We read with interest the report of Panzer-Grümayer et al1 in which the authors showed that evaluation of minimal residual disease (MRD) in childhood acute lymphoblastic leukemia (ALL) by semiquantitative molecular methods on day 15 of induction therapy can be implemented in their recently established MRD-based stratification. The authors were able to identify after only 2 weeks of treatment a patient population of 20% who may benefit from the least-intensive treatment. But for patients with higher levels of day 15 MRD (greater than or equal to 10−2 or 10−3), only the MRD-based risk groups as defined by later time points were predictive.
We performed a quantitative analysis of MRD by means of the real-time quantitative PCR (RQ-PCR). This approach is far more accurate than semiquantitative analysis of MRD, which was used by Panzer-Grümayer et al.1 We studied 17 children with B-precursor ALL treated according to protocol VIII of the Dutch Childhood Leukemia Study Group (DCLSG) in the Emma Children's Hospital/AMC (Amsterdam, Netherlands). Immunophenotyping and cytogenetic analysis were performed at diagnosis and relapse, according to standard techniques. Nine patients were in continuous complete remission (CCR), and 8 patients had relapsed (follow-up of at least 60 months after diagnosis). All patients received the same induction therapy. Bone marrow samples of all patients, taken at day 15, at the end of induction therapy and before consolidation, were analyzed for MRD with IGH- and TCRD-rearrangements as PCR targets. For quantification of residual disease, patient-specific forward primers complementary to the junctional region, in combination with consensus reverse primers and consensus Taqman probes (JH probes for IGH rearrangements and Dδ3-probe for Vδ2Dδ3 rearrangements), were used as described earlier.2
Results of MRD levels obtained with RQ-PCR at the 3 time points are shown in the Table. We found a significant difference between MRD levels at day 15 of the patients in CCR, compared with the relapsed patients (P < .05). MRD levels after 15 days of chemotherapy were in the CCR group in the range of 0%-1.9%, with one patient negative. For the patients who suffered from relapse, the MRD levels at this time point were significantly higher (2.6%-73%). In agreement with Panzer-Grümayer et al,1 all patients with rapid molecular response at day 15 (MRD level less than 10−4) are still in CCR. But in contrast to their results, we were able to identify also patients at high risk for relapse. With the quantitative RQ-PCR, it is possible to differentiate patients with relatively high tumor loads, and our results suggest that this accurate quantification is necessary for the identification for patients at high risk of relapse. There is also a clearly significant difference in MRD levels at the end of induction therapy for the patient group still in CCR, compared with the group that suffered from relapse (P < .05). MRD levels at the end of induction therapy showed levels varying from 0.24% to 2.7% for the relapsed patients. In 2 patients there was no material available at this time point. For the CCR patients the MRD levels at this time point were in the range of 0%-0.15%. In 3 patients in the CCR group MRD was not detectable. We compared our data with the recently established MRD risk-group stratification,3 based upon MRD results at the end of induction therapy and before consolidation. Based upon all available material from patients, all our CCR patients were classified into the low-risk group, and all relapsed patients into the intermediate- or high-risk groups. The study of van Dongen et al3 demonstrated that combined information on MRD from the first 3 months of treatment distinguishes patients with good prognosis from those with poor prognosis. The combined measurement of MRD levels at 2 time points was found to be more predictive than the MRD level at a single time point. This might be due to inaccuracy of the techniques applied for the quantification of MRD, or it might reflect different responses to therapy. Future studies with quantitative PCR assays, as we have used in this group of patients, will provide this information.
In our group of patients, we also determined whether the combined measurement of 2 time points was more predictive than a single time point. The precise quantification of MRD levels enabled us to determine the slope of the disappearance curve of the leukemic cells between diagnoses, at day 15 and at week 5. Significant differences for both CCR and relapsed patients were found by a linear regression, performed on log transformed data (see Figure). The slopes were tested for significance of differences (Graphpad Prism version 3.00, Graphpad, San Diego, CA). The slopes of the patient group still in remission and of the patient group that suffered a relapse were significantly different from each other (P < .001). This confirms the different clearance of residual cells in relapsed versus CCR patients.
In conclusion, our quantitative PCR results suggest that it might already be possible to discriminate between good- and poor-risk patients by precise quantification of MRD level at the end of induction therapy. But it has to be investigated in prospective studies whether indeed a clear cut-off level will be found. Furthermore, clinicians might prefer to perform stratification of therapy based on determination of MRD at 2 time points. Our results suggest that, although the start of consolidation can also be used as a second time point, quantitative MRD results at day 15 are already highly informative in this respect. The major advantage of this latter approach is the earlier availability of the prognostic information, and this information can be used to modify treatment according to risk group.
