Characterization of Extramedullary Acute Myeloid Leukemia – Results of the AML96 Trial

Myeloid Sarcoma (MS) is defined as an extramedullary mass composed of myeloid blasts occurring at an anatomical site other than the bone marrow. Furthermore, the term extramedullary manifestation (EM) is applied if it accompanies overt acute myeloid leukemia (AML) and represents non-effacing tissue infiltration. EM is reported to correspond often to the skin but can affect almost every site of the body. The prognosis of MS or EM has been discussed controversially in the past. EM at diagnosis of AML is generally thought to be a rare event. However, data defining the prevalence of EM at diagnosis of AML and its prognostic value are missing. The aim of this analysis was to provide data for estimating the prevalence of EM at diagnosis of AML and to determine its relevance by including clinical and laboratory data from patients being treated in the prospective AML96 trial of the Study Alliance Leukemia (SAL) study group.

A total of 326 patients with AML (age 17 – 83 years) and EM were treated within the AML96 trial with a median follow up of 8.8 years (95% CI, 8.4 to 9.3 years). All patients received double induction chemotherapy. Consolidation therapy contained high-dose cytosine arabinoside and for patients ≤ 60 years of age the option of autologous or allogeneic hematopoietic stem cell transplantation (HSCT). Logistic regression analyses were used to identify prognostic variables for CR rates. The method of Kaplan-Meier was used to estimate OS and EFS. Confidence interval (CI) estimation for the survival curves was based on the cumulative hazard function using the Greenwood's formula for the SE estimation. Survival distributions were compared using the log rank test.

17% of the AML patients entered into the AML96 trial were diagnosed with EM. In 313 of the 326 patients (96%) EM was evident at diagnosis. The majority of patients with EM were diagnosed with de novo AML (84%, n=273), whereas gingival infiltration (51%, n=166) displayed the main EM of AML with CNS involvement being less common (4%, n=14). The majority of patients had a cytogenetic intermediate risk profile (71%, n=221) with a total of 172 patients (56%) harboring a normal karyotype. Patients with EM had a statistically significant lower median CD34-positivity of bone marrow blasts, higher percentage of FAB subtypes M4 and M5, higher WBC counts and LDH at diagnosis and higher percentage of NPM1 mutations compared to those patients without EM (all p<.001). When comparing achievement of CR between patients with EM to patients without EM, no statistical difference between these two groups was observed. Analysis according to the NPM1/FLT3-ITD mutation status revealed highest 5-year-OS (37%, 95% CI: .24 - .508) and 5-year-EFS (36%, 95% CI: .224 - .448) in the NPM1-mut/FLT3-wt group and lowest 5-year-OS (12%, 95% CI: 0 - .261) and 5-year-EFS (4%, 95% CI: 0 - .124) in the NPM1-wt/FLT3-ITD group, p=.007 and p=.001, respectively. Of the 49 relapsed patients with EM who had a NPM1-mutation at diagnosis 48 deceased despite of intensified relapse therapies.

This analysis represents the largest study so far investigating the impact of EM AML. Patients with EM AML have distinct differences from AML patients without EM regarding their clinical and molecular characteristics at diagnosis. However these differences do not translate into differences in response to induction chemotherapy. Compared to patients without EM, survival analysis revealed differences according to the NPM1/FLT3-ITD mutation status which is also described for patients without EM AML. However, the prognosis for patients with EM who harbor a mutated NPM1 the prognosis at relapse seems to be dismal.

No relevant conflicts of interest to declare.