With intensive chemotherapy, many acute myeloid leukemia (AML) patients will enter complete remission (CR). Empirically, the bone marrow is typically examined 14 days after initiating induction therapy, and re-treatment is commonly considered if significant residual blasts remain. However, many patients receiving single induction will enter CR without additional therapy despite substantial amounts or residual marrow blasts on day 14, leaving considerable uncertainty about the value of early marrow assessments. An emerging alternative approach to predict the efficacy of induction therapy includes the assessment of peripheral blood blast (PBB) dynamics. Rapid clearance of PBB, determined either by review of manual differential counts or flow cytometry is predictive of CR (likelihood of 76–90%) and overall survival. We investigated whether mathematical modeling of early PBB dynamics using automated complete blood cell (CBC) counts and the manual differential could further refine our ability to predict CR.
We identified 111 adult patients with circulating PBB who underwent curative-intent, single-cycle induction chemotherapy for newly diagnosed AML between April 1999 and December 2011. Therapy regimens included “7+3”-like (56.8%), and regimens of similar/higher (29.7%) or lower (13.5%) intensity. PBBs were quantified as WBC count from the automated CBC times percentage of blasts by manual differential (100 cells). Cytogenetic abnormalities, NPM1/Flt3 status, day 14 and recovery bone marrow data were extracted from patient records. In the 62 patients with >3 measurable PBB counts, the rate of PBB clearance was calculated by fitting an exponential decay curve to the data points of absolute PBB counts, starting on day 1 of chemotherapy. This subgroup of patients had similar baseline parameters as the whole group, including age, WBC count at diagnosis, cytogenetic risk, percent of secondary AML, therapy regimens were similar, and CR rates were identical (69 vs. 68%).
An exponential decay curve [N(t) = N0×e-xt, with x=decay constant] resulted in an excellent goodness of fit of early PBB dynamics (mean r2=0.93). Rapid PBB clearance was highly predictive of CR achievement, with an optimal cut-off of x=1.4 (corresponding to a 4.2-log reduction in tumor burden if maintained over the course of a weeklong chemotherapy) based on the receiver operating characteristic (ROC) curve. All but 1 of the 27 patients with x>1.4 achieved CR (positive predictive value [PPV]=96%) the only non-responder with x>1.4 had a combination of negative prognostic factors including secondary AML, unfavorable cytogenetics, older age, and lower intensity treatment. PPV of PBB clearance rate of 96% for predicting CR compared favorably to alternative previously published approaches such as day of PBB clearance or percentage of day 14 bone marrow blasts (84 and 85% respectively in our study). Day 14 marrow assessments did not add prognostic information in 26/27 patients who had fast PBB clearance rate (x>1.4). In univariate analyses, CR achievement was significantly correlated with a higher PBB clearance rate, younger age, primary AML, more favorable cytogenetics, and treatment intensity. In multivariate analyses including age, primary vs. secondary AML, cytogenetics, type of therapy, and PBB clearance rate (57 patients), only the PBB clearance rate remained statistically significantly associated with CR achievement. Importantly, information from CBC differentials is routinely available in most institutions and associated costs are low. Unlike determination of the exact day of PBB clearance that coincides with profound cytopenia, the PBB clearance rate is measured while PBB are still abundant, rendering the latter method less prone to sampling and observer errors.
our findings suggest that early marrow assessment may not be necessary in AML patients who experience rapid PBB clearance upon induction treatment initiation as they have an almost 100% chance of achieving CR.
Vainstein:Neumedicines Inc: Employment.
Asterisk with author names denotes non-ASH members.