A previous letter to this journal discussed the importance of correlating increased bone marrow microvessel density with known prognostic variables in acute myeloid leukemia (AML)1 since earlier reports have not addressed this issue.2-5 One group correlated indirect evidence of angiogeneic potential (increasing levels of intracellular vascular endothelial growth factor [VEGF] protein) in newly diagnosed AML with shorter overall and disease-free survival and also found VEGF protein to be an independent prognostic variable. However, no relationship between VEGF protein with the traditional prognostic variables such as white blood cell or blast count, age, cytogenetic changes, performance status, or presence of an antecedent hematologic disorder was found.6

To address this issue, we collected 4 nonneoplastic control bone marrow specimens and 21 specimens that contained at least 80% AML from different treatment protocols. All samples were from the time of diagnosis unless otherwise noted in the data table. In a blinded fashion, we cultured 5 × 105 cells × 72 hours in 700 μL EGM (Clonetics) without human epidermal growth factor or bovine brain extract additives and 2% fetal bovine serum in 24-well plates in triplicate. The conditioned media from each well was collected, filtered, and placed over 1 × 104 human umbilical vein endothelial cells (Clonetics, Walkersville, MD) for an additional 72 hours. Endothelial cell proliferation was measured using the MTT assay.7

AML samples were classified into 2 broad, therapeutically relevant prognostic groups: favorable/intermediate (F/I) and unfavorable (U) according to published criteria.8,9Unfavorable cases (n = 12) were those with unfavorable cytogenetics, history, or morphologic evidence of myelodysplasia, or disease relapse within 6 months of diagnosis. Favorable/intermediate prognosis cases (n = 9) were those without these features.

As compared to controls (Table 1, Figure 1), 6 of 12 unfavorable but 0 of 9 F/I cases were able to support or enhance endothelial growth at least minimally. When this data were analyzed using the Duncan multiple means test for the prognosis versus the percent change, there is clear evidence of statistical difference (P < .05) between the F/I and unfavorable groups.

Table 1.
Fig. 1.

Effect of conditioned media from AML and nonneoplastic bone marrow samples on endothelial proliferation.

AML samples from patients with unfavorable prognostic features (defined in text) were significantly more likely to enhance endothelial proliferation. ○, non-neoplastic bone marrows; □, favorable/intermediate prognosis AML; ▵, unfavorable prognosis AML.

Other investigators have used AML culture supernatants as we have and demonstrated enhanced endothelial growth in vitro; however, without correlation with prognostic variables.4,10 Our preliminary results suggest that AML cases with unfavorable prognostic features are more likely to enhance endothelial proliferation in vitro than cases with favorable/intermediate prognosis. The nature of a possible relationship between prognostic variables and enhanced endothelial proliferation is intriguing. It may be that the complex karyotypic and molecular genetic changes in unfavorable prognosis AML blasts alter the expression of angioregulatory molecules such as VEGF, as suggested previously.6