FACS Analysis of Stat3/5 Signaling Reveals Ligand Sensitivity As a Significant Prognostic Factor in Pediatric AML: A Children's Oncology Group Report

Abstract 938

Pediatric acute myeloid leukemia (AML) remains a devastating disease with a relapse rate near 50%. Kinase inhibitors are being incorporated into treatments, yet the signaling abnormalities targeted by these drugs remain poorly understood. Pathways involving Stat3 and Stat5 are known to be aberrantly active in AML, and increased activity has been associated with chemoresistant disease and poor outcome. Our hypothesis is that analysis of Stat3/5 activation patterns will delineate signaling abnormalities, and thereby guide the development of targeted therapies. We performed FACS analysis of the Stat3/5 signaling pathways in 140 diagnostic bone marrow samples from patients on the AML study CCG 2961. After thawing, cells were rested in serum-free medium for 2 hr. We measured tyrosine-phosphorylated Stat3 (pY-Stat3), serine-phosphorylated Stat3 (pS-Stat3) and pY-Stat5 in unstimulated cells, cells stimulated with low (1 ng/ml), intermediate (10 ng/ml), or high (100 ng/ml) dose G-CSF, and cells stimulated with low (0.5 + 1), intermediate (5 + 10), or high (50 + 100) dose IL-6 + soluble IL-6 receptor alpha. Constitutively phosphorylated proteins were quantified as the percent of events in the positive gate in unstimulated cells. Constitutive pY-Stat3 and pY-Stat5 were found to be common and highly variable. The median percent of pY-Stat3+ events was 37.3% (range 0.1–91.8%); the median %pY-Stat5+ was 18.6% (range 0.5–86.5%). In contrast, pS-Stat3 was generally low (median 3.63%, range 0.1–78.2%). There was non-significant correlation between %pY-Stat3+ and %pY-Stat5+. There was a significant correlation between %pY-Stat3+ and %total Stat3+, with Pearson correlation coefficient R=0.523 (p<0.001). Constitutively activated Src (pY-Src) was generally low (median 4.2%), although there was a population with 20–40 %pY-Src+, which also had high constitutive %pY-Stat3, suggesting that aberrant Src activation may account for high basal pY-Stat3 in some cases. Ligand responses were expressed as the fold increase in mean fluorescence intensity over unstimulated cells (δMFI). There was a strong correlation between pY-Stat3 and pY-Stat5 δMFIs to G-CSF (R=0.91, p<0.001), suggesting overlap of the pathways from G-CSF to each STAT. Likewise, the pY-Stat3 δMFI to G-CSF significantly correlated with the pY-Stat3 δMFI to IL-6 (R=0.343, p<0.001), suggesting convergence of the pathways from each receptor to Stat3. G-CSF receptor expression positively correlated with the GCSF-induced pY-Stat3 δMFI (R=0.396, p<0.001), whereas gp130 (IL-6 receptor) expression negatively correlated with the IL-6-induced pY-Stat3 δMFI (R=-0.262, p<0.005). None of the constitutive or ligand-induced pStat measurements, pY-Src, total Stat3, or receptor expression alone were significant predictors of survival. However, we did identify patterns of ligand sensitivity that predicted outcome. Patients with G-CSF-sensitive AML cells, defined as a pY-Stat3 δMFI > 2 in response to the low or intermediate G-CSF dose (33% of patients in the cohort), had a superior 5 year event-free survival (EFS; 57.8 ± 15% v. 37.2 ± 10%, p=0.025). A similar trend was observed for IL-6 sensitivity (p=0.1). When pY-Stat3 sensitivities to G-CSF and IL-6 were considered together, patients whose blasts were equally sensitive to both ligands had the best 5 yr EFS (69 ± 20%). Patients whose blasts responded to G-CSF only, or were unresponsive to both, had intermediate outcomes (5 yr EFS 38.5 ± 16% and 41.8 ± 14%, respectively). Surprisingly, patients whose blasts responded to IL-6 only had the worst outcome, with a 5 yr EFS of 25 ± 22% (p=0.047). These ligand sensitivity results highlight the importance of understanding not only molecular changes in AML, but also functional aberrancies. For example, cases with low constitutive and ligand-induced Stat3 activation may be resistant to Stat3 pathway inhibitors, whereas cases with high basal pY-Src or aberrant IL-6 sensitivity may be effectively treated with inhibitors of these pathways. In summary, multiparameter FACS analysis of Stat3 and Stat5 activities provides biologic and prognostic insight into the role of signaling aberrancies in AML, and this knowledge will promote rational implementation of new therapies.