Time to Complete Remission As a Function of Kinetics of White Blood Cell Elimination and Recovery in Acute Myeloid Leukemia Patients Undergoing Remission Induction Chemotherapy

Background: While achieving complete remission (CR) is a major treatment milestone in acute myeloid leukemia (AML) patients (pts) undergoing induction chemotherapy (IC), it is the time to achieve CR (Tc) that has greater prognostic value in predicting subsequent survival. We are particularly interested in seeing whether trends in white blood cell (WBC) count elimination and recovery during induction phase can aid in prediction of Tc, thus enabling real-time prognostication.

Methods: Adult pts diagnosed with AML (excluding acute promyelocytic leukemia) at the Cleveland Clinic from 10/08 - 11/12 who underwent IC with 7+3 (cytarabine and anthracycline) and achieved CR were included. Pretreatment variables including age, gender, race, smoking status, body mass index at dx, peripheral blood counts at diagnosis (dx), peripheral and marrow blasts at dx, disease classification and metaphase cytogenetics (per CALGB/Alliance 8461 criteria) were assessed. For mapping WBC elimination and recovery kinetics, we collected data on total WBC, absolute neutrophil count (ANC) and absolute lymphocyte count (ALC) at several different time points after IC – day 1, day 7, day 14 (range, 13-17), day 21, day 35 (range 30-45) and the day CR was achieved. Time intervals selected for analysis included time taken for clearance of peripheral circulating blasts; time to reach nadirs for WBC, ANC and ALC; and time to ANC recovery (from ANC nadir to > 500/µL). Tc was assessed as a function of WBC, ANC and ALC measurements (at the above predefined time points); time intervals; and magnitude of drop and rise in WBC, ANC and ALC (log reductions and improvements in counts) using multivariable logistic and Cox proportional hazards models and stepwise regression using Akaike’s Information Criterion (AIC) for model selection. Due to multiple testing, parameters are reported as significant if p<0.01. Cell counts and time interval metrics were also analyzed as a function of pretreatment covariates using standard linear regression.

Results: Of 80 pts, 58% (n=46) were female, 79% (n=63) were Caucasian, 68% (n=54) were younger than 60 years, the median age at treatment was 53.5 years (range, 19 to 77) & 18% (n=14) received 2 cycles of induction therapy. Twenty one pts (26%) had secondary AML. Disease characteristics (per WHO classification) were – AML with recurrent cytogenetic abnormalities 31% (n=25), secondary AML 27% (n=22), therapy-related AML 6% (n=5), AML not otherwise specified 34% (n=26), myeloid sarcoma 1% (n=1) and unknown 1% (n=1). Cytogenetic risk groups per CALGB 8461 were – favorable 19% (n=15), intermediate 53% (n=42), unfavorable 23% (n=18), & unknown cytogenetics (n=5). In the final model of a multivariable stepwise regression analysis, none of the cell counts at the defined time points of IC predicted for Tc. Of pretreatment covariates, only male sex correlated slightly with longer Tc (p=.04). Longer times to reach ANC (p=.01) and ALC (p=.03) nadirs had borderline correlations with longer Tc. In contrast, longer times to reach ANC > 500/µL was significantly associated with delayed Tc (p<<.001). Greater fold change drop and rise in ANC counts correlated with Tc but was not significant at the p<.01 level.

Conclusions: In this study the only WBC cell kinetic parameter post IC that independently predicted Tc was time to ANC recovery, which seems intuitive. Biologically, this could represent resiliency of the underlying hematopoietic stem cell reserve which is a function of age, prior therapies or antecedent hematologic disorder. Interestingly, in our series, WBC kinetics did not correlate with the number of induction cycles required (1 vs. 2), age (using age cut-offs of 60 and 70 years) or body mass index. Although the concept of using WBC kinetics during IC as a clinical surrogate for chemotherapy responsiveness or toxicity (overdosing vs. underdosing) is appealing, it is not supported by our data.