Therapy Related Cellular Changes as a Tool for Prognosis in AML

Abstract

Background: The day of clearance of peripheral blood blast cells (PBBC) may be used as a prognostic marker in patients with acute myeloid leukemia (AML) treated with induction chemotherapy using cytarabine (ara-C) plus idarubicin (AI regimen: ara-C 1.5g/ m2 x 4 days, idarubicin 12mg/m2 x 3 days). Earlier PBBC correlates with improved overall survival (OS) in patients receiving AI. When extending this observation to other cohorts of patients receiving the AI regimen in combination with a targeted agent such as the FLT3 inhibitor sorafenib or the histone deacetylase inhibitor vorinostat, faster clearance of PBBC does not necessarily correlate with improved survival. We sought other ways of evaluating cellular changes with AI therapy, as a tool for prognosis in patients with AML. Methods: PBBC clearance (defined as PBBC ¼ 0% by CBC differential) was examined for patients with non-APL AML undergoing AI alone (n ¼ 168). We performed separate analyses to evaluate cellular changes. First we modeled the disappearance of PBBC by fitting daily blast counts to a reverse sigmoidal curve, according to the equation y(t) ¼ a(1- (1/(1+e ˇ (-c(t-b))))), where y ¼ absolute blast count (fitted data), t ¼ time, a ¼ initial absolute blasts, b ¼ time to response, c ¼ slope of blast disappearance. Second, we evaluated OS based on the concurrent change of white blood cells (WBC) and PBBC in the 2 days following chemotherapy. We evaluated the ratio of the changes and the direction of the change for both cell types. Results: We evaluated OS for groups based on 5 derived features of fitted reverse sigmoidal curves: A) initial height of curve, B) time to decrease 5%, C) time to decrease 95%, D) difference in time between 5% and 95% decrease, and E) ratio of change to height (slope of PBBC decrease). Using these derived features, we identified a method equivalent, but not superior, to using day of blast disappearance to classify patients into prognostic groups. Using derived feature D), and a cutoff of 2.7 days, OS was different between groups (p-value <0.001). In our next analysis, we found that the ratio of change of PBBC/WBC was not prognostic for OS. However, we did find the direction of change in absolute PBBC and WBC on day 1 (but not day 2) after initiation of therapy was a powerful prognostic marker for certain patients. Specifically, patients who had a decrease in PBBC and an increase in WBC on day 1 (n ¼ 6) had an excellent prognosis, while patients with an increase in both PBBC and WBC on day 1 (n ¼ 6) had a poor prognosis. Patients who had PBBC increase and WBC decrease (n ¼ 15) or both PBBC and WBC decrease (n ¼ 122), demonstrated a similar intermediate prognosis (p ¼ 0.082). Conclusions: