Pii: s0140-6736(98)04058-6

Prognostic value of minimal residual disease in acutelymphoblastic leukaemia in childhood Jacques J M van Dongen, Taku Seriu, E Renate Panzer-Grümayer, Andrea Biondi, Marja J Pongers-Willemse, Lilly Corral, Frank Stolz, Martin Schrappe, Giuseppe Masera, Willem A Kamps, Helmuth Gadner, Elisabeth R van Wering, Wolf-Dieter Ludwig, Giuseppe Basso, Marianne A C de Bruijn, Giovanni Cazzaniga, Klaudia Hettinger, Anna van der Does-van den Berg, Wim C J Hop, Hansjörg Riehm, Claus R Bartram and before consolidation treatment). At these two time points a high degree of MRD (୑10Ð2) was associated with a Background Sensitive techniques for detection of minimal three-fold higher relapse rate when compared with patients residual disease (MRD) at degrees of one leukaemic cell per with a low degree of MRD (୏10Ð4). At later time points 103–106 cells (10Ϫ3–10Ϫ6) during follow-up of children with (including the end of treatment) even a low degree of MRD acute lymphoblastic leukaemia (ALL) can provide insight was associated with a poor outcome. Positivity in patients into the effectiveness of cytotoxic treatment. However, it is in CCR after treatment was rare (<1%). With the combined not yet clear how information on MRD can be applied to MRD information from the first two follow-up time points, it was possible to recognise three different risk groups—55 Methods We monitored 240 patients with childhood ALL (43%) were in a low-risk group and had a 3-year relapse rate who were treated according to national protocols of the of only 2% (95% CI 0·05–12%); 19 (15%) were in a high-risk International BFM Study Group. 60 patients relapsed and group and had a relapse rate of 75% (55–95%); and 55 the patients in continuous complete remission (CCR) had a (43%) were in an intermediate-risk group and had a 3-year median event-free follow-up of 48 months. Bone-marrow samples were collected at up to nine time points during and Interpretation Our collaborative MRD study shows that after treatment. Standardised PCR analysis of patient- monitoring patients with childhood ALL at consecutive time specific immunoglobulin and T-cell receptor gene points gives clinically relevant insight into the effectiveness rearrangements and TAL1 deletions were used as targets for of treatment. Combined information on MRD from the first semiquantitative estimation of MRD. Amount of MRD was 3 months of treatment distinguishes patients with good classed as 10Ϫ2 or more, 10Ϫ3, and 10Ϫ4 or less.
prognoses from those with poor prognoses, and this helps in Findings MRD negativity at the various follow-up times was decisions whether and how to modify treatment.
associated with low relapse rates (3–15% at 3 years), but five-fold to ten-fold higher relapse rates (39–86% at 3 years) were found in MRD-positive patients. The distinct degrees of MRD appeared to have independent prognostic value Current cytotoxic treatment protocols induce complete (p [trend]<0·001) at all separate time points, especially at remission according to cytomorphological criteria in the first two time points (at the end of induction treatment 95–98% of children with acute lymphoblastic leukaemia(ALL).1–3 However, 25–30% of patients with childhoodALL relapse, implying that not all leukaemic cells are Department of Immunology, University Hospital eradicated. Cytomorphological criteria are not sufficient Rotterdam/Erasmus University Rotterdam, Rotterdam,Netherlands (Prof J J M van Dongen MD, M J Pongers-Willemse PhD, for adequate assessment of the remission status because M A C de Bruijn PhD); Institute of Human Genetics, University of the threshold of detection is 1–5% of leukaemic cells.
Heidelberg, Heidelberg, Germany (T Seriu MD, More sensitive techniques are required to determine Prof C R Bartram MD); Department of Pediatrics, Children’s Cancer efficacy of treatment and to improve stratification of Research Institute, St Anna Kinderspital, Vienna, Austria (Prof E R Panzer-Grümayer MD, F Stolz PhD, Prof H Gadner MD, Three types of techniques allow detection of minimal K Hettinger PhD); Department of Pediatrics, Universitá Milano, residual disease (MRD) of 10Ϫ3–10Ϫ6 (one leukaemic cell in Ospedale San Gerardo, Monza, Italy (A Biondi MD, L Corral BSc, cells): flow-cytometric immunophenotyping Prof G Masera MD, G Cazzaniga BSc); Department of Pediatrics, (detecting aberrant protein expression by the ALL cells); Medizinische Hochschule Hannover, Hannover, Germany (M Schrappe MD, Prof H Riehm MD); Dutch Childhood Leukaemia PCR analysis of breakpoint fusion regions of chromosome Study Group, The Hague, Netherlands (Prof W A Kamps MD, aberrations; and detection of clone-specific E R van Wering MD, A van der Does-van den Berg MD); Beatrix immunoglobulin (Ig) and T-cell receptor (TCR) gene Children’s Hospital, University of Groningen, Groningen, rearrangements by PCR amplification.4–7 Netherlands (W A Kamps); Department of Hematology, Oncology technique is the most widely used for MRD studies in and Tumour Immunology, Robert-Rössle-Clinic, Humboldt University of Berlin, Berlin, Germany (Prof W-D Ludwig MD); Department of Rearrangements of Ig and TCR genes result in Pediatrics, University of Torino, Torino, Italy junctional regions that can be regarded as fingerprint-like (Prof G Basso MD); Department of Epidemiology and Biostatistics, clone-specific sequences owing to deletion and random Erasmus University Rotterdam, Rotterdam, Netherlands (W C J Hop PhD) insertion of nucleotides.8 PCR-based MRD detection byclone-specific junctional regions generally reaches Correspondence to: Prof J J M van Dongen, Department ofImmunology, Erasmus University Rotterdam, PO Box 1738, sensitivities of 10Ϫ4–10Ϫ5. For this purpose, oligonucleotide primers are designed at opposite sides of the junctional THE LANCET • Vol 352 • November 28, 1998 region. To discriminate between the leukaemia-derived PCR products and PCR products of normal cells with comparable rearrangements, the amplification products are generally hybridised to a patient-specific junctional- Although rearrangements of Ig and TCR genes are clone-specific, they are not directly linked to the oncogenicprocess. Continuing rearrangements and secondaryrearrangements might change the junctional regions, which are identified at diagnosis as PCR targets for MRD studies.9–12 It seems to be important to monitor patients with ALL with two or more independent Ig/TCR targets to prevent false-negative results during follow-up.7,11 Several retrospective and small prospective studies with short follow-up indicate that the degree of MRD in ALL Figure 1: Scheme of bone-marrow sampling time points The time points of the four treatment protocols were synchronised in childhood has prognostic value, although the results of according to the treatment phase (the German sampling protocol is these studies are not fully concordant. Absence of residual given). Only the treatment blocks of the standard-risk group (SRG) and disease after remission induction is associated with good medium-risk group (MRG) protocols are indicated—therapy for high-riskgroup (HRG) is different between induction and maintenance prognosis.13–15 However, about half of patients with childhood ALL remain MRD positive after remission and 31 patients from Netherlands, because they fulfilled the induction. Consequently, the degree of MRD has following criteria: bone-marrow samples from no more than three predictive value.13,14 This is in line with cytomorphological consecutive time points and from no more than four time points studies after 1 week of treatment, which showed that a in total were missing per patient, and at least one PCR target with slow rate of reduction in numbers of blast cells in blood or a sensitivity of at least 10Ϫ4 was available for MRD detection.
bone marrow during induction therapy was related to These 242 patients did not differ from the other 383 patients higher relapse rates. However, these cytomorphological with respect to sex, white-blood-cell count, immunophenotype, techniques can identify only fewer than 30% of patients prednisone response, and distribution over treatment groups. The two groups also did not differ in relapse-free survival.
A steady disappearance of MRD during follow-up, as Of the 60 patients who relapsed, four had isolated determined by PCR, is associated with a favourable extramedullary relapses (three relapses in the central nervoussystem and one in the testis), 11 combined bone-marrow and prognosis.17 Persistence of MRD generally leads to clinical extramedullary relapses, and 45 isolated bone-marrow relapses.
relapse.18 Low degrees of MRD after treatment might be Two HRG patients did not enter remission according to associated with late relapse, but absence of MRD at the end cytomorphological criteria and were excluded from further of treatment is not sufficient to deem the patient cured.19 assessment in this study. The characteristics of the remaining 240 Despite these preliminary MRD data, it is still not clear patients (210 precursor-B-cell-ALL and 30 T-cell-ALL) are whether and how MRD information can be applied in summarised in table 1. The median event-free follow-up of the clinical decision making. We did a large prospective 180 patients in continuous complete remission was 48 months.
multicentre study to assess the clinical relevance of PCR-based MRD detection at multiple time points during Standardisation of the molecular MRD studies follow-up of 240 children with ALL who were treated The four molecular diagnostic laboratories of the I-BFM-SG according to national protocols of the International BFM study in Rotterdam, Heidelberg, Vienna, and Monza standardised (Berlin-Frankfurt-Münster) Study Group (I-BFM-SG) in the Southern blot and PCR techniques during the first part of thestudy.20 Standardisation was of technical procedures as well as Austria, Germany, Italy, and Netherlands.
reagents. DNA probes and PCR primers were centrally producedin Rotterdam and distributed to the other laboratories. Quality control was achieved by exchange of cell samples between the four laboratories. This standardisation was done in the context of a Bone-marrow samples were taken, at diagnosis (March, 1991–May, 1995) and at up to nine times during follow-up, from625 patients with childhood ALL. Recruitment for the study was Identification of PCR targets at diagnosis done in a limited number of I-BFM-SG centres, because of Because of the high frequency of oligoclonality and continuing logistics and medical ethical aspects of the study. All children rearrangements in the Ig heavy chain (IGH) genes9,10 and the were treated according to protocols of the Austrian BFM group assumed lower frequency of such rearrangements in TCR gamma (protocol ALL-BFM 90), the German BFM Group (protocol (TCRG) and TCR delta (TCRD) genes,11 we used TCRG and ALL-BFM 90), the Associazione Italiana di Ematologia ed TCRD gene junctional regions as patient-specific targets for PCR- Oncologia Pediatrica (AIEOP-ALL–91 protocol), or the Dutch based MRD detection. These two targets have the advantage that Childhood Leukemia Study Group (DCLSG, protocol ALL–8). The patients were stratified into standard-risk, medium-risk, and high-risk treatment groups (SRG, MRG, and HRG, respectively), mainly according to the presenting features—eg,leukaemic cell mass and the prednisone response.1 The nine bone- marrow sampling times of the four treatment protocols were synchronised with treatment phases (figure 1). Each centre followed the informed-consent guidelines of the local or national Mononuclear cells were isolated from the bone marrow SRG=standard-risk group. MRG=medium-risk group. HRG=high-risk group.
samples and stored in liquid nitrogen or at Ϫ70¡C for DNA extraction. 242 patients were included in the analysis: 55 patients Table 1: Characteristics of 240 patients included in the MRD from Austria, 105 patients from Germany, 51 patients from Italy, THE LANCET • Vol 352 • November 28, 1998 SRG=standard-risk group. MRG=medium-risk group. HRG=high-risk group. CCR=continuous complete remission.
Table 2: Frequencies of MRD positivity per treatment group at the various follow-up time points they occur at relatively high frequencies in both precursor-B-ALL and 10Ϫ6 were clustered with the 10Ϫ4 results, because such low and T-ALL.7,22 In addition, the so-called TAL1 deletions were MRD degrees were rarely found, since only one third of the PCR used as PCR target in T-ALL23 and rearrangements of the kappa targets had a detection limit of <10Ϫ4.
deleting element (⌲de) were used in precursor-B-ALL. The ⌲derearrangements delete the J␬ and/or C␬ gene segments of the Ig kappa (IG⌲) locus and occur in half of precursor-B-ALL.24 From time point five onwards, the number of positive patients Rearrangements in the IGK-⌲de, TCRG, TCRD, and TAL1 was too small for drawing firm conclusions. Therefore, we focused genes were studied by Southern blot analysis with the same most statistical analyses on the first five follow-up time points.
restriction enzymes and DNA probes in each laboratory.22–25 The Relapse-free survival according to the MRD results at the detected rearrangements were confirmed by PCR analysis and various time points was determined by use of Kaplan-Meier plots.
direct sequencing of the junctional regions with standardised sets Comparison of groups was done with the log-rank test or the log- of oligonucleotide primers.20,26 On the basis of the sequence data rank trend test in case of ordered groups (eg, degree of MRD).
of the junctional regions, patient-specific oligonucleotides were Multivariate analysis of the predictive value of MRD at the designed for each identified MRD-PCR target, using OLIGO 5.0 various time points, allowing for treatment group, age, sex, software (National Biosciences, Plymouth, MN).20 immunophenotype, white-blood-cell count (on a continuousscale), and country, was done with Cox-regression analysis. In this Cox analysis, data from the different treatment centres were The MRD-PCR analyses of bone-marrow samples during follow- amalgamated according to country. We also calculated whether up were generally done by single PCR analysis of 1 µg of DNA the effects of degree of MRD depended on the treatment-group (equivalent to 105–106 cells) with the standardised primer sets, classification, by investigating appropriate interaction terms in the followed by dot blotting and hybridisation with the corresponding 32P-labelled patient-specific junctional region probe.20,26 Thehybridisation signals were visualised by use of radiographic films or phosphor-imaging. However, if sensitivity of at least 10Ϫ4 wasnot reached, nested PCR followed by dot blotting and hybridisation with the corresponding 32P-labelled junctional region In 148 (62%) of 240 patients, two PCR targets (128 probe was done in case of IG⌲-⌲de and TCRD gene patients) or three PCR targets (20 patients) were used for rearrangements; in case of TCRG gene rearrangements, nested MRD detection. The PCR targets concerned 73 IG⌲-⌲de PCR was applied as an alternative with the junctional region rearrangements in 66 precursor-B-ALL, 126 TCRG gene oligonucleotide as one of the nested primers, followed by rearrangements in 83 precursor-B-ALL and in 23 T-ALL, detection of the obtained PCR products in ethidium bromide 186 incomplete TCRD gene rearrangements (Vδ2-Dδ3 stained agarose gels.12,20,27 If multiple PCR targets were available, apriority order was used for target selection—eg, incomplete TCRD and Dδ2-Dδ3) in 139 precursor-B-ALL, 21 TCRD gene rearrangements had the highest priority in precursor-B-ALL rearrangements (5 Vδ2-Dδ3, 2 Dδ2-Jδ1, and 14 Vδ-Jδ1) in because of the relatively low background in mononuclear cells (in 17 T-ALL, and 2 TAL1 deletions in 2 T-ALL. In 379 contrast to TCRG gene rearrangements) and the relatively large (93%) of 408 of these targets, sensitivities of at least 10Ϫ4 junctional regions. Preferably, two PCR targets were used per were reached: 25 targets with a sensitivity of 10−6, 113 patient of which at least one reached a sensitivity of at least 10Ϫ4.
targets with 10−5, and 241 targets with 10Ϫ4.
The sensitivity of each identified MRD-PCR target was The overall results of the four I-BFM-SG laboratories established by use of a dilution experiment, in which DNA from show that in 3–4% of patients with ALL no MRD-PCR the leukaemic cells at diagnosis was diluted in 10-fold dilution target was identified and in an additional 5–7% of cases steps into control DNA from a mixture of blood mononuclearcells of about ten different healthy donors.20,26 the identified PCR targets did not reach sensitivities of at When MRD-PCR analysis of bone-marrow follow-up samples resulted in a hybridisation signal or ethidium-bromide signal, thistime point was considered to be MRD positive. Consequently, if no signal was obtained, this time point was considered MRD For 20 of the 60 patients who had a relapse, a bone- negative, irrespective of the PCR-target sensitivity. The marrow sample at clinical relapse was not available. Four concentration of leukaemic cells in the bone-marrow samples additional patients had an isolated extramedullary relapse during follow-up was estimated by comparison of the signals with and were not considered for analysis of PCR target those of the ten-fold dilution samples of the DNA at diagnosis.
This resulted in reproducible semi-quantitative estimations of stability. Bone-marrow samples at relapse were analysed in MRD-PCR results of: ୑10Ϫ2, 10Ϫ3, 10Ϫ4, 10Ϫ5, and 10Ϫ6. We 36 patients with bone-marrow involvement. In 34 (94%) defined the following degrees of MRD: high as у10Ϫ2, of 36 of these patients at least one PCR target remained intermediate as 10Ϫ3, and low as р10Ϫ4. MRD-PCR results of 10Ϫ5 stable—ie, 51 (89%) of 57 targets. The false-negative THE LANCET • Vol 352 • November 28, 1998 a negative PCR result at time point eight (end of therapy) and relapsed 1 month and 2 months later with a positive PCR result at relapse, which suggests a rapid regrowth after cessation of therapy. The last two patients were negative for MRD after intensive induction and consolidation therapy, but relapsed 4 months later during maintenance treatment, which suggests that maintenance treatment was not able to further eradicate undetectable MRD.
Of the four isolated extramedullary relapses, two of three relapses in the central nervous system were accompanied by MRD-PCR positivity in thecorresponding bone-marrow samples as well as in the preceding bone-marrow samples. This remainedinconclusive for the testis relapse because no bone-marrow sample at the time of clinical extramedullary relapse wasavailable and the preceding bone-marrow sample wastaken 15 months earlier.
The 240 patients with childhood ALL were monitored by bone-marrow sampling during and after treatment up to 3 years after diagnosis—ie, 1 year after cessation of therapy (figure 1). A total of 1485 bone-marrow samples were analysed with one or more PCR targets. If analysis of the same bone-marrow sample, with two or three different PCR targets, resulted in different tumour-load estimations, we assumed the highest degree of MRD to be the most accurate for data analysis, since lower degrees of MRD might be caused by subclone formation via continuing Table 2 summarises the frequencies of MRD-positive bone-marrow samples of all patients by treatment groupand separately in the patients in continuous complete Figure 2: Relapse-free survival according to degree of MRD atfollow-up time points one and two remission. At the first time point about 40% of the tested Number of patients at risk are given in parentheses for each group at patients were negative for MRD—ie, half of the SRG and 24 months and 48 months after each sampling point.
MRG patients but none of the HRG patients. The results of six targets at relapse are probably due to percentages of MRD-positive SRG and MRG patientsdecreased rapidly during the induction, consolidation, and continuing rearrangements and clonal selection and reinduction treatment blocks down to about 10% just concerned one (8%) of 12 IG⌲-⌲de targets, two (11%) of before maintenance treatment. Each SRG and MRG 18 TCRG targets, and three (11%) of 27 TCRD targets.
treatment block resulted in a halving of positive patients Although false-negative results of one PCR target were (table 2). The fraction of MRD-positive patients further obtained in six of the 36 patients, in four of them a second decreased during maintenance treatment. Two-fold to target was analysed and was positive at relapse.
four-fold higher frequencies of positive samples werefound in the HRG patients when compared with SRG and MRD-PCR positivity in bone-marrow samples preceding MRG patients during the first year of treatment.
Table 3 summarises the cumulative relapse rates at In 42 (84%) of the 50 patients with a bone-marrow relapse 3 years after MRD analysis at the various follow-up time within 3 years of diagnosis (four of the 60 relapsed patients points (Kaplan-Meier analysis). MRD negativity was had an isolated extramedullary relapse and six relapsed clearly associated with low relapse rates (3–15% at 3 years), more than 3 years after diagnosis), the preceding bone- especially at the first follow-up time point. Relatively high marrow sample was MRD positive. In eight patients the relapse rates were found in MRD-positive patients preceding bone-marrow sample was negative: in two (39–86% at 3 years). High degrees of MRD (୑10−2) at the patients, probably because of false-negative results at first two time points were associated with three-fold higher relapse, and in two the preceding bone-marrow sample relapse rates than low degrees of MRD (୏10−4; figure 2).
was taken 8 months before relapse. Two other patients had Although the differences in relapse rates between the MRD=minimal residual disease. *Kaplan-Meier analysis was not done for groups smaller than eight patients. Table 3: Numbers of relapses and cumulative relapse rates at 3 years for different degrees of MRD during follow-up THE LANCET • Vol 352 • November 28, 1998 *Reference category. †Per ten-fold increase of degree of MRD.
Table 4: Multivariate analysis of treatment group and MRD information at I-BFM-SG time points one and two and nine) with a total of 424 PCR analyses using 178 PCRtargets with 10Ϫ4 sensitivity, 66 PCR targets with 10Ϫ5 sensitivity, and 19 PCR targets with 10Ϫ6 sensitivity. MRDpositivity was found in only eight samples of eight different patients, six of whom relapsed later on (table 3).
Therefore, only two samples of two different patients were Additionally, we investigated bone-marrow samples of eight patients in continuous complete remission at the two post-treatment time points with ten-fold PCR analyses using sensitive targets (sensitivity: 10Ϫ5–10Ϫ6). We did not observe any MRD positivity in these multiple PCR analyses, implying that the tested bone-marrow samples probably did not contain leukaemic cells at degrees of Multivariate analysis of relapse rates included treatmentgroup, age, sex, immunophenotype white-blood-cell count, and country, but not chromosome aberrations, because this information was far from complete. MRD wasfound to be an independent prognostic factor (p<0·001) at each of the first five time points. The results of themultivariate analysis of the prognostic factors at the firsttwo time points separately are given in table 4.
Each ten-fold increase of degree of MRD resulted in a two-fold increase in relative relapse rate (table 4). Inaddition to the degree of MRD, the treatment-group classification still retained prognostic value (overall p=0·011 and p=0·005 at time points one and two, respectively). The impact of degree of MRD did not differ significantly between the HRG, MRG, and SRG patients.
Additional adjustment for age, sex, immunophenotype white-blood-cell count, and country did not lead to Figure 3: Relapse-free survival according to the presence of appreciable changes of the effects of degree of MRD as MRD at follow-up time points three, four, and five Number of patients at risk are given in parentheses for each group at 24 compared with table 4: relative relapse rates associated and 48 months after each sampling point.
with a ten-fold increase of degree of MRD were 2·1(p<0·001) and 1·6 (p<0·001) at time points one and two, degrees of MRD were still significant at later time points, lower degrees of MRD were increasingly associated with Most HRG patients had, by definition, a poor high relapse rates (58–75% at 3 years). Five-fold to ten- prednisone response (19 of 23). 18 of 19 poor prednisone fold differences in relapse rates were seen between MRD- responders had a bone-marrow sample taken at time point positive and MRD-negative patients. The difference in 3- one. Of these patients, ten had a high degree of MRD year relapse rates after time-point five (1 year) was (୑10Ϫ2) and eight had an intermediate or low degree of impressive: 9% in MRD-negative patients versus 86% in MRD (10Ϫ3 in seven cases and 10Ϫ4 in one case). Their 3- year relapse-free survival rates were 10% and 50%,respectively (p=0·027).
MRD detection after treatmentFive of six patients who were MRD positive at the end of treatment (time point eight) had a relapse. But relapses To investigate whether information about the kinetics of were also observed in nine of 148 patients who were tumour reduction is helpful in better distinguishing negative at the end of treatment (table 3).
between patients with good prognosis and patients with A total of 251 bone-marrow samples from 168 different poor prognosis, we assessed the MRD results of the 129 patients were analysed after treatment (time points eight patients who were analysed at time points one and two.
THE LANCET • Vol 352 • November 28, 1998 MRD-based intermediate risk group contained 31% SRG, 60% MRG, and 9% HRG patients, and the MRD-based high-risk group contained 53% MRG and 47% HRG patients. Nevertheless, taking into account the treatment group classification, Cox regression revealed that the MRD-based intermediate and high-risk group had significantly higher relapse rates than the MRD-based low-risk group—ie, 11·0 (p=0·023) and 49·4 (p<0·001), Within the MRD-based low-risk and high-risk groups, MRD information at time points three, four, and five had no additional prognostic value. For the MRD-based intermediate-risk group, at time point five MRD-positive (n=9) and MRD-negative (n=39) patients differed significantly in relapse rates (67% and 10%, respectively Figure 4: Relapse-free survival of the three MRD-based risk groups, as defined by MRD information at time points one andtwo Patients in the low-risk group have MRD negativity at both time points,patients in the high-risk group have MRD degrees This large PCR-based study of MRD in childhood ALL points, and the remaining patients form the MRD-based intermediate-risk unequivocally demonstrates that monitoring patients group. The numbers of patients at risk are given in parentheses for each serially gives clinically relevant insight into the group at 24 months and 48 months after time point two.
effectiveness of treatment. For instance, during the three These 129 patients did not differ significantly in initial treatment blocks among SRG and MRG patients, distribution of age, sex, treatment group, and relapse-free the frequency of MRD-positive bone-marrow samples survival from the group of 111 patients who were not rapidly halved per treatment block down to about 10%.
Maintenance treatment further reduced the percentage of The number of patients with MRD negativity increased MRD-positive patients to less than 5% at the end of from 56 (43%) of 129 at time point one to 87 (67%) at treatment. Not only was the frequency of MRD positivity time point two (McNemar’s test: p<0·001). Among the 87 reduced by each treatment block, but also the degree of patients MRD negative at time point two the 55 who were MRD. Although the HRG patients had more MRD- already negative at time point one (group A, table 5) had a positive samples and higher degrees of MRD than did the significantly (p=0·002) lower 3-year relapse rate than did other two treatment groups during treatment, their MRD the 32 (group B) who became negative at time point two for the first time (rates 2% and 23%, respectively). The 14 MRD negativity during and after treatment was patients whose degree of MRD fell from high or associated with low relapse rates (3–15% at 3 years), and intermediate (୑10Ϫ2 or 10Ϫ3) to low (୏10Ϫ4) between time MRD positivity with high rates (39–86% at 3 years).
points one and two (group D, n=14) had lower relapse Degrees of MRD positivity also had prognostic value. At rates (p=0·002) than did patients whose tumour load the first two follow-up time points relapse rates in patients remained high or intermediate (group C, n=19). Relapse with a high degree of MRD were three times higher than rates at 3 years for group C and D were 75% and 21%, that in patients with low degree of MRD. During further respectively. The remaining group of patients (group E, follow-up even a low degree of MRD was associated with n=9) had a relapse rate at 3 years of 22%. Groups B, D, and E, which included 55 (43%) of the 129 patients, were Positivity after treatment was rare but associated with a combined into one MRD-based intermediate-risk group high rate of relapse—ie, six of the eight MRD-positive with a 3-year relapse rate of 23% (95% CI 13–36%).
patients relapsed. This is in contrast to the study by Group A, whose 3-year relapse rate was 2% (95% CI Roberts and colleagues28 who claimed that they detected 0·05–12%) was classed as low risk and group C, whose 3- MRD positivity in one or more post-treatment bone- year relapse rate was 75% (95% CI 55–95%) was classed marrow samples from 15 of 17 tested patients who were in continuous complete remission, thereby suggesting that There appeared to be an association between the nearly all patients with childhood ALL have low treatment-group classification and the MRD-based risk- concentrations of leukaemic cells (10Ϫ4−5·10Ϫ6 cells). Thishas not been confirmed by other groups. Also, our group classification. The MRD-based low-risk group sensitive ten-fold PCR analyses of bone-marrow samples consisted of 22% SRG and 78% MRG patients, the from eight patients in continuous complete remission didnot detect MRD. Our combined data indicate that post- treatment bone-marrow samples of patients treated with I- BFM-SG protocols probably do not contain MRD at 10Ϫ6–10Ϫ7 degrees, if residual leukaemic cells occur at all.
unexplained, although the possibility that the occurrence of low concentrations of leukaemic cells is dependent ontreatment regimen cannot be excluded.
Multivariate analysis at each of the first five follow-up Numbers in parentheses=number of relapses. A–E=MRD-based risk groups (A=low; B, time points (with more than 10% of the samples being Table 5: Patients (n=129) categorised according to degree of MRD positive) showed that MRD information is a prognostic factor, independent of age, sex, THE LANCET • Vol 352 • November 28, 1998 immunophenotype white-blood-cell count (on a bone-marrow relapses were preceded by an MRD-positive continuous scale), country, and treatment group. Also, the bone-marrow sample, continuous bone-marrow impact of degree of MRD was independent of the monitoring is a traumatic experience for the child,31 treatment-group classification, including the prednisone requires complex sampling logistics, and is laborious.
response. The kinetics of tumour decrease at the first two Furthermore, predicting poor prognosis from early time points appeared to be highly informative for treatment response probably provides more possibilities for identification of groups with good and poor prognosis: early treatment intervention than does detection of an patients at low-risk with MRD negativity at both time imminent relapse. Our findings indicate that the number points; patients at high risk with an intermediate or a high of bone-marrow aspirations might be reduced—eg, to degree of MRD at both time points; and the remaining three or four time points during the first year of treatment.
patients at intermediate risk. The MRD-based low-risk The exact number of bone-marrow samples and sampling group (3-year relapse rate of 2%) did not contain HRG time points should be determined for each treatment patients, whereas the MRD-based high-risk group (3-year relapse rate of 75%) contained as many MRG as HRG Replacement of bone-marrow sampling by blood patients. When considering the MRD data at time points sampling has been a topic of debate in MRD studies for three, four, and five, only time point five had additional over a decade. Initial immunophenotyping studies in T- prognostic value for the MRD-based intermediate-risk ALL and acute myeloid leukaemias indicate that MRD in group, whose 3-year relapse rate after time point five was blood is generally less than ten-fold lower than in bone 10% in MRD-negative patients and 67% in MRD-positive marrow.32,33 A PCR study published in 1997 showed that in precursor-B-ALL this difference is 11·7-fold.34 This In contrast to previous reports13–15 and the report by would imply that MRD techniques need to be about ten- Coustan-Smith and colleagues,29 our data indicate that fold more sensitive (ie, ୏10−5), when blood samples are analysis of MRD at a single time point is not sufficient for monitored. However, more information is needed to recognition of patients with poor prognosis as well as decide whether or not bone-marrow sampling can be patients with good prognosis. Time point one is useful for recognition of patients at low risk and time point two for The technical and clinical data presented here show that recognition of patients at high risk; later time points multicentre MRD studies are feasible and that a clinically (in particular time point five) provide prognostic value for relevant MRD-based risk group classification can be patients at intermediate risk. Information about achieved. This information can now be used for the design kinetics of tumour reduction helps identification of of new childhood ALL protocols with MRD-based MRD-based risk groups and thereby provides new openings for treatment stratification. The patients withMRD-based low risk make up a group of substantial size (about 43%), comparable with the frequency of Jacques J M van Dongen, E Renate Panzer-Grümayer, Andrea Biondi, andClaus R Bartram were equally responsible for the concept of the study, and survivors of childhood ALL in the early seventies, before the coordination and the interpretation of the analyses, they prepared the treatment intensification was introduced.30 This group manuscript with advice from the other authors. Taku Seriu, might profit from treatment reduction. On the other hand, Marja J Pongers-Willemse, Lilly Corral, Frank Stolz, Marianne A C de Bruijn, Giovanni Cazzaniga, and Klaudia Hettinger did the group of patients at MRD-based high risk is larger the testing and standardisation of the MRD-PCR techniques, did the MRD than any previously identified high-risk group (about 15%) analyses, and assessed the data. Elisabeth R van Wering, and has an unprecedented high 4-year relapse rate of Wolf-Dieter Ludwig, and Giuseppe Basso organised the collection of manyfollow-up samples and did immunophenotyping for classification of the about 80% (figure 4). This group might benefit from more leukaemias. Wim C J Hop and MJP-W did the statistical analyses, drew-up intensive treatment protocols (including bone-marrow and the tables, and designed the figures. Martin Schrappe, Giuseppe Masera, stem-cell transplantation) or from innovative treatment Willem A Kamps, Helmuth Gadner, Anna van der Does-van den Berg, and Hansjörg Riehm were the chairmen of the national study groups and I-BFM-SG committees, and were responsible for the treatment Our multicentre study shows that reliable PCR-based protocol design and clinical evaluation of the MRD study.
MRD detection in childhood ALL preferably needs two Igor TCR gene targets per patient to prevent false-negative results because of ongoing or secondary rearrangements.
We thank Annemarie Wijkhuijs and John Tibbe (Rotterdam,Netherlands), Dorothee Erz and Yvonne Stark (Heidelberg, Germany), Use of two MRD-PCR targets can be achieved in 90–95% Susanna Fischer (Vienna, Austria), and Elisabetta D’Aniello and Daniela of patients with childhood ALL, if IGH gene targets are Silvestri (Monza, Italy) for their excellent technical assistance. used in addition to IGK (Κde), TCRG, TCRD, and TAL1 We thank the participants of the International BFM Study Group for theirclose collaboration in the MRD study—the Dutch Childhood Leukemia gene rearrangements.7,26 Although the applied MRD-PCR Study Group (P J van Dijken, K Hählen, F G A J Hakvoort-Cammel, technology provided semi-quantative data, this study W A Kamps, E F van Leeuwen, F A E Nabben, A Postma, demonstrates that it is not only important to measure J A Rammeloo, G A M de Vaan, A J P Veerman, E Th van ‘t Veer-Korthof, and R S Weening); the German BFM Group (W Dörffel, C Niemeyer, F Berthold, M Rister, A Jobke, M Domula, H Wehinger, K Hofmann, concentrations of ୏10Ϫ4. These low degrees of MRD were F J Göbel, P Heidemann, J D Beck, N Graf, U Mittler, A Reiter, associated with progressively increasing relapse rates of J D Thaben, G Henze, R Dickerhoff, J Treuner, R Geib, P Exadactylos, 25−75% during the first year of treatment. Furthermore, U Bode, G Eggers, W Schröter, and C Tautz); the Austrian BFM Group(B Ausserer, F M Fink, R Jones, G Mann, G Müller, I Mutz, R Ploier, measuring MRD of ୏10Ϫ4 appeared to be important for N Pobegen, K Schmitt, and O Stöllinger); and the Italian AIEOP Group the MRD-based risk-group classification—ie, the (G Masera, V Conter, M G Valsecchi, F Sereni, M Aricò, E Madon, discrimination between patients at low risk and E Barisone, L Zanesco, M C Putti, M T Di Tullio, A Murano, P A Macchia, and C Favre). We also thank Daniëlle Korpershoek for secretarial support, Tar van Os for In this MRD study many follow-up samples of bone drawing up the artwork, and Joris Zwijgers for constructing the database.
marrow were collected to obtain insight into the kinetics of The collaborative study was supported by the BIOMED–1 programme ofthe European Commission—ie, the Concerted Action: investigation of tumour reduction and for finding out the relevance of minimal residual disease in acute leukaemia: international standardisation MRD detection at multiple time points. Although 84% of and clinical evaluation (grant BMH-CMT 94–1675; coordinators: THE LANCET • Vol 352 • November 28, 1998 J F San Miguel and J J M van Dongen). JJMvD, MJP-W, ERvW, and (MRD) in B-lineage acute leukemias using a simplified PCR strategy: MACdB were supported by the Dutch Cancer Foundation/Koningin evolution of MRD rather than its detection is correlated with clinical Wilhelmina Fonds (grant EUR 94–852) and the Ank Van Vlissingen outcome. Br J Haematol 1991; 79: 205–10.
Foundation; CRB was supported by the Deutsche Krebshilfe; AB, GC, and 18 Biondi A, Yokota S, Hansen-Hagge TE, et al. Minimal residual disease LC were supported by the M Tettamanti Foundation and by the Consiglio in childhood acute lymphoblastic leukemia: analysis of patients in Nazionale delle Ricerche (PF ACRO, grant 92·02140.PF·39); AB and GB continuous complete remission or with consecutive relapse. Leukemia were supported by the Associazione Italiana per la Ricerca sul Cancro 1992; 6: 282–88.
(AIRC); ERP-G, FS, and KH were supported by the Österreichische 19 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:
546–53.
20 Pongers-Willemse MJ, Seriu T, Stolz F, et al. Primers and protocols for Reiter A, Schrappe M, Ludwig W-D, et al. Chemotherapy in 998 standardized detection of minimal residual disease in acute unselected childhood acute lymphoblastic leukemia patients. Results lymphoblastic leukemia using immunoglobulin and T cell receptor gene and conclusions of the multicenter trial ALL-BFM 86. Blood 1994; 84:
rearrangements and TAL1 deletions as PCR targets. Report of the BIOMED-1 Concerted Action: investigation of minimal residual Veerman AJP, Hählen K, Kamps WA, et al. High cure rate with a disease in acute leukemia. Leukemia (in press).
moderately intensive treatment regimen in non-high-risk childhood 21 San Miguel JF, Parreira A, Wörmann B, Bartram CR, Janossy G, acute lymphoblastic leukemia: results of protocol ALL VI from the Van Dongen JJM. Investigation of minimal residual disease in acute Dutch Childhood Leukemia Study Group. J Clin Oncol 1996; 14:
non-lymphoblastic leukemia: international standardisation and clinical evaluation. In: Baert A-E, Baig SS, Bardoux C, et al, eds. European Childhood ALL Collaborative Group. Duration and intensity of Union biomedical and health research: the BIOMED–1 programme. maintenance chemotherapy in acute lymphoblastic leukaemia: overview Amsterdam: IOS Press, 1995: 372–73.
of 42 trials involving 12000 randomised children. Lancet 1996; 347:
22 Breit TM, Wolvers-Tettero ILM, Beishuizen A, Verhoeven M-AJ, Van Wering ER, Van Dongen JJM. Southern blot patterns, frequencies Van Dongen JJM, Breit TM, Adriaansen HJ, Beishuizen A, and junctional diversity of T-cell receptor δ gene rearrangements in Hooijkaas H. Detection of minimal residual disease in acute leukemia acute lymphoblastic leukemia. Blood 1993; 82: 3063–74.
by immunological marker analysis and polymerase chain reaction.
23 Breit TM, Beishuizen A, Ludwig W-D, et al. tal–1 deletions in T-cell Leukemia 1992; 6S1: 47–59.
acute lymphoblastic leukemia as PCR target for detection of minimal Bartram CR. Detection of minimal residual leukemia by the polymerase residual disease. Leukemia 1993; 7: 2004–11.
chain reaction: potential implications for therapy. Clin Chim Acta 1993; 24 Beishuizen A, Verhoeven M-AJ, Mol EJ, Van Dongen JJM. Detection 217: 75–83.
of immunoglobulin kappa light-chain gene rearrangement patterns by Campana D, Pui C-H. Detection of minimal residual disease in acute Southern blot analysis. Leukemia 1994; 8: 2228–36.
leukemia: methodologic advances and clinical significance. Blood 1995; 25 Van Dongen JJM, Wolvers-Tettero ILM. Analysis of immunoglobulin 85: 1416–34.
and T cell receptor genes. Part I: basic and technical aspects. Van Dongen JJM, Szczepa´nski T, de Bruijn MAC, et al. Detection of Clin Chim Acta 1991; 198: 1–91.
minimal residual disease in acute leukemia patients. Cytokines Mol Ther 26 Beishuizen A, De Bruijn MAC, Pongers-Willemse MJ, et al.
1996; 2: 121–33.
Heterogeneity in junctional regions of immunoglobulin kappa deleting Davis MM, Björkman PJ. T-cell antigen receptor genes and T-cell element rearrangements in B-cell leukemias: a new molecular target for recognition. Nature 1988; 334: 395–402.
detection of minimal residual disease. Leukemia 1997; 11: 2200–07.
Beishuizen A, Hählen K, Hagemeijer A, et al. Multiple rearranged 27 Seriu T, Hansen-Hagge TE, Erz DHR, Bartram CR. Improved immunoglobulin genes in childhood acute lymphoblastic leukemia of detection of minimal residual leukemia through modifications of precursor B-cell origin. Leukemia 1991; 5: 657–67.
polymerase chain reaction analyses based on clonospecific T cell 10 Steenbergen EJ, Verhagen OJHM, Van Leeuwen EF, et al. Distinct receptor junctions. Leukemia 1995; 9: 316–20.
ongoing Ig heavy chain rearrangement processes in childhood B- 28 Roberts WM, Estrov Z, Ouspenskaia MV, Johnston DA, McClain KL, precursor acute lymphoblastic leukemia. Blood 1993; 82: 581–89.
Zipf TF. Measurement of residual leukemia during remission in 11 Beishuizen A, Verhoeven M-AJ, Van Wering ER, et al. Analysis of childhood acute lymphoblastic leukemia. N Engl J Med 1997; 336:
immunoglobulin and T-cell receptor genes in 40 childhood acute lymphoblastic leukemias at diagnosis and subsequent relapse: 29 Coustan-Smith E, Behm FG, Sanchez J, et al. Immunological detection implications for the detection of minimal residual disease by PCR of minimal residual disease in children with acute lymphoblastic analysis. Blood 1994; 83: 2238–47.
leukaemia. Lancet 1998; 351: 550–54.
12 Ghali DW, Panzer S, Fischer S, et al. Heterogeneity of the T-cell 30 Stiller CA, Bunch KJ. Trends in survival for childhood cancer in Britain receptor δ gene indicating subclone formation in acute precursor B-cell diagnosed 1971–85. Br J Cancer 1990; 62: 806–15.
leukemias. Blood 1995; 85: 2795–801.
31 Van Dongen-Melman JEWM. Information booklet for parents of 13 Brisco MJ, Condon J, Hughes E, et al. Outcome prediction in children surviving cancer. Leukemia 1997; 11: 1799–806.
childhood acute lymphoblastic leukaemia by molecular quantification of 32 Van Dongen JJM, Hooijkaas H, Adriaansen HJ, Hählen K, residual disease at the end of induction. Lancet 1994; 343: 196–200.
Van Zanen GE. Detection of minimal residual acute lymphoblastic 14 Cavé H, Guidal C, Rohrlich P, et al. Prospective monitoring and leukemia by immunological marker analysis: possibilities and quantitation of residual blasts in childhood acute lymphoblastic limitations. In: Hagenbeek A, Löwenberg B, eds. Minimal residual leukemia by polymerase chain reaction study of δ and γ T-cell receptor disease in acute leukemia 1986. Dordrecht: M Nijhoff Publishers, 1986: genes. Blood 1994; 83: 1892–902.
15 Steenbergen EJ, Verhagen OJ, Van Leeuwen EF, et al. Prolonged 33 Adriaansen HJ, Jacobs BC, Kappers-Klunne MC, Hählen K, Hooijkaas persistence of PCR-detectable minimal residual disease after diagnosis H, Van Dongen JJM. Detection of residual disease in AML patients by of first relapse predicts poor outcome in childhood B-precursor acute use of double immunological marker analysis for terminal lymphoblastic leukemia. Leukemia 1995; 9: 1726–34.
deoxynucleotidyl transferase and myeloid markers. Leukemia 1993; 16 Gajjar A, Ribeiro R, Hancock ML, et al. Persistence of circulating blasts 7: 472–81.
after 1 week of multiagent chemotherapy confers a poor prognosis in 34 Brisco MJ, Sykes PJ, Hughes E, et al. Monitoring minimal residual childhood acute lymphoblastic leukemia. Blood 1995; 86: 1292–95.
disease in peripheral blood in B-lineage acute lymphoblastic leukemia. 17 Nizet Y, Martiat P, Vaerman JL, et al. Follow-up of residual disease Br J Haematol 1997; 99: 314–19.
THE LANCET • Vol 352 • November 28, 1998

Source: http://bmt-mrd.cards.cz/download/lancet.pdf