- © 2000 by American Society of Clinical Oncology
Evaluation of Minimal Residual Disease Using Reverse-Transcription Polymerase Chain Reaction in t(8;21) Acute Myeloid Leukemia: A Multicenter Study of 51 Patients
- F. Morschhauser,
- J.M. Cayuela,
- S. Martini,
- A. Baruchel,
- P. Rousselot,
- G. Socié,
- P. Berthou,
- J.P. Jouet,
- N. Straetmans,
- F. Sigaux,
- P. Fenaux and
- C. Preudhomme
- From the Service des Maladies du Sang and Laboratoire d’Hématologie ACentre Hospitalier Universitaire, and Unité L’Institut National de la Santé et de la Recherche Médicale (INSERM) U524, Lille; Laboratoire d’Hématologie Moléculaire, INSERM Unit U462, and Service d’Hématologie Pédiatrique, d’Hématologie Clinique Adulte, and de Greffe Médullaire, Hôpital St Louis, Paris; Service d’Hématologie, Centre Hospitalier Universitaire, Brest, France; and Hématologie, Clinique Universitaire St Luc, Brussels, Belgium.
- Address reprint requests to Claude Preudhomme, MD, Laboratoire d’Hématologie, Hôpital Calmette, Centre Hospitalier Universitaire, 59037 Lille, France; email cpreudhomme{at}chru-lille.fr
Abstract
PURPOSE: Most studies using various reverse-transcription polymerase chain reaction (RT-PCR) techniques reported that the detection of the AML1-ETO fusion transcript was a common finding in long-term complete remission (CR) in acute myeloid leukemia (AML) with t(8;21) translocation. However, larger prospective studies with interlaboratory quality control may be important to investigate more precisely the clinical usefulness of studying minimal residual disease with RT-PCR in t(8;21) AML.
PATIENTS AND METHODS: We collected 223 marrow samples from 51 patients with t(8;21) AML diagnosed in five centers and tested all samples by two different RT-PCR techniques (a nested technique and a one-step technique, with a sensitivity of 10−6 and 10−5, respectively) in two different laboratories.
RESULTS: Samples from 14 patients in long persistent CR (median follow-up duration, 112 months) were taken at least twice, and all were PCR-negative by both techniques. Samples were prospectively taken from 37 patients after achievement of first CR and/or second CR, before intensive consolidation treatment, and every 3 to 6 months after completion of therapy. Patients who converted to PCR negativity with the one-step technique (60%) or both techniques (48%) after CR achievement had a longer CR duration than those with persistently positive PCR results (two-sided log-rank test, P = .0001). Patients who became PCR-negative with the one-step technique before intensive consolidation (23%) had a lower relapse rate (11% v 72%) and a longer CR duration than those who remained persistently PCR-positive at that point (two-sided log-rank test, P = .0015).
CONCLUSION: Patients with AML with t(8;21) in long-term remission were all PCR-negative. In prospectively studied patients, a good correlation was found between negative PCR results and absence of relapse. Early negative results with the one-step RT-PCR technique, before consolidation treatment, seemed to carry an especially good prognosis, suggesting that RT-PCR analysis could help in choosing the type of consolidation therapy in patients with t(8;21) AML.
SEVERAL COMMON balanced cytogenetic rearrangements specific to acute myeloid leukemia (AML), including t(15;17), inv(16), and t(8;21), result in a fusion between two genes and a fusion transcript and protein.1-3 Those fusion transcripts can be detected using reverse-transcription polymerase chain reaction (RT-PCR), a sensitive method for the detection of minimal residual disease (MRD) after treatment. In AML with t(15;17) or inv(16), no evidence of the fusion transcript is generally found by RT-PCR in patients who have achieved prolonged remission after treatment.4,5
In the case of t(8;21)(q22;q22), which leads to a fusion between the AML1 and ETO genes, several single-institution studies using semiquantitative techniques with a sensitivity ranging from 10−5 to 10−7, have reported positive RT-PCR results in many of the patients who remained in prolonged hematologic and cytogenetic complete remission (CR).6-9 Thus, no consensus exists over the disappearance of AML1-ETO fusion transcript in patients during evolution.10-12 To clarify this issue, in a multicenter study, we collected 223 sequential samples from 51 patients with t(8;21) AML and tested them by two different RT-PCR techniques in two different laboratories.
PATIENTS AND METHODS
Patients
Fifty-one patients (aged 4 to 59 years) with AML according to French-American-British criteria (1976), who were diagnosed in five centers in France and Belgium with t(8,21) by conventional cytogenetics and/or AML1-ETO–positive PCR assay at diagnosis, and who had achieved CR with anthracycline/cytarabine chemotherapy, were studied.
Cytogenetic analysis was performed by standard techniques at local institutions and classified according to the International System for Human Cytogenetic Nomenclature.13 A minimum of two karyotypes from each abnormal clone was reviewed centrally. Presence of t(8;21)(q22;q22) was confirmed by karyotypic analysis in 46 patients. Twenty-two patients had additional cytogenetic abnormalities (Tables 1 and 2⇓).
Patients With t(8;21) AML in Prolonged CR (1 or 2) Retrospectively Studied: Clinical, Biologic Findings at Diagnosis, Consolidation Treatment, and Outcome
Clinical, Biologic findings at Diagnosis, Consolidation Treatment, and Outcome in Patients With AML and t(8;21) Studied Prospectively After CR1
Two patients had a normal karyotype, and three had a genetic rearrangement involving chromosome 8, without apparent chromosome 21 involvement. In those five patients and in 34 of the patients with detectable t(8;21) by conventional cytogenetics, AML1-ETO rearrangement was found by RT-PCR at diagnosis. The remaining 12 patients with t(8,21) by conventional cytogenetics were diagnosed before 1993 and could not be analyzed by RT-PCR at diagnosis, because no RNA or DNA was frozen at that time in most institutions. However, in t(8;21) AML, all AML1 breakpoints seem to be clustered in the same intron.14,15 Furthermore, a fusion transcript of expected size was always detectable using primers spanning the AML1-ETO cDNA fusion in previous studies1,7,15,16 and in the other cases with t(8;21)(q22;q22) included in this study. Thus, it is highly probable that those 12 patients were also positive for AML1-ETO fusion transcript at diagnosis.
The 51 patients analyzed included 14 patients in long persistent first CR (CR1; n = 11) or second CR (CR2; n = 3). Only long remission samples were available in patients whose median follow-up duration was 112 months (range, 46 to 181 months): patients were sampled at least twice (over a minimum period of 12 months), 24 to 174 months after CR achievement. Clinical and hematologic characteristics of those patients are listed in Table 1.
The remaining 37 patients, diagnosed between January 1993 and August 1997, were included in a prospective sampling procedure beginning at diagnosis (32 patients; Table 2) or at first relapse (14 patients, nine of whom were also studied at diagnosis; Table 3). Clinical and hematologic characteristics of those patients are listed in Tables 2 and 3⇓. Marrow and/or blood sampling was performed at 3- to 6-month intervals after CR achievement.
Salvage Treatment, Intensive Consolidation, and Outcome of Patients With AML and t(8;21) Studied Prospectively After CR2
Treatment
All patients had achieved CR with an anthracycline/cytarabine regimen that differed over the study period and between institutions (Tables 1 to 3⇑⇑). After achieving CR, patients generally received one consolidation chemotherapy course, then intensive consolidation by either further chemotherapy, autologous peripheral stem-cell transplantation, autologous purged bone marrow transplantation (BMT) or allogeneic BMT. Consolidation treatments received are listed in Tables 1 to 3⇑⇑. Doses of cytarabine of at least 1 g/m2 every 12 hours were considered as high doses, and doses less than 1 g/m2 every 12 hours were considered as conventional doses.
Bone marrow or peripheral-blood samples were collected after informed consent was obtained. Marrow sampling was preferred if the patient accepted it, based on the hypothesis that the bone marrow contains more residual tumor cells than peripheral blood and thus could be more suitable for MRD studies.17 In some cases, peripheral blood was obtained simultaneously to, or instead of, bone marrow.
RT-PCR
Mononuclear cells from bone marrow and peripheral-blood samples were separated by Ficoll Hypaque density gradient centrifugation. Aliquots of 107 cells were frozen and stored in 1.5-mL RNAse-free Eppendorf tubes at −70°C before RNA extraction. RNA was extracted with Trizol reagent (Life Technologies, Paisley, UK) according to the manufacturer’s protocol. RNA pellets were resuspended in 10 μL of RNAse-free water (diethyl pyrocarbonate–treated). RNA quality, assessed by visualization of the three main ethidium bromide–stained RNA species (7S, 18S, and 28S RNA), was checked by electrophoresis through 0.8% agarose gels. RNA quantity was estimated by comparison with a two-fold serial dilution of a standard RNA in one laboratory and by ultraviolet spectrofluorometry in the other.
Reverse transcription.
cDNA was synthesized from 1μg of total RNA in a 20-μL reaction mixture containing 10 mmol/L Tris hydrochloride (pH 8.3), 50 mmol/L potassium chloride, 5 mmol/L magnesium chloride, 10 mmol/L dithiothreitol, 1 mmol/L of each dinucleoside 5′-triphosphate, 80 units of RNAase inhibitor (Amersham, Cambridge, UK), 50 pmol of random hexamer (Pharmacia, Uppsala, Sweden), and 100 units of reverse transcriptase (MMLV; Life Technologies). After the incubation, the mixture was heated at 95°C for 3 minutes and cooled on ice for 5 minutes.
PCR amplification.
We used two different RT-PCR techniques: a one-step technique and a nested technique, both previously described.15,16 In the nested technique, PCR amplification was performed with 2 μL of cDNA corresponding to 100 ng of total RNA. In the one-step technique, PCR amplification was performed with 7 μL of cDNA. With the introduction of AmpliTaq Gold DNA polymerase (Perkin Elmer, Norwalk, CT), the previously reported generation of nonspecific amplification products with the nested technique was no more significant. Both techniques included a final control by hybridization with a unique specific junction probe labeled with gamma-32P–adenosine triphosphate. With the nested technique, the AML1-ETO fusion transcript was consistently detected at a dilution of 10−4 after one round of PCR on ethidium bromide–stained gel and of 10−6 after two rounds of PCR. The detection limit of the one-step technique was 1 in 104 cells on ethidium bromide–stained gel and 1 in 105 cells after hybridization with our specific probe in all experiments.
Quality control.
Precautions were taken in all aspects of sample handling and preparation to avoid contamination of PCRs, according to published recommendations.18,19 Experiments were conducted in two different laboratories. Analysis of the housekeeping c-abl gene served as control for the presence of amplifiable RNA. Samples were considered adequate for evaluation on the basis of intact c-abl mRNA.
RNA from the Kasumi cell line, carrying t(8;21)(q22;q22), served as positive control, and RNA from the HL60 myeloid cell line (lacking t(8;21)) and water served as negative controls. The sensitivity of both techniques was tested by performing a limiting 10-fold dilution of 1 μg of total RNA from Kasumi-1 cells in 1 μg of total RNA from HL 60 cells (ranging from undiluted to 1 in 107 HL60 cells). With both techniques, a constant level of sensitivity (10−6 with the nested technique and 10−5 with the one-step technique) could be achieved with 1 μg of total RNA in nearly all cases, as verified by the dilution assay performed simultaneously to triplicate PCR amplification of each sample. Otherwise, sample examination was not validated. For patients in long continuous CR, the initial RNA quantity was also increased to 2 μg to make sure that a negative examination was not caused by insufficient target RNA.
RESULTS
Retrospectively Studied Patients
Fourteen patients in long persistent CR1 (n = 11) or CR2 (n = 3) after chemotherapy (n = 7), allogenic BMT (n = 5), and autologous BMT (n = 2) with a median follow-up duration of 112 months (range, 46 to 182 months) had bone marrow and peripheral-blood samples collected simultaneously, at least twice over a period of at least 12 months (Table 1). All samples were PCR-negative by both techniques.
Prospectively Studied Patients
Thirty-seven patients were studied prospectively and scheduled to be sampled at CR achievement (CR1 and/or CR2), before final consolidation and every 3 to 6 months thereafter. Some patients were studied both in CR1 and CR2; therefore, overall, the study included 46 cases (32 CR1 and 14 CR2). Consolidation treatment in those 46 cases included conventional-dose cytarabine (n = 15), high-dose cytarabine (n = 15), autologous BMT (n = 2), autologous peripheral stem-cell transplantation (n = 1), HLA-matched allogeneic BMT (n = 9), HLA-matched unrelated donor allogeneic BMT (n = 3), and donor leukocyte infusions (n = 1; Tables 2 and 3⇑).
PCR status of patients in CR1 and CR2, at every time point analyzed and with both techniques, is shown in Fig 1. Median follow-up duration in this group was 24 months (range, 6 to 72 months). Data were censored at the reference date of March 1, 1999. Twenty-two patients relapsed 2 to 36 months (median, 7.5 months) after CR achievement, and three more died from infection in CR 4 to 6 months after matched unrelated BMT. The median number of bone marrow samples per patient was five (range, three to 11). There was no statistically significant difference in terms of age (Kruskal-Wallis test, P = .4725), WBC count at diagnosis (Kruskal-Wallis test, P = .1459), sex (χ2 test, P = .234), and cytogenetic abnormalities in addition to t(8;21) (χ2 test, P = .512) between patients who remained in persistent CR1 and those who relapsed.
Fig 1. Follow-up evaluation of patients monitored in (A) CR1 and (B) CR2.
RT-PCR Analysis at CR Achievement
Forty-one marrow samples were collected at CR achievement (no adequate CR sample was obtained in five cases). With the nested technique, 33 CR samples were positive after the first PCR step (25 CR1 and eight CR2 samples), and eight CR samples were positive after the second PCR step (five CR1 and three CR2 samples). With the one-step technique, 40 CR samples (29 CR1 and 11 CR2) were PCR-positive, and only one patient’s (no. 6) sample was negative.
RT-PCR Analysis Before Final Consolidation Therapy
Thirty-nine marrow samples were collected before final consolidation treatment, 1 to 5 months (median, 3 months) after CR achievement (no adequate sample was obtained in seven cases). With the nested technique, two samples were PCR-negative (patients no. 10 and 25), whereas 16 samples (10 CR1 and six CR2) were positive after the first PCR step, and 21 samples (15 CR1 and six CR2) were positive only after the second PCR step. None of the two patients with PCR-negative samples relapsed as compared with 22 of the 37 with PCR-positive samples (15 of 27 CR1 and seven of 12 CR2 patients).
Using the one-step technique, nine samples (eight CR1 and one CR2) were PCR-negative, 29 samples (18 CR1 and 11 CR2) were PCR-positive, and one CR1 patient (no. 17) was not tested because of low RNA quantity; one of the nine patients with PCR-negative samples relapsed 18 months after CR achievement (patient no. 9), compared with 21 of 29 patients with PCR-positive samples (14 of 27 CR1 and seven of 12 CR2).
Patients whose samples became PCR-negative with the one-step technique before intensive consolidation had a lower relapse rate (11% v 72%) and a significantly longer CR duration than those with persistently PCR-positive samples at that point (two-sided log-rank test, P = .0015).
RT-PCR Results During Follow-Up
In 45 of 46 cases (patient no. 7 was lost to follow-up after 9 months), samples were collected as often as planned (ie, every 3 to 6 months). Four samples obtained from patients with prolonged cytopenias were not tested because of poor RNA quantity. With the one-step technique, PCR became negative in 27 cases (18 CR1 and nine CR2) after a median of 3 months from diagnosis (range, 2 to 12 months) and always remained positive in 18 cases; we observed four relapses (patients no. 5, 9, 33, and 34) 9 to 36 months (median, 18 months) after CR achievement in the 27 cases whose samples had become negative, whereas all the 18 cases with persistently positive PCR results relapsed (relapse rate, 14.8% v 100%).
With the nested technique, PCR became negative in 22 cases (17 CR1 and five CR2) after a median of 6 months from diagnosis (range, 3 to 21 months) and remained persistently positive in 23 cases; we observed four relapses ( patients no. 5, 9, 33, and 34) in the 22 cases whose results had become negative, and 18 relapses in the 23 cases whose results remained persistently positive (relapse rate, 18% v 78%).
Patients who converted to PCR negativity with one or both techniques after CR achievement had a significantly longer CR duration than those whose results were always PCR-positive (two-sided log-rank test, P = .0001).
Comparison Between Blood and Marrow Samples
Thirty-nine blood samples were collected from 17 patients simultaneously to (12 times), or instead of (27 times), marrow samples. RT-PCR analysis on blood and marrow samples collected on the same day were discordant (negative blood testing and positive marrow testing) in two cases (patients no. 7 and 36) with the one-step technique and in seven cases with the nested technique. Blood testing was never positive in case of simultaneous negative marrow testing. All four patients with a positive blood test result with both techniques after CR achievement relapsed (patients no. 2, 18, 21, and 32). Only one of 13 patients (no. 9) with persistently negative blood examinations by both techniques relapsed, 18 months after CR1 achievement.
DISCUSSION
Most studies using various semiquantitative techniques reported that the detection of the AML1-ETO fusion transcript was a common finding in long-term CR.6-9 However, those reports lacked standardization regarding the most appropriate material for MRD studies (marrow v blood), the RT-PCR technique (nested v one-step technique), the best sampling frequency, the amount of starting RNA quantity required to avoid false-negative results, and the gene chosen as control for amplifiable RNA. Moreover, previously published studies lacked interlaboratory control, which may be important for semiquantitative RT-PCR standardization in a clinical setting.
In this report, we compared two different published techniques in two different laboratories in the same large population of patients. Technically, we followed the minimal required criteria recommended by Jurlander et al9 to validate a negative MRD scoring except for the use of the beta-actin gene as control gene. Indeed, this gene has emerged as a rather poor control for the presence of intact RNA because of the presence of numerous pseudogenes that may be amplified in case of DNA contamination of the RNA sample, even if the RNA quality or quantity is poor. With both techniques, a constant level of sensitivity (10−6 with the nested technique and 10−5 with the one-step technique) could be achieved in all validated cases.
In this multicenter study, in which we tried to avoid major recognized methodologic problems, we first found that 14 patients in long-term persistent CR analyzed 46 to 182 months after treatment (median, 112 months) were all PCR-negative irrespective of the type of consolidation treatment, confirming our preliminary results.10 Even when those CR samples were PCR amplified after increasing the initial RNA quantity from 1 to 2 μg, no positive findings were found. Our results are different from those of several previous studies that found positive PCR results in a variable proportion of t(8;21) AML patients in long-term remission. For example, Miyamoto et al8 reported positive PCR results in all 18 patients who had been in CR for 12 to 150 months (median, 45 months) after chemotherapy or autologous peripheral-blood stem-cell transplantation, whereas PCR results were negative in four patients who had been in persistent CR for more than 30 months after allogeneic BMT. Jurlander et al9 found positive PCR results in all nine patients analyzed 7.5 to 83 months after allogeneic BMT, whereas Elmaagacli et al11 found that only one of six patients in long-term remission had positive results at 112 months after allogenic BMT.
There are no clear reasons for these discrepancies. Differences in the sensitivity of assays does not seem to be an explanation because it was high with the techniques we used (10−6 and 10−5). Some of the published positive results in long-term remission might have resulted from technical bias. We tried to avoid such biases by adopting an accurate interlaboratory control procedure and consensus criteria required to score a sample as negative, but also by rejecting the β-actin gene as an adequate control gene. Our results suggest that disappearance of the fusion transcript is the rule in AML patients with t(8;21) in prolonged CR, irrespective of the consolidation treatment they received.
In patients prospectively studied after CR achievement, conversion to PCR negativity with both techniques was associated with persistent CR or late relapse. This was similar to what is observed in acute lymphoblastic leukemia20 or t(15;17) AML21 with PCR techniques of comparable sensitivity, where MRD negativity at various follow-up times is associated with a low relapse rate. The relapse rate in this series was 14% or 17%, depending on the technique used, in patients who became PCR negative, as compared with 78% to 100% in persistently MRD-positive patients. Moreover, as in acute lymphoblastic leukemia, a low quantity of MRD before consolidation treatment (< 10−5 after semiquantitative estimation with the one-step technique in our study) was associated with a low relapse rate (11%) as compared with 72% in MRD-positive patients at that point.
We confirmed in this study that bone marrow samples may be more suitable than blood samples for MRD studies. Indeed, blood examination was often negative in case of positive bone marrow examination, and a negative blood examination with both techniques was not predictive of long-term remission. Moreover, contrary to some other reports,9,12 no case showed a positive blood examination and concomitant negative bone marrow result. Interestingly, however, a positive blood examination with the one-step technique after CR1 achievement seemed to carry an adverse prognosis even if the number of patients studied was too low to reach a conclusion on that point.
Thus, in this multicenter analysis where samples were tested in two different laboratories, with two different techniques and close monitoring of the level of sensitivity, we found a good correlation between negative RT-PCR results in the bone marrow and favorable outcome. Clinical usefulness of monitoring MRD with RT-PCR still remains an issue in t(8,21) AML. Early negative PCR results with the one-step RT-PCR technique (sensitivity of 10−5) seemed to carry an especially good prognosis. Our findings support the use of PCR results in choosing consolidation therapy in a given patient with AML and t(8;21). This may be especially important because, in AML with t(8;21), none of the variables possibly identified as prognostic factors of relapse, including WBC count, age, sex, presence of additional cytogenetic abnormalities, presence of extramedullary leukemia, and CD56 expression, has been clearly confirmed.22-25 In our study, no pretreatment factor, including WBC count, age, sex, and the presence of additional cytogenetic abnormalities, had any prognostic value for CR duration.
Finally, comparison between our two techniques suggests a somewhat better clinical usefulness of the one-step technique over the nested technique, especially in the months after CR achievement. Indeed, early negativity of the one-step technique, before consolidation therapy, was associated to favorable prognosis. This could help in choosing optimal consolidation therapy in t(8;21) AML. Later, during the first year after consolidation treatment, a persistently positive single-step RT-PCR assay seemed more predictive of relapse than a positive nested RT-PCR assay (100% of patients with positive results by single-step RT-PCR relapsed v 78% of those with positive results by nested PCR). The one-order lower sensitivity of the one-step technique may be the explanation for its better clinical usefulness. Other reasons for the preferential use of the one-step technique could be its simplicity, a lower risk of contamination, and a lower cost.
Because most relapses occur during the first year in t(8;21) AML, we recommend close RT-PCR monitoring during that time. Real-time quantitative PCR will certainly help further in the evaluation of the relapse risk in patients with t(8;21) AML.
Acknowledgments
Supported by the Centre Hospitalier, Lille, France (Projet Hospitalier de Recherche Clinique 1997).
ACKNOWLEDGMENT
We thank C. Roumier, D. Pruvot, M. Wasselin, and Institut Federatif de Recherche 22 for technical assistance.
Footnotes
-
P. Fenaux and C. Preudhomme contributed equally as last authors to this work.
- Received May 25, 1999.
- Accepted September 27, 1999.
