Therapy-related AML: long-term outcome in a large cohort of AML-patients with intensive and non-intensive therapy

Abstract

Therapy-related acute myeloid leukemia (t-AML) often exhibits adverse (genetic) features. There is ongoing discussion on the impact of t-AML on long-term outcome in AML. Therefore, we retrospectively analyzed clinical and biological characteristics of 1133 AML patients (225 t-AML patients and 908 de novo AML patients) with a median follow-up of 81.8 months. T-AML patients showed more adverse genetic alterations, higher age and more comorbidities as compared to de novo AML. Median OS in intensively treated t-AML patients was 13.7 months as compared to 39.4 months in de novo AML (p < 0.001). With non-intensive therapy, OS did not differ significantly (p = 0.394). With intensive therapy, significant differences in favor of de novo AML were observed in the ELN intermediate I/II (p = 0.009) and adverse (p = 0.016) risk groups but not within favorable risk groups (APL p = 0.927, ELN favorable p = 0.714). However, t-AML was no independent risk factor for OS (p = 0.103), RR (p = 0.982) and NRM (p = 0.320) in the multivariate analysis. A limitation of our study is an ELN 2010 risk stratification due to a lack of more comprehensive molecular data according to ELN 2022. We conclude that therapeutic algorithms in t-AML, in particular with regard to allo-HSCT, should be guided by ELN genetic risk rather than classification as t-AML alone. Our data support the WHO and ICC 2022 classifications, which include t-AML as diagnostic qualifier rather than a separate subcategory.

Fig. 1. Study design.

n number of patients, t-AML therapy-related AML.

Fig. 2. Baseline characteristics in t-AML-patients.

A Karyotype and genetic alterations at baseline in t-AML vs. de novo AML. B Distribution of the most frequent primary diseases in the t-AML cohort, C Types of therapies that had been applied prior to primary diagnosis of t-AML. t translocation, inv inversion, NHL Non-Hodgkin lymphoma, cHL classical Hodgkin lymphoma, MM multiple myeloma, ALL acute lymphoblastic leukemia, LPL lymphoplasmacytic lymphoma, CLL chronic lymphocytic leukemia.

Fig. 3. OS in t-AML vs. de novo AML (logrank test).

A OS within the entire cohort of intensively treated patients: survival with t-AML is inferior to de novo AML. B OS within the entire cohort of non-intensively treated patients: no significant difference between t-AML and de novo AML. C OS of patients with APL: no significant difference between t-APL and de novo APL. D OS of patients with ELN 2010 favorable risk: no significant difference between t-AML and de novo AML. E OS of patients with ELN 2010 intermediate risk: survival with t-AML is inferior to de novo AML. F OS of patients with ELN 2010 adverse risk: survival with t-AML is inferior to de novo AML. n number of patients, OS overall survival, t-AML therapy-related AML, APL acute promyelocytic leukemia, ELN European LeukemiaNet.

Fig. 4. Baseline characteristics of t-AML patients with significant impact on OS (logrank-test).

A Type of AML therapy, B ELN adverse risk, C sex differences, D age-related survival difference, E ECOG performance score, F comorbidity, G cardiovascular disease, H chronic renal insufficiency, I diabetes mellitus, J body weight, K active cancer disease, L preceeding therapies, M previous radio-iodine therapy, N Latency period to t-AML. n Number of patients, OS overall survival, t-AML therapy-related AML, ELN European LeukemiaNet, ECOG Eastern Cooperative Oncology Group, CCI Charlson comorbidity index, BMI body mass index.

Fig. 5. Risk factors for overall survival (OS) in 225 patients with t-AML.

A Univariate Cox regression shows hazard ratio defined as risk of death from several risk factors in t-AML. Factors that were mainly determined by the previous disease are depicted in turquoise letters. B Multivariate Cox regression shows HR defined as risk of death from factors with univariate significance in t-AML and reveals ECOG-PS > 1, an adverse cytogenetic/molecular risk according to the ELN 2010 classification, intensive therapy, BMI < 18.5 kg/m² and diabetes mellitus as independent risk factors for OS in t-AML (depicted in orange letters). WBC white blood cell count, Hb hemoglobin, PLT platelets, CCI Charlson comorbidity index, ECOG Eastern Cooperative Oncology Group, BMI body-mass index, ELN European LeukemiaNet, HR hazard ratio, CI confidence interval.

Fig. 6. Risk factors for long-term survival in 995 AML patients with intensive therapy.

A Univariate Cox regression. B Multivariate Cox regression. C Cause-specific hazard ratios for non-relapse mortality in the presence of risk factors with univariate impact. D Cause-specific hazard ratios for risk of relapse in the presence of risk factors with univariate impact. T-AML is no independent risk factor in the multivariate analysis of intensively treated patients, neither for OS, nor for RR or NRM. OS overall survival, RR risk of relapse, NRM non-relapse mortality, t-AML therapy-related AML, CCI Charlson comorbidity Index, ECOG Eastern Cooperative Oncology Group, ELN European LeukemiaNet, HR hazard ratio, CI confidence interval.