Early clearance of peripheral blood blasts predicts response to induction chemotherapy in acute myeloid leukemia




Early marrow blast clearance 14 days after induction chemotherapy is an independent prognostic indicator of outcomes in acute myeloid leukemia (AML).


We evaluated the relationship between time to peripheral blood blast clearance after induction and disease status as assessed by day 14 and day 30 marrow biopsies in 162 patients with AML. Day 6 after induction was the optimal cutoff point determined by a receiver operating characteristic analysis and was selected to divide patients into early blast clearance (EBC; ≤6 days; n = 119) and delayed blast clearance (DBC; >6 days; n = 43) groups.


DBC patients were older, but otherwise the 2 groups were comparable. Marrow blast clearance on day 14 after induction chemotherapy was observed in 84% of patients in the EBC group and 60% in the DBC group. With a median follow-up of 1538 days, both relapse-free survival (RFS) (442 vs 202 days, P = .0017) and overall survival (OS) (930 vs 429 days, P < .0001) were longer in the EBC group, and a multivariable analysis showed that EBC independently predicted clearance of marrow blasts at day 14 (P = .0018), remission (P = .0179), RFS (P = .0171), and OS (P = .0122).


Early clearance of peripheral blood blasts after induction chemotherapy predicts for early marrow blast clearance, complete remission, RFS, and OS. Cancer 2012. © 2012 American Cancer Society.

Acute myeloid leukemia (AML) is characterized by a rapid clonal proliferation of neoplastic myeloblasts. Response to chemotherapy and achievement of complete remission (CR) following induction chemotherapy are prerequisites for long-term remission and cure.1 Although powerful prognostic indicators such as cytogenetics, preceding hematologic disorder, and residual disease after induction predict long-term outcomes,2-6 early predictors of response after initiation of chemotherapy are not available. Such predictors could marginalize the role of the commonly performed bone marrow biopsies at day 14 through day 16, and guide treatment-related decisions. Timing of blast clearance from blood can be considered a surrogate marker for in vivo sensitivity to chemotherapy and may predict remission. We therefore performed a retrospective analysis in a cohort of 162 adult patients with AML who were homogeneously treated at a single institution, to determine the relationship between time to peripheral blood blast clearance after induction chemotherapy and marrow blast clearance on day 14, achievement of remission as assessed by a bone marrow biopsy on day 30, relapse-free (RFS), and overall survival (OS).



Adult patients with histologically confirmed and previously untreated AML who received chemotherapy at Emory University, Atlanta, GA, between 1996 and 2005 were identified through a computerized search of the Leukemia database. A repeat bone marrow biopsy or institutional review of outside marrow slides was performed for all patients diagnosed at another institution who were referred for primary treatment of AML. Patients with acute promyelocytic leukemia and patients with no detectable circulating blasts were excluded from this analysis. Patients' basic demographic data (age, sex), disease-related information (histology, white blood cell [WBC] count at diagnosis, cytogenetics, lactate dehydrogenase level), post-induction complete blood count with manual differential, bone marrow biopsy results at day 14 and day 30, relapse, and survival information were extracted via chart review. Cytogenetic abnormalities were classified as favorable, unfavorable, intermediate, and unknown based on the Southwest Oncology Group (SWOG) criteria.7 Patients were enrolled and treated according to an institutional phase 2 trial that tested the safety and efficacy of high-dose cytosine arabinoside and daunorubicin induction chemotherapy. Induction therapy consisted of cytosine arabinoside given at 2 g/m2 intravenously, over 4 hours, daily for 6 consecutive days (high-dose cytosine arabinoside, or HiDAC), combined with daunorubicin (60 mg/m2; 45 mg/m2 for patients >60 years) daily for 3 days. Response was assessed by bone marrow biopsies performed on days 14 and 30. Patients who achieved CR received 1 additional course of induction followed by risk-adapted consolidations. Patients who did not achieve remission were allowed to receive a second cycle of induction with the same regimen. High-risk patients and patients with a matched sibling were offered an allograft, whereas others received HiDAC consolidation with or without autologous stem cell transplantation. Time to blood blast clearance was determined for each patient as the day of disappearance of blasts monitored by daily manual differentials. The complete blood counts and initial differential counts were routinely performed on an automated hematology analyzer (Coulter LH750; Beckman Coulter, Fullerton, Calif), and 100-cell manual slide differential counts were performed daily in all cases. Median time to blood blast clearance was determined for the entire group. Marrow was deemed clear from leukemic blasts (aplastic) on day 14 if the cellularity and the percentage of blasts were low (<15% and <5%, respectively). Complete remission was defined upon count recovery and using International Working Group criteria.8 This retrospective study was approved by the Emory University Institutional Review Board.

Statistical Methods

Correlation between time to disappearance of blood blasts, remission status, and marrow blast clearance on day 14 was analyzed using a logistic regression model. Multivariable logistic regression models were further applied to estimate their correlation after adjusting for effects from other covariates (age, sex, presenting WBC count, cytogenetic risk group, and lactate dehydrogenase). A receiver operating characteristic (ROC) analysis was performed to evaluate the predictive power of days from initiation of chemotherapy to disappearance of blood blasts on patients' remission and on clearance of bone marrow blasts at the day 14 marrow evaluation. Different areas under the ROC curve were compared using the chi-square test. The time to blast clearance that maximized the sum of sensitivity and specificity on the ROC curve was selected as the optimal cutoff value. In the survival analysis, the start time for each patient was the enrollment date, and patients who received hematopoietic stem cell transplantation in first CR were censored at the time of transplantation. Survival estimates were determined using the Kaplan-Meier method and compared using the log-rank test. A multivariable Cox proportional hazard model was employed to test the associations between time to blast clearance and OS and RFS, after adjusting for the other covariates. The significance levels were set at 0.05 for all tests. The SAS statistical package, version 9.2 (SAS Institute, Inc, Cary, NC), was used for all data management and analyses.



Between July 1996 and February 2005, 162 consecutive patients were diagnosed with AML, had circulating myeloblasts, and received induction chemotherapy. Median age was 53 years (range, 18-86 years). Median presenting WBC was 23,800/μL (range, 400-295,000/μL). Median time to blood blast disappearance was 5 days (range, 1-108 days) from initiation of chemotherapy. The optimal cutoff time to blast clearance with discriminating power to predict remission was 6 days (P = .0007; Fig. 1). Patients who cleared blood blasts in ≤6 days had early blast clearance (EBC), whereas those with >6 days had delayed blast clearance (DBC). A total of 119 patients cleared blood blasts in ≤6 days and 43 patients cleared blood blasts in >6 days after induction chemotherapy began. Patient characteristics were comparable between the 2 groups with regards to sex, presenting WBC, absolute peripheral blast count and percent marrow blasts at diagnosis, lactate dehydrogenase, and cytogenetic risk groups, but patients with DBC were older than those with EBC (Table 1).

Figure 1.

Receiver operating characteristic (ROC) analysis. The arrow shows the point of maximum sensitivity and specificity, with a predicted probability of remission of 81%. The corresponding cutoff value for time to blast clearance was 6 days from the time of initiation of chemotherapy.

Table 1. Patient Characteristics
CharacteristicEBC (n = 119)DBC (n = 43)P
  1. DBC indicates delayed blood blast clearance; EBC, early blood blast clearance.

Age, y; median (range)50.5 (18-86)59 (25-80).0005
Sex (male/female)64/5523/20.97
White blood cells, median (range)19,600/μL (400-295,000/μL)30,500/μL (400-195,000/μL).23
Absolute peripheral blood blast count, median (range)7235/μL (4-207,753/μL)17,000/μL (4-177,450/μL).07
Percentage marrow blasts (range)60% (20%-96%)81% (21%-95%).07
Lactate dehydrogenase, median (range)433 U/L (84-15,000 U/L)369 U/L (152-2155 U/L).77
Cytogenetic risk group, n (%)   
 Favorable20 (17%)2 (5%).23
 Intermediate64 (54%)27 (63%)
 Unfavorable30 (25%)13 (30%)
 Unknown4 (4%)1 (2%)
Secondary acute myeloid leukemia01 


Disease status after induction was assessed by bone marrow biopsies performed on day 14 (±2 days) and day 30 after induction chemotherapy in all patients. Clearance of marrow blasts on day 14 was observed in 99 of 118 patients (84%) and 25 of 42 patients (60%) in the EBC and DBC groups (P = .0018), respectively; and complete remissions were achieved in 106 of 118 patients (90%) and 23 of 42 patients (55%) in the EBC and DBC groups, respectively (P = .0122). One subject in each group was inevaluable for day 14 assessment and remission (1 early death, 1 with no marrow performed). Median time to blood blast clearance was 5.22 days (standard deviation, 2.26 days) for patients with aplastic marrow on day 14, and 8.08 days (standard deviation, 6.32 days) for those with persistent disease (P = .01).


Median follow-up was 4.2 years. Median RFS was 442 versus 202 days (P = .0017; Fig. 2), and median OS was 930 versus 429 days (P < .0001; Fig. 3) for the EBC and DBC groups, respectively. A Cox regression model showed that EBC independently predicted CR (P = .0179), RFS (P = .0171), and OS (P = .0122) (Table 2).

Figure 2.

Relapse-free survival of patients with early blood blast clearance (EBC, solid line) and delayed blood blast clearance (DBC, dashed line)

Figure 3.

Overall survival of patients with early blood blast clearance (EBC, solid line) and delayed blood blast clearance (DBC, dashed line).

Table 2. Predictors of Outcomes Following Induction Chemotherapy in Acute Myeloid Leukemia: Results of Multivariable Analysis
VariableComplete RemissionOverall SurvivalRelapse-Free Survival
OR (95% CI)PHR (95% CI)PHR (95% CI)P
  1. CI indicates confidence interval; HR, hazard ratio; OR, odds ratio.

Days to zero blasts0.844 (0.733-0.971).01791.053 (1.011-1.097).01221.048 (1.008-1.090).0171
White blood cell1.000 (0.990-1.011).93111.000 (0.996-1.004).82901.001 (0.997-1.005).6092
Age0.958 (0.927-0.991).01221.023 (1.007-1.038).00421.017 (1.002-1.032).0234
Cytogenetic risk (unfavorable vs favorable+intermediate)0.106 (0.012-0.923).02042.827 (1.315-6.079).00232.722 (1.346-5.505).0002
Sex0.424 (0.165-1.086).07381.152 (0.734-1.807).53841.251 (0.813-1.924).3089
Lactate dehydrogenase1.000 (0.999-1.001).63571.000 (1.000-1.000).34681.000 (1.000-1.000).1760


Cytogenetics and molecular markers are the best predictors of long-term outcomes in AML. However, the majority of patients present with normal cytogenetics, and these tests do not consistently predict remission. A rapid decrease in blood blasts is often considered a favorable indicator of response in patients with AML who are receiving induction chemotherapy. Clearance of marrow and blood blasts has long been recognized as a strong prognostic indicator in childhood acute lymphoblastic leukemia,9, 10 and more recently, early disappearance of marrow blasts following induction chemotherapy was found to be a strong favorable predictor of outcomes in AML.4, 6 Among patients with residual marrow blasts after induction chemotherapy, quantification of this residual disease by flow cytometry identified subsets of patients at variable risk for relapse.11, 12

Given the invasiveness of marrow aspirates and biopsies, methods that permit a more convenient detection of residual AML have been explored. These include detection of blood blasts by microscopy, flow cytometry, or quantitative polymerase chain reaction (Table 3). Early clearance of blood blasts (<5 days from start of induction chemotherapy) by serial microscopic evaluations was associated with better RFS in one study.13 Time to one-log reduction of blood blasts using flow cytometry also predicted response to induction therapy measured by bone marrow biopsy on days 14 and 30,14 as well as disease-free survival.15 Finally, time from initiation of induction to decrease of normalized blood WT1 transcripts detected by quantitative polymerase chain reaction correlated with CR, disease-free survival, and OS.16

Table 3. Summary of Trials Analyzing Correlations Between Clearance of Blood Markers and Outcomes in Acute Myeloid Leukemia
StudyPatients (n)Induction RegimenMethod for Detection of Blood BlastsMedian Time to Blood Blast Clearance (d)Presence of a Correlation Between Time to Blood Blast Clearance and Disease Status on Day 14-16Median Follow-Up (mo)Correlation
  1. EBC indicates early blood blast clearance; HiDAC, high-dose cytosine arabinoside; NE, not evaluated or not reported; OS, overall survival; Q-PCR, quantitative polymerase chain reaction; RFS, relapse-free survival.

Gianfaldoni et al15617+3Flow cytometry5YesNENENE
Lacombe et al131307+3Flow cytometry5No11YesNo
Elliott et al11867+3Morphology5NENEYesNo
Gianfaldoni et al14577+3Q-PCR (WT1)5No14YesYes
Arellano et al (this study)162HiDAC + DaunorubicinMorphology5Yes18.8YesYes

We hypothesized that a rapid decrease in blood blasts may predict early marrow blast clearance and CR in patients with AML following induction chemotherapy. To that intent, we analyzed a cohort of 162 AML patients to determine correlations between time to blood blast clearance following induction with a HiDAC-containing regimen and results of marrow biopsies obtained on day 14 after induction. Interestingly, median time to blood blast clearance was comparable to all previous studies, ie, 5 days from start of induction chemotherapy. However, the ROC analysis identified day 6 to be a better discriminating value and this time point was used to separate early from late responders. We found a direct correlation between time to blood blast clearance and marrow blast clearance as assessed on day 14, CR, RFS, and OS. However, when day 14 marrow results were incorporated into the multivariable model, blast clearance was no longer an independent predictor of survival in AML.

This study is limited by the absence of information on molecular markers such as FLT3, NPM1, and kit mutations. These markers were not fully recognized and readily available at the time these patients were treated. However, this is, to our knowledge, the first study describing correlation between blood blast clearance measured by serial microscopic evaluations and disease status assessed by day 14 and day 30 marrow biopsies, and to also correlate the time to blood blasts clearance with OS. In addition, this report describes blood blast clearance in patients receiving induction with a HiDAC-containing regimen.

In conclusion, EBC measured on daily complete blood counts predicts disease status determined by bone marrow biopsy on days 14 and 30. Unfortunately, in this analysis we were unable to identify a cutoff time to blast clearance beyond which all patients would either have or not have residual disease on day 14; and therefore monitoring time to blast clearance may not alleviate the need to perform a bone marrow biopsy on day 14. However, early blood blast clearance provides reassuring information to patients early into the course of their therapy and prior to the availability of cytogenetic and molecular markers, given the association with better RFS and OS.


We thank Drs. Adnan Jabbar, and Karen Blanchard for their contributions in data collection and organization of the data files.


No specific funding was disclosed.


The authors made no disclosure.