Multiparametric analysis of normal and postchemotherapy bone marrow: Implication for the detection of leukemia-associated immunophenotypes

Authors


  • How to cite this article: Olaru D, Campos L, Flandrin P, Nadal N, Duval A, Chautard S, Guyotat D. Multiparametric analysis of normal and postchemotherapy bone marrow: Implication for the detection of leukemia-associated immunophenotypes. Cytometry Part B 2008; 74B: 17–24.

Abstract

Background:

The knowledge of normal marrow is mandatory to assess the malignant counterpart of normal cells and define leukemia-associated immunophenotypes (LAIPs). In this study, the expression of a variety of antigens expressed in normal and postchemotherapy bone marrow (BM) was analyzed to provide a frame of reference for the identification of myeloid LAIPs.

Methods:

Multiparameter four- and six-color flow cytometry was used to define antigen combinations totally absent or present at very minimal levels in marrow cells of normal individuals (n = 20) and patients receiving chemotherapy for acute lymphoblastic leukemia (n = 20). Immature (blast) cells were gated according to CD45/SSC properties. Fifty-three acute myeloid leukemia (AML) samples were studied in six-color combinations.

Results:

In six-color flow cytometry, 47 phenotypes were totally absent from blast gate in all normal samples. Forty-one other phenotypes were identified in less than 0.05% of blast cells. There was no difference between normal and postchemotherapy BMs. The four-color panel allowed to identify only 30 phenotypes present at a frequency <0.05%. Using the six-color panel, 58% of the absent or infrequent phenotypes in normal BM were found in at least one of 53 AML samples. All AML cases exhibited at least one LAIP.

Conclusion:

Our results show that the ability to distinguish leukemic from healthy cells is considerably increased by a six-color approach. Furthermore, these absent or infrequent phenotypes in normal BM are identified in AML and can be utilized for minimal residual disease study. © 2007 Clinical Cytometry Society

Flow cytometry is a useful tool to define maturation phenotypes of bone marrow (BM) cells, thanks to the availability of a large panel of monoclonal antibodies (moAb) to differentiate antigens (1–3). By using multicolor analysis, it is possible to combine morphological parameters and multiple-antigen staining to define stage-specific maturation profiles in the myeloid and lymphoid lineages (4). Until now, routine cytometers simultaneously analyzed up to four colors. The novel generation of cytometers allows the use of up to six or even nine colors on routine analysis and the identification of multiple cell subpopulations with distinct antigen profiles.

Knowledge of normal marrow cell phenotypes is mandatory to assess their malignant counterpart, in view of the diagnosis of leukemias or follow-up of minimal residual disease (MRD). Until now, no leukemia-specific antigen suitable for flow-cytometry detection has been described. In the absence of a specific marker, the distinction between leukemic blasts and normal immature cells relies on the study of antigen combinations that allow to describe leukemia-associated immunophenotypes (LAIPs) absent from or extremely infrequent in healthy BM samples. Phenotypic abnormalities in acute myeloid leukemia (AML) include expression of markers normally not expressed on myeloid cells, coexpression of markers normally expressed at different stages of maturation as well as overexpression and underexpression of myeloid markers (5). These unusual phenotypes are detected on leukemic cells at diagnosis and can be used to monitor MRD after induction or consolidation therapy. For this purpose, these LAIPs should not be present as a “normal” phenotype in marrows recovering from induction or consolidation chemotherapy.

To our knowledge, the effects of intensive treatments with chemotherapeutic agents on myeloid differentiation have not been extensively evaluated. Data are available regarding immature nonmalignant lymphoid subsets in children undergoing myelosuppressive treatments (hematogones) (6, 7), but no study specifically addressed the issue of unusual nonmalignant myeloid phenotypes in BM after the induction treatment of acute leukemia.

The purpose of our study was to define by flow-cytometry antigen combinations that are totally absent or present at very low levels in normal BM and to assess their expression on AML cells. The second aim was to verify whether these combinations were absent from BM after intensive chemotherapy. We used a multiparametric (six-color) approach to improve the identification of such combinations and compared it with a four-color examination.

MATERIAL AND METHODS

Bone Marrow Samples

Marrow samples were collected by aspiration into heparin–lithium tubes. Twenty normal BM samples were analyzed with six-color (n = 20) and four-color (n = 10) combinations of moAbs. These samples were obtained, after informed consent, from healthy donors for BM transplantation (n = 10) and from patients with non-Hodgkin's and Hodgkin's lymphoma (n = 10) after ruling out BM involvement by histological, immunohistological, and flow-cytometry analyses.

To assess the postchemotherapy myeloid hematopoiesis, 20 BM samples were collected from 15 patients with acute lymphoblastic leukemia (ALL) in complete morphological remission after induction (12 samples) and/or consolidation chemotherapy (eight samples). This group included nine adult patients (median age 47.8 years), treated according to the GRAALL2003 protocol, and six children (median age 9.5 years), receiving the FRALLE 2000 protocol. Both treatments are similarly designed and use intensive induction or consolidation courses resulting in prolonged neutropenia (8). All samples were evaluated in six-color cytometry and 10 also in four-color.

We also analyzed with the six-color protocol 53 cases of AML diagnosed consecutively in adults at Hematology Laboratory of CHU Saint-Etienne France between November 2004 and May 2006. The median age of AML patients was 65.5 years (range 23–90 years). According to the French–American–British (FAB) Cooperative Group criteria, these cases were classified as AML M0 (n = 3), AML M1 (n = 18), AML M2 (n = 19), AML M3 (n = 2), AML M4 (n = 9), AML M5 (n = 1), and unclassified (n = 1). According to the WHO classification, 14 cases presented multilineage dysplasia and one was therapy related.

Flow Cytometry

The study was performed on erythrocyte-lysed whole BM samples after staining with directly conjugated moAbs.

The antibodies were chosen to maximize our ability to identify antigenic abnormalities within the myeloid blasts as they were described in the literature in three- or four-color combinations (5, 9, 10). The antigens with weak expression (such as CD13, CD33, and CD117) were detected using moAbs conjugated to bright fluorochromes, while strongly expressed antigens (HLADR, CD45) were analyzed with dimer conjugates. An anti-CD45 antibody was included in each tube for gating purposes (11–13).

The moAb panels used for surface antigen staining in six- and four-color combinations are presented in Tables 1 and 2, respectively.

Table 1. MoAb Panel Used for Surface Antigen Staining in Six- and Four-Color Combinations
Tube numberFITCPEPerCP-Cy5.5PE-Cy7APCAPC-Cy7
Six-color
1CD15CD117CD45CD34CD33HLADR
2CD7CD13CD45CD34CD33CD19
3CD65CD56CD45CD34CD4CD16
4CD14CD64CD45CD4CD11BHLADR
5CD38CD56CD45CD34CD33CD19
6CD61GLYCO ACD45CD33CD36HLADR
Four-color
1CD15CD117CD45CD34  
2CD15CD33CD45CD34  
3CD15HLADRCD45CD34  
4HLADRCD33CD45CD34  
5CD7CD33CD45CD19  
6CD13CD33CD45CD34  
7HLADRCD13CD45CD34  
8CD56CD33CD45CD34  
9CD65CD14CD45CD34  
10CD11bCD117CD45CD34  
11CD65CD11bCD45CD4  
12CD71CD11bCD45CD34  
13CD38CD19CD45CD34  
14CD14CD33CD45CD2  
Table 2. Six-Color Combinations for Blast Cell Populations (Defined on CD45/SSC Histogram) That are Totally Absent or Present at Very Low Frequency in Normal and Postchemotherapy Bone Marrow
 Normal BMPost-CT BM median (range) Normal BM median (range)Post-CT BM median (range)
  1. Data from 20 samples expressed as percentage of CD45+ cells; Post-CT BM = post chemotherapy bone marrow.

34+ DR+ 117+ 33− 15+<0.01<0.01 (0–0.02)34+ DR+ 117− 33+ 15+0.01 (0–0.03)<0.01 (0–0.02)
34+ DR− 117+ 33+ 15+<0.01<0.0134+ DR+ 117− 33− 15+<0.01 (0–0.02)<0.01 (0–0.03)
34+ DR− 117+ 33− 15+<0.01<0.01 (0–0.01)34+ DR− 117+ 33+ 15−0.02 (0–0.04)0.01 (0–0.03)
34+ DR− 117− 33+ 15+<0.01<0.0134+ DR− 117+ 33− 15−0.01 (0–0.05)<0.01 (0–0.05)
34+ DR− 117− 33− 15+<0.01<0.01 (0–0.01)34+ DR− 117− 33+ 15−<0.01 (0–0.02)<0.01 (0–0.02)
34+ 33+ 13+ 19+ 7+<0.01<0.0134+ DR− 117− 33− 15−0.02 (0–0.05)0.01 (0–0.05)
34+ 33+ 13− 19+ 7+<0.01<0.0134− DR− 117+ 33+ 15−0.01 (0–0.02)0.02 (0–0.05)
34+ 33+ 13− 19+ 7−<0.01<0.01 (0–0.01)34+ 33+ 13+ 19+ 7−<0.01 (0–0.04)<0.01 (0–0.03)
34+ 33− 13+ 19+ 7+<0.01<0.0134+ 33+ 13+ 19− 7+0.01 (0–0.05)0.01 (0–0.05)
34+ 33− 13− 19+ 7+<0.01<0.0134+ 33+ 13− 19− 7+<0.01 (0–0.02)0.01 (0–0.02)
34− 33+ 13− 19+ 7+<0.01<0.0134+ 33+ 13− 19− 7−0.02 (0–0.05)0.04 (0–0.07)
34+ 65+ 56+ 4+ 16+<0.01<0.0134+ 33− 13+ 19+ 7−<0.01 (0–0.01)<0.01 (0–0.01)
34+ 65+ 56+ 4+ 16−<0.01<0.0134+ 33− 13+ 19− 7+0.01 (0–0.02)<0.01 (0–0.02)
34+ 65+ 56+ 4− 16+<0.01<0.0134+ 33− 13− 19− 7+0.02 (0–0.03)<0.01 (0–0.03)
34+ 65+ 56+ 4− 16−<0.01<0.01 (0–0.02)34− 33+ 13+ 19+ 7+<0.01 (0–0.02)<0.01 (0–0.01)
34+ 65− 56+ 4+ 16+<0.01<0.0134− 33+ 13− 19+ 7−0.02 (0–0.05)0.01 (0–0.04)
34+ 65− 56+ 4+ 16−<0.01<0.0134− 33− 13+ 19+ 7+<0.01 (0–0.02)<0.01 (0–0.02)
34+ 65− 56+ 4− 16+<0.01<0.0134− 33− 13− 19+ 7+0.01 (0–0.04)<0.01 (0–0.02)
34+ 38+ 33+ 56+ 19+<0.01<0.0114− DR+ 4+ 11B+ 64−<0.01 (0–0.01)<0.01 (0–0.01)
34+ 38+ 33+ 56+ 19−<0.01<0.01 (0–0.02)34+ 65+ 56− 4+ 16+<0.01 (0–0.01)<0.01
34+ 38+ 33+ 56− 19+<0.01<0.0134+ 65+ 56− 4+ 16−<0.01 (0–0.01)<0.01
34+ 38+ 33− 56+ 19+<0.01<0.0134+ 65+ 56− 4− 16+<0.01 (0–0.01)<0.01
34+ 38+ 33− 56+ 19−<0.01<0.01 (0–0.03)34+ 65+ 56− 4− 16−0.03 (0–0.05)0.05 (0–0.08)
34+ 38− 33+ 56+ 19+<0.01<0.0134+ 65− 56+ 4− 16−<0.01 (0–0.02)0.01 (0–0.02)
34+ 38− 33+ 56+ 19−<0.01<0.0134+ 65− 56− 4+ 16+<0.01 (0–0.01)<0.01 (0–0.01)
34+ 38− 33+ 56− 19+<0.01<0.0134+ 65− 56− 4− 16+<0.01 (0–0.02)<0.01 (0–0.03)
34+ 38− 33− 56+ 19+<0.01<0.0134− 65+ 56+ 4− 16+0.01 (0–0.04)0.01 (0–0.04)
34+ 38− 33− 56+ 19−<0.01<0.0134− 65− 56+ 4+ 16+<0.01 (0–0.02)<0.01 (0–0.02)
34+ 38− 33− 56− 19+<0.01<0.0134− 65− 56+ 4+ 16−0.01 (0–0.03)<0.01 (0–0.02)
34− 38+ 33+ 56+ 19+<0.01<0.0134− 65− 56+ 4− 16+0.01 (0–0.05)0.02 (0–0.08)
34− 38+ 33− 56+ 19+<0.01<0.0133+ DR+ 61+ 36− GLYCO+<0.01 (0–0.02)<0.01 (0–0.02)
34− 38− 33+ 56+ 19+<0.01<0.0133+ DR+ 61+ 36− GLYCO−0.02 (0–0.05)0.02 (0–0.05)
34− 38− 33+ 56+ 19−<0.01<0.01 (0–0.01)33+ DR+ 61− 36− GLYCO+0.01 (0–0.03)0.01 (0–0.03)
34− 38− 33+ 56− 19+<0.01<0.0133+ DR− 61+ 36+ GLYCO+<0.01 (0–0.02)<0.01 (0–0.02)
34− 38− 33− 56+ 19+<0.01<0.0133+ DR− 61+ 36− GLYCO+<0.01 (0–0.01)<0.01 (0–0.01)
14+ DR+ 4+ 11B+ 64−<0.01<0.0133+ DR− 61− 36+ GLYCO−<0.01 (0–0.02)<0.01 (0–0.02)
14+ DR+ 4− 11B+ 64−<0.01<0.01 (0–0.03)33+ DR− 61− 36− GLYCO+<0.01 (0–0.02)<0.01 (0–0.02)
14+ DR+ 4− 11B− 64−<0.01<0.01 (0–0.02)33− DR+ 61+ 36− GLYCO+0.01 (0–0.05)0.01 (0–0.05)
14+ DR− 4+ 11B+ 64+<0.01<0.0133− DR− 61+ 36+ GLYCO+0.02 (0–0.05)0.02 (0–0.05)
14+ DR− 4+ 11B+ 64−<0.01<0.0133− DR− 61− 36+ GLYCO+0.01 (0–0.05)0.01 (0–0.05)
14+ DR− 4+ 11B− 64−<0.01<0.0133− DR− 61− 36− GLYCO+0.02 (0–0.05)0.02 (0–0.05)
14+ DR− 4− 11B+ 64−<0.01<0.01   
14+ DR− 4− 11B− 64−<0.01<0.01 (0–0.03)   
14− DR+ 4+ 11B+ 64+<0.01<0.01   
14− DR− 4+ 11B+ 64+<0.01<0.01   
14− DR− 4+ 11B+ 64−<0.01<0.01   
○33+ DR− 61− 36+ GLYCO+<0.01<0.01   

In all cases, an isotype-matched negative control with no BM reactivity was used. All antibodies were purchased from Becton Dickinson Bioscience (San Jose, USA) except for CD15, CD7, CD65, CD14, CD38, CD61, CD117, CD13, CD56, Glycophorine A, CD10, CD22, CD1a, and CD79a (Beckman Coulter, Miami, USA) and for CD2, IgM, MPO, CD64, and CD3-APC (DakoCytomation, Glostrup, Denmark).

BM samples were incubated for 30 min in AB human serum and fetal calf serum (FCS). Next, the respective combinations of antibodies in saturating concentrations were added to 500,000 cells (volume 50 μl) and incubated 15 min at room temperature in the dark. After an addition of 2 ml of FACS lysing solution (Becton Dickinson, USA), the cells were incubated for additional 5 min, then centrifuged, and washed once in 2 ml of PBS. Finally, they were resuspended in 0.5 ml of PBS. Data acquisition was performed with a FACS Canto flow cytometer (Becton Dickinson) equipped with two lasers, and an analysis was performed using FACSDiva Software (Becton Dickinson). At least 100,000 CD45-positive events were recorded (to study at least 10,000 CD34+ cells). Compensation setting was performed for each protocol according to the manufacturer's instructions, using the following tubes: unlabeled, single labeled, and isotype control for each antibody, and mixed antibodies. The unlabeled tube was used to set up baseline PMT gains. A positive population was then defined for each single-labeled tube. Finally, a compensation matrix was automatically calculated by the FACSDiva Software. Detection and compensation stability were tested daily using SetUp beads (BD Biosciences). The compensation matrix was checked in case of change of antibody lot.

Gating Strategy

The intensity of CD45 expression in combination with side scatter (SSC) properties was used to gate blast cell population and separate it from the other BM populations. Debris and fat cells were first eliminated by setting a threshold according to SSC/FSC properties. The blast population was then identified by a low expression of CD45 and low SSC properties (11, 12). The gates for positive and negative cells established for each marker (see above) were checked on corresponding fluorescence/SSC dot plots after a detailed analysis of all two-dimensional biexponential dot plots, which provided a good visualization of different cell subsets. A population was considered positive when 20% or more cells were stained. The mean fluorescence intensity (MFI) and MFI ratio (antibody MFI/isotypic control MFI) were calculated for positive cells.

Next, all logical combinations of these gates were realized: for six-color moAb panel, there were 32 combinations possible on each tube (so a total of 192 combinations of surface antigens) and for four-color panel, only eight combinations (a total of 112 combinations).

The frequencies of cells carrying the phenotype identified by each combination were determined within the normal BM and postchemotherapy BM samples. The mean percentages of positive cells in each phenotype were compared by the Student's t test.

RESULTS

Immunophenotype of Normal Bone Marrow Samples

The median percentage of blast population defined on CD45/SSC dot plot within the normal BM samples analyzed with six-color staining was 4.25% (range 3.8%–5.3%). The median percentage of CD34+ cells was 0.88% (range 0.3%–1.5%). No difference was found between these percentages in four-color or six-color staining.

From the 192 combinations possible with six-color immunostaining, 47 phenotypes were totally absent from blast gate in all normal BM and another 41 phenotypes were present at a frequency <0.05% of total cells. These phenotypes are listed in Table 3. The rest of 104 combinations was present at a frequency ranging between 0.05% and 3.58% of total cells.

Table 3. Four-Color Combinations for Blast Cell Populations (Defined on CD45/SSC Histogram) That are Totally Absent or Present at Very Low Frequency (<0.05% of CD45+ cells) in Normal and Postchemotherapy Bone Marrow
 Normal BMPost CT BM median (range) Normal BM median (range)Post CT BM median (range)
  1. Data from 10 samples expressed as % of CD45+ cells. Post CT BM = post-chemotherapy bone marrow.

34+ DR− 15+<0.01<0.01 (0–0.02)34+ 117+ 15+0.03 (0–0.04)0.03 (0–0.5)
34+ 38− 19+<0.01<0.01 (0.01)34+ 117− 15+0.02 (0–0.03)0.03 (0–0.05)
34+ 33− 56+<0.01<0.01 (0)34+ 33+ 15+0.04 (0.01–0.05)0.04 (0–0.06)
33− 14+ 2+<0.01<0.01 (0.02)34+ 33− 15+0.01 (0–0.02)<0.01 (0)
33− 14+ 2−<0.01<0.01 (0.05)34+ DR+ 15+0.04 (0–0.05)0.03 (0–0.05)
   34+ DR− 15–0.01 (0–0.04)0.03 (0–0.06)
   34+ 38− 19−0.04 (0–0.05)0.03 (0–0.05)
   34+ 33+ 56+0.01 (0–0.03)0.02 (0.05)
   34+ DR− 33+0.02 (0–0.04)0.02 (0–0.03)
   34+ DR− 33–0.01 (0–0.02)0.01 (0–0.02)
   33+ 7+ 19+0.02 (0–0.03)0.03 (0–0.04)
   33– 7+ 19+<0.01 (0–0.02)0.01 (0–0.02)
   11B+ 65− 4+0.04 (0–0.05)0.04 (0.07)
   33+ 14+ 2+0.04 (0.01–0.05)0.03 (0–0.05)
   33− 14− 2+0.04 (0.02–0.05)0.04 (0–0.05)
   34+ DR− 13+0.04 (0–0.05)0.05 (0.02–0.06)
   34+ DR− 13−0.02 (0–0.05)0.02 (0–0.04)
   34+ 14+ 65+0.01 (0–0.05)0.02 (0–0.05)
   34+ 14+ 65−0.01 (0–0.05)0.02 (0–0.04)
   34+ 14− 65+0.04 (0–0.05)0.04 (0.01–0.06)
   34+ 117+ 11b+0.03 (0–0.05)0.05 (0.01–0.07)
   34+ 117− 11b+0.01 (0–0.03)0.02 (0–0.03)
   34+ 33− 13+<0.01 (0–0.01)0.01 (0–0.02)
   34+ 33− 13−0.03 (0–0.05)0.03 (0–0.05)
   34− 33− 13+0.04 (0.02–0.05)0.04 (0.01–0.05)

The four-color panel allowed to identify only 5 of 112 possible phenotypes that were totally absent from all normal BM and 25 phenotypes present at a low frequency (<0.05% of total cells; Table 4). The frequency of the remaining 82 combinations ranged between 0.05% and 2.85% of total cells.

Table 4. Frequencies of Leukemia-Associated Immunophenotypes (LAIPs) in AML Present on >1% of Total Cells
LAIP (<0.01% in normal BM)Nr of AML (%)Median (max)LAIP (<0.05% in normal BM)Nr of AML (%)Median (max)
  1. Data are from 53 cases of AML at diagnosis. Median and maximum values are expressed as percentage of CD45+ cells.

34+ 38+ 33+ 56+ 19−14 (26)1.8 (50)34+ 33+ 13+ 19− 7+15 (28)5.6 (43.2
34+ 65+ 56+ 4− 16−7 (13)4.8 (12)34+ 65+ 56− 4− 16−15 (28)1.7 (6.5)
34+ 38+ 33− 56+ 19−7 (13)3.6 (13)34+ 65− 56+ 4− 16−15 (28)9.5 (60)
34+ DR+ 117+ 33− 15+6 (11)45 (16)34+ DR− 117+ 33− 15−13 (25)5.7 (44.7)
34+ 38− 33+ 56+ 19−6 (11)12.6 (35)34− DR− 117+ 33+ 15−13 (25)4.3 (50.2)
14− DR+ 4+ 11B+ 64+4 (8)2.1 (3.1)34+ 33+ 13− 19− 7−11 (21)15.5 (71.4)
34+ 38+ 33+ 56− 19+4 (8)2.6 (3.1)33+ DR+ 61+ 36− GLYCO−10 (19)1.5 (2.3)
34+ 38− 33− 56+ 19−4 (8)2.6 (3)33+ DR+ 61− 36− GLYCO+8 (16)1.3 (28)
34+ DR− 117+ 33+ 15+3 (6)1.9 (4.1)34+ DR− 117+ 33+ 15−7 (13)2.1 (18.3)
14+ DR+ 4− 11B− 64−3 (6)2.7 (11.6)33+ DR− 61− 36− GLYCO+6 (11)1.4 (6.4)
34+ 65+ 56+ 4+ 16−3 (6)3.6 (4.2)34+ 33+ 13− 19− 7+4 (8)6.2 (44.8)
34+ 65− 56+ 4+ 16−3 (6)3.2 (5.4)34+ 33− 13+ 19− 7+4 (8)4.4 (47.6)
34− 38+ 33+ 56+ 19+3 (6)2.9 (11)34+ DR− 117− 33− 15−4 (8)1.6 (15.4)
34− 38− 33+ 56+ 19−3 (6)1.9 (1.8)33+ DR+ 61+ 36− GLYCO+4 (8)2.7 (8.3)
34+ DR− 117− 33+ 15+1 (2)1 (1)34+ 33+ 13+ 19+ 7−3 (8)1.9 (2.3)
34+ DR− 117− 33− 15+1 (2)1.3 (1.3)34− 33+ 13+ 19+ 7+3 (8)1.5 (1.9)
34− 33+ 13− 19+ 7+1 (2)2.7 (2.7)34+ DR+ 117− 33+ 15+3 (8)2 (3)
14+ DR− 4− 11B+ 64−1 (2)6.9 (6.9)34− 65− 56+ 4+ 16−3 (8)4.4 (5.9)
14− DR− 4+ 11B+ 64+1 (2)1.9 (1.9)33+ DR− 61+ 36− GLYCO+3 (8)5.2 (7.4)
34+ 38− 33− 56− 19+1 (2)1.8 (1.8)33− DR− 61− 36− GLYCO+3 (8)1.5 (2)
34− 38− 33+ 56− 19+1 (2)1.6 (1.6)34+ 33− 13+ 19+ 7−1 (2)17.6 (17.6)
   34+ 33− 13− 19− 7+1 (2)6.1 (6.1)
   34− 33+ 13− 19+ 7−1 (2)2.5 (2.5)
   14− DR+ 4+ 11B+ 64−1 (2)1.5 (1.5)
   34+ 65+ 56− 4+ 16+1 (2)2.4 (2.4)
   34+ 65+ 56− 4+ 16−1 (2)1.1 (1.1)
   33+ DR− 61− 36+ GLYCO−1 (2)2.2 (2.2)
   33− DR− 61+ 36+ GLYCO+1 (2)1.2 (1.2)

Immunophenotype of Postchemotherapy Bone Marrow Samples

The median percentage of blast cells was 4.5% (range 3.5%–5.6%) and the median percentage of CD34+ cells was 0.75% (0.3%–1.4%) of total cells. The level of expression of each individual marker on blast cells was not different from that on the normal and postchemotherapy BM as assessed by the comparison of MFI ratios.

When studying CD45/SSC-defined blast gate, no unusual subpopulations as defined earlier were identified after chemotherapy. All combinations shown in Table 3 were present at the same frequency as in normal BM except for 14 combinations where a nonsignificant increase of 0.01%–0.03% was observed. There was no difference between either adults and children or postinduction and postconsolidation samples.

In the four-color analysis, all combinations absent of expressed at a very low frequency in normal samples were similarly absent or within the same low range.

Immunophenotype of AML at Diagnosis

The six-color panel of antibodies was used in 53 AML at diagnosis. A phenotype was considered as leukemia associated when it was expressed on at least 1% of the total population. Fifty-one of eighty-eight (58%) of the phenotypes described as absent or infrequent in normal BM (<0.05% of total cells) were found in at least one AML, at a mean percentage of 8.7% of total cells. All AML cases displayed at least one LAIP, but frequently several blast subpopulations were observed in the same sample (data not shown). Table 4 shows the examples of the most frequent LAIPs found in AML.

Representative two-dimensional dot plots of normal marrow, postchemotherapy marrow, and AML marrow are shown in Figure 1.

Figure 1.

Example of 6-color staining of normal marrow (lines A and B), postchemotherapy marrow (line C), and AML marrow (line D) with antibodies to CD45, CD34, CD65, CD56, CD4, and CD16. Line A: CD34 gating on CD34/SSC properties (normal marrow) Lines BD: left column: CD45 gating; middle column: CD65/CD56 expression after CD34 and blast gating; right column: CD4/CD16 expression after CD34 and blast gating. The respective percentages (relative to CD45+ events) of aberrant cells in the AML sample are as follows: 34+ 65+ 56+ 4− 16−: 3.3%; 34+ 65+ 56− 4− 16−: 2.6%; 34+ 65− 56+ 4− 16−: 11.1%.

DISCUSSION

A number of studies have shown the interest of immunophenotyping for the monitoring of MRD in AML patients in remission and the predictive value for relapse of persistent leukemic cells (5, 10, 14–17). Initial studies of normal and leukemic phenotypes were performed in two- or three-color analyses on the mononuclear cell fraction (MNC) of BM (5, 9). Frequent LAIPs, which were totally absent from normal BM, were described as CD34+CD11b+, CD34+CD15+, CD34+CD38−CD13+, CD34+CD14+, CD33−CD14+, CD34+, CD56+. However, leukemic cells with aberrant phenotypes cannot be detected in all AML cases by these antigen combinations. Gating on SSC and CD45 expression allows a better identification of leukemic cells (10, 17) and increases by one log the detection of LAIPs in AML in remission when compared with FSC/SSC gating. Techniques using unseparated BM after erythrocyte lysis are also increasingly used. In addition to saving time, they allow to study the total marrow population. Although a recent study showed no difference between density gradient separation and whole blood lysis (18), others have reported some discrepancies between both methods, with unexpected lack of expression of CD34 after separation (19), which could hamper the detection of abnormal cells contributing to MRD and relapse. In our panel, CD34 was used in most combinations to identify the immature fraction of BM cells. Furthermore, the leukemic stem cells with self-renewal properties and capability to grow in immunodeficient mice strongly express CD34 and may constitute the most relevant part of MRD with regard to risk of relapse (20).

Our study of normal and postchemotherapy BM using a six-color staining allowed us to define new LAIPs present at diagnosis of AML and useful for the detection of MRD. With the same number of surface antigens studied and a more limited number of tubes (6 vs. 14), we could identify more unusual phenotype combinations utilizing six-color rather than four-color combinations. We described in all AML cases at least one subpopulation of cells carrying an antigen profile absent or extremely infrequent in normal BM. So far, most studies performed with a maximum of four-color combinations of moAbs allowed to identify LAIPs only in a subgroup of patients. Alternatively, other studies considered a phenotype as leukemia associated, although it was present in as high as 3% of normal marrow cells. This increased the number of AML exhibiting at least one LAIP, but with a loss of sensitivity improper for MRD assessment. Depending on the criteria used for their definition and the panel of moAbs employed, the reported incidence of LAIP in AML is extremely variable: 24.3% (21), 51.5% (22), 75% (5), and 100% (10, 15). Our approach combines a good sensitivity with a very stringent definition of LAIP. As reported in other studies, the abnormal phenotypes were present only on a part of leukemic cells, and several different phenotypes were often detected in the same sample. This can be at least in part related to the fact that the antigen expression is usually heterogeneous in AML. Thus, the use of more complex antibody combinations reduces the proportion of cells expressing each possible phenotype. It is also likely that a leukemic population comprises various subclones with different phenotypic abnormalities. Further studies are needed to assess the significance of this heterogeneity, particularly in view of the follow up of residual disease after treatment.

The comparison of normal and leukemic phenotypes is crucial for the detection of MRD. In most studies, BM from healthy donors serves as a reference for the definition of normal phenotypes. Recently, Buccisano et al. reported on the prognostic value of the kinetics of MRD reduction in AML. This study included control samples from normal donors and from patients affected by lymphoma without marrow involvement, in steady state and after chemotherapy. However, detailed results are not provided. We analyzed 20 postchemotherapy BM samples after an induction or consolidation chemotherapy of ALL to detect potential phenotype changes induced by the drugs. Although the drug combinations used in ALL differ from those used in AML, this approach allowed to verify that postchemotherapy BM did not display significant aberrant phenotypes that could interfere with the detection of MRD in AML. It is interesting to note that our results obtained on a total of 20 normal and 20 postchemotherapy marrows were very reproducible, and it is therefore very unlikely that aberrant phenotypes detected on leukemic cells, even on a quantitatively minor population, represent outliers without biological significance.

In conclusion, we show that the six-color cytometry constantly allows the identification of LAIPs that are not expressed in normal or postchemotherapy BM and can therefore be used to follow-up MRD in AML in remission. The practical relevance of this multicolor approach and its ability to detect patients with high relapse risk needs to be tested in the context of current clinical trials.

Acknowledgements

The authors are grateful to Salima Driss and Emmanuel Ferret for technical assistance and to all the staff of the Hematology Department of Institut de Cancérologie de la Loire, Saint-Etienne, for providing samples.

Ancillary