Abstract

Acute myeloid leukemia (AML) is a heterogeneous group of hematological malignancies with individual cases showing wide variations in response to treatment. Pre-treatment karyotype analysis may be used to stratify cases into one of three prognostic classes: favorable, intermediate or adverse. Favorable karyotypes, t(15;17), t(8;21) and inv(16), are associated with the presence PML-RARα, AML1-ETO and CBFβ-MYH11 gene rearrangements respectively. The adverse cytogenetic abnormalities are monosomy 5, monosomy 7, deletion of chromosome 5q, abnormalities of chromosome 3q and a complex karyotype. All remaining abnormalities are associated with an intermediate prognosis. Approximately 5–10% of cases cannot be stratified due to failure of cytogenetic analysis. Cryptic gene rearrangements, which cannot be detected by karyotyping, may lead to assignment of cases to incorrect prognostic classes. These problems may result in sub-optimal or over-treatment of patients. Using an artificial neural network-based analysis of HOX gene expression profiles generated by real-time quantitative PCR (RT-QPCR) we have previously shown that the favorable and intermediate cytogenetic classes are characterised by low HOXA5 (Ct value > 29.5) and high HOXB3 (Ct value < 25) expression respectively (

Blood
2006
;
108
: Abstract 2318
). We have now measured HOXA5 and HOXB3 expression levels by RT-QPCR in a fresh set of 78 newly diagnosed cases of AML (31 favorable, 38 intermediate and 9 adverse karyotypes as determined by conventional cytogenetic analysis). All 31 cases with favorable cytogenetics had HOXA5 Ct values > 29.5 with twenty-nine (93.5%) having Ct values > 33. Therefore, using a HOXA5 Ct ≥; 33 to define membership of the favorable prognostic class and HOXA5 Ct < 33 to define membership of a non-favorable class, 72 cases (92.3%) were correctly assigned (29 favorable and 43 non-favorable). Of the 4 cases with non-favorable cytogenetics, originally misclassified by HOXA5 expression profiling, two were subsequently found to have cryptic rearrangements of PML-RARα and therefore should have been included within the favorable group. This increased the percentage of cases correctly assigned by HOXA5 expression profiling alone to 94.9%. We also found that within the favorable group, AML with t(15;17) or t(8;21) was characterized by low expression of HOXB3 (Ct range 30.9 to 37.4, median Ct 34.1) whereas AML with inv(16) had a distinct signature characterized by higher HOXB3 expression (Ct range 26.4 to 30.1, median Ct 28.0). In addition, we have identified a subset of patients with intermediate cytogenetics who have high white cell counts, low HOXB3 expression and an inferior response to treatment. Therefore, in AML, the measurement of expression levels of only two HOX genes, HOXA5 and HOXB3 , complements cytogenetic analysis and may improve the yield of favorable gene rearrangements detected and provide additional prognostic information for cases with intermediate or failed cytogenetics.

Author notes

Disclosure: No relevant conflicts of interest to declare.