High Level Expression of Wild Type FLT3 Is Associated with Poor Outcome and Selective Sensitivity to FLT3 Inhibitors in Childhood Acute Myeloid Leukemia: A Children’s Oncology Group Study

In AML, molecular prognostic markers (FLT3/ITD, WT1, NPM1, CEBPA, BAALC, e.g.) are increasingly used to complement classical cytogenetics in risk stratification. Perhaps the most important of these markers is FLT3/ITD, since it has profound prognostic significance and is pharmacologically targetable. In our prior studies using FLT3 inhibitors in both AML and ALL, we have seen samples demonstrating exquisite cytotoxic sensitivity despite the lack of FLT3 mutations. In these cases, we have shown the sensitivity to be due to high level expression of activated wild type (wt) FLT3 protein. Small studies in adult AML have suggested that high level wtFLT3 transcript expression may predict inferior clinical outcome. We hypothesized that quantitative expression of wtFLT3 may contribute to risk stratification in AML, and may identify patients that could benefit from treatment with FLT3 inhibitors. We examined diagnostic marrow samples from a cohort of 254 children with AML treated on CCG-2961. The samples were classified by FLT3 genotype (wtFLT3: N=216; FLT3/ITD, N=19, FLT3/ALM, N=19). We used qRT-PCR to determine the FLT3 expression level of the 254 samples and 10 normal bone marrow controls (NBM). Patients with FLT3 mutations had significantly higher FLT3 expression (median 10.9 fold NBM) than wtFLT3 patients (4.6 fold NBM, p<0.0001). Within wtFLT3 patients, FLT3 expression was highly variable, ranging from 0.003 to 95 fold NBM. A tight correlation (r=0.85) of FLT3 expression at the RNA and protein level was observed, with FLT3 protein expression measured by FACS after staining with PE-conjugated CD135 antibodies. We grouped the eligible patients into wtFLT3 expression quartiles and analyzed outcome from study entry (N=191) and from end Course 1 for patients in CR (N=151). There were no significant differences in known covariates (gender, age, race, WBC or cytogenetics) between the highest quartile and the lower quartiles. While there was no difference in induction CR rate (p=0.54) or OS from study entry (p=0.795) between the quartiles, the highest quartile (> 11.25 fold NBM, N=40) had an OS from the end Course 1 of 48 ± 18% compared to 71 ± 9% for the lower 3 quartiles (N=111, p=0.057). Various expression thresholds (from 10 fold NBM to 20 fold NBM in increments of 2 fold) were then examined. Hazard ratios (HR) for relapse and DFS were found to increase with each incremental increase in expression threshold. Using an expression threshold of 18 fold NBM, the HR for relapse and DFS were 2.3 (95% CI 1.2 to 4.7, p = 0.019) and 1.9 (95% CI 1.0 to 3.5, p = 0.042), respectively, for patients above the 18 fold threshold (N=20) vs. those below (N=131). Since FLT3/ITD is known to be a powerful predictor of poor outcome, we compared the outcome for wt FLT3 patients above the 18 fold NBM threshold with the FLT3/ITD patients treated on the CCG-2961 trial (N=53). Remarkably, the cumulative risk of relapse (60%) and DFS (30%) were essentially identical for these two groups.

Subsets of wt FLT3 patients from Low (N=13), Mid (N=11) and High (N=12) level expression groups with specimens remaining in the cell bank were randomly selected for in vitro MTT cytotoxicity testing with the selective small molecule FLT3 kinase inhibitor lestaurtinib over a dose range of 5 nM to 100 nM. The mean cytotoxic response at all doses was greatest in the High group, least in the Low group, and intermediate in the Mid group. At 50 nM lestaurtinib the cytotoxic responses were 56 ± 9%, 38 ± 7% and 21 ± 8% in the High, Mid and Low groups, respectively (p<0.0001). Remarkably, the cytotoxic response for the High group (56%) was similar to that of fifteen FLT3/ITD+ samples we previously tested under identical conditions (48%, published in

In conclusion, FLT3 mRNA and protein expression varies widely among patients with wt FLT3, and patients with the highest levels of wt FLT3 expression have a significantly increased risk of relapse and death. Furthermore, the leukemic blasts from these patients are exquisitely and selectively sensitive to FLT3 inhibition in vitro. Remarkably, high wt FLT3 expression was indistinguishable from FLT3/ITD in its strength as a poor prognostic factor and as a predictor of FLT3 inhibitor sensitivity. These data suggest that prospectively determining high level expression of wt FLT3 may be useful not only in identifying a high risk group, but also in selecting a group of patients for whom FLT3 inhibitor therapy may be indicated.