ABSTRACT: In acute myeloid leukemia, the karyotype of the leukemic cell is the most powerful predictor of treatment outcome. Approximately 30% of cases of AML have an unfavorable karyotype, and if treated with conventional chemotherapy, a complete response rate of about 50% and a 5-year overall survival of 10% to 20% are expected. Of those in the unfavorable group, almost half will have a complex karyotype, which is associated with a poorer outcome, and 40% of those will have a monosomal karyotype, which carries an even worse prognosis. The best chance for cure for patients with an unfavorable karyotype is seen in those who achieve a complete response and proceed to allogeneic transplant. For patients who are not candidates for aggressive therapy, preliminary data suggest that outcomes at least equivalent to those seen with standard chemotherapy can be obtained using azacitidine or decitabine.
The karyotype of the leukemic cell is the strongest prognostic factor for response to induction therapy and survival in acute myeloid leukemia (AML). Three cytogenetic risk groups based on treatment outcomes exist: favorable, intermediate, and unfavorable. These groups account for 20%, 50%, and 30% of cases of AML, respectively. Overlapping subgroups within the unfavorable risk category include AML with complex cytogenetics and monosomal karyotype AML. The problem of AML with unfavorable-risk cytogenetics deserves special attention: this category of disease is a difficult clinical challenge, and the most appropriate therapeutic option may vary greatly—from the most aggressive we have to offer for some patients, to palliative care for others.
Favorable, intermediate, and unfavorable cytogenetics
The Southwest Oncology Group (SWOG), the Medical Research Council (MRC), and Cancer and Leukemia Group B (CALGB) each developed and adopted cytogenetic risk classifications more than 2 decades ago based on the outcomes of large prospective clinical trials.[1-3] Because the studies involved different patient populations and different treatments, it is not surprising that the resulting classifications, while similar, are not identical. For example, as shown in Table 1, in the SWOG categorization, patients with clonal abnormalities not recognized as within one of the three major groups are assigned to a fourth category with a predicted treatment outcome somewhere between the intermediate and unfavorable risk groups. CALGB’s risk categorization differs somewhat depending on whether one is interested in complete response (CR) rate, relapse risk, or overall survival (OS) (details not shown). There are also minor differences between classification schemas in the distinction between intermediate- and unfavorable-risk cytogenetics. More recently, in 2010, the MRC revised their classification system, with the major change being the movement of patients with various translocations involving chromosome 11 [with the exception of t(9;11) and t(11;19)] from the intermediate risk group into the unfavorable group.
While these classification schemas differ slightly, they are each able to distinguish three major groups of patients. Among patients younger than 65 years treated with standard chemotherapy, those with a favorable karyotype have CR rates in the 85% to 90% range and a 5-year OS of 50% to 60%. Those with intermediate-risk cytogenetics have CR rates of 65% to 75%, and a 5-year OS of 35% to 45%. For patients with unfavorable-risk cytogenetics, a CR rate of 45% to 55% and a 5-year OS of only 10% to 20% can be expected.[1-4]
In some cases of AML, multiple unrelated cytogenetic abnormalities will be seen in a single karyotype. If the number of abnormalities is sufficient, such cases are defined as having “complex” cytogenetics. In the case of SWOG and CALGB, three or more abnormalities are required for this definition, whereas the MRC requires four or more abnormalities for the cytogenetics to be considered complex. If these multiple abnormalities accompany one of the favorable-risk translocations, inv(16) or t(8;21), then CR rates and OS are slightly poorer than those seen in patients with the favorable-risk translocations alone, but they are still better than what would be expected with intermediate or unfavorable risk. Thus, cases with inv(16) or t(8;21) and complex cytogenetics remain within the favorable-risk group and are not considered as “complex” in most discussions. All other cases with complex cytogenetics are defined as unfavorable, and make up between a third and half of all cases of unfavorable-risk disease, or 10% overall. In the analysis that led to the MRC’s recent revision of their classification system, investigators found that with each additional abnormality seen in a karyotype, the risk of failing to achieve a CR increased (hazard ratio [HR] = 1.42), as did the risk of mortality (HR = 1.19). Thus, patients with four abnormalities did worse than those with three, and those with five or more did worse than those with four. In the SWOG analysis, it was found that patients with complex cytogenetics but without involvement of either chromosomes 5 or 7 had a CR rate of 50% and an OS of 20%, while patients who had complex cytogenetics and involvement of chromosomes 5 or 7 had a CR rate of 37% and only 3% OS.
The finding of a single monosomy is relatively common in AML, with monosomy 7 being the most frequent. The presence of a single monosomy (excluding loss of a sex chromosome) is associated with a negative outcome, with a 12% OS at 4 years in the Dutch experience. Patients with two or more distinct monosomies, or a single monosomy but with an additional structural abnormality, have an even worse prognosis, with a predicted 4-year survival of less than 4%.[6,7] Thus, this category of patient (two monosomies or one monosomy plus an additional structural abnormality) has been defined as having a “monosomal karyotype.” This group of patients has a very dismal prognosis if treated with conventional chemotherapy.
There is obvious overlap among patients with unfavorable, complex, and monosomal karyotypes. All patients with complex karyotypes, except those with inv(16) and t(8;21) fall within the unfavorable risk group and comprise between a third and half of the total. Also, virtually all (98%) of the patients with monosomal karyotypes fall in the unfavorable risk group and most (95%) have complex abnormalities. Monosomal karyotype AML accounts for about 40% of unfavorable-risk AML and two-thirds of cases with complex cytogenetics (see Figure 1). A recent SWOG analysis found that the 4-year OS of patients with complex cytogenetics but without a monosomal karyotype was 13%, but with a monosomal karyotype it was only 3%.
Biology of Unfavorable-Risk Cytogenetics
Specific mutations associated with unfavorable-risk AML
Unfavorable-risk AML likely represents many different biologic syndromes.
A number of specific genetic abnormalities are included in this category of disease. For example, t(6;9) is an uncommon translocation seen in about 1% of AMLs that results in a chimeric fusion gene between DEK (6p23) and CAN (9q34). Inv(3) or t(3;3) is seen in about 1% of AMLs and involves RPN1 (3q21) and EVI1 (3q26.2). Exactly how t(6;9) and inv(3) lead to AML is so far unclear. Rare cases of de novo AML are associated with the same t(9;22) translocation seen in chronic myeloid leukemia (CML); these cases do not appear to be CML in blast crisis since there is no prodromal syndrome and no splenomegaly. Deletions or mutations at 17p are seen in 5% to 9% of adult AMLs and are associated with the loss or dysfunction of the tumor suppressor gene TP53. Such losses or mutations are particularly common in patients with complex or monosomal karyotypes; in a recent study, 70% of such cases had TP53 abnormalities. TP53 abnormalities are also associated with losses of part or all of chromosomes 5 or 7, the most common cytogenetic changes in unfavorable-risk AML. These losses or deletions have led many to speculate that there might be tumor suppressor genes on chromosomes 5 and/or 7 whose loss leads to AML. Despite much work, no specific tumor suppressor gene has been identified and no simple explanation is available for the propensity for loss of part or all of these two chromosomes.
Integrated genetic profiling
An increasing number of non-random mutations have been found in AML of all risk categories. A recent analysis tested for mutations in TET2, ASXL1, DNMT3A, CEBPA, PHF6, WT1, TP53, EZH2, RUNX1, and PTEN in a series of 502 patients. This genetic profiling did not appear to offer additional diagnostic information for patients with unfavorable-risk cytogenetics. However, there were several categories of patients who are normally considered to have intermediate-risk disease who could be reclassified as high risk based on this analysis, including those without FLT3 mutations who have mutated TET2, mutated ASXL1, mutated PHF6, or MLL partial tandem duplications (PTD), and those who have FLT3 mutations with mutant TET2, DNMT3A, or MLL PTD.
Clonal architecture of unfavorable-risk AML
Several clinical and recent laboratory insights are providing a somewhat better picture of the development of at least some cases of unfavorable-risk AML. We know that the incidence of AML with unfavorable-risk cytogenetics increases as patients age, rising from around 30% in patients less than 55 years of age to more than 55% in those over age 75. Additionally, in a comparison of treatment-related AMLs vs de novo AMLs, the chromosomal abnormalities that are over-represented in treatment-related AMLs are abn(17p), abnormalities in chromosomes 5 and 7, complex karyotypes, and monosomal karyotypes. This listing of abnormalities almost exactly duplicates those that increase most with aging. Laboratory studies employing genome-wide copy-number analysis have found frequent somatic copy-number alterations not detectable on routine cytogenetic analysis, and perhaps not surprisingly, these additional abnormalities were more commonly found in patients with abnormal karyotypes. Very recently, the group from St. Louis performed whole-genome sequencing of paired samples of skin and bone marrow from patients with AML that evolved from a prior myelodysplastic syndrome (MDS). They were able to show that nearly all of the bone marrow cells in patients with MDS are clonally derived, and that the evolution of AML from MDS is the result of multiple cycles of mutation acquisition and clonal selection. The picture we are left with, then, is that there may be some cases of unfavorable-risk AML, such as those associated with t(6;9), in which a limited number of mutations may be sufficient to result in the disease. For such cases, one can hope that small-molecule inhibitors might be found. However, most cases of unfavorable-risk AML are likely the result of the accumulation of large numbers of mutations as patients age, with the development of progressively more disordered hematopoiesis and the eventual emergence of leukemia with multiple subclones. The development of effective small-molecule inhibitors for such AML patients will be a challenge.