Different diagnostic methods add information to define complete remission (CR) in patients with acute myeloid leukemia (AML). The detection of minimal residual disease (MRD) for predicting prognosis and for therapeutic planning still are under discussion.
The authors studied 216 patients with AML at the time of initial diagnosis and during follow-up and correlated cytomorphology, interphase fluorescence in situ hybridization (FISH), and flow cytometry results to evaluate response status. They further tested the prognostic impact of those results, especially in patients who achieved a morphologic CR.
Interphase FISH was found to be correlated significantly with the clinical course at the time of complete cytomorphologic remission and was more reliable than morphology for defining CR. Furthermore, interphase FISH was correlated with immunophenotyping results at all times during follow-up.
Acute myeloid leukemia (AML) is a heterogeneous disease with a variety of clinical, cytogenetic, cytomorphologic, and molecular subytpes.1, 2 Several parameters, such as age, leukocyte count, and especially cytogenetics predefine clinical outcomes.3–5 Furthermore, survival is influenced in part by whether patients achieve complete remission (CR) after induction therapy.4, 6, 7 Currently, a CR is defined cytomorphologically by bone marrow blasts < 5%, platelets ≥ 100 g/L, and leukocytes ≥ 1.5 g/L.8–10
Cytomorphologic CR is achieved in approximately 50–80% of all patients with AML. However, > 50% of all complete responders develop recurrent disease.4, 11 A more sensitive method for the detection of residual disease would be helpful to separate the “real” complete responders from those patients who have residual disease at a level below the detection of cytomorphology and who are more likely to develop recurrent disease.
However, cytomorphology alone does not allow the quantification of residual leukemia in patients with < 5% bone marrow blasts during follow-up for AML.12, 13 Therefore, in 2002, Cheson et al. suggested the development of a modern diagnostic panel in the routine workup of patients with AML that also would include conventional cytogenetics, polymerase chain reaction (PCR) analysis, and multiparameter flow cytometry.8 Because of the greater sensitivity of those methods compared with cytomorphology, they are used increasingly during follow-up for the quantification of minimal residual disease (MRD) in patients who achieve hematologic CR. To our knowledge, the usefulness of those diverse methods in determining CR and their thresholds for prognostic predictions and for therapeutic planning still are under discussion and need top be validated.
In the current investigation, we evaluated 216 patients with AML at the time of diagnosis and during follow-up and correlated their cytomorphologic, interphase fluorescence in situ hybridization (FISH), and flow cytometric findings with their response status. We also tested the prognostic impact of those findings, especially in patients with who achieved a morphologic CR.
MATERIALS AND METHODS
We analyzed 216 patients with AML at diagnosis who had undergone cytomorphologic evaluation of bone marrow samples in combination with cytogenetics and interphase FISH. In addition, 354 follow-up assessments applying morphology and FISH in parallel were available for these patients. Patients were diagnosed at our institution, or their samples were sent to our reference laboratory from other centers. All patients received standardized therapy; they were either included in the German AML Cooperative Group study or received treatment on the same protocol.
All patients underwent cytomorphologic bone marrow examination performed on Pappenheim-stained smears. Cytochemistry with myeloperoxidase and nonspecific esterase was performed on all samples.14 The cytomorphologic classification of AML was performed according to World Health Organization and French–American–British criteria.15
Cytomorphologic CR was defined by < 5% blasts in bone marrow according to standard criteria.8 Conventional cytogenetics were performed in all samples as described previously,16 and we used the nomenclature of the International System for Human Cytogenetic Nomenclature.17 With regard to prognosis, patients were divided into three cytogenetic subgroups. The first group included patients who had balanced translocations with a favorable prognosis: t(15;17) with PML-RARA, t(8;21) with AML1-ETO, and inv(16) or t(16;16) with CBFB-MYH11. The second group included patients who had an intermediate prognosis–patients with normal karyotype and all other abnormalities. The third group included patients with an unfavorable karyotype—losses or structural deletions of chromosomes 5 or 7, 11q23/mixed-lineage leukemia (MLL) rearrangements, inv(3)/t(3;3), and complex aberrant karyotype. For further comparisons, our evaluation focused only on patients who demonstrated abnormalities that were detectable by interphase FISH; therefore, patients with a normal karyotype were not included in the study.
Interphase FISH was performed on bone marrow samples from all patients at the time of diagnosis to confirm the cytogenetic results.16 At least 100 cells were evaluated in every sample. Interphase FISH was performed with the following centromeric probes: chromosome enumeration probe (CEP) 4, CEP 7, CEP 8, CEP 9, CEP 11, CEP X/CEP Y, and CEP X. The following locus-specific probes were used: MLL, LSI 5P15.2/LSI, LSI 5g31, CEP 7/LSI 7Q31, LSI RB, AML1, LSI IGH, LSI 20Q13.2, AML1-ETO, CBFB, and TEL. The same probes also were applied at all follow-up investigations (Abbott, Wiesbaden-Delkenheim, Germany).
Multiparameter flow cytometric analysis was performed as described previously at the time of diagnosis and for MRD.4 To account for differences in the percentages of leukemic cells covered by the respective leukemia-associated immunophenotypes (LAIPs) at diagnosis, the logarithmic differences (LD) in LAIP-positive cells between diagnosis and follow-up assessment were determined in all patients.4, 16 In patients who had reciprocal fusion transcripts PML-RARA, AML1-ETO, and CBFB-MYH11 and in patients who had 11q23/MLL rearrangements, quantitative and qualitative PCR analyses also were performed as described previously.18 In the current study, we did not focus in detail on the correlation between cytomorphology and PCR, because that issue has been addressed elsewhere.18, 19
Diagnosis was defined as Day 0. Follow-up checkpoints were grouped into 5 intervals: Days 8–60, Days 61–120, Days 121–240, Days 241–365, and > Day 365.
Spearman rank correlation was used to analyze correlations between percentages of bone marrow blasts, FISH-positive bone marrow cells, and flow cytometrically quantified levels of MRD in bone marrow. Event-free survival (EFS) and overall survival (OS) were calculated according to the method of Kaplan and Meier,20 and differences were analyzed using the log-rank test. The impact of dichotomous and continuous parameters on EFS and OS was analyzed in Cox regression models.
A total of 216 patients were included at Day 0. In total, 68 patients had favorable cytogenetics, 42 patients had intermediate cytogenetics, and 106 patients had an unfavorable karyotype. Patient characteristics are shown in Table 1.
Table 1. Characteristics of the 216 Patients with Acute Myeloid Leukemia at the Time of Diagnosis
During follow-up 354 assessments were performed with both cytomorphology and interphase FISH in parallel: Interval 1, Days 8–60 (n = 116 assessments); Interval 2, Days 61–120 (n = 93 assessments); Interval 3, Days 121–240 (n = 67 assessments); Interval 4, Days 241–365 (n = 40 assessments); and Interval 5, > Day 365 (n = 38 assessments). Each patient was analyzed at least once during follow-up.
Multiparameter flow cytometry was performed at the time of diagnosis in 142 of 216 patients. During follow-up, varying numbers of patients were analyzed by immunophenotyping for MRD (Days 8–60, 73 patients; Days 61–120, 51 patients; Days 121–240, 35 patients; Days 241–365, 14 patients; and > Day 365, 20 patients). Thus, there were 173 immunophenotyping analyses during follow-up.
Correlation between Interphase FISH and Cytomorphology and Flow Cytometry
The results of cytomorphology were correlated with interphase FISH and flow cytometry results at the time of diagnosis and at the different intervals during follow-up for the evaluation of CR and MRD (Table 2). The correlation between the percentage of aberrant interphase nuclei and the percentage of bone marrow blasts was found to be highly significant during the first, second, third, and fifth follow-up intervals (Days 8–60, Days 61–120, Days 121–240, and > Day 365; P < 0.0001 at all intervals). Significant correlations between the percentage of aberrant interphase nuclei during follow-up and the LD, as determined by flow cytometry, were found for the Day 8–60 interval (P = 0.001), the Day 61–120 interval (P = 0.020), and the Day 121–240 interval (P < 0.001).
Table 2. Correlation between Interphase Fluorescence in Situ Hybridization and Bone Marrow Cytomorphology and Flow Cytometry (Spearman Rank Test) at the Time of Diagnosis and at Follow-Up
Flow cytometry (LD)
No. of patients
LD: logarithmic difference; NA: not applicable.
> Day 365
Correlation between Cytomorphology, Interphase FISH, and Flow Cytometry Results and EFS and OS during Cytomorphologic CR in Univariate Analysis
We investigated whether there were any correlations between age, therapy-associated or secondary AML after myelodysplastic syndrome, leukocyte count, or favorable/unfavorable cytogenetics and EFS or OS in univariate analyses. EFS was dated from the initiation of therapy. The MRD time points were distributed equally within the intervals. Cytomorphology, interphase FISH, and immunophenotyping also were included in those univariate analyses.
Age was found to be correlated with both EFS (P = 0.002) and OS (P = 0.026). Favorable cytogenetics were correlated with EFS (P = 0.000001) and with OS (P = 0.000008). Unfavorable cytogenetics were correlated with EFS (P = 0.000001) and with OS (P = 0.000008). Evidence of therapy-related AML or secondary AML or leukocyte count was not found to be correlated significantly with EFS or OS.
We analyzed whether the presence of > 0% but < 5% bone marrow blasts during cytomorphologic CR were correlated with EFS. The correlation between the percentage of bone marrow blasts and EFS was significant only during the Day 121–240 interval (P = 0.03). During all other intervals, the correlation with EFS was not significant. There was no correlation noted between bone marrow blasts < 0% and < 5% during hematologic CR and OS.
We also investigated whether the presence of cells that produced positive interphase FISH results in patients in cytomorphologic CR as a continuous variable was correlated with OS or EFS at different time points (Table 3). A highly significant correlation was observed between interphase FISH and EFS for the Day 121–240 interval (P = 0.00005) and the Day 241–365 interval (P = 0.012). During the Day 121–240 interval (P = 0.015) and the Day 241–365 interval (P = 0.005), interphase FISH was found to be correlated significantly with OS.
Table 3. Correlation between Interphase Fluorescence in Situ Hybridization Results and Event-Free Survival and Overall Survival in Univariate and Multivariate Analyses
No. of patients
FISH: fluorescence in situ hybridization; NS: not significant.
0% of cells
≥ 1% of cells
0% of cells
≥ 1% of cells
0% of cells
≥ 1% of cells
0% of cells
≥ 1% of cells
> Day 365
0% of cells
≥ 1% of cells
To visualize the prognostic impact of positive FISH results during cytomorphologic CR, Kaplan–Meier plots were generated and are shown in Figures 1–4. The LD, as determined by flow cytometry, was found to be correlated significantly as a continuous variable with EFS during the Day 121–240 interval (P = 0.032). The correlation between flow cytometry and OS was significant in the Day 61–120 interval (P = 0.029).
The following parameters demonstrated significant correlations in univariate analyses: Age and cytogenetics were correlated significantly with EFS and OS. The percentage of bone marrow blasts was correlated significantly with EFS during the Day 121–240 interval. The correlation of interphase FISH with EFS was significant in the Day 121–240 interval and the Day 241–365 interval; and the correlation of interphase FISH with OS was significant during the Day 121–240 interval and the Day 241–365 interval. The LD, as determined by flow cytometry, was correlated significantly with EFS during the Day 121–240 interval and with OS during the Day 61–120 interval. Therefore, these parameters were included in the multivariate analyses.
The correlation between cytomorphology, as measured by the percentage of blasts in bone marrow, with EFS also was found to be significant during the Day 121–240 interval in multivariate analysis (P = 0.025). The presence of positive cells in interphase FISH analysis from patients in cytomorphologic CR, as a continuous variable, was found to be very significantly correlated with OS during the Day 240–365 interval (P = 0.0002) and during the Day 241–365 interval (P = 0.015) in multivariate analysis. Interphase FISH results were found to be correlated significantly with OS in multivariate analysis during the Day 241–365 interval (P = 0.01).
The flow-cytometric LD as a continuous variable was correlated significantly with EFS during the Day 121–240 interval in multivariate analysis (P = 0.044). The correlation between flow cytometry and OS was significant in the Day 61–120 interval in multivariate analysis (P = 0.037).
Correlation of the Relative Risk with Positive Results from Cytomorphology, Interphase FISH, and Flow Cytometry
We analyzed whether the relative risk demonstrated any significant correlations with cytomorphology, interphase FISH, or flow cytometry. A positive result with cytomorphology was found to be correlated significantly with the relative risk with respect to EFS during the Day 121–140 interval (P = 0.031). There were no significant correlations noted between positive results with cytomorphology and the relative risk with respect to OS.
Positive results with interphase FISH were correlated significantly with the relative risk during the Day 121–240 interval (P = 0.000) and during the Day 241–365 interval (P = 0.012) with respect to EFS (Table 4). Furthermore, significant correlations between positive FISH results and the relative risk during the Day 8–60 interval (P = 0.023), the Day 121–240 interval (P = 0.015), and the Day 241–365 interval (P = 0.005) were found with respect to OS.
Table 4. Correlation of the RR Associated with a Positive FISH Result with Respect to Event-Free Survival
Flow cytometry was found to be correlated significantly with the relative risk with respect to EFS during the Day 121–240 interval (P = 0.032) and with respect to OS during the Day 61–120 interval (P = 0.029) (Table 5).
Table 5. Correlation of the RR Associated with a Positive FISH Result with Respect to Overall Survival
The objectives of modern treatment concepts for AML include both modifying the intensity of therapy with respect to individualized risk and minimizing the risk of disease recurrence and treatment-related morbidity and mortality. Pretreatment risk stratification is based predominately on cytogenetics and age. However, according to these criteria, the majority of patients belong to the intermediate-risk group. Currently, pretherapeutic parameters for predicting the efficacy of standard therapies in this group are not available.4 Therefore, posttherapeutic parameters may provide useful information for further treatment decisions. To our knowledge to date, cytomorphology has been the gold standard for monitoring response to therapy; however, it is not considered sensitive enough to provide reliable parameters on which to base therapeutic decisions.7, 21
In addition to cytomorphology, we tested the conventional cytogenetic method for its validity as an MRD marker. Marcucci et al. demonstrated that abnormal cytogenetics at cytomorphologic disease remission were correlated significantly with shorter OS and disease-free survival.22 A cytogenetic follow-up study of 71 patients who had AML with an abnormal karyotype by Freireich et al. demonstrated that all patients who had ≥ 1 abnormal metaphase in CR developed recurrent disease. However, the sensitivity of the cytogenetic method as an MRD parameter was low, with a false-negative rate of 49%.23 In a cytogenetic study of 197 patients with AML, Konopleva et al. demonstrated that the presence of at least 1 normal metaphase on Day 21 after the first induction course was an independent, positive prognostic parameter.24
Furthermore, it also has been demonstrated that molecular diagnostics represent reliable prognostic parameters. Increases in the expression of the fusion transcripts PML-RARA, CBFB-MYH11, and AML1-ETO were predictive of recurrence in their corresponding AML subentities.18, 25
Some more recent studies also have established immunophenotyping as an MRD parameter. The cytometric response after induction therapy or after consolidation therapy was found to be correlated significantly with prognosis.4, 7, 26
In addition, metaphase FISH and interphase FISH both represent efficient and easily applicable methods in AML diagnostics. However, to our knowledge, their prognostic impact during follow-up has been tested to date only in studies of very limited size and has not been established clearly. The metaphase FISH analyses by Nylund et al. of 11 patients with AML and by El Rifai et al. of 22 patients showed that persistent or increased aberrant metaphase nuclei during CR predicted recurrent disease in nearly all patientss.27, 28 Bernell et al. reported a recurrence rate of 100% in 11 patients with AML who had persistence of aberrant interphase nuclei in hematologic CR compared with a recurrence rate of only 29% in patients who had negative FISH results.29 Amare et al. performed interphase FISH analyses in 14 patients with AML and PML-RARA during cytogenetic CR and reported a recurrence rate of 71% in patients who had > 5% aberrant nuclei compared with a recurrence rate of only 29% in patients who had < 5% aberrant cells.30 However, based on their follow-up of 13 patients with AML and inv(16), Mancini et al. concluded that interphase FISH had low sensitivity for predicting recurrence in patients who were in hematologic CR.31
For future studies, Cheson et al. suggested that the CR criteria for AML no longer should be restricted to cytomorphology and peripheral blood parameters alone but should be redefined to also include conventional cytogenetics, PCR analysis, and multiparameter flow cytometry.8 Therefore, first, we evaluated whether interphase FISH was qualified for inclusion in the remission criteria for AML by performing interphase FISH analyses in 216 patients with AML at diagnosis and at 354 time points during the follow-up of patients who were in cytomorphologic CR to evaluate the correlation between FISH and other diagnostic methods and to define its validity as an MRD parameter.
We were able to demonstrate that interphase FISH was correlated with the percentage of blasts at the time of diagnosis. At all analyzed intervals during follow-up, the correlation between interphase FISH and blast percentage also was significant.
However, with cytomorphologic methods, only blasts ≥ 5% can be detected as malignant cells. However, mature cells also may belong to the malignant clone, and they can be identified with interphase FISH. Conversely, cytomorphology after chemotherapy may lead to false estimations of increased proportions of blasts in regenerating bone marrow.
In addition, 142 patients were analyzed by flow cytometry at the time of diagnosis. Interphase FISH results were correlated with immunophenotyping results at all intervals during follow-up. The correlation of interphase FISH results with immunophenotyping results was stronger in patients who had higher percentages of aberrant interphase nuclei than in patients who had lower proportions of aberrant cells. This finding was anticipated and correlates with the higher sensitivity of immunophenotyping compared with interphase FISH.
Comparison of the relative risk of positive results with the three analyzed methods with respect to EFS and OS demonstrated the strongest correlations for interphase FISH. In the current analysis, we demonstrated clearly that interphase FISH at cytomorphologic CR was correlated significantly with the clinical course. The occurrence of aberrant interphase nuclei was found to be correlated significantly with EFS in both univariate and multivariate analyses. A significant correlation with OS also was found in univariate and multivariate analyses. Therefore, interphase FISH appears to be more reliable for defining CR than morphology and should be tested as an MRD parameter in patients with AML.
Because multiparameter flow immunophenotyping represents a newly established method for detecting MRD, interphase FISH is validated indirectly as an MRD parameter through its correlation with immunophenotyping. Interphase FISH is more sensitive than cytomorphology but cannot be performed in all patients. Immunophenotyping can be performed in all patients and is the most sensitive method. Cheson et al. suggested the inclusion of cytogenetic and molecular genetic CR in addition to immunophenotyping as response criteria in patients with AML.8 The current results suggest also the inclusion of interphase FISH into this modified definition of CR of AML because it demonstrated high prognostic power.
We recommend using cytomorphologic evaluation of bone marrow and peripheral blood in all patients at the time they are diagnosed with AML. If abnormalities are detected by standard cytogenetics for which interphase FISH probes are available, then FISH must be performed at diagnosis with the same probes. This will allow the respective monitoring of residual disease during the clinical course.
Multiparameter flow cytometry should be included in the assessment of all patients at the time of diagnosis to determine the presence of leukemia-associated, aberrant immunophenotypes for further MRD diagnostics. Only with the prospective combination of these methods, in addition to cytomorphology, will we be able to redefine the CR criteria for individual patients with AML and provide information for further MRD diagnostics and for individual therapeutic decisions based on the patient's individual risk of disease recurrence.