Chapter 15: Chronic Myeloid Leukemia

Chronic myeloid leukemia (CML) is a clonal hematopoietic stem cell disorder. The disease is driven by the BCR-ABL1 chimeric gene product, a constitutively active tyrosine kinase, resulting from a reciprocal balanced translocation between the long arms of chromosomes 9 and 22, t(9;22) (q34;q11.2), cytogenetically detected as the Philadelphia chromosome (Ph) (Fig. 15-1). Untreated, the course of CML may be biphasic or triphasic, with an early indolent or chronic phase, followed often by an accelerated phase and a terminal blastic phase. Before the era of selective BCR-ABL1 tyrosine kinase inhibitors (TKIs), the median survival in CML was 3–7 years, and the 10-year survival rate was 30% or less. Introduced into CML therapy in 2000, TKIs have revolutionized the treatment, natural history, and prognosis of CML. Today, the estimated 10-year survival rate with imatinib mesylate, the first BCR-ABL1 TKI approved, is 85%. Allogeneic stem cell transplantation (SCT), a curative but risky treatment approach, is now offered as second- or third-line therapy after failure of TKIs.

CML accounts for 15% of all cases of leukemia. There is a slight male preponderance (male:female ratio 1.6:1). The median age at diagnosis is 55–65 years. It is uncommon in children; only 3% of patients with CML are younger than 20 years. CML incidence increases slowly with age, with a steeper increase after the age of 40–50 years. The annual incidence of CML is 1.5 cases per 100,000 individuals. In the United States, this translates into 4500–5000 new cases per year. The incidence of CML has not changed over several decades. By extrapolation, the worldwide annual incidence of CML is about 100,000 cases. With a median survival of 6 years before 2000, the disease prevalence in the United States was 20,000–30,000 cases. With TKI therapy, the annual mortality has been reduced from 10–20% to about 2%. Therefore, the prevalence of CML in the United States is expected to continue to increase (about 80,000 in 2013) and reach a plateau of approximately 180,000 cases around 2030. The worldwide prevalence will depend on the treatment penetration of TKIs and their effect on reduction of worldwide annual mortality. Ideally, with full TKI treatment penetration, the worldwide prevalence should plateau at 35 times the incidence, or around 3 million patients.

There are no familial associations in CML. The risk of developing CML is not increased in monozygotic twins or in relatives of patients. No etiologic agents are incriminated, and no associations exist with exposures to benzene or other toxins, fertilizers, insecticides, or viruses. CML is not a frequent secondary leukemia following therapy of other cancers with alkylating agents and/or radiation. Exposure to ionizing radiation (e.g., nuclear accidents, radiation treatment for ankylosing spondylitis or cervical cancer) has increased the risk of CML, which peaks at 5–10 years after exposure and is dose-related. The median time to development of CML among atomic bomb survivors was 6.3 years. Following the Chernobyl accident, the incidence of CML did not increase, suggesting that only large doses of radiation can cause CML. Because of adequate protection, the risk of CML is not increased in individuals working in the nuclear industry or among radiologists in recent times.

The t(9;22) (q34;q11.2) is present in more than 90% of classical CML cases. It results from a balanced reciprocal translocation between the long arms of chromosomes 9 and 22. It is present in hematopoietic cells (myeloid, erythroid, megakaryocytes, and monocytes; less often mature B lymphocytes; rarely mature T lymphocytes, but not stromal cells), but not in other cells in the human body. As a result of the translocation, DNA sequences from the cellular oncogene ABL1 are translocated next to the major breakpoint cluster region (BCR) gene on chromosome 22, generating a hybrid oncogene, BCR-ABL1. This fusion gene codes for a novel oncoprotein of molecular weight 210 kDa, referred to as p210^BCR-ABL1 (Fig. 15-1B). This BCR-ABL1 oncoprotein exhibits constitutive kinase activity that leads to excessive proliferation and reduced apoptosis of CML cells, endowing them with a growth advantage over their normal counterparts. Over time, normal hematopoiesis is suppressed, but normal stem cells can persist and may reemerge following effective therapy, for example with TKIs. In Ph-positive acute lymphocytic leukemia (ALL) and in rare cases of CML, the breakpoint in BCR is more centromeric, in a region called the minor BCR region (mBCR). As a result, a shorter sequence of BCR is fused to ABL1, with a consequent smaller BCR-ABL1 oncoprotein, p190^BCR-ABL1. When occurring in Ph-positive CML, this translocation may predict for a worse outcome. A third rare breakpoint in BCR occurs telomeric to the major BCR region and is called micro-BCR (μ-BCR). It juxtaposes a larger fragment of the BCR gene to ABL1 and produces a larger p230^BCR-ABL1 oncoprotein, which is associated with a more indolent CML course.

The constitutive activation of BCR-ABL1 results in autophosphorylation and activation of multiple downstream pathways that modify gene transcription, apoptosis, skeletal organization, and degradation of inhibitory proteins. These transduction pathways may involve RAS, mitogen-activated protein (MAP) kinases, signal transducers and activators of transcription (STAT), phosphatidylinositol-3-kinase (PI3k), MYC, and others. These interactions are mostly mediated through tyrosine phosphorylation and require binding of BCR-ABL1 to adapter proteins such as GRB-2, CRK, CRK-like (CRK-L) protein, and Src homology containing proteins (SHC). BCR-ABL1 TKIs bind to the BCR-ABL1 kinase domain (KD), preventing the activation of transformation pathways and inhibiting downstream signaling. As a result, proliferation of CML cells is inhibited and apoptosis induced, leading to the reemergence of normal hematopoiesis. A plethora of signaling pathways have been implicated in BCR-ABL1-mediated cellular transformation. The emerging picture is a complex and redundant transformation network. An additional layer of complexity is related to differences in signal transduction between CML differentiated cells and early progenitors. Beta-catenin, Wnt1, Foxo3a, transforming growth factor β, interleukin-6, PP2A, SIRT1, and others have been implicated in CML stem cell survival.

Experimental models have established the causal relationship between the Ph-related BCR-ABL1 molecular events and the development of CML. In animal models, expression of BCR-ABL1 in normal hematopoietic cells produced CML-like disorders or lymphoid leukemia, demonstrating the leukemogenic potential of BCR-ABL1 as a single oncogenic abnormality.

The cause of the BCR-ABL1 molecular rearrangement is unknown. Molecular techniques that detect BCR-ABL1 at a level of 1 in 10⁸ identify this molecular abnormality in the blood of up to 25% of normal adults and 5% of infants, but 0% of cord blood samples. This suggests that BCR-ABL1 is not sufficient to cause overt CML in the overwhelming majority of individuals in whom it occurs. Because CML develops in only 1.5 of 100,000 individuals annually, it is evident that additional molecular events or poor immune recognition of the rearranged cells are needed to cause overt CML.

CML is defined by the presence of BCR-ABL1 abnormality in a patient with a myeloproliferative neoplasm. In some patients with a typical morphologic picture of CML, the Ph abnormality is not detectable by standard cytogenetic analysis, but fluorescence in situ hybridization (FISH) and molecular studies (polymerase chain reaction [PCR]) detect BCR-ABL1. These patients have a course similar to Ph-positive CML and respond to TKI therapy. Many of the remaining patients have atypical morphologic or clinical features and belong to other diagnostic groups, such as atypical CML or chronic myelomonocytic leukemia. These individuals do not respond to TKI therapy and have a poor prognosis with a median survival of about 2–3 years. Detection of mutations in the granulocyte colony-stimulating factor receptor (CSF3R) in chronic neutrophilic leukemia and in some cases of atypical CML and of mutations in SETBP1 in atypical CML confirmed that they are distinct entities.

The mechanisms associated with the transition of CML from a chronic to accelerated-blastic phase are poorly understood. They are often associated with characteristic chromosomal abnormalities such as a double Ph, trisomy 8, isochromosome 17 or deletion of 17p (loss of TP53), 20q–, and others. Molecular events associated with transformation include mutations in TP53, retinoblastoma 1 (RB1), myeloid transcriptions factors like Runx1, and cell cycle regulators like p16. A plethora of other mutations or functional abnormalities have been implicated in blastic transformation, but no unifying theme has emerged other than that BCR-ABL1 itself induces genetic instability that leads to the acquisition of additional mutations and eventually to blastic transformation. In this frame of thinking, one critical effect of TKIs is their ability to stabilize the CML genome, leading to a much reduced transformation rate. In particular, the previously observed sudden blastic transformations (i.e., abrupt transformation to blastic phase in a patient who had been in cytogenetic response) have become uncommon, occurring rarely in younger patients in the first 1–2 years of TKI therapy (usually sudden lymphoid blastic transformations). Sudden transformations beyond the third year of TKI therapy are rare in patients who continue on TKI therapy. Moreover, initial experience suggests that the course of CML has become significantly more indolent, even without cytogenetic responses, in patients on TKI-based therapy compared to previous experience with hydroxyurea/busulfan.

Among patients developing resistance to TKIs, several resistance mechanisms have been observed. The most clinically relevant one is the development of different ABL1 kinase domain mutations that prevent the binding of TKIs to the catalytic site (ATP binding site) of the kinase. More than 100 BCR-ABL1 mutations have now been described, many of which confer relative or absolute resistance to imatinib. This has resulted in the development of second-generation TKIs (i.e., dasatinib, nilotinib, bosutinib) and of a third-generation TKI (ponatinib) with selective efficacy against T315I, a mutation of the gatekeeper residue of the kinase that causes resistance to all other TKIs.

The presenting signs and symptoms in CML depend on the availability of and access to health care procedures, including physical exams and screening tests. In the United States, because of the easy access to health care screening and physical exams, 50–60% of patients are diagnosed on routine blood tests and have minimal symptoms at presentation, such as fatigue. In geographic locations where access to health care is more limited, patients often present with high CML burden including splenomegaly, anemia, and related symptoms (abdominal pain, weight loss, fatigue), as well as a higher frequency of high-risk CML. Presenting findings in patients diagnosed in the United States are shown in Table 15-1.

Symptoms

Most patients with CML (90%) present in the indolent or chronic phase. Depending on the timing of diagnosis, patients are often asymptomatic (if the diagnosis is discovered during health care screening tests). Common symptoms, when present, are manifestations of anemia and splenomegaly. These may include fatigue, malaise, weight loss (if high leukemia burden), or early satiety and left upper quadrant pain or masses (from splenomegaly). Less common presenting findings include thrombotic or vasoocclusive events (from severe leukocytosis or thrombocytosis). These include priapism, cardiovascular complications, myocardial infarction, venous thrombosis, visual disturbances, dyspnea and pulmonary insufficiency, drowsiness, loss of coordination, confusion, or cerebrovascular accidents. Bleeding diatheses findings include retinal hemorrhages, gastrointestinal bleeding, and others. Patients who present with, or progress to, the accelerated or blastic phases have additional symptoms including unexplained fever, significant weight loss, severe fatigue, bone and joint aches, bleeding and thrombotic events, and infections.

Physical findings

Splenomegaly is the most common physical finding, occurring in 20–70% of patients depending on health care screening frequency. Other less common findings include hepatomegaly (10–20%), lymphadenopathy (5–10%), and extramedullary disease (skin or subcutaneous lesions). The latter indicates CML transformation if a biopsy confirms the presence of sheets of blasts. Other physical findings are manifestations of complications of high tumor burden described earlier (e.g., cardiovascular, cerebrovascular, bleeding). High basophil counts may be associated with histamine overproduction causing pruritus, diarrhea, flushing, and even gastrointestinal ulcers.

Hematologic and marrow findings

In untreated CML, leukocytosis ranging from 10–500 × 10⁹/L is common. The peripheral blood differential shows left-shifted hematopoiesis with predominance of neutrophils and the presence of bands, myelocytes, metamyelocytes, promyelocytes, and blasts (usually ≤5%). Basophils and/or eosinophils are frequently increased. Thrombocytosis is common, but thrombocytopenia is rare and, when present, suggests a worse prognosis, disease acceleration, or an unrelated etiology. Anemia is present in one-third of patients. Cyclic oscillations of counts are noted in 25% of patients without treatment. Biochemical abnormalities include a low leukocyte alkaline phosphatase score and high levels of vitamin B₁₂, uric acid, lactic dehydrogenase, and lysozyme. The presence of unexplained and sustained leukocytosis, with or without splenomegaly, should lead to a marrow examination and cytogenetic analysis.

The bone marrow is hypercellular with marked myeloid hyperplasia and a high myeloid-to-erythroid ratio of 15–20:1. Marrow blasts are 5% or less; when higher, they carry a worse prognosis or represent acceleration (if they are ≥15%). Increased reticulin fibrosis (by Snook’s silver stain) is common, with 30–40% of patients demonstrating grade 3–4 reticulin fibrosis. This was considered adverse in the pre-TKI era. With TKI therapy, reticulin fibrosis resolves in most patients and is not an indicator of poor prognosis. Collagen fibrosis (Wright-Giemsa stain) is rare at diagnosis. Disease progression with a “spent phase” of myelofibrosis (myelophthisis, or burnt-out marrow) was common with busulfan therapy (20–30%) but is rare with TKI therapy.

Cytogenetic and molecular findings

The diagnosis of CML is straightforward and depends on documenting t(9;22)(q34;q11.2), which is found in 90% of cases. This is known as the Philadelphia-chromosome abnormality (discovered in Philadelphia) and was initially identified as a shortened chromosome, later identified to be chromosome 22 (22q–) (Fig. 15-1). Some patients may have complex translocations (variant Ph) involving three or more translocations that include chromosomes 9 and 22 and one or more other chromosomes. Others may have a “masked Ph,” involving translocations between chromosome 9 and a chromosome other than 22. The prognosis of these patients and their response to TKI therapy are similar to those in patients with Ph. About 5–10% of patients may have additional chromosomal abnormalities in the Ph-positive cells. These usually involve trisomy 8, a double Ph, isochromosome 17 or 17p deletion, 20q–, or others. This is referred to as clonal evolution and was historically a sign of adverse prognosis, particularly when trisomy 8, double Ph, or chromosome 17 abnormalities were noted.

Techniques such as FISH and PCR are now used to aid in the diagnosis of CML. They are more sensitive approaches to estimate the CML burden in patients on TKI therapy. They can be done on peripheral samples, and thus are less painful and more convenient. Patients with CML at diagnosis should have a FISH analysis to quantify the percentage of Ph-positive cells, if FISH is used to replace marrow cytogenetic analysis in monitoring response to therapy. FISH may not detect additional chromosomal abnormalities (clonal evolution); thus, a cytogenetic analysis is usually recommended at the time of diagnosis. The BCR-ABL1 RNA message is usually one of two variants: e13a2 (formerly b2a2) and e14a2 (formerly b3a2). About 2–5% of patients may have other RNA fusion types (e.g., e1a2, e13a3, or e14a3). In these patients, the routine PCR primers may not amplify the BCR-ABL1 transcripts, thus leading to false-negative results. Therefore, molecular studies at diagnosis are important to document the type and presence of BCR-ABL1 transcripts to avoid erroneously “undetectable” BCR-ABL1 transcripts on follow-up studies, with the misconception of a complete molecular response.

Both FISH and PCR studies can be falsely positive at low levels or falsely negative because of technical issues. Therefore, a diagnosis of CML must always rely on a marrow analysis with routine cytogenetics. The diagnostic bone marrow confirms the presence of the Ph chromosome, detects clonal evolution, i.e., chromosomal abnormalities in the Ph-positive cells (which may be prognostic), and also quantifies the percentage of marrow blasts and basophils. In 10% of patients, the percentage of marrow blasts and basophils can be significantly higher than in the peripheral blood, suggesting poorer prognosis or even disease transformation.

Monitoring patients on TKI therapy by cytogenetics, FISH, and molecular studies has become an important standard practice to assess response to therapy, emphasize compliance, evaluate possible treatment resistance, change TKI therapy, and order mutational analysis studies. It is thus important to recognize the comparability of these measures in monitoring response. A partial cytogenetic response is defined as the presence of 35% less Ph-positive metaphases by routine cytogenetic analysis. This is roughly equivalent to BCR-ABL1 transcripts by the International Scale (IS) of 10% or less. A complete cytogenetic response refers to the absence of Ph-positive metaphases (0% Ph positivity). This is approximately equivalent to BCR-ABL1 transcripts (IS) of 1% or less. A major molecular response refers to BCR-ABL1 transcripts (IS) ≤0.1%, or roughly a 3-log or greater reduction of CML burden from baseline. A complete molecular response usually refers to BCR-ABL1 transcripts (IS) <0.0032% (undetectable by current techniques), roughly equivalent to a more than 4.5-log reduction of CML burden from baseline.

Findings in CML transformation

Progression of CML is usually associated with leukocytosis resistant to therapy, increasing anemia, fever and constitutional symptoms, and increased blasts and basophils in the peripheral blood or marrow. Criteria of accelerated-phase CML, historically associated with median survival of less than 1.5 years, include the presence of 15% or more peripheral blasts, 30% or more peripheral blasts plus promyelocytes, 20% or more peripheral basophils, cytogenetic clonal evolution (presence of chromosomal abnormalities in addition to Ph), and thrombocytopenia <100 × 10⁹/L (unrelated to therapy). About 5–10% of patients present with de novo accelerated phase or blastic phase. The prognosis of de novo accelerated phase with TKI therapy has improved significantly, with an estimated 8-year survival rate of 75%. The median survival of accelerated phase evolving from chronic phase has also improved from a historical median survival of 18 months to an estimated 4-year survival rate of 70% on TKI therapy. Therefore, the criteria for accelerated-phase CML should be revisited because most have lost much of their prognostic significance. Blastic-phase CML is defined by the presence of 30% or more peripheral or marrow blasts or the presence of sheets of blasts in extramedullary disease (usually skin, soft tissues, or lytic bone lesions). Blastic-phase CML is commonly myeloid (60%) but can present uncommonly as erythroid, promyelocytic, monocytic, or megakaryocytic. Lymphoid blastic phase occurs in about 25% of patients. Lymphoblasts are terminal deoxynucleotide transferase positive and peroxidase negative (although occasionally with low positivity up to 3–5%) and express lymphoid markers (CD10, CD19, CD20, CD22). However, they also often express myeloid markers (50–80%), resulting in diagnostic confusion. This is important because, unlike other morphologic blastic phases, lymphoid blastic-phase CML is quite responsive to anti-ALL-type chemotherapy (e.g., hyper-CVAD [cyclophosphamide, vincristine, doxorubicin, and dexamethasone]) in combination with TKIs.

Before the imatinib era, the annual mortality in CML was 10% in the first 2 years and 15–20% thereafter. The median survival time in CML was 3–7 years (with hydroxyurea-busulfan and interferon α). Without a curative option of allogeneic SCT, the course of CML was inexorable toward transformation to, and death from, accelerated or blastic phases. The disease stability was unpredictable, with some patients demonstrating sudden transformation to a blastic phase. With imatinib therapy, the annual mortality in CML has decreased to 2% in the first 12 years of observation. Half of the deaths are from factors other than CML, such as old age, accidents, suicides, other cancers, and other medical conditions (e.g., infections, surgical procedures). The estimated 8- to 10-year survival rate is now 85%, or 93% if only CML-related deaths are considered (Fig. 15-2). The course of CML has also become quite predictable. In the first 2 years of TKI therapy, rare sudden transformations are still noted (1–2%), usually lymphoid blastic transformations that respond to combinations of chemotherapy and TKIs followed by allogeneic SCT. These may be explained by the intrinsic mechanisms of sudden transformation already existing in the CML clones before the start of therapy that were not amenable to TKI inhibition, in particular imatinib. Second-generation TKIs (nilotinib, dasatinib) used as frontline therapy have reduced the incidence of transformation in the first 2–3 years from 6–8% with imatinib to 2–4% with nilotinib or dasatinib. Disease transformation to accelerated or blastic phase is rare on continued TKI therapy, estimated at <1% annually in years 4–8 of follow-up on the original imatinib trials. Patients usually develop resistance in the form of cytogenetic relapse, followed by hematologic relapse and subsequent transformation, rather than the previously feared sudden transformations without the warning signals of cytogenetic-hematologic relapse.

Before the imatinib era, several pretreatment prognostic factors predicted for worse outcome in CML and have been incorporated into prognostic models and staging systems. These have included older age, significant splenomegaly, anemia, thrombocytopenia or thrombocytosis, high percentages of blasts and basophils (and/or eosinophils), marrow fibrosis, deletions in the long arm of chromosome 9, clonal evolution, and others. Different risk models and staging systems, derived from multivariate analyses, were proposed to define different risk groups. As with the introduction of cisplatin into testicular cancer therapy, the introduction of TKIs into CML therapy has nullified or lessened the prognostic impact of most of these prognostic factors and the significance of the CML models (e.g., Sokal, Hasford, European Treatment and Outcome Study [EUTOS]). Treatment-related prognostic factors have emerged as the most important prognostic factors in the era of imatinib therapy. Achievement of complete cytogenetic response has become the major therapeutic endpoint and is the only endpoint associated with improvement in survival. Achievement of a major molecular response is associated with decreased risk of events (relapse) and CML progression, may predict for differences in event-free survival (depending on the definition of an event) and for small differences in transformation rates, but has not been associated with survival prolongation. Among patients in complete cytogenetic response, survival is similar independent of whether they achieve a major molecular response or not. This may be due to the efficacy of salvage TKI therapies, which are and should be implemented at the first evidence of cytogenetic relapse. Achievement of complete molecular response (undetectable BCR-ABL1 transcripts), particularly when durable (>2 years), may offer the possibility of durable molecular response (molecular cure rather than functional cure) in the context of investigational trials and may allow temporary therapy interruption in women eager to have babies. The lack of achievement of major or complete molecular responses should not be considered as “failure” of a particular TKI therapy and/or an indication to change the TKI or to consider allogeneic SCT.

Pretreatment prognostic factors and prognostic models have lost much of their clinical relevance to define prognosis and to select different therapies. However TKI-associated therapeutic responses have gained major clinical relevance and dictate appropriate and careful monitoring of patients to optimize their treatment.

Routine physical exams and blood tests in the United States and advanced countries result in early detection of CML in most patients. About 50–70% of patients with CML are diagnosed accidentally, and high-risk CML as defined by prognostic models (e.g., Sokal risk groups) is found in only 10–20% of patients. This is not the same situation in emerging nations (e.g., India, China, African countries, the Middle East), where most patients are diagnosed following evaluation for symptoms and many present with high tumor burdens, such as massive splenomegaly, and advanced phases of CML (high-risk CML documented in 30–50%). Therefore, the prognosis of such patients on TKI therapy may be worse than the published experience.

The high cost of TKI therapies (annual costs of $90,000–$140,000 in the United States; lower but variable in the rest of the world) makes the general affordability of such treatments difficult. Although TKI treatment penetration is high in nations where cost of therapy is not an issue (e.g., Sweden, European Union), it may be less so in other nations, even in advanced ones like the United States, where out-of-pocket expenses may be prohibitive to a subset of patients (perhaps 10–20%). Based on the sales of imatinib worldwide and charity free drug supplies, it is estimated that less than 30% of patients are treated with imatinib (or other TKIs) consistently. Although the estimated 10-year survival in CML is 85% in single-institution studies (e.g., M.D. Anderson Cancer Center), in national studies in countries with TKI affordability (Sweden) (Figs. 15-2 and 15-3) or in company-sponsored studies (where all patients have access to TKIs throughout their care), the estimated 10-year survival worldwide, even 12 years after the introduction of TKI therapies, is likely to be less than 50%. The Surveillance, Epidemiology, and End Results (SEER) data from the United States report an estimated 5-year survival rate of 60% in the era of TKIs.

The current high cost of TKI therapies poses two additional considerations. The first are the treatment pathways and guidelines in nations where TKIs may not be affordable by patients or the health care system. In these conditions, there are trends of pathways advocating frontline allogeneic SCT (a one-time cost of $30,000–$50,000) despite the associated mortality and morbidities. The second is the choice of frontline TKI therapy once imatinib becomes available in generic forms (hopefully at much lower annual prices, e.g., $2,000–$10,000). This will depend on the maturing data in randomized studies of second-generation TKIs versus imatinib in relation to important long-term outcome endpoints, particularly survival, but also event-free survival and transformation-free survival.