- Email: [email protected]
Immunological detection of minimal residual disease in children with acute lymphoblastic leukaemia
ARTICLES
Immunological detection of minimal residual disease in children with acute lymphoblastic leukaemia Elaine Coustan-Smith, Frederick G Behm, Joaquin Sanchez, James M Boyett, Michael L Hancock, Susana C Raimondi, Jeffrey E Rubnitz, Gaston K Rivera, J Torrey Sandlund, Ching-Hon Pui, Dario Campana
Summary Background The clinical significance of submicroscopic levels of leukaemic cells in bone-marrow aspirates from children with acute lymphoblastic leukaemia (ALL) remains controversial. We prospectively determined the frequency and prognostic importance of minimal residual disease detected by a rapid immunological assay in bone-marrow aspirates of children with ALL. Methods 158 children with newly diagnosed ALL received 6 weeks of remission-induction chemotherapy. Once complete clinical remission was attained the patients received 2 weeks of consolidation therapy followed by continuation therapy. Bone-marrow aspirates were collected after induction therapy and during weeks 14, 32, and 56 of continuation therapy, and then at 120 weeks (end of therapy). Cells with leukaemia-associated immunophenotypes were investigated by multiparameter flowcytometry capable of detecting one leukaemic cell among 10 000 normal cells. Findings The proportion of patients with detectable leukaemic cells was 23% at remission induction and 17% at week 14 of continuation therapy, decreasing to 5% and 4% at weeks 32 and 56. None of the 65 samples examined at completion of therapy (week 120) showed evidence of disease. Detectable residual disease at the end of remission induction correlated with adverse genetic abnormalities—the Philadelphia chromosome and MLL gene rearrangements—but not with other presenting features. Detectable leukaemia was associated with subsequent relapse regardless of the time at which bone-marrow samples were examined (p<0·002 for all comparisons). For example, 3-year cumulative incidence (SE) of relapse in patients with and without detectable leukaemia at remission induction was 32·5% (10·6) and 7·5% (4·0), respectively (p<0·001); for tests done at week 14, it was 42·1% (14·6) and 6·6% (3·5), p<0·001. These correlations remained significant after adjusting for adverse presenting features. A higher degree of marrow infiltration by leukaemic cells (⭓0·1%) in week 14 samples identified a subset of patients with an especially poor prognosis. Departments of Hematology-Oncology (E Coustan-Smith FIBMS, J Sanchez MD, J E Rubnitz MD, Prof G K Rivera MD, J T Sandlund MD, Prof C-H Pui MD, D Campana MD), Pathology and Laboratory Medicine (F G Behm MD, S C Raimondi PhD, C-H Pui), and Biostatistics and Epidemiology (Prof J M Boyett PhD, M L Hancock MS), St Jude Children’s Research Hospital; and the University of Tennessee College of Medicine, Memphis, Tennessee, USA (F G Behm, J M Boyett, S C Raimondi, J E Rubnitz, G K Rivera, J T Sandlund, C-H Pui, D Campana) Correspondence to: Dr Dario Campana, Department of HematologyOncology, St Jude Children’s Research Hospital, 332 North Lauderdale, Memphis TN 38105, USA (e-mail: [email protected])
550
Interpretation Immunological detection of residual leukaemic cells at any point in the treatment course is a powerful predictor of relapse in children with ALL. Alternative treatment should be considered for cases with persistent disease beyond the first 3 months of continuation therapy.
Lancet 1998; 351: 550–54
Introduction Nearly a quarter of children with acute lymphoblastic leukaemia (ALL) who achieve complete remission by standard criteria eventually relapse, and most die from the disease.1 Attempts to monitor the effectiveness of therapy and to detect impending (subclinical) relapse have generally relied on the examination of the morphology of sequentially collected bone-marrow cells. Patients with fewer than 5% morphologically identifiable lymphoblasts in bone-marrow samples are judged to be in clinical remission, although they may still harbour as many as 1010 leukaemic cells.2 In theory, the detection of leukaemic cells when their concentrations are below the resolution of morphological techniques (ie, minimal residual disease) would allow better estimates of the total body burden of leukaemic cells and, in turn, could improve clinical management practices and advance cure rates. A variety of lymphoblast features—gene fusion transcripts, antigen-receptor gene rearrangements, and immunophenotype—can be used to detect minimal residual disease in patients with ALL.2 Use of PCR to identify fusion transcripts and clonal antigen-receptor gene rearrangements is attractive because of its high sensitivity (one target cell per 103 to 106 cells). Specific fusion transcripts can be used as a PCR target in only one-third of cases of childhood ALL, but clonal antigenreceptor gene rearrangements occur in virtually all cases. 2 In general, studies of these gene rearrangements have demonstrated a high risk of relapse in cases with PCR signals corresponding to one or more leukaemic cells per 103 bone-marrow cells at the end of remission induction,3–5 or increasing levels of residual leukaemia during continuation therapy.6–12 The reliability of PCR assays depends on the use of stringent quantitation protocols,3,4,10,12,13 and analysis of multiple genetic targets to prevent false-negative results due to changes in the pattern of gene rearrangement during the disease course.14–16 These requirements, however, further complicate an already laborious procedure. Thus, the suitability of these assays for routine clinical studies may be limited. We have developed a rapid assay for minimal residual disease that relies on multiparameter flow-cytometric identification of various combinations of surfacemembrane, cytoplasmic, and nuclear molecules expressed by leukaemic lymphoblasts but not by normal
THE LANCET • Vol 351 • February 21, 1998
ARTICLES
bone-marrow cells.2 In addition to its high sensitivity (one leukaemic cell per 104 normal bone-marrow cells), this method allows precise quantitation of leukaemic cell populations and discriminates between viable and dead or dying cells. We describe a prospective study in which we applied our immunological method to monitor residual disease in bone-marrow aspirates collected sequentially from 158 children with newly diagnosed ALL.
Methods From December, 1991, to March, 1997, 316 children with newly diagnosed ALL were enrolled in studies XIIIA (n=166) and XIIIB (n=150) at our institution. At diagnosis, a leukaemia-associated immunophenotype was identified in 166 (57·2%) of the 290 patients in whom full immunophenotypic studies were done. Minimal residual disease was studied in 158 (95·2%) patients; of the remaining eight patients, five did not attain complete clinical remissions within the planned 6-week period of induction chemotherapy, and three had inadequate samples for flow-cytometric studies. The presenting features of the patients with leukaemia-associated immunophenotypes were not significantly different from those of the remaining patients, with the exception of a higher proportion of boys (65·1 vs 49·2%, p<0·01), MLL gene rearrangements (9·6 vs 1·6%, p<0·01) and T-cell phenotype (28·9 vs 0·8%, p<0·001). The 3year cumulative incidence of relapse was 11·3% (SE 3·0) in patients with leukaemia-associated phenotypes and 4·8% (2·4) in the remaining patients (p=0·04). All procedures were approved by the St Jude Institutional Review Board, with informed consent obtained from the parents or guardians of each child. Diagnostic immunophenotyping and chromosomal and genetic analysis were done with standard techniques. After initial treatment with methotrexate or mercaptopurine, or both, all patients received 6 weeks of remission-induction chemotherapy with prednisone, vincristine, daunorubicin, Lasparaginase, teniposide, and cytarabine.17,18 After they attained a complete clinical remission, patients received 2 weeks of consolidation therapy with high-dose methotrexate and mercaptopurine, followed by risk-directed continuation therapy. Lower-risk cases were treated with mercaptopurine and methotrexate for 120 weeks, with a pulse of prednisone or dexamethasone plus vincristine every 4 weeks. For patients at high-risk of relapse, continuation therapy consisted of alternating pairs of drugs: etoposide plus cyclophosphamide, mercaptopurine plus methotrexate, methotrexate or etoposide plus cytarabine, or prednisone or dexamethasone plus vincristine. All patients received intensive age-adjusted doses of triple intrathecal therapy with methotrexate, hydrocortisone, and cytarabine for 1 year; cranial irradiation was reserved for patients at very high-risk of relapse (18 Gy) or those with central-nervous-system status three at diagnosis (24 Gy) at 1 year of continuation therapy. Six patients judged to have very high-risk ALL underwent allogeneic bone-marrow transplantation.
CD13, and CD33 (Dako, Carpinteria), anti-IgM (Southern Biotechnology Assoc, Birmingham, AL, USA), KORSA3544 (a gift from Dr Mori, Saitama, Japan), and CD21 (Immunotech, Westbrook, Maine). Isotype-matched non-reactive fluorochrome-conjugated antibodies were used as controls. The staining procedure has been described.2,19,20 For intracellular staining, cells were treated with Permeafix (Ortho Diagnostic Systems, Raritan).2 A FACScan flow cytometer with Lysis II or Cell Quest software (Becton Dickinson) was used in all analyses. We defined two gates after analysing 10 000 cells. Gate 1 surrounded the light-scatter dot plot that represented lymphoid cells; this area included all leukaemic cells present at diagnosis in the cases studied. Gate 2 identified the cell population expressing the immunophenotypic features of immature lymphoid cells (eg, CD19 and CD34). Electronic events falling within both gates 1 and 2 were then selectively recorded. Acquisition was terminated when virtually all cells in the tube (eg, 2–5 x 105) had passed through the flow cytometer or 10 000 events that represent immature lymphoid cells were recorded, whichever occurred first. The expression of leukaemiaassociated immunophenotypes among the selected population was then analysed. Flow-cytometric data were recorded within 24 h after sample collection and processing, with no knowledge of patients clinical status or diagnostic variables, except immunophenotype. The marker combinations used in this study were selected because they are not expressed (or expressed extremely weakly) in normal or regenerating bone-marrow cells.2 Thus, in preliminary experiments with artificial mixtures of leukaemic and normal cells, we could consistently detect one leukaemic cell per 104 normal nucleated bone-marrow cells with each antibody combination used in the study, and correctly determine different ratios of leukaemic-to-normal cells.2,20 To assess the precision of the assay, we compared results of multiple measurements of residual leukaemia in identical cell preparations, in 23 tests with mixtures containing one leukaemic cell in 104. The coefficient of variation was 15%. In 22 tests with mixtures containing one leukaemic cell in 10 3, the coefficient of variation was 10%.
Statistical analyses Distributions of presenting features according to the degree of residual disease during clinical remission were compared by Fisher’s exact test. The probability of surviving without ALL relapse or without an adverse event (ie, relapse or death due to any cause) was estimated by the Kaplan-Meier method, applied to follow-up observations through July 8, 1997; associated SEs were calculated by the Peto method. Follow-up information was gathered within 6 months of the analysis for 96% of patients. The prognostic importance of the levels of residual disease was determined first by analyses stratified by treatment only, and then by stratified analyses with competing covariates. Comparisons of continuous complete-remission distributions were made by the Mantel-Haenszel test and the exact log-rank test, as necessary. Cumulative incidence functions of leukaemia
Flow cytometry
Marker combinations
Bone-marrow aspirates were collected in preservative-free heparin after 6 weeks of induction therapy; during weeks 14, 32, and 56 of continuation therapy; and at the end of therapy (week 120). Cells with leukaemia-associated immunophenotypes (ie, those found on leukaemic cells but not on normal bone-marrow cells) were investigated by multiparameter flow-cytometry, with various combinations of monoclonal and polyclonal antibodies against surface, cytoplasmic or nuclear leucocyte antigens (table 1). The following antibodies, conjugated to fluorescein isothiocyanate, phycoerythrin, peridinin chlorophyll protein, or biotin, were used: anti-terminal deoxynucleotidyl transferase (TdT; Supertechs, Bethesda); CD34 and CD56 (Becton Dickinson, San Jose); CD10 (Caltag, S San Francisco); CD19,
Early lymphoid
Leukaemia-associated
TdT TdT and CD34 CD10 CD19 and CD34 or CD10 CD19 and CD34 or CD10 CD19 and CD34 or CD10 CD19 and CD34 CD34 and CD10
CD3 (cytoplasmic or surface) IgM (cytoplasmic) KORSA3544 CD13 CD33 CD65 CD21 CD56
THE LANCET • Vol 351 • February 21, 1998
Frequency*
27·8% 21·5% 20·9% 20·2% 11·4% 10·8% 8·9% 5·1%
*Percentage of cases in this series expressing each marker at diagnosis (>50% positive cells). 42 (26·6%) of 158 cases studied had more than one leukaemiaassociated marker.
Table 1: Leukaemia markers used to study minimal residual disease by double or triple colour flow-cytometry
551
ARTICLES
relapse were estimated as described by Kalbfleisch and Prentice, and compared by Gray’s test. For these calculations, missing residual disease determinations during continuation therapy (4% of all tests) were imputed when the degree of residual disease at earlier and later tests were identical.
Results We identified cells with leukaemia-associated immunophenotypes in 23·4% of the bone marrow samples taken at the end of remission induction (table 2). The proportion of samples with residual disease remained relatively high (17·1%) at week 14 of continuation therapy, but had decreased strikingly (to 5·3%) by week 32, with an additional decrease to 3·9% apparent at 56. None of the 65 patients studied at the completion of therapy had detectable leukaemic blast cells. The percentage of leukaemic cells among mononucleated bone-marrow cells ranged from 0·01% to 19%, with a median of about 0·1% in all positive samples. Nine of the samples had more than 1% of leukaemic cells. An additional two had more than 10% leukaemic cells by flow cytometry. In both of these samples, two expert observers could not identify leukaemic lymphoblasts by morphology; in one sample, all lymphoid cells appeared to be normal differentiated lymphocytes, while in the other, 3·5% of cells had immature undifferentiated appearance and were regarded to be normal haematopoietic progenitor cells (haematogones). Rates of residual disease detection did not differ significantly in comparisons based on sex, age, race, leucocyte count, central-nervous-system status, immunophenotype or leukaemic cell ploidy. There were, however, striking differences in subgroups of patients with exceptionally poor prognoses.1 For example, at the end of remission-induction chemotherapy, residual disease was detected in all eight patients with the Philadelphia chromosome (p<0·0001) and in six of 12 cases with MLL gene rearrangements (p=0·033). By contrast, only one of 12 patients with rearrangements of the TEL gene and four of 29 with hyperdiploidy (51–65 chromosomes), both of which are favourable prognostic signs,1 had detectable levels of residual disease at the end of remission-induction chemotherapy. At each time point, the detection of minimal residual disease was significantly associated with a greater likelihood of treatment failure and of leukaemic relapse (p<0·002, table 3). In the group tested at the end of induction therapy, 3-year estimates of continuous complete remission (SE) were 68·5% (14·5) for patients with positive findings and 89·8% (5·9) for those with negative findings. Three of the seven failures in the subgroup without evidence of residual leukaemia were due to causes other than ALL relapse (infection or secondary leukaemia). Moreover, one of the four patients with negative findings who relapsed had detectable leukaemic cells in subsequent testing, while in two other patients the resurgent blast cells lacked the leukaemia-specific markers observed at diagnosis. By contrast, all eight failures in the positive group were ALL relapses, with cells retaining leukaemia-associated markers at the time of relapse. The value of residual leukaemia detected at the time of remission induction to predict subsequent relapse remained significant after adjustment for established prognostic factors, such as age younger than 1 year and older than 9 years (p<0·001), leucocyte count of more than 50⫻109/L
552
Time of testing
Number tested
Number with residual disease
Median % of leukaemia cells (lowest and (highest value)
End of remission induction Continuation therapy Week 14 Week 32 Week 56 End of therapy
128
30 (23·4%)
0·13 (0·01–19·0)
105 114 103 65
18 (17·1%) 6 (5·3%) 4 (3·9%) 0
0·02 (0·01–14·9) 0·10 (0·01–0·44) 0·03 (0·01–0·06) ..
Table 2: Detection of residual disease with immunological markers in children with ALL in clinical remission
(p<0·001) and presence of adverse genetic features (Philadelphia chromosome and MLL gene rearrangements, p<0·006). Because previous studies with PCR had indicated that patients with levels of residual disease higher than one in 1000 normal cells on the date of complete remission were at a significantly higher risk of relapse,3–5 we reanalysed our data after segregating positive cases into two groups according to levels of residual disease (<0·1% and ⭓0·1%). The risk of treatment failure or leukaemia relapse were not significantly related to the proportion of leukaemic cells (data not shown). Detection of minimal residual disease during continuation chemotherapy was also strongly predictive of leukaemic relapse (p<0·002, table 3). At week 14, the cumulative incidence of leukaemic relapse after 3 years of the test was 6·6% (3·5) for patients with negative findings and 42·1% (14·6) for those with detectable disease. The prognostic value of the flow-cytometric assay remained significant after adjustment for competing covariates, including age (p<0·001), leucocyte count (p<0·001), and adverse genetic features (p=0·04). Even after exclusion of patients with Philadelphia chromosome and those with MLL gene rearrangements from the analysis, detection of minimal residual disease remained a significant predictor of leukaemia relapse (p= 0·015). Of note, among the seven patients with MLL gene rearrangements, only the two with positive findings subsequently relapsed. Relapse also occurred in two of the four patients with hyperdiploid leukaemia who had positive findings at week 14. At this time, measurements of residual disease levels identified patients at a very high risk of relapse (figure). Indeed, four of five patients with higher levels of Time of testing
End of remission induction MRD positive MRD negative Continuation therapy Week 14 MRD positive MRD negative Week 32 MRD positive MRD negative Week 56 MRD positive MRD negative
Continuous complete remission(SE)*
p†
Cumulative incidence of leukaemia relapse (SE)*
p†
68·5 (14·5) 89·8 (5·9)
<0·002
32·5 (10·6) 7·5 (4·0)
<0·001
58·0 (21·7) 88·3 (6·0)
<0·001
42·1 (14·6) 6·6 (3·5)
<0·001
25·0 (15·3) 83·5 (6·8)
<0·001
75·0 (27·5) 8·0 (3·8)
<0·001
25·0 (15·3) 86·6 (7·7)
<0·001
50·0 (32·5) 6·6 (3·7)
<0·002
MRD=minimal residual disease. *Estimated percentages at 3 years after each sampling. †p based on log rank and Gray’s test, respectively.
Table 3: Continuous complete remission and cumulative incidence of leukaemic relapse according to detection of residual disease
THE LANCET • Vol 351 • February 21, 1998
Cumulative risk of relapse (%)
ARTICLES
100
p<0·001
Positive (⭓0·1%)
80 60
60 (26)% 35 (18)%
40 20
Positive (⬍0·1%)
7 (4)%
Negative
0 0
1 2 3 4 5 Time after week 14 of continuation therapy (years)
Kaplan-Meier estimates of cumulative incidence of relapse of ALL in children Estimates are according to levels of residual disease in bone-marrow samples collected on week 14 of continuation therapy. Numbers in parentheses are SEs.
leukaemia infiltration (⭓0·1%) have subsequently relapsed. The remaining patient has not yet completed therapy but has retained more than 0·1% leukaemic lymphoblasts in four follow-up assays (through week 79). Of the 65 children studied at the end of treatment, none had detectable residual disease. However, four eventually relapsed, at 7 to 15 months after the negative test. One of these patients had positive findings throughout the first year of therapy (up to week 56 of continuation therapy), while two had phenotypic switches resulting in loss of leukaemia-specific markers.
Discussion The lack of prospective studies in large cohorts of similarly treated patients has hampered efforts to assess the clinical significance of minimal residual disease in children with ALL who are in complete clinical remission. Our main conclusion is that the presence or absence of detectable residual leukaemia during clinical remission strongly influences treatment outcome, suggesting that flow-cytometric assays could be used to augment current methods of risk assessment. We found that detection of residual disease at any of four different intervals during treatment was associated with a high likelihood of subsequent relapse. The extent of marrow infiltration by leukaemic cells closely predicted treatment outcome only when measurements were taken from samples from week 14. At points beyond 6 months of continuation therapy, any amount of residual disease carried a very poor prognosis. The flow-cytometric techniques we used are well suited for clinical investigations of residual disease because of their rapidity and wide availability.2,21,22 As demonstrated here, they reliably discriminate between leukaemic and normal cells, even when the former cells assume an apparently normal morphology. This point is well illustrated by our two patients with greater than 10% leukaemic cells disguised as differentiated lymphocytes or haematogones. Both patients had a relapse of leukaemia 4 months and 13 months after the positive immunological observation. Although the leukaemia-associated phenotypes used as targets in the flow-cytometric assay were expressed in only 57% of all cases studied at diagnosis, this percentage is likely to increase with the continuous development of new antibodies against leucocyte
THE LANCET • Vol 351 • February 21, 1998
antigens. For example, monoclonal antibodies reacting specifically with cells carrying E2A-PBX1 or MLL gene rearrangements are now available for flow-cytometric studies of residual disease,23,24 and antibodies to molecules overexpressed in leukaemic lymphoblasts, such as WT1 and BCL2, have shown promise in preliminary experiments.25,26 These reagents may well allow all children with ALL to be monitored for residual leukaemia by flow cytometry. In any case, the antibody combinations used in our analysis were applicable to most high-risk cases, including those with the Philadelphia chromosome, MLL gene rearrangements, or a T-cell immunophenotype, as these subtypes of ALL regularly express leukaemia-associated immunophenotypes.2 The value of both immunological and molecular techniques in monitoring residual leukaemia may be limited by evolution of the leukaemic clone during the course of the disease.2,14–16 Indeed, in our series, failure to detect impending relapse in three patients was associated with the loss of leukaemia-associated immunophenotypes in the resurgent clone. However, in all three cases, only one combination suitable for monitoring was detected at diagnosis, suggesting that the addition of new markers to the current arsenal should significantly reduce or eliminate the occurrence of false-negative results due to immunophenotypic switches. Four of 65 patients in our study relapsed 7–15 months later even though they lacked detectable leukaemic cells at the end of therapy. Persistent leukaemic cells in these patients might have been detected by a more sensitive technique.12 However, in the study by Ito and colleagues,27 PCR amplification of IgH genes, purported to be about one log more sensitive then immunological techniques, failed to identify residual disease in eight of nine patients who relapsed after completion of therapy. Thus, the absence of residual disease at the end of therapy by either immunological or molecular criteria does not guarantee a durable remission. In summary, immunological investigation of minimal residual disease provides clinically meaningful information that could be used to improve risk assessment strategies and treatment selection in the management of ALL in children. We suggest that patients with detectable leukaemic cells at the end of remission induction should be closely monitored for disease resurgence. Alternative treatment should be considered for patients with persistent disease during early continuation therapy. Contributors Elaine Coustan-Smith was responsible for the flow-cytometric analysis of residual disease and for the overall study coordination. Frederick Behm, Susana Raimondi, and Jeffrey Rubnitz provided the information on presenting immunophenotypic, karyotypic, and genetic features. Joaquin Sanchez contributed to the flow-cytometric studies. James Boyett and Michael Hancock undertook the statistical design and analysis of the results. Gaston Rivera and Torrey Sandlund contributed to the protocol design and the clinical conduct of the study. Ching-Hon Pui was responsible for the overall clinical conduct of the study and contributed to data analysis and preparation of the paper. Dario Campana was responsible for the concept of the study, data analysis, and with advice from the other authors, preparation of the paper.
Acknowledgments This work was supported by grants CA60419, CA21765, and CA20180 from the National Cancer Institute, and by the American Lebanese Syrian Associated Charities (ALSAC). JS was the recipient of grant 97/5486 from the Fondo de Investigacion Sanitaria of Spain. We thank Taijiro Mori for the gift of KORSA3544 and John Gilbert for critical review of the manuscript.
553
ARTICLES
References 1 2
3
4
5
6
7
8
9
10
11
12
13 14
Pui C-H. Childhood leukemias. N Engl J Med 1995; 332: 1618–30. Campana D, Pui C-H. Detection of minimal residual disease in acute leukemia: methodological advances and clinical significance. Blood 1995; 85: 1416–34. Wasserman R, Galili N, Ito Y, et al. Residual disease at the end of the induction therapy as a predictor of relapse during therapy in childhood B-lineage acute lymphoblastic leukemia. J Clin Oncol 1992; 10: 1879–85. Brisco MJ, Condon J, Hughes E, et al. Outcome prediction in childhood acute lymphoblastic leukaemia by molecular quantification of residual disease at the end of induction. Lancet 1994; 343: 196–200. Jacquy C, Delepaut B, Van Daele S, et al. A prospective study of minimal residual disease in childhood B-lineage acute lymphoblastic leukaemia: MRD level at the end of induction is a strong predictive factor of relapse. Br J Haematol 1997; 98: 140–46. Neale GAM, Menarguez J, Kitchingman GR, et al. Detection of minimal residual disease in T-cell acute lymphoblastic leukemia using polymerase chain reaction predicts impending relapse. Blood 1991; 78: 739–45. Biondi A, Yokota S, Hansen-Hagge TE, et al. Minimal residual disease in childhood acute lymphoblastic leukemia: analysis of patients in continuous complete remission or with consecutive relapse. Leukemia 1992; 6: 282–88. Nizet Y, Van Daele S, Lewalle P, et al. Long-term follow-up of residual disease in acute lymphoblastic leukemia patients in complete remission using clonogeneic IgH probes and the polymerase chain reaction. Blood 1993; 82: 1618–25. Kitchingman GR. Residual disease detection in multiple follow-up samples in children with acute lymphoblastic leukemia. Leukemia 1994; 8: 395–402. Cave H, Guidal P, Rohrlich P, et al. Prospective monitoring and quantitation of residual blasts in childhood acute lymphoblastic leukemia by polymerase chain reaction study of omega and gamma T-cell receptor genes. Blood 1994; 83: 1892–901. Seriu T, Yokota S, Nakao M, et al. Prospective monitoring of minimal residual disease during the course of chemotherapy in patients with acute lymphoblastic leukemia, and detection of contaminating tumor cells in peripheral blood stem cells for autotransplantation. Leukemia 1995; 9: 615–23. Roberts WM, Estrov Z, Ouspenskaia MV, Johnston DA, McClain KL, Zipf TF. Measurement of residual leukemia during remission in childhood acute lymphoblastic leukemia. N Engl J Med 1997; 336: 317–23. Cross NCP. Quantitative PCR techniques and applications. Br J Haematol 1995; 89: 693–97. Beishuizen A, Verhoeven M-AJ, van Wering ER, Hahlen K, Hooijkaas H, van Dongen JJM. Analysis of Ig and T-cell receptor genes in 40 childhood acute lymphoblastic leukemias at diagnosis and subsequent relapse: implications for the detection of minimal residual
554
15
16
17
18
19
20
21
22
23
24
25
26
27
disease by polymerase chain reaction analysis. Blood 1994; 83: 2238–47. Steward CG, Goulden NJ, Katz F, et al. A polymerase chain reaction study of the stability of Ig heavy-chain and T-cell receptor omega gene rearrangements between presentation and relapse of childhood B-lineage acute lymphoblastic leukemia. Blood 1994; 83: 1355–62. Steenbergen EJ, Verhagen OJHM, Van Leeuwen EF, et al. Frequent ongoing TCR delta rearrangements in childhood precursor acute lymphoblastic leukemia: implications for monitoring minimal residual disease. Blood 1995; 86: 692–702. Pui C-H, Boyett JM, Hughes WT, et al. Human granulocyte colonystimulating factor after induction chemotherapy in children with acute lymphoblastic leukemia. N Engl J Med 1997; 336: 1781–87. Pui C-H, Rivera GK, Hancock ML, et al. Risk-adapted treatment for acute lymphoblastic leukemia: findings from St Jude Children’s Research Hospital. In: Buchner T, Hiddeman W, Wormann B, Schellong G, Ritter J, Crentzig U, eds. Acute leukemia VI: prognostic factors and treatment strategies; haematology and blood transfusions. Berlin:Springer-Verlag, 1997: 629–37. Campana D, Coustan-Smith E, Janossy G. The immunological detection of minimal residual disease in acute leukemia. Blood 1990; 76: 163–69. Coustan-Smith E, Behm FG, Hurwitz CA, et al. N-CAM (CD56) expression by CD34+ malignant myeloblasts has implications for minimal residual disease detection in acute myeloid leukemia. Leukemia 1993; 7: 853–58. Orfao A, Ciudad J, Lopez-Berges MC, et al. Acute lymphoblastic leukemia (ALL): detection of minimal residual disease (MRD) by flow cytometry. Leuk Lymph 1994; 13: 87–90. Vervoordeldonk SF, Merle PA, Behrendt H, et al. Triple immunofluorescence staining for prediction of relapse in childhood B acute lymphoblastic leukemia. Br J Haematol 1996; 92: 922–28. Sang BC, Shi L, Dias P, et al. Monoclonal antibodies specific to the acute lymphoblastic leukemia t(1;19)-associated E2A/pbx1 chimeric protein: characterization and diagnostic utility. Blood 1997; 89: 2909–14. Behm FG, Smith FO, Raimondi SC, et al. Human homologue of the rat chondroitin sulfate proteoglycan, NG2, detected by monoclonal antibody 7.1, identifies childhood acute lymphoblastic leukemia with t(4;11)(q21;q23) or t(11;19)(q23;p13) and MLL gene rearrangements. Blood 1996; 87: 1134–39. Menssen HD, Renkl HJ, Rodeck U, et al. Presence of Wilms’ tumor gene (wt1) transcripts and the WT1 nuclear protein in the majority of human acute leukemias. Leukemia 1995; 9: 1060–67. Coustan-Smith E, Kitanaka A, Pui C-H, et al. Clinical relevance of BCL-2 overexpression in childhood acute lymphoblastic leukemia. Blood 1996; 87: 1140–46. Ito Y, Wasserman R, Galili N, et al. Molecular residual disease status at the end of chemotherapy fails to predict subsequent relapse in children with B-lineage acute lymphoblastic leukemia. J Clin Oncol 1993; 11: 546–50.
THE LANCET • Vol 351 • February 21, 1998
Immunological detection of minimal residual disease in children with acute lymphoblastic leukaemia Elaine Coustan-Smith, Frederick G Behm, Joaquin Sanchez, James M Boyett, Michael L Hancock, Susana C Raimondi, Jeffrey E Rubnitz, Gaston K Rivera, J Torrey Sandlund, Ching-Hon Pui, Dario Campana
Summary Background The clinical significance of submicroscopic levels of leukaemic cells in bone-marrow aspirates from children with acute lymphoblastic leukaemia (ALL) remains controversial. We prospectively determined the frequency and prognostic importance of minimal residual disease detected by a rapid immunological assay in bone-marrow aspirates of children with ALL. Methods 158 children with newly diagnosed ALL received 6 weeks of remission-induction chemotherapy. Once complete clinical remission was attained the patients received 2 weeks of consolidation therapy followed by continuation therapy. Bone-marrow aspirates were collected after induction therapy and during weeks 14, 32, and 56 of continuation therapy, and then at 120 weeks (end of therapy). Cells with leukaemia-associated immunophenotypes were investigated by multiparameter flowcytometry capable of detecting one leukaemic cell among 10 000 normal cells. Findings The proportion of patients with detectable leukaemic cells was 23% at remission induction and 17% at week 14 of continuation therapy, decreasing to 5% and 4% at weeks 32 and 56. None of the 65 samples examined at completion of therapy (week 120) showed evidence of disease. Detectable residual disease at the end of remission induction correlated with adverse genetic abnormalities—the Philadelphia chromosome and MLL gene rearrangements—but not with other presenting features. Detectable leukaemia was associated with subsequent relapse regardless of the time at which bone-marrow samples were examined (p<0·002 for all comparisons). For example, 3-year cumulative incidence (SE) of relapse in patients with and without detectable leukaemia at remission induction was 32·5% (10·6) and 7·5% (4·0), respectively (p<0·001); for tests done at week 14, it was 42·1% (14·6) and 6·6% (3·5), p<0·001. These correlations remained significant after adjusting for adverse presenting features. A higher degree of marrow infiltration by leukaemic cells (⭓0·1%) in week 14 samples identified a subset of patients with an especially poor prognosis. Departments of Hematology-Oncology (E Coustan-Smith FIBMS, J Sanchez MD, J E Rubnitz MD, Prof G K Rivera MD, J T Sandlund MD, Prof C-H Pui MD, D Campana MD), Pathology and Laboratory Medicine (F G Behm MD, S C Raimondi PhD, C-H Pui), and Biostatistics and Epidemiology (Prof J M Boyett PhD, M L Hancock MS), St Jude Children’s Research Hospital; and the University of Tennessee College of Medicine, Memphis, Tennessee, USA (F G Behm, J M Boyett, S C Raimondi, J E Rubnitz, G K Rivera, J T Sandlund, C-H Pui, D Campana) Correspondence to: Dr Dario Campana, Department of HematologyOncology, St Jude Children’s Research Hospital, 332 North Lauderdale, Memphis TN 38105, USA (e-mail: [email protected])
550
Interpretation Immunological detection of residual leukaemic cells at any point in the treatment course is a powerful predictor of relapse in children with ALL. Alternative treatment should be considered for cases with persistent disease beyond the first 3 months of continuation therapy.
Lancet 1998; 351: 550–54
Introduction Nearly a quarter of children with acute lymphoblastic leukaemia (ALL) who achieve complete remission by standard criteria eventually relapse, and most die from the disease.1 Attempts to monitor the effectiveness of therapy and to detect impending (subclinical) relapse have generally relied on the examination of the morphology of sequentially collected bone-marrow cells. Patients with fewer than 5% morphologically identifiable lymphoblasts in bone-marrow samples are judged to be in clinical remission, although they may still harbour as many as 1010 leukaemic cells.2 In theory, the detection of leukaemic cells when their concentrations are below the resolution of morphological techniques (ie, minimal residual disease) would allow better estimates of the total body burden of leukaemic cells and, in turn, could improve clinical management practices and advance cure rates. A variety of lymphoblast features—gene fusion transcripts, antigen-receptor gene rearrangements, and immunophenotype—can be used to detect minimal residual disease in patients with ALL.2 Use of PCR to identify fusion transcripts and clonal antigen-receptor gene rearrangements is attractive because of its high sensitivity (one target cell per 103 to 106 cells). Specific fusion transcripts can be used as a PCR target in only one-third of cases of childhood ALL, but clonal antigenreceptor gene rearrangements occur in virtually all cases. 2 In general, studies of these gene rearrangements have demonstrated a high risk of relapse in cases with PCR signals corresponding to one or more leukaemic cells per 103 bone-marrow cells at the end of remission induction,3–5 or increasing levels of residual leukaemia during continuation therapy.6–12 The reliability of PCR assays depends on the use of stringent quantitation protocols,3,4,10,12,13 and analysis of multiple genetic targets to prevent false-negative results due to changes in the pattern of gene rearrangement during the disease course.14–16 These requirements, however, further complicate an already laborious procedure. Thus, the suitability of these assays for routine clinical studies may be limited. We have developed a rapid assay for minimal residual disease that relies on multiparameter flow-cytometric identification of various combinations of surfacemembrane, cytoplasmic, and nuclear molecules expressed by leukaemic lymphoblasts but not by normal
THE LANCET • Vol 351 • February 21, 1998
ARTICLES
bone-marrow cells.2 In addition to its high sensitivity (one leukaemic cell per 104 normal bone-marrow cells), this method allows precise quantitation of leukaemic cell populations and discriminates between viable and dead or dying cells. We describe a prospective study in which we applied our immunological method to monitor residual disease in bone-marrow aspirates collected sequentially from 158 children with newly diagnosed ALL.
Methods From December, 1991, to March, 1997, 316 children with newly diagnosed ALL were enrolled in studies XIIIA (n=166) and XIIIB (n=150) at our institution. At diagnosis, a leukaemia-associated immunophenotype was identified in 166 (57·2%) of the 290 patients in whom full immunophenotypic studies were done. Minimal residual disease was studied in 158 (95·2%) patients; of the remaining eight patients, five did not attain complete clinical remissions within the planned 6-week period of induction chemotherapy, and three had inadequate samples for flow-cytometric studies. The presenting features of the patients with leukaemia-associated immunophenotypes were not significantly different from those of the remaining patients, with the exception of a higher proportion of boys (65·1 vs 49·2%, p<0·01), MLL gene rearrangements (9·6 vs 1·6%, p<0·01) and T-cell phenotype (28·9 vs 0·8%, p<0·001). The 3year cumulative incidence of relapse was 11·3% (SE 3·0) in patients with leukaemia-associated phenotypes and 4·8% (2·4) in the remaining patients (p=0·04). All procedures were approved by the St Jude Institutional Review Board, with informed consent obtained from the parents or guardians of each child. Diagnostic immunophenotyping and chromosomal and genetic analysis were done with standard techniques. After initial treatment with methotrexate or mercaptopurine, or both, all patients received 6 weeks of remission-induction chemotherapy with prednisone, vincristine, daunorubicin, Lasparaginase, teniposide, and cytarabine.17,18 After they attained a complete clinical remission, patients received 2 weeks of consolidation therapy with high-dose methotrexate and mercaptopurine, followed by risk-directed continuation therapy. Lower-risk cases were treated with mercaptopurine and methotrexate for 120 weeks, with a pulse of prednisone or dexamethasone plus vincristine every 4 weeks. For patients at high-risk of relapse, continuation therapy consisted of alternating pairs of drugs: etoposide plus cyclophosphamide, mercaptopurine plus methotrexate, methotrexate or etoposide plus cytarabine, or prednisone or dexamethasone plus vincristine. All patients received intensive age-adjusted doses of triple intrathecal therapy with methotrexate, hydrocortisone, and cytarabine for 1 year; cranial irradiation was reserved for patients at very high-risk of relapse (18 Gy) or those with central-nervous-system status three at diagnosis (24 Gy) at 1 year of continuation therapy. Six patients judged to have very high-risk ALL underwent allogeneic bone-marrow transplantation.
CD13, and CD33 (Dako, Carpinteria), anti-IgM (Southern Biotechnology Assoc, Birmingham, AL, USA), KORSA3544 (a gift from Dr Mori, Saitama, Japan), and CD21 (Immunotech, Westbrook, Maine). Isotype-matched non-reactive fluorochrome-conjugated antibodies were used as controls. The staining procedure has been described.2,19,20 For intracellular staining, cells were treated with Permeafix (Ortho Diagnostic Systems, Raritan).2 A FACScan flow cytometer with Lysis II or Cell Quest software (Becton Dickinson) was used in all analyses. We defined two gates after analysing 10 000 cells. Gate 1 surrounded the light-scatter dot plot that represented lymphoid cells; this area included all leukaemic cells present at diagnosis in the cases studied. Gate 2 identified the cell population expressing the immunophenotypic features of immature lymphoid cells (eg, CD19 and CD34). Electronic events falling within both gates 1 and 2 were then selectively recorded. Acquisition was terminated when virtually all cells in the tube (eg, 2–5 x 105) had passed through the flow cytometer or 10 000 events that represent immature lymphoid cells were recorded, whichever occurred first. The expression of leukaemiaassociated immunophenotypes among the selected population was then analysed. Flow-cytometric data were recorded within 24 h after sample collection and processing, with no knowledge of patients clinical status or diagnostic variables, except immunophenotype. The marker combinations used in this study were selected because they are not expressed (or expressed extremely weakly) in normal or regenerating bone-marrow cells.2 Thus, in preliminary experiments with artificial mixtures of leukaemic and normal cells, we could consistently detect one leukaemic cell per 104 normal nucleated bone-marrow cells with each antibody combination used in the study, and correctly determine different ratios of leukaemic-to-normal cells.2,20 To assess the precision of the assay, we compared results of multiple measurements of residual leukaemia in identical cell preparations, in 23 tests with mixtures containing one leukaemic cell in 104. The coefficient of variation was 15%. In 22 tests with mixtures containing one leukaemic cell in 10 3, the coefficient of variation was 10%.
Statistical analyses Distributions of presenting features according to the degree of residual disease during clinical remission were compared by Fisher’s exact test. The probability of surviving without ALL relapse or without an adverse event (ie, relapse or death due to any cause) was estimated by the Kaplan-Meier method, applied to follow-up observations through July 8, 1997; associated SEs were calculated by the Peto method. Follow-up information was gathered within 6 months of the analysis for 96% of patients. The prognostic importance of the levels of residual disease was determined first by analyses stratified by treatment only, and then by stratified analyses with competing covariates. Comparisons of continuous complete-remission distributions were made by the Mantel-Haenszel test and the exact log-rank test, as necessary. Cumulative incidence functions of leukaemia
Flow cytometry
Marker combinations
Bone-marrow aspirates were collected in preservative-free heparin after 6 weeks of induction therapy; during weeks 14, 32, and 56 of continuation therapy; and at the end of therapy (week 120). Cells with leukaemia-associated immunophenotypes (ie, those found on leukaemic cells but not on normal bone-marrow cells) were investigated by multiparameter flow-cytometry, with various combinations of monoclonal and polyclonal antibodies against surface, cytoplasmic or nuclear leucocyte antigens (table 1). The following antibodies, conjugated to fluorescein isothiocyanate, phycoerythrin, peridinin chlorophyll protein, or biotin, were used: anti-terminal deoxynucleotidyl transferase (TdT; Supertechs, Bethesda); CD34 and CD56 (Becton Dickinson, San Jose); CD10 (Caltag, S San Francisco); CD19,
Early lymphoid
Leukaemia-associated
TdT TdT and CD34 CD10 CD19 and CD34 or CD10 CD19 and CD34 or CD10 CD19 and CD34 or CD10 CD19 and CD34 CD34 and CD10
CD3 (cytoplasmic or surface) IgM (cytoplasmic) KORSA3544 CD13 CD33 CD65 CD21 CD56
THE LANCET • Vol 351 • February 21, 1998
Frequency*
27·8% 21·5% 20·9% 20·2% 11·4% 10·8% 8·9% 5·1%
*Percentage of cases in this series expressing each marker at diagnosis (>50% positive cells). 42 (26·6%) of 158 cases studied had more than one leukaemiaassociated marker.
Table 1: Leukaemia markers used to study minimal residual disease by double or triple colour flow-cytometry
551
ARTICLES
relapse were estimated as described by Kalbfleisch and Prentice, and compared by Gray’s test. For these calculations, missing residual disease determinations during continuation therapy (4% of all tests) were imputed when the degree of residual disease at earlier and later tests were identical.
Results We identified cells with leukaemia-associated immunophenotypes in 23·4% of the bone marrow samples taken at the end of remission induction (table 2). The proportion of samples with residual disease remained relatively high (17·1%) at week 14 of continuation therapy, but had decreased strikingly (to 5·3%) by week 32, with an additional decrease to 3·9% apparent at 56. None of the 65 patients studied at the completion of therapy had detectable leukaemic blast cells. The percentage of leukaemic cells among mononucleated bone-marrow cells ranged from 0·01% to 19%, with a median of about 0·1% in all positive samples. Nine of the samples had more than 1% of leukaemic cells. An additional two had more than 10% leukaemic cells by flow cytometry. In both of these samples, two expert observers could not identify leukaemic lymphoblasts by morphology; in one sample, all lymphoid cells appeared to be normal differentiated lymphocytes, while in the other, 3·5% of cells had immature undifferentiated appearance and were regarded to be normal haematopoietic progenitor cells (haematogones). Rates of residual disease detection did not differ significantly in comparisons based on sex, age, race, leucocyte count, central-nervous-system status, immunophenotype or leukaemic cell ploidy. There were, however, striking differences in subgroups of patients with exceptionally poor prognoses.1 For example, at the end of remission-induction chemotherapy, residual disease was detected in all eight patients with the Philadelphia chromosome (p<0·0001) and in six of 12 cases with MLL gene rearrangements (p=0·033). By contrast, only one of 12 patients with rearrangements of the TEL gene and four of 29 with hyperdiploidy (51–65 chromosomes), both of which are favourable prognostic signs,1 had detectable levels of residual disease at the end of remission-induction chemotherapy. At each time point, the detection of minimal residual disease was significantly associated with a greater likelihood of treatment failure and of leukaemic relapse (p<0·002, table 3). In the group tested at the end of induction therapy, 3-year estimates of continuous complete remission (SE) were 68·5% (14·5) for patients with positive findings and 89·8% (5·9) for those with negative findings. Three of the seven failures in the subgroup without evidence of residual leukaemia were due to causes other than ALL relapse (infection or secondary leukaemia). Moreover, one of the four patients with negative findings who relapsed had detectable leukaemic cells in subsequent testing, while in two other patients the resurgent blast cells lacked the leukaemia-specific markers observed at diagnosis. By contrast, all eight failures in the positive group were ALL relapses, with cells retaining leukaemia-associated markers at the time of relapse. The value of residual leukaemia detected at the time of remission induction to predict subsequent relapse remained significant after adjustment for established prognostic factors, such as age younger than 1 year and older than 9 years (p<0·001), leucocyte count of more than 50⫻109/L
552
Time of testing
Number tested
Number with residual disease
Median % of leukaemia cells (lowest and (highest value)
End of remission induction Continuation therapy Week 14 Week 32 Week 56 End of therapy
128
30 (23·4%)
0·13 (0·01–19·0)
105 114 103 65
18 (17·1%) 6 (5·3%) 4 (3·9%) 0
0·02 (0·01–14·9) 0·10 (0·01–0·44) 0·03 (0·01–0·06) ..
Table 2: Detection of residual disease with immunological markers in children with ALL in clinical remission
(p<0·001) and presence of adverse genetic features (Philadelphia chromosome and MLL gene rearrangements, p<0·006). Because previous studies with PCR had indicated that patients with levels of residual disease higher than one in 1000 normal cells on the date of complete remission were at a significantly higher risk of relapse,3–5 we reanalysed our data after segregating positive cases into two groups according to levels of residual disease (<0·1% and ⭓0·1%). The risk of treatment failure or leukaemia relapse were not significantly related to the proportion of leukaemic cells (data not shown). Detection of minimal residual disease during continuation chemotherapy was also strongly predictive of leukaemic relapse (p<0·002, table 3). At week 14, the cumulative incidence of leukaemic relapse after 3 years of the test was 6·6% (3·5) for patients with negative findings and 42·1% (14·6) for those with detectable disease. The prognostic value of the flow-cytometric assay remained significant after adjustment for competing covariates, including age (p<0·001), leucocyte count (p<0·001), and adverse genetic features (p=0·04). Even after exclusion of patients with Philadelphia chromosome and those with MLL gene rearrangements from the analysis, detection of minimal residual disease remained a significant predictor of leukaemia relapse (p= 0·015). Of note, among the seven patients with MLL gene rearrangements, only the two with positive findings subsequently relapsed. Relapse also occurred in two of the four patients with hyperdiploid leukaemia who had positive findings at week 14. At this time, measurements of residual disease levels identified patients at a very high risk of relapse (figure). Indeed, four of five patients with higher levels of Time of testing
End of remission induction MRD positive MRD negative Continuation therapy Week 14 MRD positive MRD negative Week 32 MRD positive MRD negative Week 56 MRD positive MRD negative
Continuous complete remission(SE)*
p†
Cumulative incidence of leukaemia relapse (SE)*
p†
68·5 (14·5) 89·8 (5·9)
<0·002
32·5 (10·6) 7·5 (4·0)
<0·001
58·0 (21·7) 88·3 (6·0)
<0·001
42·1 (14·6) 6·6 (3·5)
<0·001
25·0 (15·3) 83·5 (6·8)
<0·001
75·0 (27·5) 8·0 (3·8)
<0·001
25·0 (15·3) 86·6 (7·7)
<0·001
50·0 (32·5) 6·6 (3·7)
<0·002
MRD=minimal residual disease. *Estimated percentages at 3 years after each sampling. †p based on log rank and Gray’s test, respectively.
Table 3: Continuous complete remission and cumulative incidence of leukaemic relapse according to detection of residual disease
THE LANCET • Vol 351 • February 21, 1998
Cumulative risk of relapse (%)
ARTICLES
100
p<0·001
Positive (⭓0·1%)
80 60
60 (26)% 35 (18)%
40 20
Positive (⬍0·1%)
7 (4)%
Negative
0 0
1 2 3 4 5 Time after week 14 of continuation therapy (years)
Kaplan-Meier estimates of cumulative incidence of relapse of ALL in children Estimates are according to levels of residual disease in bone-marrow samples collected on week 14 of continuation therapy. Numbers in parentheses are SEs.
leukaemia infiltration (⭓0·1%) have subsequently relapsed. The remaining patient has not yet completed therapy but has retained more than 0·1% leukaemic lymphoblasts in four follow-up assays (through week 79). Of the 65 children studied at the end of treatment, none had detectable residual disease. However, four eventually relapsed, at 7 to 15 months after the negative test. One of these patients had positive findings throughout the first year of therapy (up to week 56 of continuation therapy), while two had phenotypic switches resulting in loss of leukaemia-specific markers.
Discussion The lack of prospective studies in large cohorts of similarly treated patients has hampered efforts to assess the clinical significance of minimal residual disease in children with ALL who are in complete clinical remission. Our main conclusion is that the presence or absence of detectable residual leukaemia during clinical remission strongly influences treatment outcome, suggesting that flow-cytometric assays could be used to augment current methods of risk assessment. We found that detection of residual disease at any of four different intervals during treatment was associated with a high likelihood of subsequent relapse. The extent of marrow infiltration by leukaemic cells closely predicted treatment outcome only when measurements were taken from samples from week 14. At points beyond 6 months of continuation therapy, any amount of residual disease carried a very poor prognosis. The flow-cytometric techniques we used are well suited for clinical investigations of residual disease because of their rapidity and wide availability.2,21,22 As demonstrated here, they reliably discriminate between leukaemic and normal cells, even when the former cells assume an apparently normal morphology. This point is well illustrated by our two patients with greater than 10% leukaemic cells disguised as differentiated lymphocytes or haematogones. Both patients had a relapse of leukaemia 4 months and 13 months after the positive immunological observation. Although the leukaemia-associated phenotypes used as targets in the flow-cytometric assay were expressed in only 57% of all cases studied at diagnosis, this percentage is likely to increase with the continuous development of new antibodies against leucocyte
THE LANCET • Vol 351 • February 21, 1998
antigens. For example, monoclonal antibodies reacting specifically with cells carrying E2A-PBX1 or MLL gene rearrangements are now available for flow-cytometric studies of residual disease,23,24 and antibodies to molecules overexpressed in leukaemic lymphoblasts, such as WT1 and BCL2, have shown promise in preliminary experiments.25,26 These reagents may well allow all children with ALL to be monitored for residual leukaemia by flow cytometry. In any case, the antibody combinations used in our analysis were applicable to most high-risk cases, including those with the Philadelphia chromosome, MLL gene rearrangements, or a T-cell immunophenotype, as these subtypes of ALL regularly express leukaemia-associated immunophenotypes.2 The value of both immunological and molecular techniques in monitoring residual leukaemia may be limited by evolution of the leukaemic clone during the course of the disease.2,14–16 Indeed, in our series, failure to detect impending relapse in three patients was associated with the loss of leukaemia-associated immunophenotypes in the resurgent clone. However, in all three cases, only one combination suitable for monitoring was detected at diagnosis, suggesting that the addition of new markers to the current arsenal should significantly reduce or eliminate the occurrence of false-negative results due to immunophenotypic switches. Four of 65 patients in our study relapsed 7–15 months later even though they lacked detectable leukaemic cells at the end of therapy. Persistent leukaemic cells in these patients might have been detected by a more sensitive technique.12 However, in the study by Ito and colleagues,27 PCR amplification of IgH genes, purported to be about one log more sensitive then immunological techniques, failed to identify residual disease in eight of nine patients who relapsed after completion of therapy. Thus, the absence of residual disease at the end of therapy by either immunological or molecular criteria does not guarantee a durable remission. In summary, immunological investigation of minimal residual disease provides clinically meaningful information that could be used to improve risk assessment strategies and treatment selection in the management of ALL in children. We suggest that patients with detectable leukaemic cells at the end of remission induction should be closely monitored for disease resurgence. Alternative treatment should be considered for patients with persistent disease during early continuation therapy. Contributors Elaine Coustan-Smith was responsible for the flow-cytometric analysis of residual disease and for the overall study coordination. Frederick Behm, Susana Raimondi, and Jeffrey Rubnitz provided the information on presenting immunophenotypic, karyotypic, and genetic features. Joaquin Sanchez contributed to the flow-cytometric studies. James Boyett and Michael Hancock undertook the statistical design and analysis of the results. Gaston Rivera and Torrey Sandlund contributed to the protocol design and the clinical conduct of the study. Ching-Hon Pui was responsible for the overall clinical conduct of the study and contributed to data analysis and preparation of the paper. Dario Campana was responsible for the concept of the study, data analysis, and with advice from the other authors, preparation of the paper.
Acknowledgments This work was supported by grants CA60419, CA21765, and CA20180 from the National Cancer Institute, and by the American Lebanese Syrian Associated Charities (ALSAC). JS was the recipient of grant 97/5486 from the Fondo de Investigacion Sanitaria of Spain. We thank Taijiro Mori for the gift of KORSA3544 and John Gilbert for critical review of the manuscript.
553
ARTICLES
References 1 2
3
4
5
6
7
8
9
10
11
12
13 14
Pui C-H. Childhood leukemias. N Engl J Med 1995; 332: 1618–30. Campana D, Pui C-H. Detection of minimal residual disease in acute leukemia: methodological advances and clinical significance. Blood 1995; 85: 1416–34. Wasserman R, Galili N, Ito Y, et al. Residual disease at the end of the induction therapy as a predictor of relapse during therapy in childhood B-lineage acute lymphoblastic leukemia. J Clin Oncol 1992; 10: 1879–85. Brisco MJ, Condon J, Hughes E, et al. Outcome prediction in childhood acute lymphoblastic leukaemia by molecular quantification of residual disease at the end of induction. Lancet 1994; 343: 196–200. Jacquy C, Delepaut B, Van Daele S, et al. A prospective study of minimal residual disease in childhood B-lineage acute lymphoblastic leukaemia: MRD level at the end of induction is a strong predictive factor of relapse. Br J Haematol 1997; 98: 140–46. Neale GAM, Menarguez J, Kitchingman GR, et al. Detection of minimal residual disease in T-cell acute lymphoblastic leukemia using polymerase chain reaction predicts impending relapse. Blood 1991; 78: 739–45. Biondi A, Yokota S, Hansen-Hagge TE, et al. Minimal residual disease in childhood acute lymphoblastic leukemia: analysis of patients in continuous complete remission or with consecutive relapse. Leukemia 1992; 6: 282–88. Nizet Y, Van Daele S, Lewalle P, et al. Long-term follow-up of residual disease in acute lymphoblastic leukemia patients in complete remission using clonogeneic IgH probes and the polymerase chain reaction. Blood 1993; 82: 1618–25. Kitchingman GR. Residual disease detection in multiple follow-up samples in children with acute lymphoblastic leukemia. Leukemia 1994; 8: 395–402. Cave H, Guidal P, Rohrlich P, et al. Prospective monitoring and quantitation of residual blasts in childhood acute lymphoblastic leukemia by polymerase chain reaction study of omega and gamma T-cell receptor genes. Blood 1994; 83: 1892–901. Seriu T, Yokota S, Nakao M, et al. Prospective monitoring of minimal residual disease during the course of chemotherapy in patients with acute lymphoblastic leukemia, and detection of contaminating tumor cells in peripheral blood stem cells for autotransplantation. Leukemia 1995; 9: 615–23. Roberts WM, Estrov Z, Ouspenskaia MV, Johnston DA, McClain KL, Zipf TF. Measurement of residual leukemia during remission in childhood acute lymphoblastic leukemia. N Engl J Med 1997; 336: 317–23. Cross NCP. Quantitative PCR techniques and applications. Br J Haematol 1995; 89: 693–97. Beishuizen A, Verhoeven M-AJ, van Wering ER, Hahlen K, Hooijkaas H, van Dongen JJM. Analysis of Ig and T-cell receptor genes in 40 childhood acute lymphoblastic leukemias at diagnosis and subsequent relapse: implications for the detection of minimal residual
554
15
16
17
18
19
20
21
22
23
24
25
26
27
disease by polymerase chain reaction analysis. Blood 1994; 83: 2238–47. Steward CG, Goulden NJ, Katz F, et al. A polymerase chain reaction study of the stability of Ig heavy-chain and T-cell receptor omega gene rearrangements between presentation and relapse of childhood B-lineage acute lymphoblastic leukemia. Blood 1994; 83: 1355–62. Steenbergen EJ, Verhagen OJHM, Van Leeuwen EF, et al. Frequent ongoing TCR delta rearrangements in childhood precursor acute lymphoblastic leukemia: implications for monitoring minimal residual disease. Blood 1995; 86: 692–702. Pui C-H, Boyett JM, Hughes WT, et al. Human granulocyte colonystimulating factor after induction chemotherapy in children with acute lymphoblastic leukemia. N Engl J Med 1997; 336: 1781–87. Pui C-H, Rivera GK, Hancock ML, et al. Risk-adapted treatment for acute lymphoblastic leukemia: findings from St Jude Children’s Research Hospital. In: Buchner T, Hiddeman W, Wormann B, Schellong G, Ritter J, Crentzig U, eds. Acute leukemia VI: prognostic factors and treatment strategies; haematology and blood transfusions. Berlin:Springer-Verlag, 1997: 629–37. Campana D, Coustan-Smith E, Janossy G. The immunological detection of minimal residual disease in acute leukemia. Blood 1990; 76: 163–69. Coustan-Smith E, Behm FG, Hurwitz CA, et al. N-CAM (CD56) expression by CD34+ malignant myeloblasts has implications for minimal residual disease detection in acute myeloid leukemia. Leukemia 1993; 7: 853–58. Orfao A, Ciudad J, Lopez-Berges MC, et al. Acute lymphoblastic leukemia (ALL): detection of minimal residual disease (MRD) by flow cytometry. Leuk Lymph 1994; 13: 87–90. Vervoordeldonk SF, Merle PA, Behrendt H, et al. Triple immunofluorescence staining for prediction of relapse in childhood B acute lymphoblastic leukemia. Br J Haematol 1996; 92: 922–28. Sang BC, Shi L, Dias P, et al. Monoclonal antibodies specific to the acute lymphoblastic leukemia t(1;19)-associated E2A/pbx1 chimeric protein: characterization and diagnostic utility. Blood 1997; 89: 2909–14. Behm FG, Smith FO, Raimondi SC, et al. Human homologue of the rat chondroitin sulfate proteoglycan, NG2, detected by monoclonal antibody 7.1, identifies childhood acute lymphoblastic leukemia with t(4;11)(q21;q23) or t(11;19)(q23;p13) and MLL gene rearrangements. Blood 1996; 87: 1134–39. Menssen HD, Renkl HJ, Rodeck U, et al. Presence of Wilms’ tumor gene (wt1) transcripts and the WT1 nuclear protein in the majority of human acute leukemias. Leukemia 1995; 9: 1060–67. Coustan-Smith E, Kitanaka A, Pui C-H, et al. Clinical relevance of BCL-2 overexpression in childhood acute lymphoblastic leukemia. Blood 1996; 87: 1140–46. Ito Y, Wasserman R, Galili N, et al. Molecular residual disease status at the end of chemotherapy fails to predict subsequent relapse in children with B-lineage acute lymphoblastic leukemia. J Clin Oncol 1993; 11: 546–50.
THE LANCET • Vol 351 • February 21, 1998