Extramedullary disease (EMD) of acute myeloid leukemia (AML) was described decades ago but the incidence of this phenomenon and its prognostic impact are not clear. It is also uncertain whether every site of EMD has the same significance. This study explored these questions using a large cohort of AML patients treated on consecutive Eastern Cooperative Oncology Group (ECOG) frontline clinical trials.


For AML patients enrolled into ECOG clinical trials, the presence of EMD was captured at baseline on case report forms. From patients with newly diagnosed AML, age 15 and above, who were treated on 11 consecutive different clinical trials, we identified those with or without EMD, defined by physical examination, laboratory findings and imaging, without necessarily a biopsy. We used descriptive statistics to summarize patient and disease characteristics. Univariate analyses of potential prognostic factors were done. The Kaplan-Meier method was used to estimate median overall survival (OS) within each prognostic category and differences were explored using the log-rank test. Cox proportional hazards models were used to examine the effect of one-unit increases in continuous variables on OS and for multivariable analyses. Multivariate models were built using backward selection. Factors (or groups of factors) significant at the 0.10 level in univariate analyses were tested for inclusion in the model, and retained if they were significant at the 0.05 level.


Of the 3,522 patients enrolled in 11 different AML clinical trials, we excluded 281; for diagnosis of other types of leukemia (n=220), no EMD evaluation at baseline (n=41) or no survival data (n=20). The overall incidence of EMD was 23.8% (770 out of the remaining 3,241 patients). The sites involved were: lymph nodes 367 (11.3%) patients, spleen 234 (7.2%), liver 173 (5.3%), skin 146 (4.5%), gingiva 104 (3.2%), central nerve system (CNS) 32 (1%), peripheral nerve system (PNS) 8 (0.2%) and other sites 134 (4.1%). In 65 cases (8.4%) EMD was confirmed by biopsy. Most of the patients (64.4%) had only one site of EMD, 163 (21.2%) had 2 sites, 75 (9.7%) - 3 sites, 28 (3.6%) - 4 sites, and 4 patients (0.5%) each had 5 or 6 sites.

EMD patients compared to those without EMD; tend to be younger (median age 45.7 vs 52.9 years); male (57.7% vs 52%); poorer ECOG performance status (76.4% with ECOG 0-1 vs 85.9%) and with higher white blood cell count (WBC) at diagnosis (median of 41.6/µl vs 10.2/µl).

In univariate analysis, having EMD was associated with a shorter OS (P=0.006). Examination of individual EMD sites revealed that skin (P=0.002), spleen (P=0.0002) and liver (P=0.0007) but not CNS (P=0.35), PNS (P=0.53), nodal involvement (P=0.85) and gingival hypertrophy (P=0.14), were associated with poorer OS.

Using proportional hazards models for continuous factors, each additional site of EMD conferred a 9.4% increase in the risk of death.

In a multivariable model, after adjusting for known prognostic factors (such as: age, WBC count and cytogenetic risk group), the presence of EMD, number of EMD sites and any specific EMD site were not independently prognostic. Of 165 patients with known favorable cytogenetics EMD was present in only 22 patients (13%). This group appeared to have a worse prognosis, but the numbers are too small for a definitive assessment (figure 1).


Extramedullary disease is common in newly diagnosed AML with an incidence of almost 24%, but CNS involvement is very rare (0.95%). In contrast to published data, no site of EMD was found to have an independent prognostic impact in multivariate analysis. This large dataset emphasizes the importance of evaluating large number of patients and considering all known risk factors in the multivariate analysis.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution