The Impact of Therapy-Related and Secondary AML in Relation to Karyotype and Molecular Markers: A Study of the AMLSG.

Background: Patients (pts) with therapy-related AML (t-AML) or secondary AML (s-AML) after a myelodysplastic syndrome are considered to have an inferior outcome compared to de novo AML. Whether this is due to an adverse cytogenetic risk profile or by the fact of t/s-AML per se has not yet been determined.

Aims: To evaluate the frequency and prognostic impact of cytogenetic abnormalities and molecular markers in t/s-AML in comparison to de novo AML in a large cohort of adult AML pts.

Methods: Patients were entered on seven AMLSG treatment trials. The key inclusion criteria for this analysis were availability of cytogenetics and molecular markers. Genotypic groups were defined as follows: Acute promyelocytic leukemia (APL) [t(15;17)], core binding factor (CBF)-AML [t(8;21), inv(16)/t(16;16)], cytogenetically normal (CN)-AML, t(11q23), adverse cytogenetics outside a complex karyotype (AC) [abn(3q), −5/5q-, -7/7q-, abn(12p), abn(17p)], complex karyotype (CK, ≥3 aberrations) and all remaining aberrations (various).

Results: Between 1993 and 2006, 3083 adults (age range: 16–85 years) were registered and in 2604 karyotype and information about the type of AML were available; 196 (7.5%) had t-AML, 374 (14.5%) s-AML, and 2034 (78%) de novo AML. The median age was lower in de novo AML (50 years) compared to t-AML (58 years) and s-AML (57 years) (p<0.0001). Median WBC and LDH levels were higher (p<.0001) in de novo AML compared to t/s-AML. The distribution of genotypic groups according to type of AML revealed highest incidences of t-AML in the groups t(11q23) (23%), CK (17%), AC (14%), and of s-AML in the groups CK (23%), various (22%), and AC (21%), whereas incidences of de novo AML were highest in APL (92%), CBF (87%) and CN (82%). Response to induction therapy was higher in de novo AML (CR 72%) compared to t-AML (59%) and s-AML (52%). The logistic regression model for response to induction therapy revealed s-AML (OR, 0.53; p<0.001), age diff: of 10 years (OR, 0.71; p<0.0001), log10(WBC) (OR, 0.74; p<0.0001), and cytogenetics with CBF (OR, 2.77; p<0.0001), CK (OR, 0.22; p<0.0001), AC (OR, 0.29; p<0.0001), and various (OR, 0.69; p=0.01) in comparison to CN-AML as significant variables. Cox regression model on overall survival (OS) revealed s-AML (HR, 1.29; p=0.0004), age diff: of 10 years (HR, 1.40; p<0.0001), log10(WBC) (HR, 1.33; p<0.0001), and cytogenetics with CBF (HR, 0.53; p<0.0001), CK (HR, 3.20; p<0.0001), t(11q23) (HR, 1.76; p<0.0001) and AC (HR 1.97, p<0.0001) in comparison to CN-AML as significant variables. To better define the role of t/s-AML, subgroup analyses were performed according to genotypic groups. Of note, survival rates of de novo AML, t- and s-AML were almost identical in the genetically defined groups APL (p=0.93), CBF-AML (p=0.34), CK (p=0.33), AC (p=0.38) and t(11q23) (p=0.23), whereas differences were found in the groups various (p=0.0005) and CN-AML (p=0.02). Analyses of CN-AML including the FLT3-ITD and NPM1 status revealed significant differences in the subgroups defined by FLT3-ITDpos (p=0.05) and double negative (p=0.05), whereas in the group with a favorable genotype NPM1mut/FLT3-ITDneg survival rates were nearly identical (p=0.65).

Conclusions: The prognostic impact of t/s-AML is restricted to patients with CN-AML exhibiting an unfavourable genotype and patients within the heterogeneous group exhibiting various aberrations.