Novel flow-cytometric analysis based on BCD5+ subpopulations for the evaluation of minimal residual disease in chronic lymphocytic leukaemia


Karim Maloum, MD, PhD, Service d'Hématologie Biologique, Groupe Hospitalier Pitié-Salpêtrière, 47 Bd de l'Hôpital de la Pitié-Salpêtrière, 75013 Paris, France. E-mail:


Summary. We describe a new flow-cytometric analysis using quadruple labelling with anti-CD19, CD20, CD5, CD79b monoclonal antibodies and sequential gating. We determined a novel criteria defined by BCD5+CD79b–/low/total BCD5+ cells ratio (BCD5+R), and compared it with the previous definition of phenotypic remission, based on CD19+CD5+ coexpression, and with complementarity-determining region 3 polymerase chain reaction (CDR3 PCR) and clonotypic PCR (cPCR). A series of 54 peripheral blood samples from 21 chronic lymphocytic leukaemia (CLL) patients in complete haematological remission and a series of 16 from normal volunteers were analysed. In normal controls, the BCD5+R was always < 0·2. The sensitivity of the BCD5+R was 1 × 10−4vs 5 × 10−2 for CDR3 PCR and 1 × 10−5 for cPCR. Among the 54 CLL samples, 35 had a BCD5+R < 0·2 and showed polyclonal CDR3 PCR, whereas the cPCR was positive in 12 out of 20 tested. In the remaining 19 samples, BCD5+R was > 0·2, CDR3 PCR was monoclonal in 16 out of 19 and cPCR positive in 14 out 14 tested, including one out of three samples with polyclonal CDR3 amplification. Even though cPCR remains the most sensitive method to evaluate MRD, this new, sensitive and specific flow cytometric parameter, the BCD5+R, is more suitable than CDR3 PCR for routine clinical MRD assessment in CLL.

Chronic lymphocytic leukaemia (CLL) is a haematological malignancy known to have variable prognosis. In contrast with acute leukaemia, in which aggressive therapy aimed at obtaining cure is mandatory, the goal of conventional therapy for CLL is to control the disease and rarely results in the complete eradication of detectable tumour cells (Clavio et al, 1998; Magnac et al, 1999). The application of novel therapies, such as haematopoietic stem cell transplantation (Khouri et al, 1994; Provan et al, 1996; Sutton et al, 1998) and monoclonal antibodies (Osterborg et al, 1997; Huhn et al, 2001) have resulted in a significant proportion of patients attaining much more profound responses. Minimal residual disease (MRD) study constitutes an important issue in monitoring these new therapeutic strategies in CLL. The most sensitive method used for the detection of MRD in CLL remains the clonotypic polymerase chain reaction (cPCR), which is able to detect a single CLL cell in > 105 normal cells (Owen et al, 1997; Magnac et al, 1999), but needs generation of primers that are specific to each individual patient. Unfortunately, cPCR is very expensive and very labour intensive. Development of simple, rapid and sensitive methods to assess MRD in CLL is necessary for monitoring and management of therapy. The recent advances in flow cytometry have enabled the use of these technologies for the detection of residual CLL cells in the blood or the bone marrow. The development of sequential gating strategies and the use of multicolour fluorescence enable the identification of very small populations of cells. CLL cells are characterized by the presence of CD5 and CD23 molecules and the absence or low expression of the CD20 molecule, membrane surface immunoglobulin (sIg) and CD79b molecule. Until now, patients were considered in phenotypic remission (PhR) when the percentage of CD19+CD5+/total CD19+ cells was < 25% in blood (Vuillier et al, 1992; Cabezudo et al, 1997). Recently, expansion of CD19+CD5+ cells in peripheral blood was described during haematopoietic reconstitution after stem cell transplantation (Bomberger et al, 1998). These findings enabled the determination of other phenotypic combinations, using the most specific markers to detect CLL cells in MRD assessment, particularly low expression of CD20 and CD79b molecules.

We report the results of MRD assessment obtained with a new flow-cytometric scoring system based on analysis of BCD5+ subpopulations, and compared them with those obtained by using conventional flow cytometry and molecular methods.

Patients and methods

Patients.  In total, 21 stage B-CLL patients (16 men, five women; median age, 58 years, range 35–80 years) in complete remission (CR) according to the National Cancer Institute (NCI) criteria (Cheson et al, 1996) were entered into the study. They had been treated by fludarabine with or without cyclophosmamide (n = 13), multiple chemotherapy regimens followed by autologous (n = 7), or allogeneic stem cell transplantation (n = 1).

Samples.  In total, 54 peripheral blood mononuclear cell (PBMC) samples obtained from 21 CLL patients and 16 PBMC samples obtained from normal volunteers were analysed. Furthermore, three patients with B-lymphoproliferative disorders other than B-CLL in CR after allogeneic BMT (n = 2) or after chemotherapy alone (n = 1) were included in the control group. PBMC were isolated by Ficoll-Hypaque density gradient purification and then were cryopreserved or immediately used for immunophenotyping and DNA extraction to study IgH gene rearrangements. Informed consent was obtained from all patients.

Sensitivity of the flow-cytometric method compared with IgH PCR was evaluated by analysing diluted leukaemic cells into normal PBMC. CLL cells were mixed with normal PBMC to obtain seven serial 10-fold dilutions from 1:0 to 1:106.

Flow cytometry.  Between 5 × 105 and 1 × 106 cells were incubated with the following antibody combinations: (1) anti-CD3 fluorescein isothiocyanate (FITC)/anti-CD16/CD56 phycoerythrin (PE), anti-CD4 peridinin chlorophyll (PerCP), anti-CD8 allophycocyanin (APC) (Becton Dickinson, Le-Pont-de-Claix, France); (2) anti-kappaFITC/anti-lambdaPE (DAKO, Trappes, France), anti-CD19APC (Becton Dickinson); (3) anti-CD10FITC (Coulter, Villepinte, France), anti-CD23PE (Becton Dickinson), anti-CD5Cy-Chrome (Pharmingen, Le-Pont-de-Claix, France) and anti-CD19APC (Becton Dickinson); (4) anti-FMC7FITC (Immunotech, Villepinte, France), and anti-CD19PC5 (Immunotech); (5) anti-CD20FITC (Coulter), anti CD79bPE (Immunotech), anti-CD5Cy-Chrome (Pharmingen) and anti-CD19APC (Becton Dickinson); and (6) irrelevant Ig coupled to FITC, and PE, PE-PC5 and APC. After incubation with the antibodies, cells were acquired using a Becton Dickinson FACScalibur device (San Jose, CA, USA) and analysed with cellquestTM v.3.3 and paint-a-gatepro/TM v.3.0 software. A minimum of 5000 B lymphocytes was analysed in each test. In all cases, B cells were identified using a sequential gating strategy, setting a region on CD19 versus side scatter (SSC). Analyses of the whole B-cell subpopulations were performed using the multilabelling of gated CD19 cells.

IgH PCR analysis.  DNA was extracted from at least 5 × 106 cells. The complementarity-determining region 3 (CDR3) was amplified using a PCR with a 3′ JH consensus primer and a 5′ FR3 consensus primer. In the cases in which CDR3 amplification was found to be polyclonal (smear), the sequence of this region was performed. The specific probe of each clone corresponding to the NDN region was then synthesized, and was used as the 3′ primer in a cPCR in which the 5′ primer was chosen to amplify the CDR3-FR1 segment in each case. Briefly, 40 cycles of amplification (90°C for 45 s, 62°C for 45 s and 72°C for 45 s) were performed in the cPCR. This technique can detect at least one leukaemic cell among 105 normal cells (Magnac et al, 1999). This high sensitivity was confirmed in the present work. In all DNA samples tested for cPCR, CDR3 PCR was carried out and, in all cases, showed polyclonal smears that were considered as controls for amplification.


Flow cytometry

The presence of the CD5 molecule associated with a low expression of CD20 and CD79b was required to characterize the malignant B cells. In a first step, we determined the phenotypic profile of normal B cells by analysing PBMC obtained from 16 normal donors. Results showed that CD19+ cells constituted 5–22% of total peripheral lymphocytes (median, 10%). The CD19+CD5+ cells represented 6–47% of total B cells (median, 27%) and the percentage of CD19+CD5+CD79b–/low comprised 0–5% of total B cells. In addition, analysis of PBMC from patients with lymphoproliferative disorders other than B-CLL in CR after either BMT or chemotherapy, showed that a high proportion of circulating BCD19+ cells expressed the CD5 molecule (35–70%) and more than 95% expressed the CD79b molecule. These results, as well as previously published reports of B cell reconstitution after autologous and allogeneic stem cell transplantation showing polyclonal CD19+CD5+ cell expansion (Bomberger et al, 1998; Esteve et al, 2001), enabled the early definition of PhR in CLL to be reconsidered, based only on the coexpression of CD19 and CD5 molecules.

We determined a new flow-cytometric parameter defined as the ratio of CD19+CD5+CD79b–/low cells/Total CD19+CD5+ cells (BCD5+R). In our controls, this ratio was always < 0·20, and this value was considered as the cut-off (Table I).

Table I.  Flow-cytometric analysis of normal B cells in peripheral blood.
Normal donors (n = 16)CD19+ cells*CD19+CD5+ cells**CD19+CD5+CD79/low cells**BCD5+R
  • *

    Percentage of positive cells among total lymphocytes (sideways scatter-forward scatter, SCC-FSC).

  • **Percentage of positive cells among CD19+ cells.

Median (range)10% (5–23)27% (6–47)2% (0–5)0·06 (0–0·16)

Analysis of the 54 CLL samples showed that 35 were in PhR with a BCD5+ < 0·2. The remaining 19 samples did not achieve PhR as the BCD5+R was > 0·2. As shown in Table II, in the group of samples with BCD5+R < 0·2, CD19+ cells varied from 5% to 37% of total lymphocytes of which 16–86% coexpressed the CD5 molecule, but only 0–5% were CD5+ and CD79b–/low. In the group of samples with BCD5+R > 0·2, CD19+ cells represented a median of 11% (1–37%) of total lymphocytes, of which 35–86% coexpressed the CD5 molecule and 10–75% were CD5+/CD79b–/low. Figure 1 illustrates one representative case showing persistent residual B-CLL cells with a BCD5+R > 0·2, as analysed by cellquest software.

Table II.  MRD assessment using flow cytometry.
PatientsCD19 cells*CD19+CD5+cells**CD19+CD5+CD79/low cells**
Median (range)Median (range)Median (range)
  • *

    Percentage of positive cells among total lymphocytes (SCC-FSC).

  • **Percentage of positive cells among CD19+ cells.

BCD5+R < 0·2 (n = 35)13% (5–37)43% (16–86) 1% (0–9)
BCD5+R > 0·2 (n = 19)11% (1–37)56% (35–86)24% (10–75)
Figure 1.

cellquestTM software analysis of chronic lymphocytic leukemia (CLL) cells using a quadruple labelling with CD19APC, CD20FITC, CD79bPE and CD5Cy-Chrome. (A) Total lymphocytes were identified using combination of forward and side scatter. (B) Total B cells were identified using CD19 and side scatter. (C) CLL cells were separated from normal B cells according to their CD5bright, CD20dim and CD79bdim expression: C1, CD19+/CD20dim cells; C2, CD19+/CD79bdim cells; C3, CD5+/CD79bdim cells.

Altogether, these results demonstrated that the first definition of the PhR based only on CD19 and CD5 coexpression did not distinguish with certainty between real PhR with polyclonal expansion of CD19+CD5+ cells and the persistence of authentic CLL cells.

Sensitivity of BCD5+R compared with IgH PCR methods

The sensitivity of BCD5+R was assessed by analysing diluted leukaemic cells into normal PBMC as described in Patients and methods, and results were compared with those obtained by IgH-PCR assessment. Results showed that BCD5+R could detect one CLL cell among 104 normal cells, whereas CDR3 amplification could detect five CLL cells among 102 normal cells, and cPCR, one CLL cell among 105 normal cells. Figure 2 shows the flow-cytometric analysis of the different dilutions from 100% leukaemic cells to one leukaemic cell among 104 normal lymphocytes, in duplicate.

Figure 2.

paint-a-gatePRO/TM software analysis of diluted CLL cells into normal cells in duplicate (upper and lower panels): (A) 100% CLL cells (BCD5+R ≈1 in the two replicates), (B) 10% CLL cells (BCD5+R = 0·76 and 0·79), (C) 1% CLL cells (BCD5+R = 0·33 in the two replicates), (D) 0·1% CLL cells (BCD5+R ≈0·21 in the two replicates), (E) 0·01% CLL cells (BCD5+R ≈0·21 in the two replicates) and (F) normal peripheral blood mononuclear cells (PBMC) (BCD5+R = 0·14 and 0·15).

IgH PCR in CLL samples compared with conventional flow cytometry and the BCD5+R

Comparison between CDR3-PCR and immunophenotype.  In the 35 CLL samples showing PhR with a BCD5+R < 0·2, CDR3-PCR did not detect any residual leukaemic clones. Of note, if we had used the conventional immunophenotype, we would have observed PhR in only six samples among 35. In contrast, for the 19 samples with BCD5+R > 0·2, CDR3-PCR showed persistence of the clonal rearrangement in 16 out of 19, and a polyclonal pattern in the remaining three samples. All these results are included in Table III.

Table III.  Comparison between BCD5+R, CDR3 PCR and cPCR.
 BCD5+R < 0·2BCD5+R > 0·2
  1. *In one sample with BCD5+R > 0·2, CDR3 amplification was polyclonal, whereas cPCR was positive.

CLL samples (n = 54)3519
CDR3 PCR-positive 016
CDR3 PCR-negative35 3*
cPCR-positive12/20 tested14/14 tested
cPCR-negative 8/20 tested 0/14 tested

Comparison between cPCR and immunophenotype.  In the group of 35 samples in PhR as defined by our ratio (< 0·2), cPCR was performed in 20. Results showed that a specific clone was detected in 12 samples and no clone was found in eight of them.

In the group of the 19 samples with BCD5+R > 0·2, a specific clone was detected in all 14 tested, including the one sample tested for cPCR out of the three with polyclonal CDR3 amplification (Table III).


Minimal residual disease evaluation in CLL constitutes the primary end-point of several new therapeutic trials, such as immunotherapy with Campath 1H (Osterborg et al, 1997) or Rituximab (Huhn et al, 2001), chemotherapy regimens completed by immunotherapy or stem cell transplantation. The development of easy, rapid and very sensitive techniques is required to evaluate routinely these new treatment strategies. Several groups have demonstrated the high sensitivity of cPCR (Magnac et al, 1999; Noy et al, 2001) but this method remains time consuming, which is critical in the clinical routine. A recent report described a rapid flow-cytometric method using quadruple labelling with CD19, CD5, CD20 and CD79b monoclonal antibodies (Rawstron et al, 2001). The authors showed that this technique could detect one leukaemic cell among 104−105 normal cells. In the past decade, phenotypic remission in CLL has been based on the percentage of CD19+CD5+ lymphocytes. Recent publications concerning lymphoid reconstitution after BMT (Bomberger et al, 1998) and our results demonstrate the increased polyclonal BCD5+ subpopulation. These data lead us to use more specific markers to better define residual CLL cells. We herein report a new flow cytometric parameter for MRD assessment in CLL, based on the analysis of the different CD19+CD5+ cell subpopulations. This new cytometric parameter corresponds to the ratio of CD19+CD5+CD79b–/low/total CD19+CD5+ lymphocytes. This ratio is independent of circulating lymphocyte count and is also independent of CD20 expression, which can vary according to the cell status, i.e. fresh or cryopreserved. In our experience, low expression of CD20 molecule in CLL was constant when analysing fresh cells, as reported by Pedersen et al (2002). However, when using cryopreserved cells, CD20 expression varied from one CLL sample to another (data not shown). For this reason, we chose to study the CD79b molecule in combination with CD5 and CD19 to distinguish CD19+CD5+CD79b–/low CLL cells from CD19+CD5+CD79b+ normal cells.

We compared the BCD5+R with conventional flow cytometric analysis and molecular methods to evaluate MRD. Our results showed that samples in CR as assessed by CDR3 amplification would have been considered as PhR in only six out of 35 samples if we had used conventional flow cytometric methods, whereas BCD5+R showed PhR in all of these samples. CR was confirmed by cPCR in some of them. These results were in agreement with the recent data showing a polyclonal expansion of CD19+CD5+ lymphocytes during haematopoietic recovery after stem cell transplantation (Bomberger et al, 1998; Esteve et al, 2001). They were also in accordance with our data that was obtained with CD19+CD5+ cells in peripheral blood from normal controls, which can reach 47% of total CD19 cells.

Compared with molecular biology techniques, BCD5+R has a high sensitivity. Indeed it is more sensitive than CDR3 amplification when it was > 0·2, whereas CDR3 PCR did not detect any clones in at least one case where cPCR was positive. Dilution studies also confirmed that cPCR remains the most sensitive method, followed by BCD5+R, which detected at least one CLL cell among 104 normal cells, and then CDR3-PCR, which detected less than one CLL cell among 102 normal cells.

Altogether these results demonstrated that our new flow-cytometric technique is not only more sensitive but also more specific than conventional flow cytometry. Its high sensitivity and rapidity justify its use in clinical routine for MRD evaluation in CLL, and reduces the use of cPCR to patients with BCD5+R < 0·2, who are more likely to achieve CR.

Some authors have shown that the persistence of residual leukaemic cells detected by very high-sensitivity methods is indicative of progressive disease after immunotherapy and/or autograft (Rawstron et al, 2001; Esteve et al, 2002). It remains to be established in a larger series of patients, whether the presence of residual leukaemic cells in blood, using only BCD5+R and/or cPCR, can consistently predict relapse and, therefore, justify the intensification of therapy in positive cases, or whether it only enables the detection of leukaemic cell populations whose proliferative potential has been altered by treatment. The exact clinical significance of the reappearance of the malignant clone, detected by cPCR and/or BCD5+R after complete molecular remission, also needs to be determined.