 genes bearing gain-of-function mutations in mouse hematopoietic stem cells causes the rapid development of T-cell acute lymphoblastic leukemia/lymphoma (T-ALL), the disease that is specifically associated with NOTCH1 mutations in man, proving that these mutations are leukemogenic. However, similar experiments performed with other transcription factor genes bearing leukemia-associated mutations that decrease or interfere with a normal transcriptional activity usually fail to produce acute leukemia or do so only after very long periods of time. In fact, when expressed in hematopoietic stem cells, some mutated transcription factors cause bone marrow failure, suggesting that their primary effect is to block differentiation rather than to cause proliferation. Further evidence that transcription factor mutations are not sufficient to cause acute leukemia has been gleaned from retrospective studies using blood samples (so-called Guthrie cards) that are obtained routinely from newborns. Analyses of these samples by using sensitive polymerase chain reaction (PCR)-based assays have shown that certain transcription factor mutations were present at birth in children who developed acute leukemia as much as 10 to 12 years later. Thus, even acute leukemia may have a long prodrome, during which "pre-leukemic" clones bearing an initial mutation in a transcription factor gene must acquire other somatic mutations for full-blown disease to appear. These insights have provided support for a model in which the development of acute leukemia requires at least two complementary events, a class 1 mutation that drives proliferation and a class 2 mutation in a transcription factor that arrests differentiation (Fig. 21-1).

The class 1 mutations in at least some acute leukemias appear to be gain-of-function mutations in tyrosine kinases, similar to the mutations that are prevalent in the myeloproliferative disorders (Chapter 20). It has long been recognized that a subset of acute lymphoblastic leukemia/lymphoma of B-cell origin (B-ALL) is associated with the presence of the Philadelphia chromosome and a BCR-ABL fusion gene. At the cytogenetic level, this is the same lesion found in chronic myelogenous leukemia (CML), a classic myeloproliferative disorder that, if untreated, usually transforms to blast crisis and a picture identical to that of acute leukemia. At a molecular level, it turns out that the form of BCR-ABL protein in B-ALL is usually 190 kDa in size, slightly smaller than the 210 kDa form usually found in CML. However, the 190 kDa form of BCR-ABL activates the same pathways (RAS, JAK/STAT, and PI-3 kinase/AKT) as the 210 kDa form, differing mainly in having a stronger tyrosine kinase activity. When expressed in the hematopoietic stem cells of mice, both the 210 kDa and 190 kDa forms of BCR-ABL produce a myeloproliferative disorder similar to CML.

These observations suggested that BCR-ABL–positive ALLs and B-cell blast crises arising in CML might share the same complementary class 2 mutation in some critical transcription factor. Recently, this prediction has been confirmed with the detection of mutations in a gene called Ikaros in up to 90% of these two forms of acute leukemia. Ikaros is a transcription factor that regulates early stages of lymphocyte development. The Ikaros mutations result in expression of truncated forms of the protein that interfere with its normal function and thereby prevent the normal differentiation of early lymphoid progenitors. Compellingly, the Ikaros mutations found in the blast crisis stage are absent from the preceding stable phase of CML, linking them directly to the transformation of the myeloproliferative disorder to acute leukemia. It is hypothesized that other transformations, such as the progression of myelodysplastic syndromes to acute myeloid leukemia (AML), are also due to as-yet-unknown acquired class 2 mutations that lead to the loss of some critical transcription factor activity that is required for differentiation.

The mutations that conspire to produce acute leukemia are still being discovered, but it is clear that many different combinations of class 1 and class 2 mutations may be involved, including some that are described later, as well as additional mutations in tumor suppressor genes. The complexity of these molecular details can be daunting, but they are important for two clinically relevant reasons:

• Prognosis. The mutations found in a particular acute leukemia are often predictive of outcome with conventional therapies.

• Targeted therapies. In some instances, the mutated proteins are themselves the therapeutic target.

As targeted therapies become more prevalent, the molecular subtyping of acute leukemia will take on even greater importance. Indeed, it may soon be routine to completely sequence the genomes of acute leukemia cells (as well as other cancers), providing a complete understanding of all of the genetic changes that are responsible for an individual patient's tumor.

Acute lymphoblastic leukemia and lymphoblastic lymphoma encompass different clinical presentations of similar disease entities that are defined by the unbridled proliferation of immature lymphoid cells referred to as lymphoblasts. Acute lymphoblastic leukemia presents in the bone marrow and the peripheral blood, whereas lymphoblastic lymphoma presents as a mass within the bone or soft tissues. These different clinical manifestations relate in part to the origin of the tumor. B-cell tumors almost always arise in the bone marrow (where B cells develop) and present as "leukemias," whereas T-cell tumors often arise in the thymus (where T cells develop) and present as lymphomatous masses. For simplicity we will discuss these two forms of the disease together under the rubric of acute lymphoblastic leukemia/lymphoblastic lymphoma, or ALL.

ALL is the most common cancer of children. There are about 2500 to 3000 new cases of pediatric ALL in the United States per year. The disease also occurs throughout life in adults but makes up a much smaller fraction of the cancer burden in this segment of the population. The factors leading to the pathogenic mutations that drive ALL development are unknown. It most commonly appears sporadically in previously healthy, normal children. There are two major ALL subdivisions, ALL of B-cell origin (B-ALL) and ALL of T-cell origin (T-ALL). These two types of ALL are morphologically identical but differ in terms of their clinical characteristics, immunophenotype, and genetics.

B-CELL ACUTE LYMPHOBLASTIC LEUKEMIA/LYMPHOBLASTIC LYMPHOMA

B-ALL is the most common type of ALL, comprising roughly 85% of cases. The peak incidence is about age 3 years, which closely coincides with the time of peak bone marrow production of B cells during life. Groups that are at relatively increased risk include children of White or Hispanic origin as well as those with Down syndrome (trisomy 21).

The most common clinical features in B-ALL are related to the replacement of the marrow by lymphoblasts and the resulting pancytopenia. The patient usually presents acutely with several weeks of fatigue or listlessness due to increasing anemia, often accompanied by easy bruising due to thrombocytopenia. Neutropenia may lead to infection, producing fever and sometimes localizing signs. Inspection of the peripheral blood film reveals the presence of lymphoblasts, which may be few or numerous, as well as anemia, thrombocytopenia, and granulocytopenia of variable severity. Unusual patients with lymphomatous presentations may complain of pain in a single long bone or come to attention due to involvement of the skin. B-ALL tends to spread via the meninges to the central nervous system and also may involve other immunologically privileged sites such as the ovary and testis. Splenomegaly, hepatomegaly, and lymphadenopathy may also be present due to tumor infiltration, but these features are usually not prominent.

The diagnosis can be strongly suspected based on morphologic inspection of the blood or the marrow. By definition, the diagnosis in those with leukemic presentations requires that lymphoblasts, cells with fine chromatin, small nucleoli, and scant agranular cytoplasm (Fig. 21-2), comprise at least 25% of the marrow cellularity. However, definitive diagnosis requires immunophenotyping, which is usually carried out by flow cytometry. The tumors cells are positive for terminal deoxynucleotidyl transferase (TdT, an enzyme that is expressed only in immature B and T cells) and certain B-lineage proteins such as CD19, and negative for surface immunoglobulin, which appears only on mature B cells (Fig. 21-3). In unusual lymphomatous presentations, the neoplastic nature of the process is obvious from the effacement of normal tissue by sheets of lymphoblasts, and the diagnosis is usually established by performance of immunohistochemistry on tissue sections. The principal diagnostic difficulties arise in children with severe viral infections, in which the immune response may increase the production of normal precursor B cells in the bone marrow while suppressing hematopoiesis, sometimes resulting in modest granulocytopenia. It is usually not difficult to distinguish such processes from B-ALL because reactive populations of lymphoblasts include cells from all stages of early B-cell development and never replace the marrow. In contrast, neoplastic lymphoblasts tend to have a uniform immunophenotype representing one or another stage of early B-cell development.

Subtyping of B-ALL is based on cytogenetics, and particular cytogenetic abnormalities are strongly predictive of clinical outcome; for these reasons, cytogenetic studies should be obtained in all patients with ALL. The most common abnormalities, the involved genes, and their clinical correlates are listed in Table 21-2. Of note, B-ALL associated with BCR-ABL fusion genes, rearrangements of the MLL gene, or hypodiploidy are associated with poor outcomes with conventional therapies. Outcomes are also worse for patients under 1 or over 10 years. This is partly due to age-dependent differences in the frequency of particular cytogenetic abnormalities (Table 21-2). High white cell counts are also associated with worse outcomes, possibly because they identify patients with unusually high tumor burdens. Overall, B-ALL in children has an excellent prognosis, with complete remissions being obtained in >95% and cures in 75% to 85% of patients. In adults the outcome is more guarded, with cures being obtained in a minority of patients.

T-CELL ACUTE LYMPHOBLASTIC LEUKEMIA/LYMPHOBLASTIC LYMPHOMA

T-ALL makes up roughly 15% of ALL. The peak incidence is at about age 15 years, which coincides roughly with the age at which the thymus reaches its greatest size. For unknown reasons, there is a 2:1 male predominance. About two-thirds of T-ALL present as mediastinal lymphomas associated with large masses centered in the thymus, the organ that gives birth to normal T cells. The enlarging tumor mass often compresses structures such as the major airways and blood vessels, producing cough, shortness of breath, and superior vena cava syndrome, which is characterized by swelling and redness of the face and upper extremities due to blockage of venous return to the heart. The remaining third of cases present with predominantly bone marrow involvement and a leukemic picture identical to that seen with most B-ALL. T-ALL shows a somewhat greater tendency than B-ALL to produce organomegaly and lymphadenopathy, and like B-ALL, often spreads to the central nervous system.

The diagnosis is based on morphologic demonstration of lymphoblasts effacing tissues such as the thymus or bone marrow (Fig. 21-4). As in B-ALL, immunophenotyping is essential to confirm the diagnosis. T-ALL expresses terminal deoxynucleotidyl transferase and variable combinations of T-lineage antigens such as CD1a, CD3, CD4, and CD8 (Fig. 21-5). Unlike in B-ALL, cytogenetics has not proved helpful in predicting patient outcome. In addition to very common activating mutations in NOTCH1, other genetic mechanisms that induce T-ALL include chromosomal rearrangements or other aberrations that lead to increased expression of several other transcription factors, including TAL1, LMO1, and LMO2. Many of these dysregulated factors appear to interfere with the activity of E2A, a transcription factor that is required for proper B- and T-cell development. Activating mutations in tyrosine kinases are being sought but appear to be uncommon, and the identity of the class 1 mutations in most T-ALLs is uncertain. T-ALL also very commonly has loss-of-function mutations in CDKN2A, a complex locus on chromosome 9q that encodes two tumor suppressors, ARF, which inhibits p53, and p16, which inhibits several kinases that promote cell division.

In the past, T-ALL was considered to have a poorer prognosis than B-ALL, but with newer chemotherapy regimens, this difference no longer holds. Overall, 75% to 80% of T-ALL patients in the pediatric age group are cured, whereas the cure rates in adults are in the 40% to 50% range.

TREATMENT OF ALL

ALLs, regardless of molecular subtype, are very aggressive tumors with high growth rates and a proclivity to infiltrate many different organs. Unless effective therapy is given, patients succumb to the disease in weeks to a few months, usually due to complications related to bone marrow failure. Currently, childhood B-ALL and T-ALL are treated with an identical regimen that has several phases. The initial goal is to induce remission and clear the central nervous system of any infiltrating lymphoblasts. This is achieved with high doses of intravenous and intrathecal¹ chemotherapy, sometimes with cranial irradiation as well. Several studies have shown that complete clearance of the tumor at the end of induction (4 weeks), as assessed by either very sensitive flow cytometry or PCR-based assays, predicts whether patients will remain in remission, independent of other risk factors. Remission is then consolidated with multiple rounds of high-dose chemotherapy, which is followed with daily low-dose maintenance chemotherapy for up to another 2 to 2.5 years. Young adults also respond well to this very intensive pediatric treatment regimen. It remains to be seen whether use of intensive regimens in selected older adults will also be beneficial.

Along with therapy directed toward the malignant blast cell, supportive measures to prevent complications that are caused or exacerbated by chemotherapy are essential components of treatment. These consist of:

• Transfusions of both red cells and platelets, to reverse cytopenias stemming from leukemia and chemotherapy-induced myelosuppression.

• Careful monitoring for signs and symptoms of infection. In the setting of neutropenia, fever is often the only sign of bacterial or fungal infections, which can be rapidly fatal. Febrile patients require immediate culturing of blood and other body fluids followed by prompt treatment with broad-spectrum antimicrobial agents.

• Meticulous attention to fluid, electrolyte, and acid-base balances. Highly proliferative lymphoblasts are rapidly killed and lysed by chemotherapy, releasing large amounts of metabolites and electrolytes such as uric acid, potassium, and phosphate. These substances must be carefully monitored, and appropriate measures must be taken to prevent toxic buildup during treatment. For example, uric acid released into the blood and filtered by the kidney may precipitate in the renal tubules, leading to acute renal failure. This complication can be prevented by allopurinol, a drug that blocks the conversion of the soluble purines hypoxanthine and xanthine to poorly soluble uric acid.

• Psychological support and encouragement for both the patient and the family. It is essential that caregivers provide clear and unambiguous information, allay fears, and foster hope throughout the arduous and prolonged period from the time of diagnosis through the completion of therapy.

Unconventional therapies are needed for ALL patients with cytogenetic abnormalities that predict a poor prognosis, and for those who either fail to go into remission, or recur during therapy or after therapy is completed. These alternative approaches include intensified chemotherapy and bone marrow transplantation, which is the treatment of choice for patients with B-ALL associated with BCR-ABL fusion genes or MLL gene rearrangements. Imatinib and dasatinib, tyrosine kinase inhibitors that are effective in chronic myelogenous leukemia and other myeloproliferative disorders, also have been used in patients with BCR-ABL-positive B-ALL. Most tumors respond well to tyrosine kinase inhibitors initially but recur within a few months. Analysis of recurrent tumors usually shows the presence of BCR-ABL mutations that confer resistance to these inhibitors, indicating that the growth of the tumor still depends on signals generated by BCR-ABL. Recent studies have shown that administration of tyrosine kinase inhibitors in combination with conventional chemotherapy agents produces sustained remissions in most patients with BCR-ABL-positive B-ALL, and it appears that at least some of these patients may be cured.

AML encompasses a group of molecularly distinct entities that share a single morphologic feature, the accumulation of undifferentiated blasts of myeloid origin in the bone marrow. Very rarely, tumors identical to AML present as soft tissue masses variously referred to as myeloid sarcoma or granulocytic sarcoma, but such tumors inevitably progress to full-blown AML and are best thought of as unusual presentations of the same disease.

AML is the most common acute leukemia of adults. There are roughly 13,000 cases per year in the United States, with only 10% to 15% of these occurring in children. The incidence of AML begins to rise in middle age and peaks in those over 80 years of age. The risk is increased in those exposed to agents that damage DNA, such as radiotherapy and chemotherapy given for other forms of cancer, as well as patients with rare inherited defects in DNA repair, such as Fanconi anemia (Chapter 4). Certain inherited severe neutropenias, such as that caused by elastase mutations (Chapter 18), are also associated with an increased risk for AML. The incidence is slightly increased in smokers, presumably because of the presence of carcinogens in cigarette smoke, and is increased to a greater degree in individuals exposed to certain toxins, such as benzene. As already noted in Chapter 20, patients with a pre-existing myeloproliferative disorder or a myelodysplastic syndrome are at high risk for progression to AML.

The presentation of AML is similar to ALL, but with some subtle differences. Most signs and symptoms are related to anemia, thrombocytopenia, and neutropenia. Splenomegaly and lymphadenopathy are seen less frequently than in ALL, and involvement of the central nervous system is much less common. On the other hand, AML is more likely to infiltrate soft tissues such as the gums, particularly tumors with monocytic differentiation. Myeloid blasts are also larger and stickier than lymphoblasts. Accordingly, patients with very high blast counts sometimes present with pulmonary or central nervous system symptoms, owing to sludging of blasts in capillary beds. Other patients, particularly those with acute promyelocytic leukemia, present with disseminated intravascular coagulation, which ranges from subclinical, being detected only by laboratory tests of coagulation, to a severe coagulopathy associated with bleeding complications that may be fatal.

The diagnosis is based on morphologic inspection of the marrow and peripheral blood coupled with histochemistry, immunophenotyping, and cytogenetics. In AML the number of blasts in the peripheral blood varies widely, from 0 to >200,000/mm³. Thus, the absence of blasts in the peripheral blood (so-called aleukemic leukemia) does not exclude the diagnosis. The marrow is usually markedly hypercellular and contains little or no fat, but the marrow cellularity may also be normal (50% fat) or even decreased, especially in older patients. Regardless of the cellularity, increased numbers of blasts are present. The morphologic findings in AML vary considerably, as illustrated in Fig. 21-6. Individual tumors may be composed predominantly of myeloblasts (granulocytic precursors), monoblasts or early monocytic forms (monocytic precursors), megakaryoblasts, or (rarely) erythroblasts. In many cases, mixtures of blasts with varying degrees of granulocytic and monocytic differentiation are seen. The block in differentiation that is characteristic of AML also varies from nearly complete to partial. In the vast majority of cases, the number of blasts exceeds 20% of the marrow cellularity, which is taken to be diagnostic of AML. Rarely, the number of blasts is <20%, but cytogenetic studies show the presence of an aberration specific for AML (described in the following paragraph), allowing the diagnosis to be made. An additional morphologic finding that is specific for AML is the presence of Auer rods (Fig. 21-6A), red-purple needle-like inclusions derived from coalescence of abnormal primary granules that are commonly seen in neoplastic myeloblasts or progranulocytes. Histochemical stains for myeloperoxidase (a marker of myeloblasts) and nonspecific esterase (a marker of monoblasts) helps to confirm the diagnosis, as does flow cytometry, which typically reveals the presence of myeloid markers such as CD33, CD13, and CD117 and/or monocytic markers such as CD14. Many cases are also positive for CD34, a marker found on hematopoietic stem cells.

Subtyping of AML is based on cytogenetic findings, molecular features, and clinical context, each of which predicts prognosis and guides therapy. Selected molecular subtypes that are relatively common or clinically distinct are listed in Table 21-3. The cytogenetics of AML is complex; over 100 recurrent chromosomal aberrations have been described. Increasing numbers of other types of recurrent mutations also are being described; those that are most common, mutations in the transcription factor C/EBPα and the nuclear transport protein NPM, now define distinct subsets of AML. Finally, AML following genotoxic therapy is genetically and clinically unique and has its own disease category. Of note, gain-of-function mutations in various tyrosine kinases that likely function as pro-growth class 1 mutations have been described in a subset of cases. The most commonly mutated genes are FLT3 and c-KIT, both of which encode receptor tyrosine kinases that are normally expressed in early myeloid progenitors.

AML is generally treated with a standard induction regimen of high-dose chemotherapy followed by consolidation with additional chemotherapy. The drugs and schedules are different from those used to treat ALL, emphasizing the importance of making this diagnostic distinction. The critically important supportive measures used in ALL patients also apply to the management of those with AML.

Overall, patients with AML fare less well than those with ALL. Disease types with particularly poor prognoses, such as AML occurring after genotoxic therapy, may respond better to alternative, more aggressive therapies such as bone marrow transplantation (Chapter 26); however, morbidity and mortality associated with bone marrow transplantation increase sharply with age, excluding many older patients from this option.

Several forms of AML therapy that specifically target mutated oncoproteins are also in current use or under study. Acute promyelocytic leukemia associated with the t(15;17) translocation merits special mention in this regard. This form of AML often presents with leukopenia, few or no circulating blasts, and disseminated intravascular coagulation, the latter caused by procoagulants released from the abnormal azurophilic granules that are present in the neoplastic promyelocytes (Fig. 21-7). At a molecular level, the t(15;17) results in the formation of a chimeric gene encoding a PML-RARα fusion oncoprotein. Normal RARα (retinoic acid receptor-α) is a transcription factor that associates with a partner protein called RXR and binds to special DNA sequences called retinoid response elements (Fig. 21-8). The binding of retinoic acid (vitamin A) to the RARα/RXR complex causes a conformational change that leads to the recruitment of proteins called coactivators, which acetylate histones attached to DNA and turn on the transcription of adjacent genes. In myeloid progenitors, the genes regulated by RARα/RXR include some that are required for granulocytic differentiation. PML-RARα retains the ability to bind to RXR and DNA but is much less sensitive to retinoic acid. Thus, by occupying retinoid response elements, PML-RARα interferes with gene transcription and blocks differentiation.

Proof of this mechanism comes from treatment of patients with acute promyelocytic leukemia with high doses of all-trans retinoic acid (ATRA), which binds and activates PML-RARα/RXR complexes. Within 1 to 2 days of the start of ATRA treatment, the neutrophil count rises, sometimes reaching levels of 50,000/mm³, and disseminated intravascular coagulation stops. This new cohort of neutrophils harbors the t(15;17), proving that they are the differentiated progeny of the neoplastic promyelocytes. Because neutrophils are short-lived cells with no capacity for cell division, ATRA rapidly clears the bulk of the neoplastic cells. Patients treated with ATRA avoid the cytotoxic complications of conventional chemotherapy but still suffer from other complications. Most notably, shortly after beginning ATRA treatment, some patients develop dyspnea and pulmonary infiltrates due to exudation of fluid from capillaries, a pathologic change that is caused in some way by the differentiating myeloid cells. This complication responds well to temporary cessation of ATRA and treatment with glucocorticoids, which decreases the adhesiveness of neutrophils to endothelium (Chapter 18).

Differentiation therapy with ATRA is sufficient to get most patients with acute promyelocytic leukemia into remission but has not proven to be curative when given alone. To date, it has been used most widely in combination with conventional chemotherapy. More recently, it has been noted that arsenic salts (specifically arsenic trioxide) also target PML-RARα but through a different mechanism that involves increased degradation of the fusion protein. Early trials using a combination of ATRA and arsenic trioxide show response rates of over 90%, and almost all responsive patients have remained in remission, suggesting that this unique combination of targeted therapies will become the treatment of choice in this particular form of AML. Other targeted therapies under evaluation include tyrosine kinase inhibitors, which are being tested on AMLs associated with activating mutations in FLT3, and antibodies specific for myeloid surface proteins (eg, CD33) linked to toxins or radiochemicals.