Professor DanielCatovsky Academic Department of Haematology and Cytogenetics, The Royal Marsden Hospital, Fulham Road, London SW3 6JJ.
The clinical significance of detecting minimal residual disease (MRD) in B-lineage acute lymphoblastic leukaemia (ALL) was evaluated by quantitative flow cytometry using a combination of TdT with CD10 and CD19. 53 patients with B-cell precursor ALL were followed during and after completion of treatment (median follow-up 23 months). Nine patients relapsed and MRD had been detected in six of them, 5–15 weeks before relapse despite morphological complete remission. 43 patients remain in clinical remission and in none of these was MRD detected. Disease-free survival based on the detection of MRD by flow cytometry showed a statistically significant difference between both groups (P < 0.0001). The absence of MRD correlates with a low relapse rate, whereas the presence of MRD predicted early relapse. This study has shown that flow cytometry can improve the morphologic assessment of bone marrow (BM) remission status in B-lineage ALL. The finding of < 5% blasts in BM aspirates did not correlate with ‘true’ remission in a proportion of cases as residual leukaemic blasts were detected by flow cytometry in nine samples from six patients. On the other hand, the presence of > 5% blasts assessed by morphology was not necessarily a feature of relapse in five patients as these cells were shown to have a phenotype identical to normal TdT-negative B-cell precursors. Quantitative flow cytometry was more informative than conventional morphology to assess remission status and showed a strong correlation with clinical outcome. This methodology is useful to define MRD in the majority of patients with B-lineage ALL and should be tested in prospective clinical trials.
By conventional criteria, complete remission (CR) in acute leukaemia is defined by < 5% blasts in the bone marrow (BM). If these blasts are malignant, the patient would still harbour 1011–1012 leukaemic cells while in morphological CR. From this point until overt relapse, the leukaemic burden is unknown and patients are treated by standard protocols regardless of the degree of residual leukaemia.
The concept of minimal residual disease (MRD) was introduced to estimate more accurately the true number of leukaemic cells and, in turn, to improve clinical management and cure rates ( Campana & Pui, 1995). The potential benefit of detecting MRD during continuing chemotherapy is that this will predict subsequent relapse. This may therefore facilitate dose intensification and strategies such as bone marrow transplantation. The monitoring of remission status is done conventionally by morphology. The detection limit of this technique is 1–5%. Experiments with mixtures of leukaemic and normal mononuclear cells have demonstrated that a leukaemic contamination of < 1% cannot be detected unequivocally by morphology ( Janossy et al, 1988 ). Therefore more sensitive and objective techniques have been developed including PCR and FISH.
In precursor B-ALL the results using markers such as CD10 and TdT have been disappointing because both antigens are expressed in normal and leukaemic cells ( Janossy et al, 1979 ). However, this problem could be overcome by exploiting differences in relative antigen expression in leukaemic cells ( Lavabre-Bertrand et al, 1994a , b). We have shown quantitative differences in the expression of TdT, CD10 and CD19 between B-lineage ALL and normal bone marrow lymphoid precursors ( Farahat et al, 1995b ). Normal BM precursors express strong TdT and weak CD10 and CD19 whereas leukaemic B cells express weak TdT and strong CD10 and CD19 ( Farahat et al, 1995b ).
We have now followed these quantitative studies to examine MRD in leukaemic patients mostly after treatment-induced morphological CR. Samples with values similar to those of leukaemic B cells with two or three of the above markers (e.g. weak TdT, strong CD19 and CD10 expression) were considered positive for MRD, whereas samples with values similar to normal B-cell precursors were defined as immunologically free of disease.
MATERIALS AND METHODS
This study was carried out on 149 BM samples from 53 patients with B-lineage ALL (TdT+, CD19+) treated at the Royal Marsden Hospital during the period 1994–97. There were 45 cases of common-ALL (CD10+), five pre-B ALL (cyt μ+) and three early precursor B-ALL (CD10−). There were 15 adults and 38 children; 33 males and 20 females. Diagnosis was based on morphology, cytochemistry ( Bennett et al, 1981 ) and immunophenotyping ( Catovsky & Matutes, 1992). 32 patients were treated by combination chemotherapy according to the UKALL trial protocols; the remaining 21 patients had bone marrow transplantation, autologous in 12 and allogeneic in nine. The timing of the sample was decided by the treating clinicians and, as the adults and children were treated by different protocols and/or modalities, they were tested at variable times.
The morphological assessment of the BM samples was carried out independently by the diagnostic laboratory. All samples were tested by combinations of TdT with CD10 and CD19. Mononuclear cells were isolated by density gradient centrifugation with Lymphoprep (Nycomed SA, Oslo, Norway). Cells were washed twice with Hanks balanced salt solution before immunostaining.
Double-colour flow cytometry
The cells were first stained for membrane antigens using phycoerythrin derivative (RD1) antibodies MoAbs: J5 (CD10) and B4 (CD19) (Coulter Corporation, Miami, Fla.). 0.5–1 × 106 cells were used per tube. To each tube 50 μl of 2% AB serum were added followed by the appropriate volume of RD1-labelled MoAb against CD10 or CD19. RB1-labelled mouse immunoglobulin (Coulter Clone) was used as negative control. The tubes were vortexed and incubated for 10 min at room temperature. The cells were washed once with PBS azide and the supernatant discarded.
TdT nuclear staining
After membrane staining, the cells were fixed and permeabilized by incubating for 10 min at room temperature with 2 ml of a mixture of equal volumes of 4% paraformaldehyde in PBS and 1:10 dilution of Becton Dickinson's FACS lysing solution in distilled water (Becton Dickinson, San Jose, Calif.). The cells were centrifuged, the supernatant discarded and then washed with 2 ml of 0.5% Tween 20 in PBS in an immunofuge for 2 min. The cells were then incubated with 10 μl of fluorescein-labelled anti-TdT MoAb (Harlan Seralab, Loughborough) for 20 min at room temperature. FITC-labelled mouse immunoglobulin was used as negative control (Seralab). After two washes with 0.5% Tween 20 the cells were resuspended in 0.5 ml Isoton and analysed by flow cytometry.
Analysis by flow cytometry and quantification
All samples were acquired on a FACScan flow cytometer (Becton Dickinson) using LYSYS ll software. The fluorescence intensity was measured with FL1 and FL2 detectors and amplifiers set on a logarithmic scale. Each sample was acquired twice, once acquiring 5000 mononuclear cells and the other 5000 lymphocytes, of which 1000 or more were B lymphocytes depending on the percentage of these cells in the bone marrow. The latter cells were acquired by setting a live gate around the area of small to large lymphocytes in the forward scatter, side scatter dot plot. Data were analysed after setting a gate around the lymphoid area. The mean fluorescence intensity (MFI) of the positive cells was converted into antibody binding capacity (ABC) or the number of molecules per cell by using the Quantum Simply Cellular Microbeads Kit (Sigma, St Louis, Mo., U.S.A.). This kit has a mixture of four types of microbeads coated with different amounts of goat anti-mouse immunoglobulin with a precalibrated antibody binding capacity. The microbeads react with directly labelled mouse MoAb and serve as a set of standards to calibrate the fluorescence scale of the flow cytometer for each antibody thus converting the MFI into the number of molecules of antigens expressed per cell. For each sample the ABC value of the isotypic control was subtracted from the ABC value of the positive cells to compensate for any increased fluorescence background.
Definition of MRD
Samples were considered positive for MRD if, by quantification, they had values similar to those of B-lineage ALL (TdT < 100 × 103, CD10 > 50 × 103 and CD19 > 11 × 103 molecules per cell) regardless of the percentage and were considered MRD-free if the values were similar to those of normal BM (TdT > 100 × 103, CD10 < 50 × 103 and CD19 < 11 × 103 molecules per cell) ( Farahat et al, 1995a , b). Thus, any percentage over 0 was considered positive for residual leukaemia.
For analysis of survival the Kaplan-Meier test was used. A two-sided P value < 0.05 was considered statistically significant.
By morphology, 121/149 BM aspirate samples were considered to be in complete remission (CR) and 18 were found to have residual blasts (≥ 5%); findings in 10 were equivocal (< 5% blasts, but reported as suspicious of persistent leukaemia). By flow cytometry, 101 samples were disease free and 48 showed evidence of MRD. Of the latter 48 MRD-positive samples, 19 were at day 8 of treatment, 13 at day 28 and the remaining 17 later in the evolution. Discrepancies between flow cytometry and morphology were seen in 21 samples. Of these, morphology was consistent with CR in 15 and flow cytometry showed MRD. Follow-up studies showed that seven of these became MRD negative by flow cytometry and four remained MRD positive. The remaining six samples were reported as suspicious of relapse by morphology but they were clear of leukaemia by flow cytometry.
For the purpose of this analysis we separated the 10 patients who experienced relapse or death (Fig 1) from the 43 patients who remained in remission. Morphological examination of the BM suggested that 9/10 relapsed patients reached morphological CR during the course of treatment (Fig 1). Patient 8 could not be assessed morphologically due to persistent bone marrow hypoplasia and was never in proper remission. At the time of morphological CR, MRD was detected by flow cytometry in 6/10 relapsed patients (cases 1–6, Fig 1). These six patients (two children and four adults) subsequently relapsed 38–109 d from the detection of MRD. The detection of MRD by flow cytometry in two of these cases is illustrated in Fig 3 (case 1) and Fig 4 (case 4). In the six patients who relapsed (nos. 1–6), nine samples were reported as CR by morphology with 3–5% blasts; however, all these samples had MRD detected by flow cytometry ( 1 Table I). In four of the relapsed patients MRD was not detected.
Table 1. Table I. Differences between morphology and immunology in patients with MRD. * Case numbers correspond to the relapsed patients included in Fig 1 .† Out of gated lymphoid cells. Any percentage in this column is considered positive for MRD.
Patient 7 relapsed after 32 months in remission, despite the fact that MRD was not detected in two samples, including one tested 11 months before relapse. Patient 8 did not attain remission due to BM hypoplasia, but was interpreted as disease free by flow cytometry and subsequently died 3 months later with active disease. Patient 10 suffered an extramedullary relapse 16 months after the onset of remission. No MRD was detected immunologically and the BM was disease free at the time of relapse.
The follow-up of the 43 patients who did not relapse and were still in CR when last seen is shown in Fig 2. MRD was detected in these patients by flow cytometry only at day 8 in 13 and at day 28 in 10; none of the samples tested positive after day 28. Remission in these patients lasted for periods up to 38+ months (median follow-up 23 months). Samples from five patients in this group (cases 4, 15, 26, 27 and 37, Fig 2) were reported by morphology as suspicious of relapse, with values of 5% blasts or just over. On the other hand, analysis by flow cytometry showed an increase (between 30% and 50%) in normal TdT-negative B-cell precursors (CD10+ CD19+) ( Farahat et al, 1995b ) in all those samples. All these five patients are still in CR 10–28 months from the onset of remission.
In order to examine disease-free survival we divided the patients in two groups according to whether MRD was detected or not by flow cytometry after the onset of morphological remission. This analysis (Fig 5) showed a statistically significant difference between the two groups (P < 0.0001). The absence of MRD by flow cytometry after the achievement of morphological remission (usually after day 28) correlated with good prognosis since 43/47 patients (91%) in which no MRD was detected remained in clinical remission for periods up to 38 months, whereas the presence of MRD predicted relapse in 6/9 patients in whom MRD was detected.
Despite advances in treatment, relapse remains a major problem in acute leukaemia. The development of methods to monitor individual responses, to detect impending relapses prior to clinical manifestation, or to determine the quality of BM scheduled for autologous transplantation, still represents a challenge. We evaluated the clinical significance of MRD detection by quantitative flow cytometry and found this methodology better than conventional morphology in assessing remission status, and correlates with patient outcome in B-lineage ALL.
Fifty-three patients with B-ALL precursor were followed during treatment. The detection of MRD with the combination of TdT, CD10 and CD19 predicted relapse in 6/9 patients within 5–15 weeks but failed to detect MRD in three patients who subsequently relapsed. The possible reasons for these false negative results are threefold: (i) the sample may not have been representative due to the non-homogenous distribution of leukaemia in the BM ( Hann et al, 1977 ; Jacobs, 1977; Martens et al, 1987 ); this may be the case in patient 8 who was not in remission due to BM hypoplasia; (ii) the possibility of a change in phenotype at the time of relapse ( Campana et al, 1990b ; Raghavachar et al, 1988 ; Abshire et al, 1992 ; van-Wering et al, 1995 ). Although none of our 53 cases had a phenotypic switch, this potential problem could be overcome by using combinations of markers and not relying on a single one; and (iii) the presence of leukaemic cells in numbers below the sensitivity of detection. Findings in patient 7, who relapsed after 2 years in remission, may illustrate this problem which may theoretically be overcome by more frequent testing of BM samples.
Quantitative flow cytometry was suitable for detecting MRD in all B-lineage ALL cases tested and not only those expressing aberrant phenotypes. Most of the reports in the literature are based on the aberrant antigen expression on the leukaemic blasts ( Campana et al, 1990a ). In the group of 43 patients who did not relapse after morphological remission we did not detect MRD by flow cytometry combining TdT with CD10 or CD19. Although seven of them had MRD detected at days 8 and 28, there was no correlation between the detection of leukaemia at this early stage and subsequent relapse. On the other hand, all six patients (four of them adults) in whom MRD was detected later, relapsed.
We conclude that the detection of MRD by flow cytometry is a relatively simple and sensitive technique for the follow-up of ALL patients which may help predict clinical outcome. The sensitivity of this method, in the range of 10−3–10−4, although lower than that of PCR 10−4–10−6 ( Campana et al, 1990b ; van Dongen et al, 1992 ), appears to identify patients at a high risk of relapse. Many studies by PCR showed residual cells detected immediately after remission induction and in a substantial number of patients up to 24 months post-remission despite the patients remaining in remission. Thus PCR positivity does not always correlate with long-term responses ( Yokota et al, 1991 ; Nizet et al, 1993 ; Deane & Hoffbrand, 1993). To overcome the lower sensitivity of the immunological techniques, more frequent sampling of the patients may be needed especially in the first 2 years of treatment where relapses are more common. The main question which remains to be answered is whether the detection of MRD will have a role in selecting treatment strategies in individual patients. The data of this study suggest that in the group of patients destined to early relapse, detection of MRD by flow cytometry predicts outcome. Failure to achieve true CR by day 28 in the majority of those who relapsed would put them in a poor-risk group. The value of this simple methodology should be tested into a larger trial for MRD, particularly for adult ALL.
Note added in proof. In the February 1998 issue of the Lancet (351, 550–554), E. Coustan-Smith et al from St Jude Children's Hospital published a prospective study which confirms the clinical value of detecting minimal residual disease in childhood ALL by immunological methods and flow cytometry, reported in our study.
This work was supported by a grant from the Egyptian Government (N.F.) and by Trust Funds from the Royal Marsden NHS Trust. We are grateful to Mr R. A'Hern for help with the statistical analysis.