|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2003 American Society for Clinical Oncology Importance of Minimal Residual Disease Testing During the Second Year of Therapy for Children With Acute Lymphoblastic Leukemia
From the Children’s Cancer Institute Australia for Medical Research, Sydney Children’s Hospital Randwick, and Children’s Hospital at Westmead, Sydney; Flinder’s Medical Centre, Adelaide; Royal Children’s Hospital, Melbourne, Australia; and the Australian and New Zealand Children’s Cancer Study Group. Address reprint requests to Murray Norris, PhD, Children’s Cancer Institute Australia for Medical Research, PO Box 81, Randwick, 2031, Sydney, Australia; email: m.norris{at}unsw.edu.au
Purpose: A high level of minimal residual disease (MRD) after induction chemotherapy in children with acute lymphoblastic leukemia (ALL) is an indicator of relative chemotherapy resistance and a risk factor for relapse. However, the significance of MRD in the second year of therapy is unclear. Moreover, it is unknown whether treatment intervention can alter outcome in patients with detectable MRD. Patients and Methods: We assessed the prognostic value of MRD testing in bone marrow samples from 85 children at 1, 12, and 24 months from diagnosis using clone-specific polymerase chain reaction primers designed to detect clonal antigen receptor gene rearrangements. These children were part of a multicenter, randomized clinical trial, which, in the second year of treatment, compared a 2-month reinduction-reintensification followed by maintenance chemotherapy with standard maintenance chemotherapy alone. Results: MRD was detected in 69% of patients at 1 month, 25% at 12 months, and 28% at 24 months from diagnosis. By univariate analysis, high levels of MRD at 1 month, or the presence of any detectable MRD at 12 or 24 months from diagnosis, were highly predictive of relapse. Multivariate analysis showed that MRD testing at 1 and 24 months each had independent prognostic significance. Intensified therapy at 12 months from diagnosis did not improve prognosis in those patients who were MRD positive at 12 months from diagnosis. Conclusion: Clinical outcome in childhood ALL can be predicted with high accuracy by combining the results of MRD testing at 1 and 24 months from diagnosis.
THE USE of multiagent chemotherapy has markedly increased the relapse-free survival rate for children with acute lymphoblastic leukemia (ALL).1–3 Despite this success, however, relapse remains a continuing problem for 20% to 30% of patients. We and others4–6 have shown that high levels of minimal residual disease (MRD) detected by polymerase chain reaction (PCR) at the end of induction chemotherapy signifies a relative resistance to chemotherapy and a high risk of relapse. However, high levels of residual disease at this time identify only 30% to 50% of children destined to relapse, and the majority of these relapses occur early, in the first 18 months of therapy.5,6 Because relapse of ALL occurs most frequently in the third and subsequent years after diagnosis, a more effective method to identify these patients is required. Furthermore, the high incidence of therapy-related side effects in children treated with some ALL protocols has indicated a need for tests with both high positive and high negative predictive value for relapse, which can also be used to reduce treatment intensity and thus the side effects of therapy in low-risk patients.7,8 The clinical significance of persistent residual disease in the second year of therapy for childhood ALL is unclear.6,9–11 In the first large-scale study of MRD in the second year of therapy, van Dongen et al6 recently found persistent MRD in only 4% of 154 patients 24 months after diagnosis. These patients, although few in number, had a high risk of relapse. In contrast, Roberts et al12 used a more sensitive PCR-based technique and demonstrated persistent low levels of MRD in more than 80% of a small group of childhood ALL patients, many of whom remained in long-term remission. In this study, relapsing patients were all distinguished by gradually increasing levels of MRD 6 to 12 months before clinical relapse. These differing findings have led to uncertainty about the value of measuring MRD in the second year of therapy for childhood ALL. The principle that ALL patients with high-risk, chemotherapy-resistant disease can be identified early in the disease course, and then successfully treated by intensification of therapy, has been demonstrated.13,14 However, the impact of intensified chemotherapy on patients with persistent MRD in first-remission ALL has not been assessed. In this study, we compared the prognostic value of MRD detected in bone marrow at 1, 12, and 24 months from diagnosis in first-remission childhood ALL, and assessed the impact of intensified therapy at 12 months from diagnosis in patients with persistent MRD.
Patients and Bone Marrow Samples A total of 597 children with newly diagnosed ALL were consecutively enrolled on the Australia and New Zealand Children’s Cancer Study Group ALL Study VI Protocol. Children older than 18 years or younger than 1 year were considered ineligible. Patients who were not treated according to protocol, who did not achieve remission at the end of the induction, or who had t(9;22) evident in the diagnostic bone marrow sample were also ineligible and stratified to alternative treatment protocols. Eighty-five patients were included in the MRD analysis and followed for the presence of PCR-detectable MRD throughout the course of their therapy. These 85 patients met the criteria of having bone marrow samples available from both the end-of-induction time point and from at least one later time point during therapy, and having at least one rearrangement of the immunoglobulin heavy chain (IgH) or T-cell receptor gamma (TCR  ) gene loci at diagnosis that exhibited the required specificity and sensitivity (< 10-3) when used for MRD detection.
Treatment
MRD Testing
MRD testing of remission bone marrow samples collected 1 month after diagnosis was performed with primers specific for IgH or TCR
Study Design and Statistical Analysis The samples were collected prospectively and stored. PCR results were not used to determine therapy for any patient, and the treating oncologists, patients, and parents were blinded to the MRD results. Laboratory staff that performed the MRD testing was blinded to the clinical outcome of the patients. The diagnosis and remission bone marrow samples were analyzed at the Children’s Cancer Institute, Sydney, Australia, and Flinder’s Medical Centre, Adelaide, Australia.
Relapse-free survival analysis was performed according to the method of Kaplan and Meier, and comparisons of outcome between subgroups were performed by the log-rank test for univariate comparisons, using two-tailed tests. Associations between baseline clinical characteristics of patients in the MRD group and the overall study cohort were made using
Clinical Characteristics In terms of clinical and laboratory characteristics at diagnosis, or for clinical outcomes, there were no significant differences between the cohort of 85 children who were monitored for MRD and the overall group of 597 children treated on ALL Study VI (Table 1  ). The 5-year relapse-free survival rates for the MRD study group and for the overall ALL Study VI group were 60% and 64%, respectively, which did not differ significantly (P = .99). The median follow-up for those patients in continuous clinical remission was 55 months and 56 months for the MRD and overall Study VI groups, respectively. Cumulative relapse-free survival was analyzed in the MRD study group according to well-established prognostic variables. High total WBC count at diagnosis was associated with a significantly increased relapse rate in comparison with children with a low total WBC count (P < .05), whereas patients defined clinically as high risk at time of diagnosis similarly had a significantly increased relapse rate (P < .05) compared with low-risk patients. However, patient age at diagnosis failed to achieve statistical significance as a predictor of relapse (P = .48).
MRD and Outcome MRD was detected in 69% of patients at the end of induction therapy (1 month), in 25% of patients at the point of treatment randomization (12 months), and in 28% of patients at the end of therapy (24 months). The level of residual leukemic blasts in the bone marrow at the end of 1 month of induction therapy was quantitated using real-time PCR and found to be strongly associated with risk of relapse (Fig 1 ). However, a high level of MRD at 1 month had reduced predictive value for relapse as the patient group was subdivided around a lower MRD threshold (Fig 1). The presence of any detectable MRD in marrow samples taken at either 12 or 24 months after diagnosis was also a highly significant predictor of relapse (Fig 2).
The value of detecting residual disease was analyzed in the subgroup of patients who, on the basis of presenting clinical features, were not classified as being at high risk of relapse at diagnosis. For patients with a low or average risk of relapse on the basis of clinical criteria (Table 1  ), a high level of residual leukemic blasts (> 10-3) 1 month after diagnosis was associated with significantly reduced relapse-free survival in comparison with a low level of residual blasts, with a relative hazard of 9.8 (95% CI, 3.4 to 27.8). Similarly, the presence of detectable MRD at either 12 or 24 months predicted a poor outcome in this group of patients, with relative hazards of 3.3 (95% CI, 1.5 to 7.5) and 5.5 (95% CI, 2.3 to 13.1), respectively.
Improved prognostic significance was achieved when the results of MRD assays at different time points were combined. In Fig 3
The increased predictive value provided by combining the 1- and 24-month assay results suggested that these two time points were identifying different subgroups of patients with a poor prognosis. To determine whether detection of MRD during the second year of therapy yielded prognostic information independent of testing at end of induction, multivariate analyses were performed. When outcome was adjusted for the effect of a high level of detectable MRD at 1 month (> 10-3) after diagnosis, the presence of detectable MRD at 24 months remained a significant indicator of poor relapse-free survival (relative hazard, 2.8; 95% CI, 1.2 to 6.4) (Table 2 ). In this analysis, the relative hazard associated with a high level of detectable MRD at 1 month was 4.9 (95% CI, 2.1 to 11.4). In a separate analysis, the relation between a high level of detectable MRD at 1 month and outcome was adjusted for the effect of detectable disease at 12 months after diagnosis. Although a high level of detectable MRD at 1 month remained a significant indicator of poor relapse-free survival (relative hazard, 5.9; 95% CI, 2.6 to 13.3), the presence of detectable disease at 12 months had no added prognostic value. When three established prognostic indicators (WBC count, clinical risk category, and age at diagnosis) were combined with MRD testing at 1 and 24 months as variables in the Cox regression model, MRD detection at 1 and 24 months each remained significant indicators of poor relapse-free survival (relative hazard, 5.3; 95% CI, 2.1 to 13.7 and relative hazard, 2.6; 95% CI, 1.1 to 6.1, respectively) (Table 2).
Patients in the overall study group were assigned at 12 months from diagnosis to either standard-risk therapy (arm A) or to a more intensive treatment arm (arm B). Intensification of chemotherapy at 12 months from diagnosis did not significantly improve the outcome for the ALL Study VI group, for which the relapse-free survival rates were 71% for arm A and 66% for arm B (N = 597, P = .23). Similarly, there was no significant difference in relapse-free survival rates for arms A and B for the MRD group (62% and 63%, respectively [n = 82, P = .61]), or for those children who were MRD positive at 12 months from diagnosis (41% and 27%, respectively [n = 20, P = .93]).
This study demonstrates that the detection of residual leukemia in bone marrow samples taken early in (1 month), during (12 months), and at the end of therapy (24 months) is highly predictive of outcome. The association between MRD and reduced relapse-free survival was also evident in those children classified clinically as low risk, and it was independent of standard risk factors identified at diagnosis. In particular, the importance of measuring MRD not only at the end of induction but also at the end of therapy was highlighted by the independent prognostic significance of these two measurements. Thus, testing for the presence of MRD in the second year of therapy identified a cohort of children who relapsed after the completion of therapy and whose poor prognosis was not identified by testing for residual disease at the end of induction. This is the first report of significant clinical information being afforded by residual leukemia detection at 24 months from diagnosis, and the results have major implications for the biology, prognosis, and treatment of childhood ALL. The finding that MRD assays at 1 and 24 months after diagnosis are independently predictive of outcome suggests the existence of distinct leukemic populations exhibiting either intrinsic or acquired chemotherapy resistance. Quantitation of the leukemia burden immediately after induction therapy essentially provides an in vivo measure of the inherent chemosensitivity of the leukemic clone.19 However, poor early response does not explain all relapses. A comparison of clinical and MRD-based risk stratification performed on a cohort of patients from the German Berlin-Frankfurt-Munster 90 ALL protocol showed that 47% of relapses would not have been predicted by their poor chemosensitivity, as measured by MRD, at 1 and 3 months from diagnosis.20 The results of our study suggest that acquired resistance to therapy may also be important in ultimately determining a cure. We found a higher incidence of MRD positivity at the 1-, 12-, and 24-month time points than has been reported in other studies.4–6 These differences may be explained by the lower overall relapse-free survival rate as well as the number of mononuclear cells examined in our study. Thus, at the 12- and 24-month time points in the present study, five replicate PCR reactions were routinely performed, each containing 2 µg of DNA and representing a total number of cells in excess of 2 million. This amount of DNA is approximately 10-fold higher than that used in most MRD studies, which have routinely examined only 1 µg of DNA in a single PCR reaction.5,6 In our study, only two of 22 (9%) patients with detectable MRD at 24 months from diagnosis demonstrated positivity in all five replicate DNA samples from the 24-month remission bone marrow, indicating that the analysis of only 1 µg of DNA would have yielded considerably fewer positive MRD results. These data highlight the importance of examining at least 10 µg of DNA at each time point to afford a greater degree of sensitivity in detecting MRD. Certainly, a level of sensitivity of 10-6 can only validly be achieved if greater than 106 genomes are examined. It is of interest that the prospective study of Roberts et al12 also examined 10 µg of bone marrow DNA from each patient for the presence of MRD, and, using a highly sensitive technique, these investigators demonstrated PCR-detectable residual disease in 22 of 24 ALL patients examined after the completion of therapy. The higher rates of MRD positivity at 12 and 24 months may have also been affected by the higher overall relapse rate in our study population. Collectively, these results suggest that the PCR methodology, the number of remission marrow cells being examined, and the overall relapse rate are critical factors in determining the incidence of MRD positivity. In addition, the data support the hypothesis that there may be a biologically relevant concentration at which persistent MRD correlates with high risk of relapse and a level below which relapse is unusual. The ability to monitor residual disease and define molecular relapse during remission with a high degree of sensitivity and specificity suggests that the results of MRD assays may be used to alter therapy. However, in our study, a short, 9-week intensification of chemotherapy at 12 months from diagnosis, with similar drug doses and scheduling to that used to induce and consolidate initial remission, was not a successful strategy for preventing relapse in patients who were MRD positive at 12 months from diagnosis. Although MRD testing at three time points throughout the 2-year duration of therapy can identify patients with molecular evidence of chemotherapy-resistant disease well before it is clinically evident, the results of our study suggest that more aggressive treatment, such as allogeneic bone marrow transplant for selected patients with persistent MRD, may be necessary to achieve molecular remission. Treatment intensification in the first 6 months after diagnosis, on the basis of the result of the day 7 marrow response to chemotherapy, improved survival rates in some high-risk ALL patients.14 It is unclear whether the arm B treatment intensification in our study failed as a result of the timing of the intervention or the nature of the drugs used. The relapse-free survival rate in our overall ALL Study VI group of 64% was inferior to that of some reports and equivalent to others.21 However, the representativeness of the study cohort to patients with leukemia in general, in terms of both well-established prognostic indicators and predictive value of MRD testing at the end of induction, confirms the comparability of our study to other studies that evaluated the role of MRD testing in ALL. Indeed, our results suggest a strategy for identifying the significant proportion of patients who are at risk of relapse but who are not detected by early MRD testing.20 As with other factors predicting relapse in children with ALL, the prognostic significance of MRD testing in the second year of therapy will need to be further evaluated in the context of future chemotherapy protocols; in particular, those studies in which therapy is intensified on the basis of the early MRD levels. In conclusion, our results demonstrate that assaying MRD at 1 and 24 months from diagnosis in childhood ALL can identify almost all patients who will relapse either during or after the completion of therapy. The detection of residual disease at the end of therapy provides additional prognostic information independent of that obtained at end of induction, thereby identifying a subgroup of children at high risk of relapse. Improved outcome is most likely to result from the use of MRD assays immediately after induction as a method of targeting therapy to individual patients, combined with the development of novel treatment strategies for children with persistent MRD at the completion of current therapeutic protocols.
We thank Dianne Tucker, MBBS, Ashutosh Lal, MBBS, and Scott Williams for their administrative, clinical, or technical support of this project.
Supported by research grants from the National Health and Medical Research Council, Government Employees Medical Research Fund, and the New South Wales Cancer Council, Australia.
1. Rivera GK, Pinkel D, Simone JV, et al: Treatment of acute lymphoblastic leukemia: 30 years’ experience at St. Jude Children’s Research Hospital. N Engl J Med 329:1289–1295, 1993
2. Pui CH, Evans WE: Acute lymphoblastic leukemia. N Engl J Med 339:605–615, 1998 3. Smith M, Arthur D, Camitta B, et al: Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. J Clin Oncol 14:18–24, 1996[Abstract] 4. 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 343:196–200, 1994[CrossRef][Medline]
5. Cave H, van der Werff ten Bosch J, Suciu S, et al: Clinical significance of minimal residual disease in childhood acute lymphoblastic leukemia. European Organization for Research and Treatment of Cancer-Childhood Leukemia Cooperative Group. N Engl J Med 339:591–598, 1998 6. van Dongen JJM, Seriu T, Panzergrumayer ER, et al: Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet 352:1731–1738, 1998[CrossRef][Medline]
7. Winick N, Shuster JJ, Bowman WP, et al: Intensive oral methotrexate protects against lymphoid marrow relapse in childhood B-precursor acute lymphoblastic leukemia. J Clin Oncol 14:2803–2811, 1996
8. Mattano LA, Sather HN, Trigg ME, et al: Osteonecrosis as a complication of treating acute lymphoblastic leukemia in children: A report from the Children’s Cancer Group. J Clin Oncol 18:3262–3272, 2000
9. Cave H, Guidal C, Rohrlich P, et al: Prospective monitoring and quantitation of residual blasts in childhood acute lymphoblastic leukemia by polymerase chain reaction study of delta and gamma T-cell receptor genes. Blood 83:1892–1902, 1994
10. Nizet Y, Vandaele 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 82:1618–1625, 1993 11. Goulden NJ, Knechtli CJ, Garland RJ, et al: Minimal residual disease analysis for the prediction of relapse in children with standard-risk acute lymphoblastic leukaemia. Br J Haematol 100:235–244, 1998[CrossRef][Medline]
12. Roberts WM, Estrov Z, Ouspenskaia MV, et al: Measurement of residual leukemia during remission in childhood acute lymphoblastic leukemia. N Engl J Med 336:317–323, 1997
13. Arico M, Valsecchi MG, Camitta B, et al: Outcome of treatment in children with Philadelphia chromosome-positive acute lymphoblastic leukemia. N Engl J Med 342:998–1006, 2000
14. Nachman J, Sather H, Sensel M, et al: Augmented post-induction therapy for children with high-risk acute lymphoblastic leukemia and a slow response to initial therapy. N Engl J Med 338:1663–1671, 1998 15. Waters K: Australian and New Zealand trials in acute lymphoblastic leukemia of childhood. Int J Pediat Haem/Oncol 5:187–197, 1998 16. Norris MD, Kwan E, Haber M, et al: Detection of evolving immunoglobulin heavy-chain gene rearrangements in acute lymphoblastic leukemia: A PCR-based assay employing overlapping DJH primers. Leukemia 9:1779–1782, 1995[Medline] 17. Brisco MJ, Condon J, Sykes PJ, et al: Detection and quantitation of neoplastic cells in acute lymphoblastic leukaemia, by use of the polymerase chain reaction. Br J Haematol 79:211–217, 1991[Medline] 18. Kwan E, Norris MD, Zhu L, et al: Simultaneous detection and quantitation of minimal residual disease in childhood acute lymphoblastic leukaemia using real-time polymerase chain reaction. Br J Haematol 109:430–434, 2000[CrossRef][Medline] 19. Gaynon PS, Desai AA, Bostrom BC, et al: Early response to therapy and outcome in childhood acute lymphoblastic leukemia: a review. Cancer 80:1717–1726, 1997[CrossRef][Medline] 20. Goulden N, Oakhill A, Steward C: Practical application of minimal residual disease assessment in childhood acute lymphoblastic leukaemia. Br J Haematol 112:275–281, 2001[CrossRef][Medline] 21. Chessells JM: The management of high-risk lymphoblastic leukaemia in children. Br J Haematol 108:204–216, 2000[CrossRef][Medline] Submitted October 16, 2001; accepted October 26, 2002.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2003 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 tests. For multivariate analysis, the Cox proportional hazards regression model was applied. Statistical analyses were performed using StatView 5.0.1 (SAS Institute, Inc, Cary, NC). Relapse-free survival probabilities and relative hazards are given with 95% confidence intervals (CIs). The positive predictive value of MRD testing for individual or combined time points was determined by dividing the number of patients who relapsed and were MRD positive at a given time point, by the total number of MRD-positive patients at that time point. Conversely, the negative predictive value was calculated by dividing the number of MRD-negative patients in continuous remission for a given time point by the total number of MRD-negative patients at that time point. 
