CME. The Role of Multiparameter Flow Cytometry for Detection of Minimal Residual Disease in Acute Myeloid Leukemia

Hematopathology / Minimal Residual Disease in AML The Role of Multiparameter Flow Cytometry for Detection of Minimal Residual Disease in Acute Myeloid Leukemia Adhra Al-Mawali, PhD, 1-3 David Gillis, MBBS, 2 and Ian Lewis, MBBS, PhD 1,3 Key Words: Acute myeloid leukemia; Immunophenotyping; Multiparameter flow cytometry; Minimal residual disease; Leukemia-associated phenotypes CME DOI: 10.1309/AJCP5TSD3DZXFLCX Upon completion of this activity you will be able to: define what is meant by minimal residual disease (MRD). describe the factors that influence flow cytometry for detecting MRD. list the relevant types of aberrations in acute myeloid leukemia. discuss the current strategy for MRD studies. discuss the limitations for MRD detection using immunophenotyping. Abstract The presence of minimal residual disease (MRD) in the bone marrow (BM) of patients with acute myeloid leukemia (AML) following chemotherapy has been established by many studies to be strongly associated with relapse of leukemia. In addition, detection of MRD is the major objective of many of the newer diagnostic techniques used in malignant hematology. Because of the wide availability and conceptual straightforwardness of immunophenotyping, flow cytometry is the most accessible method for MRD detection. This review is not an overview of all MRD studies, but rather discusses the possibilities for optimizing MRD detection, the use of multiparameter flow cytometry (MFC) techniques in MRD detection, and the implications for future patient treatment. This review focuses on MRD detection in AML using MFC and discusses the reported correlations of MRD, clinical and biologic features of the disease, and outcome. In addition, it discusses the laboratory and clinical aspects of this approach. The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this educational activity for a maximum of 1 AMA PRA Category 1 Credit per article. The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose. Questions appear on p 143. Exam is located at www.ascp.org/ajcpcme. Patients with acute myeloid leukemia (AML) have approximately 2 malignant cells at the time of diagnosis. 1 The disease is considered to be in complete remission (CR) when fewer than 5% of malignant cells in the bone marrow (BM) are morphologically detectable following chemotherapy. 2 However, the patients may still harbor as many as 0 leukemic cells, and the level of leukemia cells is largely unknown. 3 The sensitivity of morphologic studies can be improved by cytochemical staining, but the detection limit of these methods remains at a level of 1 leukemic cell in a normal cell population ztable 1z. 4 The leukemic population undetectable by morphologic methods has been defined as minimal residual disease (MRD). 2 In other words, MRD is a term used when there is evidence (immunophenotypic, molecular, or cytogenetic) that leukemic cells remain in the BM but there are insufficient cells to be detected by routine examination under the microscope. Despite a high remission rate (approaching 80% in younger adults) after intensive chemotherapy, only 30% to 40% of patients survive 5 years after diagnosis. 5 Many patients will experience relapse, which is caused by the presence of MRD. Using an accurate combination of monoclonal antibodies (MoAbs) with multiparameter flow cytometry (MFC) allows more specific detection of leukemic cells, which can be applied for quantification of MRD. In the past 2 decades, tremendous effort has been made to describe and characterize residual leukemic cells and relate their presence to prognosis and therapy. The goal of studies of MRD detection 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 16 Am J Clin Pathol 2009;131:16-26 16 DOI: 10.1309/AJCP5TSD3DZXFLCX

Hematopathology / Review Article ztable 1z Methods for Detection of Minimal Residual Disease in AML Technique Advantages Disadvantages Sensitivity Standard morphologic studies Low sensitivity; not suitable for low level 1%-5% of leukemic cells Cytogenetics Labor-intensive; slow; requires metaphase 5% chromosome preparations FISH Dividing cells not required; large number of cells Labor-intensive; limited sensitivity 0.3%-5% can be analyzed in a short time; interphase FISH precludes need for high-quality metaphases (cf standard cytogenetics) Gene rearrangements by Low risk of contamination; patient-specific Labor-intensive; slow; limited sensitivity 1%-5% Southern blotting PCR analysis Can be identified with limited set of primers; False-positive results; applied in <50% of 10 4 to 10 5 high stability of DNA; relatively easy; rapid AML cases; relatively expensive (1-3 d); no (or very low <10 6 ) background in normal cells; sensitive; patient specific Flow cytometric immuno- Applicable for most cases (>80%); quantification Not as specific as PCR; presence of 10 4 phenotyping simple; single cell analysis; cell viability can be subpopulation in AML; immunodetermined; information on normal cells; phenotypic shifts can occur between relatively easy; cheap; rapid (1-2 d); relatively diagnosis and relapse patient-specific AML, acute myeloid leukemia; FISH, fluorescence in situ hybridization; PCR, polymerase chain reaction. specific biologic indications. The ultimate purpose of these types of studies is to discriminate patients who respond successfully to treatment and who are likely not to need additional therapy from patients at high risk of relapse who require therapeutic intervention to reduce the risk. This review focuses on the practical application of MFC for MRD investigation in AML and its clinical and biologic importance. Immunologic Detection of MRD by MFC Detection of MRD by flow cytometry in AML presents some specific difficulties owing to immunophenotypic heterogeneity. 6 AML cells usually spread across many areas of the dot plot instead of forming a tight cluster when compared with acute lymphoblastic leukemia (ALL), which is usually homogeneous. Immunophenotypic detection of MRD in AML can be performed by defining aberrant marker expression, denoted as leukemia-associated phenotypes (LAPs), on malignant cells at diagnosis. 7-11 These LAPs are not present or only very infrequently present on normal blood or BM cells. For AML, the most relevant types of aberrations include the following: (1) asynchronous antigen expression (simultaneous expression of early and late markers in 1 cell, such as coexpression of the CD34 and CD15 antigens); (2) lineage infidelity (expression of the lymphoid-associated markers, ie, CD2, CD3, CD5, CD7, CD10, and CD19 on myeloid blast cells); (3) antigen overexpression (abnormally increased expression of a certain antigen per cell); (4) aberrant light-scatter properties (the expression of lymphoidassociated antigens in blast cells displaying a relatively high forward scatter and side scatter, corresponding to normal myeloid cells) 7,11-14 ; and (5) absence of lineage-specific antigens (absence of antigen expression such as CD13 and CD33 on myeloid blasts). Current immunologic strategies for detecting MRD rely on combinations of leukocyte markers normally not seen in peripheral blood and BM. 12 These LAPs can be identified by 3-, 4-, or 5-color staining techniques performed with antibodies conjugated to different fluorochromes. In reality, using more than 2-color staining allows more precise detection of LAPs. Blasts should be identified by a CD45/side scatter gating strategy. In addition, back-gating strategies using CD34 and/ or CD117 should be used to better define the blast population. CD45/CD34/CD117, when possible, should be used in combination with different myeloid and lymphoid markers to increase the sensitivity of LAP detection. Incorporating various permutations and combinations allows the maximum advantage to be attained in defining the LAP. Appropriate isotype-matched negative control samples should be used in the panel of MoAbs to assess background fluorescence intensity, thresholds for positivity should be based generally on isotype and internal negative control samples, and not fewer than 250,000 events should be acquired. The data should be analyzed for numeric coexpression ( 10% overlap on the blast population) for 2 markers of interest at a time and for numeric evidence of unusual phenotypes. The current strategy for MRD studies relies on the following steps zfigure 1z: (1) identification of the unique phenotype by a multicolor staining technique, (2) definition of a patient-specific immunophenotype or LAP, and (3) tracking residual leukemic cells after CR is achieved and during subsequent follow-up, using the specific patient LAPs zimage 1z. Am J Clin Pathol 2009;131:16-26 17 17 DOI: 10.1309/AJCP5TSD3DZXFLCX 17

Al-Mawali et al / Minimal Re s i d u a l Disease in AML Screening panel of MoAbs by multiple staining assays at presentation of BM samples AML ALL Secondary panel to identify LAP using 5-color staining and CD45 gating strategy LAP+ patient LAP patient Factors That Influence Flow Cytometry for Detecting MRD The applicability and feasibility of using MFC for MRD detection depends on 2 factors. First, the frequency of LAPs that can be identified in AML, which determines the number of patients who can be monitored for MRD by MFC, and, second, the level of sensitivity that can be achieved using MFC. Frequency of LAPs Definition of specific LAP+ per patient and the % of LAP+ at diagnosis Use LAP+ during follow-up BMs to track MRD Establish % MRD after every cycle of treatment Correlate % MRD with relapse, RFS, and OS The proportion of leukemia cases that can currently be monitored for MRD by MFC varies from laboratory to laboratory. Factors influencing this variability include the following: (1) the number of markers tested, (2) the inclusion of normal and regenerating BM samples postchemotherapy in determining normal vs abnormal phenotypes and levels, and (3) the rigor with which the laboratory defines LAPs and MRD. 15 MRD studies reported in the literature indicate that in AML, 60% to 88% of patients have an aberrant phenotype at diagnosis. We have shown that LAPs can be identified in 94% of patients with AML when an extensive and comprehensive panel of MoAbs using 5-color MFC is used. 11 Adriaansen et al 16 identified subsets of myeloblasts that expressed terminal deoxynucleotidyl transferase in 75% of 45 AML cases; however, in most of these cases, terminal deoxynucleotidyl transferase was positive in fewer than 20% of blasts. Reading et al 9 detected LAPs in 85% of 272 cases of AML. Macedo et Not suitable to track MRD zfigure 1z Algorithm used for identifying leukemia-associated phenotype (LAP) in patients with acute myeloid leukemia (AML) and for detection of minimal residual disease (MRD). Once LAP is identified, it will serve to establish a phenotype to trace residual leukemia after a complete remission is recognized morphologically. Adapted from Campana and Pui 12 and Campana and Coustan-Smith 15 and previous literature on MRD. ALL, acute lymphoblastic leukemia; BM, bone marrow; MoAbs, monoclonal antibodies; OS, overall survival; RFS, relapse-free survival; +, positive;, negative. al 7 found that 29 (73%) of 40 AML cases analyzed displayed the existence of at least 1 aberrant phenotype. San Miguel et al 13 reported that 46 (87%) of 53 AML cases had an aberrant phenotype; this figure was confirmed in another report on 126 AML cases. 17 In the experience of Venditti et al, 18,19 70% of 113 newly diagnosed AML cases had an aberrant phenotype. In addition, an incidence as high as 89% has been also reported, 20 possibly because of the use of a large variety of MoAbs ztable 2z. In our study, 11 if we were to include a broader definition of LAP incorporating abnormal intensity of antigen expression, we could potentially increase the percentage of LAP detection from 94% to 100%. In our experience, the most common LAPs identified were CD117+/CD15+, CD117+/CD65+, CD34+/CD15+, and CD34+/CD65+, and these were present in 25 (49%), 22 (43%), 20 (39%), and 15 (29%) of 51 cases studied from 2005 to 2007, respectively 11 ztable 3z. The lowest incidence of aberrant immunophenotypes was reported by Drach et al, 21 who observed LAPs in 35 (51%) of 68 AML cases. The use of double, but not triple or quadruple, staining assays and a limited panel of antibodies explains this lower frequency. Altogether, these results demonstrate that LAP detection by MFC is a feasible approach for MRD detection in a significant proportion of AML cases. In this respect, further development is needed with the increasing knowledge about the antigenic composition of the leukemic cells. 18 Am J Clin Pathol 2009;131:16-26 18 DOI: 10.1309/AJCP5TSD3DZXFLCX

Hematopathology / Review Article A C1 C2 C1 C2 C1 C2 C1 C2 C1 C2 CD117 PE C3 CD117+CD15+ 24% C4 CD117 PE C3 CD117+CD15+ 0.78% C4 CD117 PE C3 CD117+CD15+ 0.42% C4 CD117 PE C3 CD117+CD15+ 0.64% C4 CD117 PE C3 CD117+CD15+ 52% C4 CD15 FITC CD15 FITC CD15 FITC CD15 FITC CD15 FITC CD34 PE B G1 G3 G2 G4 CD34+CD7+ 32% CD7 FITC CD34 PE G1 G3 G2 G4 CD34+CD7+ 0.12% CD7 FITC CD34 PE G1 G3 G2 G4 CD34+CD7+ BM: AML at Diagnosis After Induction After Consolidation I After Consolidation II Follow-up 0.15% CD7 FITC CD34 PE G1 G3 G2 G4 CD34+CD7+ 0.13% CD7 FITC CD34 PE G1 G3 G2 G4 CD34+CD7+ 0.09% CD7 FITC zimage 1z Detection of minimal residual disease (MRD) in consecutive bone marrow (BM) samples from a patient with relapse (A) and a patient still in remission (B). Cells from acute myeloid leukemia (AML) at diagnosis and follow-up BM samples are stained with a combination of monoclonal antibodies, which identifies a leukemia-associated phenotype (LAP). The gating strategy starts by defining the WBC compartment characterized by CD45 expression and SS Log. A, On these selected cells, the cell population with a primitive marker expression, in this case CD117, and low side scatter (SSC) is subsequently gated for detection of the cells with aberrant phenotypes. The patient experienced relapse within 6 months after achieving complete remission (CR). The aberrant phenotype was CD117+/CD15+ expression on the CD34+/CD117+ cells. The MRD percentages were 0.78%, 0.42%, and 0.64% after induction and consolidation I and consolidation II chemotherapy, respectively. The dot plot at the extreme right shows relapsed material with LAP expression similar to that at diagnosis. B, On these selected cells, the cell population with a primitive marker expression, in this case CD34, and low SSC is subsequently gated for detection of the cells with aberrant phenotypes. The patient was still in CR after 24 months. The LAP includes CD34+/CD7+ expression. The MRD percentages were 0.12%, 0.15%, 0.13%, and 0.09% after induction, consolidation I and consolidation II chemotherapy, and at follow-up, respectively. FITC, fluorescein isothiocyanate; PE, phycoerythrin; SS, side scatter. ztable 2z Various LAPs and Their Frequencies Asynchronous Lineage Antigen Aberrant Absence of No. of MFC LAPs Antigen Infidelity Overexpression Light-Scatter Lineage-Specific Report Colors Used (%) Expression (%) * (%) * (%) * Properties (%) * Antigens (%) * Al-Mawali et al 14 3 64 52 35 ND ND 35 Al-Mawali et al 11 5 94 73 20 ND ND 7 Babusíková et al 2 56 13 42 1 NR NR Bahia et al 20 3 89 82.4 34.3 NR NR NR Drach et al 21 2 51 13 32 NR NR 16 Macedo et al 7 2 and 3 73 62.5 37.5 17.5 32.5 NR Reading et al 9 2 and 3 85 70 54 NR NR NR Laane et al 3 93.5 83 55 16 NR 9 San Miguel et al 17 3 NR 78 29 21 17 NR Feller et al 22 4 60 55 45 2.5 NR NR Voskova et al 23 2 82 50 64.3 14.3 50 27.8 Kern et al 100 14.2 26.4 32.9 NR 26.4 AML, acute myeloid leukemia; LAPs, leukemia-associated phenotypes; MFC, multiparameter flow cytometry; ND, not done; NR, not reported. * Percentage of total AML cases unless otherwise indicated. Percentage of total number of LAPs. Babusíková O, Glasová M, Koníková E, et al. Leukemia-associated phenotypes: their characteristics and incidence in acute leukemia. Neoplasma. 1996;43:367-372. Laane E, Derolf AR, Björklund E, et al. The effect of allogeneic stem cell transplantation on outcome in younger acute myeloid leukemia patients with minimal residual disease detected by flow cytometry at the end of post-remission chemotherapy. Haematologica. 2006;91:833-836. Am J Clin Pathol 2009;131:16-26 19 19 DOI: 10.1309/AJCP5TSD3DZXFLCX 19

Al-Mawali et al / Minimal Re s i d u a l Disease in AML The Quality of LAP for MRD Detection Analytic sensitivity refers to how good LAP detection by MFC is at correctly identifying LAPs. Specificity, on the other hand, refers to how good the test is at correctly identifying cells that are normal. Thus, the quality of LAP for MRD detection depends on the following: (1) Specificity: Specificity depends on the percentage of LAP expression on normal BM cells; a high specificity can be achieved by including primitive markers (CD34, CD133, and CD117) if present on AML. In most cases, LAP expression on normal BM cells is less than 0.1%. 7-11 (2) Sensitivity: The sensitivity of MRD detection depends, among other factors, on the percentage of LAP expression on the leukemic blast population at diagnosis and the number of cells analyzed. For this reason only LAPs that are expressed on more than 10% on the leukemic blast population should be considered. 9 (3) Stability: LAPs may undergo phenotypic shifts. During the disease, marker expression in AML may disappear, resulting in false negativity. Especially dim expression of markers is susceptible to change. Furthermore, CD19 expression has been shown to disappear on several occasions in the course of disease. (These definitions are adapted from Dutch/Belgium task force for MRD detection in AML in cooperation with the European Working Group on Clinical Cell Analysis.) Points for Consideration for Detection of LAPs Several issues need to be considered for LAPs detection. There is a need for testing a wide and comprehensive antibody panel with progenitor markers included (such as CD34, CD117, and CD133). In our experience, 11 using maximum log difference analyses from the median frequency in normal and regenerating BM samples resulted in 5 most sensitive LAPs in our series of 54 newly diagnosed AML cases, and these were CD2, CD56, CD7, CD11b, and CD19 in normal BM and CD56, CD11b, CD2, CD7, and CD19 in regenerating BM, in order of their sensitivity based on maximum log difference, which means that these LAPs are the most sensitive markers for tracking MRD when present in AML samples. These markers represented a reasonable percentage of our cohort: CD2 (9%), CD56 (17%), CD7 (28%), CD11b (14%), and CD19 (4%). Furthermore, our data showed that CD2, CD56, CD7, CD19, and CD11b are potentially the most useful markers for MRD detection because these markers allowed the maximum analytic sensitivity when normal and regenerating BM samples were taken into account as baseline. Of the 54 cases, 24 (44%) expressed at least one of these markers, and, thus, they should be included in any diagnostic panel of AML for MRD monitoring. The correct interpretation of antigen expression patterns as compared with normal vs regenerating hematopoietic cells also needs to be considered. Another issue for consideration in LAP detection is the concept that the leukemic blasts are ztable 3z Frequencies of LAPs in 51 Patients With AML Range of Positive No. of Cells in AML Bone LAPs (n = 200) Cases Marrow (%) Lineage infidelity (n = 40) CD34+/CD2+ 5 26-92 CD34+/CD7+ 12 11-88 CD34+/CD10+ 1 22 CD34+/CD19+ 2 24-65 CD117+/CD2+ 5 13-91 CD117+/CD7+ 11 18-90 CD117+/CD10+ 1 17 CD117+/CD19+ 2 15-53 CD34+/Glycophorin A+ 0 0 CD117+/Glycophorin A+ 1 30 Asynchronous antigen expression (n = 146) CD34+/CD11b+ 11 11-80 CD34+/CD14+ 0 0 CD34+/CD15+ 20 10-36 CD34+/CD56+ 5 29-84 CD34+/CD64+ 6 10-49 CD34+/CD65+ 15 11-63 CD117+/CD11b+ 2-62 CD117+/CD14+ 0 0 CD117+/CD15+ 25 10-55 CD117+/CD56+ 6 11-80 CD117+/CD64+ 9 11-70 CD117+/CD65+ 22 10-62 CD33+/CD15+ 2 80-94 CD33+/CD64+ 5 18-97 CD33+/CD65+ 5 10-99 CD33+/CD11b+ 3 20-26 CD33+/CD56+ 2 10-78 Lack of lineage-specific antigen (n = 14) CD33++/CD13 5 93-99 CD33 /CD13++ 9 25-98 AML, acute myeloid leukemia; LAPs, leukemia-associated phenotypes; +, positive; ++, strong positive;, negative. certainly composed by distinct phenotypic clones. 12,19,21 In this regard, Macedo et al 7 found that most patients with AML with a LAP have more than 1 aberration. Although this finding needs to be interpreted with caution, multiple staining assays to confirm the coexistence of more than 1 aberrant phenotype may be a tool to identify the population of interest in a more appropriate way and then to negate the phenomenon of phenotypic switch at relapse. Level of Sensitivity In theory, MFC may allow a maximum level of sensitivity of 1 leukemic cell in 10 6 after an accurate and prolonged cleansing of the fluidics system. 15 However, a more realistic figure for practical applications is 1 target cell in 10 4 normal cells. 15 By using serial dilution of leukemic cells in normal peripheral blood and BM cells, it was shown that MFC is able reliably to detect cells displaying LAPs up to a level of 10 4. 11 However, it should be noted that the level of sensitivity varies (from 10 3 to 10 5 ) depending on the type of phenotypes being analyzed, the combination of MoAb reagents used for their detection, 15 and the sample under study. 20 Am J Clin Pathol 2009;131:16-26 20 DOI: 10.1309/AJCP5TSD3DZXFLCX

Hematopathology / Review Article The introduction of MFC analysis has improved the sensitivity and specificity of leukemic cell detection. 24-27 In the experience of some groups, serial dilution experiments of leukemic cells with normal BM cells determined that the detection limit of MFC ranges between 10 4 and 10 5. 3,11,14,19,28 Advantages and Drawbacks of MFC for Detecting MRD Polymerase chain reaction (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 Fms-like tyrosine kinase 3 (FLT3) gene and partial tandem duplications within the MLL gene, in addition to the fusion transcripts AML1-ETO, CBFb- MYH11, and PML-RARa. 29 Although PCR techniques seem to be promising for monitoring MRD in AML, they are currently applicable only on leukemias that bear specific DNA markers such as fused genes. The most common chromosomal translocations, ie, t(8;21), t(15;17), and inv(16), comprise only about 30% of patients with AML. 30 However, this still results in approximately 70% 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 patients with acute promyelocytic leukemia (APL). 31 In contrast, recent data indicate that immunologic monitoring may be applicable to more than 90% of patients with AML when pursuing a comprehensive approach. 10,11,14,29 Additionally, sensitivity of MRD detection can range as high as 1 leukemic cell per 10 4 to 10 5 normal cells. Furthermore, rapid and accurate quantification can be achieved. Factors that reduce the sensitivity of immunophenotyping include the following: (1) the lack of antigen specificity for malignant cells because these cells represent the counterparts of normal cells that, in many cases, have identical or similar antigen profiles; (2) the existence of several subpopulations, some of them as minor clones, that are difficult to identify; (3) the incapability to identify phenotypic switch; and (4) the need to count a large number of cells and for technical expertise. The phenomenon of immunophenotypic switch is relatively rare in AML but poses a significant problem in using the LAPs as a strategy for MRD detection. 32,33 The analysis of LAPs by 5-color MFC should help in addressing this problem; we have demonstrated detection of more than 1 LAP in 78% of AML cases, which increases the probability of detection for MRD analysis. 11 The large number of cells that need to be counted and the technical expertise required preclude the practicality of this procedure for some laboratories. 34 The heterogeneity of AML and the large number of myeloid progenitors in normal BM are major concerns in favor of a relative insensitivity of the techniques for MRD detection. However, the introduction of MFC analysis has improved the sensitivity and specificity of leukemic cell detection. 24-27 Clinical Studies of MRD Investigation in AML Although studies of MRD detection by MFC in AML are limited as compared with ALL, 35-38 several reports have been published providing evidence that study of MRD is a valuable tool for predicting relapse. 13,17-19,21,22,33,39,40 Drach et al 21 evaluated MRD in 68 patients with AML. In follow-up assessment of BM samples collected consecutively after induction therapy, they observed that only patients in whom a LAP continued to be detected experienced early relapse. On the other hand, patients who became MRD-negative remained in CR with a median duration of remission of 52 weeks after the first negative finding. They concluded that the persistence of MRD was indicative of impending relapse. In a pediatric series of 39 AML patients, Sievers et al 40 detected MRD in more than 50% of patients in morphologic CR; the estimated risk of relapse during the MRD-positive period was 2.8 times greater than during time in which MRD was negative (P =.02). In another study, 22 AML patients in a trial setting were evaluated for residual leukemic cells. MRD was monitored in follow-up samples taken from the BM of 72 patients after 3 different cycles of chemotherapy and from autologous peripheral blood stem cell (PBSC) products. The percentage of MRD in BM after the first cycle (n = 51), second cycle (n = 52), and third cycle (n = 30) and in PBSC products (n = 39) strongly correlated with relapse-free survival (RFS). At a cutoff level of 1% after the first cycle and median cutoff levels of 0.14% after the second cycle, 0.11% after the third cycle, and 0.13% for PBSC products, the relative risk of relapse was higher by factors of 6.1, 3.4, 7.2, and 5.7, respectively, for patients in the higher MRD groups. In addition, the absolute number per milliliter of MRD cells was highly predictive of the clinical outcome. After treatment had ended, an increase of the percentage of MRD predicted impending relapse, with MRD assessment intervals of 3 months or less. Thus, Feller et al 22 concluded that MRD assessment at different stages of disease is highly reliable in predicting relapse risk and survival even after the completion of chemotherapy, and sequential evaluation of MRD frequencies can be used to predict relapse. In our study, 41 5-color MFC and receiver operating characteristic (ROC) analysis were used to determine the optimal threshold that can separate patients into 2 groups in terms of leukemic residual cells and relapse status after induction and consolidation chemotherapy. We analyzed 54 AML cases. LAPs were detected in 51 (94%) of 54 cases. MRD was evaluated in the BM during morphologic CR from 25 and 22 patients after induction and consolidation chemotherapy, Am J Clin Pathol 2009;131:16-26 21 21 DOI: 10.1309/AJCP5TSD3DZXFLCX 21

Al-Mawali et al / Minimal Re s i d u a l Disease in AML respectively. The threshold discriminating MRD-negative from MRD-positive cases was set at 0.15% residual leukemic cells using ROC analysis, a level that allowed optimal sensitivity and specificity for prediction of relapse at postinduction (P =.05) and postconsolidation (P =.009) time points. The postinduction MRD level influenced not only RFS (P =.004), but also overall survival (OS) (P =.003). Multivariate analysis showed that the postinduction MRD level was a powerful independent prognostic factor for RFS (P =.037) and OS (P =.026). Our study concluded that a threshold MRD level of 0.15% was the optimal value for discriminating risk categories in AML. In addition, postinduction MRD assessment was able to better predict disease outcome than consolidation, and, thus, MRD analysis by MFC could be used for refining the selection of therapeutic strategies and improving clinical outcome in individual patients. San Miguel et al 13 examined the prognostic role of MRD detection in 53 patients with AML with LAPs at diagnosis. The patients had achieved morphologic CR after standard AML regimens, and the levels of MRD were studied serially at the end of induction and intensification therapy. The authors established that the amount of MRD at the end of induction and intensification was directly correlated with relapse and RFS, in line with our findings. Specifically, they found that patients with MRD cells of 0.5% or more after induction therapy had a higher rate of relapse (67%) compared with patients with fewer than 0.5% residual cells (20%) (P =.002). Similarly, these patients had a shorter RFS (P =.01). At the end of intensification, the value of 0.2% residual leukemic cells grouped patients in 2 categories with relapse rates of 69% and 32%, respectively (P =.02); the patients with 0.2% or more residual leukemic cells also had a lower RFS (P =.04). Detection of MRD cells of 0.5% or more after induction and 0.2% after intensification was also correlated with a shorter duration of OS. Finally, the impact of the level of MRD on RFS was established in multivariate analysis. However, Venditti et al, 18 in their study of 56 patients with AML with LAPs, found that the level of MRD after consolidation therapy was the best predictor of outcome. In fact, an MRD level of 0.035% after consolidation was significantly correlated with a high relapse rate (77% vs 17%) (P <.001). This MRD level was also significantly correlated with poor or intermediate cytogenetics, MDR1 phenotype, and short duration of RFS and OS (P =.014; P =.031, P =.00022, and P =.00014, respectively). The prognostic role of MRD positivity after consolidation therapy was confirmed in multivariate analysis. It is important to note that in their study, autologous stem cell transplantation was shown not to have a role in changing the negative effect of high MRD levels after consolidation. Indeed, the relapse rate after transplantation was 70% vs 28% in patients with a negative result at the end of consolidation (P =.031). Although the concluding message is the same, the findings of Venditti et al 18 and of our study 41 are discrepant. In fact, we and San Miguel et al 13 demonstrated a correlation with relapse for level of MRD after induction; in contrast, Venditti et al 18 found that the level of MRD after consolidation, but not after induction, was the important prognostic factor. The different therapeutic regimens used in the 2 studies may explain this variance. In the study be San Miguel et al, 13 the induction therapy consisted of 1 or 2 courses of an anthracycline and cytarabine (3 + 7 regimen), followed by an identical consolidation course. This was followed by 1 or 2 intensification courses consisting of intermediate- or high-dose cytarabine and daunorubicin or idarubicin. However, in the study by Venditti et al, 18 the patients were treated on the European Organization for Research and Treatment of Cancer/Gruppo Italiano Malattie Ematologiche dell Adulto (GIMEMA) AML-10 and AML-13 protocols, which are 3 drug-based regimens, associating an anthracycline with cytarabine and etoposide. In addition, in the AML-10 protocol, cytarabine administration is prolonged through 10 days instead of the conventional 7 days during the induction phase. Thus, one would believe that in the study of San Miguel et al 13 and our study, less intensive therapy was associated with a milder purging effect, which in turn may account for different levels of MRD associated with increased risk of relapse. This may also explain why the number of residual leukemic cells after induction correlated with the rates of relapse and RFS, whereas, in the experience of Venditti et al, 18 such a correlation was found only after consolidation. San Miguel et al, 17 in a further report, confirmed the importance of the level of MRD after induction therapy. The authors found that early response to chemotherapy, as determined by immunophenotyping in the first BM sample in morphologic CR after induction therapy, might identify patients at low risk of relapse. The study included 126 patients with AML who had aberrant phenotypes at diagnosis. Based on the level of MRD after induction therapy, 4 categories of risk were identified: (1) very low risk (<10 4 cells), (2) low risk (10 4 to 10 3 cells), (3) intermediate risk (10 3 to 10 2 cells), and (4) high risk (>10 2 cells). The authors concluded that MRD detection after first cycle of chemotherapy and achieving morphologic CR is an important tool for risk assessment in patients with AML. In our study, 41 even with a smaller cohort of patients, we were able to show that MRD detection greater than 0.15% is an independent predictor of poor prognosis after induction chemotherapy. Our studies support MRD detection as a prognostic indicator of relevance in clinical practice. The findings of recent studies demonstrate that the precise evaluation of MRD by MFC has prognostic significance 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 22 Am J Clin Pathol 2009;131:16-26 22 DOI: 10.1309/AJCP5TSD3DZXFLCX

Hematopathology / Review Article may help design 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, thereby propagating the potential benefits of MRD monitoring to all patients. analytic procedure to be used. In other words, one of the most relevant aspects when performing or requesting flow cytometric immunophenotyping of AML is to have a clear idea of the type of information provided by this test that could help in answering a specific clinical question. Potential Limitations for MRD Detection Using Immunophenotyping The literature varies in the reported incidence of LAPs expressed in leukemic samples, and a clear example is the described incidence for cross-lineage and asynchronous antigen expression in AML 7,9,12,23,42-45 and its potential prognostic value. A critical analysis of these reports shows that there is a wide range of reagents and methods used for sample preparation and data acquisition and analysis. In addition, a lack of standardized criteria for data interpretation is a common finding that introduces variability. Hence, morphologic examination continues to be the standard of reference for the immunophenotypic characterization of leukemia, which, according to Paietta et al, 46 should be considered a relic of the past. A significant number of factors that affect the results of MRD analysis have been identified. 47-53 Among them, technical aspects such as the type and quality of the sample, the reagents and sample preparation protocols, instrument setup and calibration, and the potential component of bias introduced during data analysis or with the interpretation of the results correspond to the most common sources of variability. Also, the inability to identify immunophenotypic switch and lineage switch, a phenomenon that may occur at relapse, although at a low rate in patients with AML, is a limitation. 13,17 Based on this knowledge, several groups 47-53 have discussed these aspects in detail and have reported their consensus opinions and recommendations. However, in many of these reports, the type of information provided by the flow cytometric immunophenotyping of AML is not analyzed in depth. Such a detailed analysis would certainly contribute to understanding the answers provided by flow cytometry to the specific questions posed in current practice that are related to MRD in AML, including detection of MRD in CD34 and CD117 leukemic cells; accurate quantification of overexpression of myeloid or stem cell markers, as it is not clear from the literature what is the true definition or level of overexpression of an antigen; and the true threshold of MRD positivity or negativity, which is likely to vary between centers. Moreover, although in few cases, however, the differentiation of leukemic blasts and monocytes in M4 and M5 using MFC is not clear, as usually in these cases the blast cells form a continuum and merge with a large monocytoid population. Therefore, these areas need further investigation. In general, it may be considered that, at least to a certain extent, the type of information requested influences the Points for Consideration A number of important points for consideration and questions, which have biologic and clinical relevance, remain unanswered. Eight of them are listed here. 1. There is a need for changing the current definition of CR, which is still based on the morphologic appearance of BM, whatever the method used. Investigation of MRD identifies patients who will experience relapse regardless of a standard morphologic CR. Following the proposal of Pui and Campana 54 in pediatric ALL and the paradigmatic experience of the GIMEMA group in the treatment of APL, 55 it may be useful in AML if CR is redefined according to immunologic and molecular criteria. 2. Which level of MRD correlates with cure? 3. The value of immunophenotyping compared with molecular monitoring of MRD needs to be studied. The 2 techniques may be able to be efficiently integrated because approximately 30% of AML cases have distinguishable molecular markers in contrast with the high incidence of LAPs. The combined use of fluorescence in situ hybridization and MFC for MRD detection established that only 25% to 30% of the residual leukemic cells as identified by an aberrant phenotype had the same cytogenetic markers as observed at diagnosis in at least 90% of the blasts. 19 Theoretically, molecular defects and chromosomal abnormalities may take place at precise time points during the development of the leukemic phenotype, as established in APL. 56 4. The role of additional therapy in the case of a high level of MRD after induction and/or consolidation therapy needs to be determined. 5. Analysis of peripheral blood samples to determine whether MRD may be successfully monitored in this source, as is already demonstrated in pediatric ALL, 36,37 should be evaluated. 6. The use of a CD45 gating strategy needs to be studied. A significant step forward has been the introduction of the simultaneous detection of more fluorochromes and the use of CD45 gating; because CD45 is differentially expressed not only between different lineages of hematopoietic differentiation, but also within these lineages during the process of maturation, it is capable in combination with side scatter 57 of clearly differentiating blasts from other cells. Future clinical trials have to Am J Clin Pathol 2009;131:16-26 23 23 DOI: 10.1309/AJCP5TSD3DZXFLCX 23

Al-Mawali et al / Minimal Re s i d u a l Disease in AML validate the translation of these data into improved prognostic parameters. 7. The value of studying MRD in CD34 AML requires further consideration because approximately 20% of AML samples are completely negative for CD34 or CD117. More effort is needed to identify stem cells in CD34 AML. 8. One of the main causes of false-positive MRD results by flow cytometry is the use of inappropriate markers to distinguish leukemic cells from normal cells. The range of normality needs to be established by extensive studies of BM samples not only from healthy people but also from regenerating BM, ie, from patients at various stages of treatment or posttransplantation. Future Directions for MRD Detection Precise information on MRD may, in the future, provide the biologic basis for therapeutic decision making and, thus, allow specific treatment to be tailored to the needs of individual patients. Ultimately, the clinical usefulness of MRD monitoring in the routine management of patients will need to be confirmed in prospective trials involving large numbers of patients, uniformly treated and monitored within well-defined protocols. Quantitation of MRD, especially with sensitive PCR and MFC assays, has not only enabled the study of the kinetic behavior of leukemic cells during chemotherapy and remission but has also increased our understanding of the biology of AML and leukemia in general and opened up new areas of scientific interest that will need further inquiry. In AML, novel treatments are needed if one hopes to improve cure rates substantially. Thus, the highest value of MRD assays may possibly lie in the rapid measurement of the effect of novel therapies on the leukemic clone. Summary The ideal assay system for the detection of small numbers of leukemic cells that persist after chemotherapy in BM or blood samples should fulfill the following criteria: (1) The method should be applicable in most cases of the disease under investigation. (2) The method should be specific for the neoplastic cell. (3) The method should be sensitive. (4) The method should allow quantitation of tumor burden for prognostic purposes. 58 Based on the applicability and sensitivity of the MFC immunophenotyping approach, it could be concluded that this is a well-suited approach for the specific detection of minimal numbers of leukemic cells and, hence, could help obtain a more precise and early evaluation of the effectiveness of new treatment strategies and better assess the CR status in patients with AML. 59-62 Our experience in the use of MFC for detection of MRD in AML has shown that the method is feasible, rapid, and sensitive. 41 To ensure the reliability of MRD testing, the procedures for cell collection, separation, staining, and analysis must be followed carefully. Most important, the sequence in which antibodies are added to cells and the times of incubation must be rigorously standardized because variations in these procedures can alter the intensity of cell labeling. 15 Moreover, the reproducibility of MRD analysis can be affected by changes between batches of antibodies, instrument instability, and variations in fluorochromes overlapping into different channels. These variations must be monitored by frequent staining of normal samples and by regular testing of the instrument settings and the need for correct fluorochrome compensation. In addition to the aforementioned technical limitations specific to flow cytometric detection of MRD, other factors may affect any method of MRD detection. For example, the anatomic distribution of leukemic cells during clinical remission may be uneven, 63,64 leading to sampling variability and false-negative results. Based on published data, it is expected that the quantification of MRD will significantly improve the evaluation of prognosis in patients with AML and, ultimately, have a major role as a stratification parameter to guide the riskadapted therapy of the disease. Therefore, it seems reasonable to intensify therapy for patients who have a slow early response to treatment and have detectable MRD during clinical remission. 15 Alternatively, the excellent clinical outcome of MRD-negative cases raises the possibility of using MRD assays to identify candidates for considering reduction in treatment intensity. Moreover, MRD monitoring during follow-up after therapy is useful, providing an adequate observation in patients with a tendency toward MRD recurrence. In addition, MFC presents applicability with a clinically relevant sensitivity to almost all patients with AML and, consequently, is considered a perfect tool for the use of MRD quantification in large AML populations including all subtypes of the disease. Technical improvements, like the use of 5 or more colors in MFC and the identification of additional LAPs and leukemia-specific genetic targets that can be quantified by PCR will lead to a further increase in the validity and sensitivity of both methods for MRD quantification. Future studies will define which methods at which time points are most useful for the management of patients with AML. We still do not know whether early detection of relapse and subsequent changes in therapeutic strategies will improve cure rates, but there is reason to believe that this might be the case. First, it is well established that the residual leukemic cell burden and the curability of cancer are related. Second, the likelihood of the emergence of drug-resistant leukemic cells by mutation increases as the number of cell divisions 24 Am J Clin Pathol 2009;131:16-26 24 DOI: 10.1309/AJCP5TSD3DZXFLCX

Hematopathology / Review Article increases and, hence, relates to the total leukemic cell burden. 65 Thus, timely detection of MRD would identify patients who need more intensive therapy to remain in remission. However, known prognostic factors are not 100% predictive and MRD studies might well complement and enhance their informative value. From the Institute of Medical and Veterinary Science, Divisions of 1 Haematology and 2 Human Immunology, Hanson Institute; and 3 University of Adelaide, Adelaide, Australia. Address reprint requests to Dr Al-Mawali: Institute of Medical and Veterinary Science (IMVS), Divisions of Haematology and Human Immunology, Frome Rd, Adelaide, SA, PO Box 14 Rundle Mall, Adelaide SA 5000, Australia. References 1. Lilleyman J, Hann I, Blanchette V. Pediatric Hematology. 2nd ed. London, England: Churchill Livingstone; 1999:369-385. 2. Golub T, Weistein H, Grier H. Acute myeloid leukemia. In: Pizzo P, Poplack D, eds. Principles and Practice of Pediatric Oncology. 3rd ed. Philadelphia, PA: Lippincott- Raven; 1997:463-482. 3. Kern W, Schoch C, Haferlach T, et al. Monitoring of minimal residual disease in acute myeloid leukemia. Crit Rev Oncol Hematol. 2005;56:283-309. 4. Baer MR. Assessment of minimal residual disease in patients with acute leukemia. Curr Opin Oncol. 1998;10:17-22. 5. Löwenberg B, Griffin JD, Tallman MS. Acute myeloid leukemia and acute promyelocytic leukemia. Hematology Am Soc Hematol Educ Program. 2003:82-101. 6. Campana D. Determination of minimal residual disease in leukaemia patients. Br J Haematol. 2003;121:823-838. 7. Macedo A, Orfao A, Vidriales MB, et al. Characterization of aberrant phenotypes in acute myeloblastic leukemia. Ann Hematol. 1995;70:189-194. 8. Macedo A, Orfao A, Gonzalez M, et al. Immunological detection of blast cell subpopulations in acute myeloblastic leukemia at diagnosis: implications for minimal residual disease studies. Leukemia. 1995;9:993-998. 9. Reading CL, Estey EH, Huh YO, et al. Expression of unusual immunophenotype combinations in acute myelogenous leukemia. Blood. 1993;81:3083-3090. 10. Kern W, Danhauser-Riedl S, Ratei R, et al. 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;88:646-653. 11. Al-Mawali A, Gillis D, Hissaria P, et al. Incidence, sensitivity, and specificity of leukemia-associated phenotypes in acute myeloid leukemia using specific five-color multiparameter flow cytometry. Am J Clin Pathol. 2008;129:934-945. 12. Campana D, Pui CH. Detection of minimal residual disease in acute leukemia: methodologic advances and clinical significance. Blood. 1995;85:1416-1434. 13. San Miguel JF, Martinez A, Macedo A, et al. Immunophenotyping investigation of minimal residual disease is a useful approach for predicting relapse in acute myeloid leukemia patients. Blood. 1997;90:2465-2470. 14. Al-Mawali A, Gillis D, Hissaria P, et al. The presence of leukaemia-associated phenotypes is an independent predictor of induction failure in acute myeloid leukaemia [published online ahead of print November 22, 2007]. Int J Lab Hematol. doi:10.1111/j.1751-553x.2007.01003.x. 15. Campana D, Coustan-Smith E. Detection of minimal residual disease in acute leukemia by flow cytometry. Cytometry. 1999;38:139-152. 16. Adriaansen HJ, van Dongen JJ, Kappers-Klunne MC, et al. Terminal deoxynucleotidyl transferase positive subpopulations occur in the majority of ANLL: implications for the detection of minimal disease. Leukemia. 1990;4:404-410. 17. San Miguel JF, Vidriales MB, Lopez-Berges C, et al. 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;98:1746-1751. 18. Venditti A, Buccisano F, Del Poeta G, et al. Level of minimal residual disease after consolidation therapy predicts outcome in acute myeloid leukemia. Blood. 2000;96:3948-3952. 19. Venditti A, Tamburini A, Buccisano F, et al. Clinical relevance of minimal residual disease detection in adult acute myeloid leukemia. J Hematother Stem Cell Res. 2002;11:349-357. 20. Bahia DM, Yamamoto M, Chauffaille Mde L, et al. Aberrant phenotypes in acute myeloid leukemia: a high frequency and its clinical significance. Haematologica. 2001;86:801-806. 21. Drach J, Drach D, Glassl H, et al. Flow cytometric determination of atypical antigen expression in acute leukemia for the study of minimal residual disease. Cytometry. 1992;13:893-901. 22. Feller N, van der Pol MA, van Stijn A, et al. MRD parameters using immunophenotypic detection methods are highly reliable in predicting survival in acute myeloid leukaemia. Leukemia. 2004;18:1380-1390. 23. Voskova D, Valekova L, Fedorova J, et al. Leukemic cells and aberrant phenotypes in acute leukemia patients: a flow cytometry analysis. Neoplasma. 2003;50:422-427. 24. Terstappen LW, Safford M, Konemann S, et al. Flow cytometric characterization of acute myeloid leukemia, part II: phenotypic heterogeneity at diagnosis. Leukemia. 1992;6:70-80. 25. Terstappen LW, Loken MR. Myeloid cell differentiation in normal bone marrow and acute myeloid leukemia assessed by multi-dimensional flow cytometry. Anal Cell Pathol. 1990;2:229-240. 26. Terstappen LW, Safford M, Loken MR. Flow cytometric analysis of human bone marrow; III: neutrophil maturation. Leukemia. 1990;4:657-663. 27. Terstappen LW, Mickaels RA, Dost R, et al. Increased light scattering resolution facilitates multidimensional flow cytometric analysis. Cytometry. 1990;11:506-512. 28. Kern W, Schnittger S. Monitoring of acute myeloid leukemia by flow cytometry. Curr Oncol Rep. 2003;5:405-412. 29. Kern W, Voskova D, Schoch C, et al. 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;104:3078-3085. 30. Yin JA, Tobal K. Detection of minimal residual disease in acute myeloid leukaemia: methodologies, clinical and biological significance. Br J Haematol. 1999;106:578-590. 31. Kern W, Haferlach C, Haferlach T, et al. Monitoring of minimal residual disease in acute myeloid leukemia. Cancer. 2008;112:4-16. Am J Clin Pathol 2009;131:16-26 25 25 DOI: 10.1309/AJCP5TSD3DZXFLCX 25