• acute myeloid leukemia;
  • multiparameter flow cytometry;
  • minimal residual disease;
  • leukemia associated phenotypes;
  • receiver operating characteristics


  1. Top of page
  2. Abstract
  6. Acknowledgements


Multiparameter flow cytometry (MFC) has been shown to be a useful approach for detection of minimal residual disease (MRD). The aim of the study was to determine the optimal threshold that can separate patients into two groups in terms of leukemic residual cells and relapse status after induction and consolidation chemotherapy.


Five-color MFC and receiver operating characteristics (ROC) analysis were used to determine the optimal threshold. This study analyzed 54 acute myeloid leukemia (AML) patients.


LAPs were detected in 51/54 (94%) patients. MRD was evaluated in the bone marrow (BM) in morphologic complete remission from 25 and 22 patients after induction and consolidation, respectively. The threshold discriminating MRD from MRD+ cases was set at 0.15% residual leukemic cells, a level that allowed optimal sensitivity and specificity for prediction of relapse, both at postinduction (P = 0.05) and postconsolidation (P = 0.009) time points using ROC analysis. MRD level postinduction not only influenced relapse-free survival (RFS) (P = 0.004) but also overall survival (OS) (P = 0.003). Multivariate analysis showed that MRD level postinduction was a powerful independent prognostic factor for both RFS (P = 0.037) and OS (P = 0.026).


Using the ROC analysis, the threshold of 0.15% was defined as the optimal value in discriminating risk categories in AML, and postinduction MRD assessment is able to better predict disease outcome than consolidation. Therefore, MRD analysis by MFC could be used for refining the selection of therapeutic strategies and improving clinical outcome in individual patients. © 2008 Clinical Cytometry Society

Despite a high remission rate (approaching 80% in younger adults), a significant number of patients with acute myeloid leukemia (AML) eventually relapse (1). Relapse is indicative of the existence of residual leukemic cells below the threshold of morphologic detection. This cell population is defined as minimal residual disease (MRD) and can be detected by several techniques: polymerase chain reaction (PCR) (2), multiparameter flow cytometry (MFC) (3), and fluorescence in situ hybridization (4). PCR-based quantification of MRD has high sensitivity, and the proportion of AML cases amenable to PCR detection may be significantly increased by targeting length mutations of the FLT3 gene and partial tandem duplications within the MLL gene in addition to the fusion transcripts AML1-ETO, CBFB-MYH11, and PML-RAR (5). However, this still results in ∼50% of cases not having an identifiable leukemia-specific genetic alteration, and these patients therefore are not subject to PCR-based monitoring of MRD. In fact, in practical terms, MRD detection by PCR is only routinely applied in acute promyelocytic leukemia patients. Using accurate combination of monoclonal antibodies (MoAbs) and MFC allows more specific detection of leukemic cells and can be used for quantification of MRD.

The exact time following remission induction at which the presence of MRD indicates a higher risk for relapse is unclear. It seems to be dependent on leukemia subtype and the technique used for detection of MRD. In addition, the precise value of serial MRD determinations by immunophenotyping to predict relapse in AML is still not well defined.

Recent data indicate that immunologic monitoring for detection of MRD may be applicable to >80% patients with AML (5, 6) and it can reach up to 94% (7). Additionally, sensitivity of MRD detection by MFC can range up to 1 leukemic cell per 104–105 normal cells (5–7).

Previous studies on MRD by immunophenotyping used three- or four-color MFC (5, 8–10). These studies differ in the optimal threshold for MRD detection. They were performed in a retrospective fashion, and on highly selected populations. In addition, different results were obtained for induction and consolidation chemotherapies. In the present study, we used five-color MFC for a more sensitive and accurate quantification of MRD in AML in a prospective, unselected population. Five-color MFC was used to investigate the level of MRD after standard induction and consolidation chemotherapy in AML patients in first complete remission (CR) using a CD45 gating strategy and extensive panel of MoAbs.

Although studies of MRD detection by MFC in AML are still limited when compared with acute lymphoblastic leukemia (ALL) (11–14), several reports have been published providing evidence that study of MRD are a useful tool for predicting relapse (8–10, 15–20).

The findings of these studies reveal that the precise evaluation of MRD by MFC has a prognostic significance in predicting relapse and may have a major impact in the clinical management of patients with AML. In particular, the understanding of the clinical significance of MRD at different stages of treatment may help at designing modified therapeutic programs according to patient risk category. Above all, the main challenge for expert MRD investigators is to simplify methods while maintaining or increasing their reliability, therefore propagating the potential benefits of MRD monitoring to all patients.

Our aim was to determine the optimal threshold value that can split patients into two groups in terms of residual leukemic cells and relapse status. The secondary aim was to establish the time point of choice, that is, postinduction or postconsolidation, by using receiver operating characteristics (ROC) analysis with optimal sensitivity and specificity able to better predict outcome.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Patient Samples

Fresh bone marrow (BM) samples from 54 consecutive, unselected AML patients were obtained at diagnosis between January 2005 and April 2007. Diagnosis of patients was based on morphology, immunophenotyping, and cytogenetics (21). Patient characteristics are shown in Table 1. Patients were classified in three risk groups for cytogenetic abnormalities according to Grimwade et al. (22). Favorable risk was defined by the presence of t(15;17) or t(8;21) or inv(16) and poor risk was defined by the presence of either five unrelated abnormalities, monosomy 5 or 7, abnormalities of the long arm of chromosome 5, or inv(3) (q21:q26). Patients who did not meet the criteria for poor or favorable risk were classified as intermediate risk.

Table 1. Patient Clinical Characteristics Analyzed for MRD After Induction and Consolidation Chemotherapy
Patient characteristicsTotal (%)
No. patients25
Age at diagnosis, mean (range)48 (20–75)
WBC count at diagnosis × 109/L, median (range)5.1 (0.86–179)
BM blasts % by morphology, median (range)50 (6.10–85)
FAB classification, n (%) 
 M18 (32)
 M27 (28)
 M32 (8)
 M43 (12)
 M54 (16)
 M61 (4)
Cytogenetic risk group 
 Favorable7 (28)
 Intermediate12 (48)
 Poor6 (24)
Consolidation I21
Consolidation II13
Consolidation III4
Consolidation IV2
Consolidation V2
AuSCT/AlloSCT3 (12)/6 (24)

We have previously described our results demonstrating the sensitivity and specificity of the LAPs detected using normal and regenerating BMs (7). We have demonstrated that specific LAPs are present at very low level in normal and regenerating BMs, and consequently can be used as tracking markers for MRD analysis in AML.

Of the 54 patients diagnosed with AML, 31 received chemotherapy with 27 achieving CR, two patients were LAP and thus only 25 were analyzed at different time points during therapy for MRD detection. The patients' characteristics are shown in Table 1. For samples at diagnosis, 20,000 events were acquired to identify the LAPs, whereas for follow-up samples for MRD evaluations, at least 250,000 events were acquired to identify these rare populations. For the follow-up samples, only the combinations for the respective LAPs from the time of the diagnosis were applied.

Chemotherapeutic Treatment

The chemotherapy received during induction and consolidation chemotherapy was similar in all patients. AML-M3 cases were treated with all-trans-retinoic acid (ATRA) 45 mg/m2/day until CR plus idarubicin 12 mg/m2/day on days 2, 4, 6, and 8. Non-M3 patients were treated with induction chemotherapy consisting of idarubicin 9 mg/m2/day on days 1–3, etoposide 75 mg/m2 on days 1–7 and cytarabine. Cytarabine was given as either low dose (200 mg/m2/day on days 1–7) or high dose (6 g/m2/day on days 1, 3, 5, and 7). Consolidation therapies consisted of idarubicin 9 mg/m2/day on days 1–2, etoposide 75 mg/m2 on days 1–5 and cytarabine (100 mg/m2/day on days 1–5).

Flow Cytometry

The immunophenotypic analysis was performed on erythrocyte-lysed whole BM samples with direct conjugated MoAbs (Table 2). Antigen expression was analyzed using five-color combinations of the following MoAbs conjugated with fluorescein isothiocyanate (FITC), phycoerythrin (PE), phycoerythrin-TR (ECD), phycoerythrin-cyanin 5 (PC-5), and phycoerythrin cyanin 7 (PC-7) at diagnosis. Two MoAb panels were used in the study: screening and secondary panels. The screening panel (2/3 color) is our standard diagnostic panel used to distinguish acute lymphoblastic from AML. Subsequently, in AML, the secondary panel (five color) was used to identify the LAPs and evaluate MRD. Blasts were identified by CD45/SS log gating strategy. Backgating strategies using CD34 and CD117 were used to better define the blast population. CD45/CD34/CD117 were used in combination with different myeloid and lymphoid markers in a five-color combination to increase the sensitivity of the LAP detection. This was done to obtain a maximum advantage, incorporating various permutations and combinations, of the five-color MFC in defining the LAPs as we previously described (7, 23). The screening panel was CD7FITC/CD13PE/CD45PC5, CD19FITC/CD10PE/CD45PC5, CD34FITC/CD33PE/CD45PC5, CD14FITC/CD56PE/CD45PC5, HLA-DRFITC/CD45PC5, CD15FITC/CD117PE/CD45PC5, MPO FITC/CD45PC5, TdT FITC/CD45PC5, and CD45PC5/SNEG. The five-colour secondary panel was demonstrated in Table 3.

Table 2. Monoclonal Antibodies Used for Immunostaining of AML Patients
  • a

    S Neg, which is a combination of FITC IgG1 and PE IgG2b, were used for all the antibodies. BD Simultest™ control γ12a (IgG1 FITC/IgG2b PE) is a two-color direct immunofluorescent for use as a negative control.

CD2SFCI3Pt2H9IgG1 κFITCCoulter
CD73A1E-12H7IgG2b κFITCCoulter
CD10SS2/36IgG1 κPEDako
CD11bBear1IgG1 κPECoulter
CD13L138IgG1 κPEBD
CD19HD37IgG1 κFITCDako
CD33P67.6IgG1 κPEBD
CD45Immu19.2IgG1 κPC5ImmunoTech
CD56N901 (NKH-1)IgG1 κPECoulter
CD1239F5IgG1 κPEBD
CD235a(Glyco A)JC159IgG1 κFITCDako
S Nega IgG1 κ IgG2b κFITC/PEBD
Control ECD679.1Mc7IgG1ECDImmunoTech
Control PC-5679.1Mc7IgG1PC5ImmunoTech
Control PC-7679.1Mc7IgG1PC7Coulter
Table 3. The Secondary Panel
Control FITCControl PECD45Control PC-5Control PC-7

Appropriate isotype-matched negative controls were used in the panel of MoAbs to assess background fluorescence intensity. The respective combinations of antibodies were added to ∼106 mononuclear cells and incubated for 10 min. Red cell lysis was carried out using a Beckman Coulter Multi Q-Prep (semiautomated procedure).

Data Acquisition and Analysis

Data acquisition was performed on a Coulter FC500 flow cytometer. Analysis of list mode data was performed using CD45/sides scatter log by CXP Software. Thresholds for positivity were based generally on isotype and internal negative controls. The positivity threshold was 20% for all markers. For all follow-up samples, not less than 250,000 events were acquired and not less than 20 events in a cluster were considered as positive.

Study Conduct

Prior to therapy, all patients gave their informed consent for participation in the current evaluation after having been advised about the purpose and investigational nature of the study as well as potential risks. The study design was approved by the Royal Adelaide Hospital Research Ethics Committee prior to its initiation.

Statistical Analysis

CR was defined according to criteria reported by Cheson et al. (24) (<5% BM blast cells and recovery of hematologic parameters). Relapse was defined as the reappearance of circulating blasts not attributable to “overshoot” following recovery from myelosuppressive therapy, or 5% blasts in the BM not attributable to another cause, or development of extramedullary leukaemia. Relapse-free survival (RFS) was defined only for patients who achieved CR and was measured from the documented date of CR until date of relapse or last follow-up date available or death regardless of cause. Overall survival (OS) was measured from date of diagnosis until date of death, regardless of cause of death or last follow-up date available.

Kaplan–Meier curves were constructed for RFS and OS comparing MRD positive and MRD negative groups. The log-rank test was performed to determine whether there was a significant difference between the survival curves of the two groups. Finally, to adjust for potential confounding covariates, a Cox proportional hazards model was built using a backwards variable selection procedure to determine whether the two groups, as an indicator variable, remained significant as a prognostic factor once other covariates adjusted the model. First, univariate model for each clinical characteristic at baseline were fit. Second, univariate models incorporating an artificial time-dependent covariate expressed as the product of the covariate and the log of the time variable were fit to assess whether the proportional hazards assumption was met. If the proportional hazards assumption was not met for a particular variable, then the artificial time-dependent covariate was included in all of the subsequent models containing that variable. Thereafter, variables reflecting a P from the likelihood ratio test in the univariate models of <0.25 were incorporated together in a full model (25). Variables reporting a P of >0.05 from the corresponding Wald statistic in the full model were subsequently dropped one at a time in determining the final model. Statistical analyses were performed using SAS version 9.1 (Cary, NC) in Discipline of Public Health, University of Adelaide.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Clinical Characteristics

Fifty-four AML BM samples were analyzed from consecutive, newly diagnosed patients using a comprehensive panel of MoAbs with five-colour staining for expression of LAPs. LAPs were identified in 51 (94%) patients. Thirty one (57%) received chemotherapy, and of these 27 (87%) achieved CR, two of them were LAP negative, and therefore 25 were followed-up for MRD monitoring postinduction therapy. One patient died after induction and two did not receive consolidation chemotherapy, and hence 22 patients were evaluable for MRD postconsolidation therapy as summarized in Figure 1. The median follow-up duration for patients was 18.3 months (range: 1.4–51.1).

thumbnail image

Figure 1. Flow chart of cases analyzed by MFC at diagnosis of AML. After induction therapy, 27 patients achieved CR. After induction and consolidation therapies, 25 and 22 patients, respectively, were analyzed by flow cytometry. [Color figure can be viewed in the online issue, which is available at]

Download figure to PowerPoint

Establishing the Optimal Threshold That can Split Patients into Two Groups with Residual Leukemic Cells in Terms of Relapse Status

ROC analysis was carried out using Prism™ software (GraphPad Software, San Diego, CA) to determine the optimal threshold yielding the best separation of AML patients into two groups for MRD and relapse status with optimal sensitivity and specificity.

Choosing the cutoff value closer to an area of (0, 1) from the ROC curve revealed that the optimal threshold is 0.145 for postconsolidation and 0.155 for postinduction (Fig. 2). These threshold values represent the optimal sensitivity (89%) and specificity (73%) with a likelihood ratio of 3.26 for postconsolidation and a sensitivity of 100% and specificity of 39% with a likelihood ratio of 1.63 for postinduction. The likelihood ratio of 1.63 represents the highest likelihood ratio if all cutoffs are taken into account.

thumbnail image

Figure 2. ROC analysis from 11 relapsed patients and 9 who had no relapse (postconsolidation) and 13 relapsed versus nine who had no relapse (postinduction). (A) A ROC curve of sensitivity (Y-axis) versus its false positive specificity (X-axis) obtained at each cutoff level. The discrete points on the empirical ROC curve and its 95% confidence interval is 0.66 to 1.02 (postconsolidation) and 0.33 to 0.82 (postinduction). (B) ROC curve table representing 100% − specificity vs. sensitivity %. (C) The sensitivity and specificity table of the test at various cutoff values between MRD levels for patients who did not have relapse (as a normal) and MRD levels for patients who underwent relapse (abnormal). As indicated from both ROC curves and tables, the optimal cutoff point for both optimal sensitivity and specificity is MRD level of ∼0.15% with a likelihood ratio of 3.26 (postconsolidation) and 1.63 (postinduction).

Download figure to PowerPoint

Therefore, we utilized the MRD level of 0.15% residual leukemic cells as a threshold value to discriminate MRD from MRD+ cases both after induction and after consolidation (Fig. 3). Consequently, patients with a residual leukemic cells ≤ 0.15% were categorized as MRD, whereas those with residual leukemic cells > 0.15% were grouped as MRD+.

thumbnail image

Figure 3. Determination of the optimal threshold value (0.15%) capable to split patients into two groups according to relapse status for both postinduction (upper plots) and postconsolidation (lower plots). X-axis reports the relapse status for AML patients and Y-axis reports the corresponding residual leukemic cells as percentages. [Color figure can be viewed in the online issue, which is available at]

Download figure to PowerPoint

Detection of MRD in BM Postinduction

After induction, the median level of the residual leukemic cells was 0.07% (range: 0.001–0.78%). At this time point, 72% of the patients (18/25) were MRD and 28% (7/25) were MRD+. Five of seven (71%) patients in the MRD+ group had disease relapse at a median time of 12.9 months (range: 1.4–18.2) and two died postinduction chemotherapy, whereas in the MRD group 8/18 (44%) relapsed at a median time of 24.7 months (range: 9.5–51.1) (P = 0.048), 9/18 (50%) patients remained in CR at a median time of 18.5 months (range: 1.8–25.1) and one patient died after induction chemotherapy. The probability of RFS and OS was illustrated in (Fig. 4) for patients in the MRD group, compared with those in the MRD+ group (P = 0.004 and 0.003), respectively.

thumbnail image

Figure 4. RFS and OS of AML patients according to MRD levels after induction chemotherapy. Patients were grouped according MRD threshold value of 1.5 × 10−3 residual leukemic cells, as determined at postinduction and evaluated for RFS (upper plot) and OS (lower plot).

Download figure to PowerPoint

Detection of MRD in BM Postconsolidation

One patient died after induction therapy (was MRD+), two patients did not receive consolidation and 22 proceeded to receive consolidation; of these, two patients died after consolidation (were MRD) and 20 were evaluable for consecutive MRD evaluation. At postconsolidation analysis, the median level of residual leukemic cells in the whole series was 0.115 (range: 0.0–0.56). Thirteen (59%) of the patients were MRD and 9 (41%) were MRD+. Three of 13 (23%) and 8 of 9 (89%) patients in MRD and in MRD+ group underwent a relapse at a median time of 24.5 months (range: 18.2–31.8) and 15.5 months (range: 5.7–51.1), respectively (P = 0.009). The probability of RFS and OS was depicted in (Fig. 5) for patients in the MRD group, compared to those in the MRD+ group (P = 0.06 and 0.98, respectively). Figure 6 demonstrates a representative example for MRD detection in consecutive BM samples of a relapsing patient and of a patient still in remission.

thumbnail image

Figure 5. RFS and OS of AML patients according to MRD levels after consolidation chemotherapy. Patients were grouped according MRD threshold value of 1.5 × 10−3 residual leukemic cells, as determined at postconsolidation and evaluated for RFS (upper plot) and OS (lower plot).

Download figure to PowerPoint

thumbnail image

Figure 6. MRD detection in consecutive BM samples of a relapsing patient (A–E) and of a patient still in remission (F–J). The relapse example demonstrated the aberrant phenotype of CD117+CD15+ expression on CD34+CD117+ cells (A). The MRD% was 0.78, 0.42, and 0.64 after induction (B), consolidation I (C) and consolidation II chemotherapy (D), respectively. Relapse at 6 months shows LAP expression similar to diagnosis material (E). The example showing ongoing remission demonstrated the aberrant phenotype CD34+CD7+ (F). The MRD% was 0.12, 0.15, 0.13 and 0.09 after induction (G), consolidation I (H), consolidation II (I) chemotherapy, and later follow-up (J) respectively. [Color figure can be viewed in the online issue, which is available at]

Download figure to PowerPoint

Of 25 patients evaluable for MRD in our series, three patients underwent allogeneic stem cell transplantation (AlloSCT) in the first CR. Of these, two were MRD+ postinduction and became MRD postconsolidation and the third patient was MRD postinduction and remained MRD postconsolidation. The decision to transplant was not based on MRD results.

Relation Between Level of MRD Postinduction and Consolidation Chemotherapy

The dynamics of MRD fluctuations after induction and consolidation therapy in the 22 patients completing both treatment phases was studied. Based on the comparative analysis of MRD levels detected at the two time points, we separated the series into four groups: (1) Ind+Cons+ [three patients]: MRD+ both after induction and consolidation; (2) Ind+Cons [two patients]: MRD+ after induction converted into MRD after consolidation; (3) IndCons+ [six patients]: MRD after induction converted into MRD+ at the end of consolidation; and (4) IndCons [11 patients]: who were MRD at both time points.

Of the 22 patients who were evaluable in terms of comparison between induction and consolidation therapy; 13 (59%) were MRD at the end of consolidation; 11 of them were already negative after induction and two became MRD only after consolidation. On the contrary, nine patients were MRD+ at the end of consolidation; six of them, MRD after induction, progressed into an MRD+ status in spite of administration of the consolidation cycle. The remaining three patients were in a MRD+ status throughout the induction and consolidation cycles.

The analysis of RFS and OS rates showed that the MRD status at the end of induction was the most significant predictor of outcome, regardless of the levels of MRD after consolidation. There was significant difference in RFS among the four groups (P = 0.025). In addition, patients with an Ind+Cons status had different probability of RFS (P = 0.408) and borderline significance for OS (P = 0.054) than those never exceeding the threshold of 0.15% residual leukemic cells (IndCons).

Prognostic Impact of Conventional Parameters in Univariate Analysis

The prognostic impact of poor cytogenetics, gender, FAB subtypes (M0–M2) vs. (M4–M5), and transplant as dichotomous variables as well as age, WBC count at diagnosis, percentage of BM blasts at diagnosis as continuous variables was analyzed using RFS and OS as dependent variables.

For both analyses, after induction and consolidation therapy, none of the variables were significant as shown in Table 4. However, FAB subtype group (M0–M2) in RFS and OS, and WBC count at diagnosis were close to significance (P < 0.25) and thus were pooled in multivariate model.

Table 4. Prognostic Impact of Conventional Parameters and MRD Detection in Univariate Analysis
VariablesRFS (P value)OS (P value)
Cytogenetics adverse risk group0.30.7
% blast at diagnosis0.40.8
FAB classification (M0-M2) vs. (M4-M5)0.090.06
MRD post Induction0.0040.0026
MRD post Consolidation0.060.98
WBC count at diagnosis0.070.4

Prognostic Determinants in Multivariate Analysis

Finally, all the relevant prognostic variables with a statistical significance >0.25 in univariate analysis (25) being WBC count, FAB subtype and MRD status after induction and consolidation were pooled into a multivariate model to determine to what extent they affected independently the outcome of treatment. In this analysis, postinduction MRD+ status was found to be an independent variable significantly associated with a higher frequency of relapse (P < 0.05) and a shorter duration of OS (P = 0.026) and RFS (P = 0.037) with an estimated hazard ratio of 4.7 (95% CI: 1.1–20.5) for RFS and 5.2 (95% CI: 1.2–22.2) for OS. The prognostic impact of postinduction MRD status remained significant even after adjustment for WBC, MRD postconsolidation and FAB subtype (group M0–M2 and M4–M5) (P < 0.05).

Thus after induction therapy, the MRD frequency correlated with the clinical outcome, as measured in the 25 evaluable patients. In the total group again MRD frequency inversely correlated both with RFS and OS (P < 0.05) (Table 5).

Table 5. Relative Risk of a Relapse Defined by MRD Frequency in BM
 Threshold value (%)n% of patients with MRD > threshold valueUnivariate analysis for RFS, OS (P values)Relative risk of relapse (95% CI)Multivariate analysis for RFS (P value)Relative risk of death (95% CI)Multivariate analysis for OS (P value)
  1. In multivariate analysis, postinduction MRD status was independently associated with shorter duration of OS and RFS.

BM Postinduction0.152528%(0.0041, 0.0026)4.7 (1.10–20.48)0.0375.2 (1.22–22.16)0.029
BM Postconsolidation0.152241%(0.06, 0.98)3.1 (0.80–12.18)0.101.6 (0.33–7.67)0.57


  1. Top of page
  2. Abstract
  6. Acknowledgements

The goal of MRD studies is to optimize the clinical management of the postremission phase in patients with acute leukemia, offering the opportunity to guide the therapeutic decisions based on specific biologic findings. The ultimate purpose of these types of studies is to distinguish patients who respond successfully to standard treatment, from those at high risk of relapse who require therapeutic intervention to reduce risk of relapse.

Previous studies reporting on the prognostic impact of MRD level in patients with AML in CR mainly focused on only a proportion of patients with AML using three- or four-color MFC (8–10, 12). In our study, LAPs were able to be detected in majority of patients (94%) when five-color MFC and a comprehensive panel of MoAbs were applied. Moreover, in 78% of cases, leukemic cells simultaneously displayed more than one LAP. This was important when immunophenotypic shifts occurred during treatment (20, 26, 27). The phenomenon of immunophenotypic switch is a relatively rare but significant problem in using the LAPs as a strategy for MRD detection (20, 26). The analysis of LAPs by five-color MFC should help in addressing this problem as we detected more than 1 LAP in 78% of cases, which increases the probability of detection for MRD analysis. However, a broader definition of LAP incorporating abnormal intensity of antigen expression would probably extend the percentage of LAP detection from 94 to 100%.

The introduction of four-color flow cytometry and CD45 gating has been shown to improve detection of leukemic cells among normal BM cells with a log increase of sensitivity for MRD quantification over three-color flow cytometry (28). In our study, we demonstrate that the detection of LAPs by five-color flow cytometry with CD45 gating can be used in the assessment of MRD in a group of unselected leukemic patients presenting to a teaching hospital hematology unit. Moreover, the presence of MRD postinduction is a powerful prognostic indicator for relapse. These results suggest a good treatment outcome in patients with an early response in terms of MRD clearance analogous to the findings in ALL (29–31).

In addition, the increased detection of LAPs by using five color MFC and its utility for monitoring MRD was demonstrated. The sensitivity of MFC increases in parallel to the number of aberrantly expressed antigens. We have shown that MRD level of 0.15% is the most optimal threshold value that can discriminate patients in terms of residual leukemic cells and their relapse risk after both induction and consolidation therapies. Our data also show that patients who have detectable MRD after induction chemotherapy have a worse prognosis. This result was significant even after adjusting for traditional risk factors on a multivariate analysis. However, because of the limited sample size, some of the confounding factors did not stand out such as cytogenetics. The fact that detection of MRD postinduction was a significant independent prognostic feature, even with small number of patients in a sequentially selected group of patients, indicates that MRD detection must be considered in routine clinical practice to be a useful prognostic factor.

Three other groups have reported prognostic impact of MRD frequency using immunophenotypic detection methods (9, 10, 15, 32). Considering the threshold levels for distinguishing two patient groups with different clinical outcome, Venditti et al. were not able to establish such a threshold level after one cycle of induction chemotherapy (17). On the other hand, San Miguel et al. (9) and Sievers et al. (32) established a threshold level of 0.2% after one or two cycles of induction chemotherapy, which is similar to our threshold level of 0.15% established after one cycle of induction chemotherapy. Furthermore, after consolidation chemotherapy two patient groups with different RFS were defined at a threshold level of 0.035% in the study of Venditti et al. and 0.2% in the study of San Miguel et al. We found a trend toward significance using a threshold level of MRD% of 0.145 after consolidation therapy.

ROC analysis was very useful in this setting in discriminating two groups with different levels of residual leukemic cells with the highest likelihood ratio and optimal sensitivity and specificity. For that reason, we recommend applying it to MRD investigations.

Data from previously published studies, as well as our study, indicate that MRD levels above 0.15% define the population of patients with the highest risk of relapse (8, 9, 12, 32). Therefore, MRD analysis by flow-cytometry may be used for refining the selection of therapeutic strategies and improving clinical outcome in individual patients.

The MFC assay is applicable to the majority of patients, relatively cheap, can be performed easily, reliably, and rapidly in clinical laboratories equipped with a flow cytometer.

To confirm these single institution results and to further improve response-adapted management of patients with AML, large prospective multicenter studies are needed in which MRD levels are thoroughly followed and used to allocate patients to different predefined therapies.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors gratefully acknowledge the support of Dr. Nancy Briggs, Discipline of Public Health, University of Adelaide, for her assistance with statistical analysis, and also Dr. Pravin Hissaria, Division of Human Immunology, IMVS, Adelaide, South Australia, for his valuable comments on the manuscript.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Mayer RJ,Davis RB,Schiffer CA,Berg DT,Powell BL,Schulman P,Omura GA,Moore JO,McIntyre OR,Frei EIII. Intensive postremission chemotherapy in adults with acute myeloid leukemia. Cancer and leukemia Group B. N Engl J Med 1994; 331: 896903.
  • 2
    Raanani P,Ben-Bassat I. Detection of minimal residual disease in acute myelogenous leukemia. Acta Haematol 2004; 112: 4054.
  • 3
    Vidriales MB,San-Miguel JF,Orfao A,Coustan-Smith E,Campana D. Minimal residual disease monitoring by flow cytometry. Best Pract Res Clin Haematol 2003; 16: 599612.
  • 4
    Schmidt HH,Strehl S,Thaler D,Strunk D,Sill H,Linkesch W,Jager U,Sperr W,Greinix HT,Konig M,Emberger W,Haas OA. RT-PCR and FISH analysis of acute myeloid leukemia with t(8;16)(p11;p13) and chimeric MOZ and CBP transcripts: Breakpoint cluster region and clinical implications. Leukemia 2004; 18: 11151121.
  • 5
    Kern W,Voskova D,Schoch C,Hiddemann W,Schnittger S,Haferlach T. Determination of relapse risk based on assessment of minimal residual disease during complete remission by multiparameter flow cytometry in unselected patients with acute myeloid leukemia. Blood 2004; 104: 30783085.
  • 6
    Kern W,Danhauser-Riedl S,Ratei R,Schnittger S,Schoch C,Kolb HJ,Ludwig WD,Hiddemann W,Haferlach T. Detection of minimal residual disease in unselected patients with acute myeloid leukemia using multiparameter flow cytometry for definition of leukemia-associated immunophenotypes and determination of their frequencies in normal bone marrow. Haematologica 2003; 88: 646653.
  • 7
    Al-Mawali AH,Gillis D,Hissaria P,Lewis ID. Incidence, sensitivity and specificity of leukaemia associated phenotypes in acute myeloid leukaemia patients using specific five-colour multiparameter flow cytometry. Am J Clin Pathol 2008; 129: 934945.
  • 8
    Feller N,van der Pol MA,van Stijn A,Weijers GW,Westra AH,Evertse BW,Ossenkoppele GJ,Schuurhuis GJ. MRD parameters using immunophenotypic detection methods are highly reliable in predicting survival in acute myeloid leukaemia. Leukemia 2004; 18: 13801390.
  • 9
    San Miguel JF,Vidriales MB,Lopez-Berges C,Diaz-Mediavilla J,Gutierrez N,Canizo C,Ramos F,Calmuntia MJ,Perez JJ,Gonzalez M,Orfao A. Early immunophenotypical evaluation of minimal residual disease in acute myeloid leukemia identifies different patient risk groups and may contribute to postinduction treatment stratification. Blood 2001; 98: 17461751.
  • 10
    Venditti A,Buccisano F,Del Poeta G,Maurillo L,Tamburini A,Cox C,Battaglia A,Catalano G,Del Moro B,Cudillo L,Postorino M,Masi M,Amadori S. Level of minimal residual disease after consolidation therapy predicts outcome in acute myeloid leukemia. Blood 2000; 96: 39483952.
  • 11
    Coustan-Smith E,Behm FG,Sanchez J,Boyett JM,Hancock ML,Raimondi SC,Rubnitz JE,Rivera GK,Sandlund JT,Pui CH, et al. Immunological detection of minimal residual disease in children with acute lymphoblastic leukaemia. Lancet 1998; 351: 550554.
  • 12
    Coustan-Smith E,Ribeiro RC,Rubnitz JE,Razzouk BI,Pui CH,Pounds S,Andreansky M,Behm FG,Raimondi SC,Shurtleff SA,Downing JR,Campana D. Clinical significance of residual disease during treatment in childhood acute myeloid leukaemia. Br J Haematol 2003; 123: 243252.
  • 13
    Coustan-Smith E,Sancho J,Hancock ML,Razzouk BI,Ribeiro RC,Rivera GK,Rubnitz JE,Sandlund JT,Pui CH,Campana D. Use of peripheral blood instead of bone marrow to monitor residual disease in children with acute lymphoblastic leukemia. Blood 2002; 100: 23992402.
  • 14
    Ciudad J,San Miguel JF,Lopez-Berges MC,Vidriales B,Valverde B,Ocqueteau M,Mateos G,Caballero MD,Hernandez J,Moro MJ,Mateos MV,Orfao A. Prognostic value of immunophenotypic detection of minimal residual disease in acute lymphoblastic leukemia. J Clin Oncol 1998; 16: 37743781.
  • 15
    San Miguel JF,Martinez A,Macedo A,Vidriales MB,Lopez-Berges C,Gonzalez M,Caballero D,Garcia-Marcos MA,Ramos F,Fernandez-Calvo J,Calmuntia MJ,Diaz-Mediavilla J,Orfao A. Immunophenotyping investigation of minimal residual disease is a useful approach for predicting relapse in acute myeloid leukemia patients. Blood 1997; 90: 24652470.
  • 16
    Venditti A,Maurillo L,Buccisano F,Tamburini A,Del Poeta G,Del Principe MI,Panetta P,Consalvo MI,Mazzone C,Tendas A,Trawinska M,Forte V,Amadori S. Multidimensional flow cytometry for detection of minimal residual disease in acute myeloid leukemia. Leuk Lymphoma 2003; 44: 445450.
  • 17
    Venditti A,Tamburini A,Buccisano F,Del Poeta G,Maurillo L,Panetta P,Scornajenghi KA,Cox C,Amadori S. Clinical relevance of minimal residual disease detection in adult acute myeloid leukemia. J Hematother Stem Cell Res 2002; 11: 349357.
  • 18
    Drach J,Drach D,Glassl H,Gattringer C,Huber H. Flow cytometric determination of atypical antigen expression in acute leukemia for the study of minimal residual disease. Cytometry 1992; 13: 893901.
  • 19
    Sievers EL,Lange BJ,Buckley JD,Smith FO,Wells DA,Daigneault-Creech CA,Shults KE,Bernstein ID,Loken MR. Prediction of relapse of pediatric acute myeloid leukemia by use of multidimensional flow cytometry. J Natl Cancer Inst 1996; 88: 14831488.
  • 20
    Plata E,Choremi-Papadopoulou H,Viglis V,Yataganas X. Flow-cytometric detection of minimal residual disease with atypical antigen combinations in patients with de novo acute myeloid leukemia. Ann Hematol 2000; 79: 543546.
  • 21
    Lowenberg B,Downing JR,Burnett A. Acute myeloid leukemia. N Engl J Med 1999; 341: 10511062.
  • 22
    Grimwade D,Walker H,Harrison G,Oliver F,Chatters S,Harrison CJ,Wheatley K,Burnett AK,Goldstone AH. The predictive value of hierarchical cytogenetic classification in older adults with acute myeloid leukemia (AML): Analysis of 1065 patients entered into the United Kingdom Medical Research Council AML11 trial. Blood 2001; 98: 13121320.
  • 23
    Al-Mawali AH,To LB,D G,Hissaria P,Mundy J,Lewis ID. The presence of leukaemia-associated phenotypes is an independent predictor of induction failure in acute myeloid leukaemia. Int J Lab Hematol (in press). Doi: 10.1111/j.1751-553X.2007.01003.x.
  • 24
    Cheson BD,Bennett JM,Kopecky KJ,Buchner T,Willman CL,Estey EH,Schiffer CA,Doehner H,Tallman MS,Lister TA,Lo-Coco F,Willemze R,Biondi A,Hiddemann W,Larson RA,Lowenberg B,Sanz MA,Head DR,Ohno R,Bloomfield CD. Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol 2003; 21: 46424649.
  • 25
    HosmerDW, LemeshowS, editors. Applied Logistic Regression. New York: Wiley; 2000.
  • 26
    Bresters D,Reus AC,Veerman AJ,van Wering ER,van der Does-van den Berg A,Kaspers GJ. Congenital leukaemia: The Dutch experience and review of the literature. Br J Haematol 2002; 117: 513524.
  • 27
    Chen W,Karandikar NJ,McKenna RW,Kroft SH. Stability of leukemia-associated immunophenotypes in precursor B-lymphoblastic leukemia/lymphoma: A single institution experience. Am J Clin Pathol 2007; 127: 3946.
  • 28
    Kern W,Voskova D,Schnittger S,Schoch C,Hiddemann W,Haferlach T. Four-fold staining including CD45 gating improves the sensitivity of multiparameter flow cytometric assessment of minimal residual disease in patients with acute myeloid leukemia. Hematol J 2004; 5: 410418.
  • 29
    Borowitz MJ,Pullen DJ,Winick N,Martin PL,Bowman WP,Camitta B. Comparison of diagnostic and relapse flow cytometry phenotypes in childhood acute lymphoblastic leukemia: Implications for residual disease detection: A report from the children's oncology group. Cytometry Part B Clin Cytom 2005; 68B: 1824.
  • 30
    Digiuseppe JA. Acute lymphoblastic leukemia: Diagnosis and detection of minimal residual disease following therapy. Clin Lab Med 2007; 27: 533549,vi.
  • 31
    Bastida Vila P,Palacio Garcia C,Solsona Riera M,Ortega Aramburu JJ,Sanchez de Toledo Codina J. [Minimal residual disease in acute lymphoblastic leukemia: A new concept of complete remission]. An Pediatr (Barc) 2005; 63: 390395.
  • 32
    Sievers EL,Lange BJ,Alonzo TA,Gerbing RB,Bernstein ID,Smith FO,Arceci RJ,Woods WG,Loken MR. Immunophenotypic evidence of leukemia after induction therapy predicts relapse: Results from a prospective Children's Cancer Group study of 252 patients with acute myeloid leukemia. Blood 2003; 101: 33983406.