To the Editor:
In a recent issue of Blood, Loh et al1 reported a low frequency of TEL-AML1 fusion [translocation t(12;21)(p13;q22)] in relapsed childhood acute lymphoblastic leukemia (ALL) treated initially on four different DFCI ALL consortium protocols from 1981 until 1995. The incidence of 1 TEL-AML1–positive ALL among 28 B-cell precursor (BCP-) ALL relapses (3.6%) is exceptionally low, although in subsequent reports the frequency was approximately 10% (5 of 49; ALL subtype not specified)2 to 13% (4 of 30; BCP-ALL).3 These results are in contrast to previously communicated Japanese and German retrospective studies on the frequency of TEL-AML1 fusion in relapsed childhood ALL (19% to 28%) questioning the favorable risk feature of TEL-AML1rearrangement in initial ALL.4-6 These data underline the necessity of prospective analysis to clearly determine the relapse rate of TEL-AML1–positive BCP-ALL following specific frontline treatment.
In our recently published study,6 (the only analysis with a substantial number of patients to date) on incidence and prognostic significance of TEL-AML1 positivity in relapsed childhood ALL, 22.5% of relapsed BCP-ALL were TEL-AML1–positive (32 of 142 patients). TEL-AML1–positive patients had a significantly longer duration of complete remission (CR), and the majority relapsed off-therapy and were younger at initial diagnosis. Furthermore, probability of event-free survival (EFS) for TEL-AML1–positive children enrolled in multicentric Berlin-Frankfurt-Münster (BFM) ALL relapse trials ALL-REZ BFM 90/96 was significantly better.
The ongoing study used nested reverse transcriptase-polymerase chain reaction (RT-PCR) to determine the frequency and the prognostic impact of TEL-AML1 fusion in relapsed childhood ALL.6 All samples were subjected to duplex-PCR including the simultaneous amplification of an ABL transcript to confirm the presence of intact cDNA. Positive results were confirmed by analyzing at least one additional sample independently, and in 16 cases by fluorescence in situ hybridization (FISH). Updating our data, we have analyzed bone marrow samples from a total of 340 children with recurrent ALL retrospectively (117 patients) and prospectively (223 patients). Sixty-two of 310 (20%) relapsed BCP-ALL wereTEL-AML1–positive. None of the 30 relapsed T-cell lineage ALL showed a TEL-AML1 fusion. In prospective analysis performed since September 1996, 40 of 202 relapsed BCP-ALL samples wereTEL-AML1–positive (19.8%), and all 21 T-cell ALL wereTEL-AML1–negative. The frequency of TEL-AML1positivity at first relapse of BCP-ALL was 16.7% (28 of 168), and even in samples available at second relapse, 12 of 34 patients wereTEL-AML1–positive. A similar frequency was obtained for patients with first relapse of BCP-ALL screened prospectively and initially treated uniformly on a single BFM frontline protocol ALL BFM 90 (26 of 123; 21.1%). All TEL-AML1–positive patients had been assigned to standard or intermediate-risk arms of trial ALL BFM 90, whereas 15 of the TEL-AML1–negative patients belonged to the high-risk group. None of the TEL-AML1–positive patients have thus received epipodophyllotoxins reported to be involved in therapy-related secondary leukemia. The frequency of TEL-AML1fusion in relapsed BCP-ALL enrolled in ALL-REZ BFM trials treated initially on CoALL frontline protocols was similar (5 of 29; 17.7%).
Restricting the statistical analysis to children with first relapse of BCP-ALL enrolled in relapse trials ALL-REZ BFM 90/96, the proportion ofTEL-AML1 positivity was 18.5% (43 of 233).TEL-AML1–positive and –negative patients differed significantly with respect to duration of first CR (median, 45.8v 27.1 months; P < .0001), and age at initial diagnosis (median, 54 v 76 months; P = .02). The overwhelming majority of TEL-AML1–positive relapses occurred off-therapy (37 of 44; 84%) in contrast to 83 of 197 (42%;P < .0001) TEL-AML1–negative relapses. Identical results are obtained if only patients (35 TEL-AML1–positives among 193 patients) treated initially on BFM frontline trials ALL BFM 90 and 95 are considered.
Published data on the frequency of TEL-AML1 positivity in relapsed childhood BCP-ALL disclose an enormous variation between 3.6% to 28%. Major shortcomings of the majority of these retrospective studies are the inclusion of very limited numbers of relapsed ALL patients and short follow-up periods. The analysis of small numbers may be subject to strong inaccuracies and deviations per se, and, when performed retrospectively, to selection biases that may lead to misinterpretations. From our consecutive and almost population-based BFM trials on relapsed ALL with large numbers of patients analyzed forTEL-AML1, it has become very clear thatTEL-AML1–positive leukemia is biologically characterized by a long duration of first CR and that the vast majority of relapses occur off-therapy (median timepoint of relapse, 46; range, 12.4 to 123.5 months). Because cure rates with modern ALL protocols are between 70% and 80% worldwide and TEL-AML1 is only known since 1995, it may be superfluous to speculate at this point whether the differences on the incidence of TEL-AML1–positive relapsed ALL in the various studies are caused by the efficacy of frontline therapy (ie, applied drugs and dose intensity), selection biases, or methodological aspects. From our data it is also not likely that TEL-AML1positivity at relapse may represent a therapy-induced secondary leukemia because TEL-AML1 could be detected in all available samples from patients at first manifestation or subsequent relapse. From our point of view, the most likely explanation for the conflicting results on the incidence of TEL-AML1 positivity among patients with relapsed BCP-ALL is that they are based on small numbers of relapse patients and insufficiently long observation times of patients in first CR. Prospective screening of TEL-AML1 fusion in newly diagnosed and relapsed ALL is required to disclose the true relapse rate and to clarify the question of which specific modality during frontline treatment might have any additional impact on the incidence of TEL-AML1 positivity at relapse.
We are in complete agreement with Seeger et al that prospective analysis is required to resolve the conflicting literature reports on the prognostic impact of the TEL/AML1 gene rearrangement in pediatric ALL. We also support the view that, because of considerable variation in the reported prognostic significance of theTEL/AML1 gene rearrangement in retrospective analyses, alterations in therapeutic approaches to this subset of ALL patients should await prospective evaluation.
We concur that the reported variation in studies to date may be caused by any of the many biases inherent in retrospective analyses. However, it is unlikely that the favorable prognosis in our analysis is caused by short follow-up. In the study of patients with newly diagnosed ALL by McLean et al,1-1 there was 100% relapse-free survival forTEL/AML1-positive patients with a median follow-up of 8.3 years. Follow-up for this cohort has now been extended to 11.5 years with no TEL-AML1-positive relapses. In our study of patients with relapsed ALL,1-2 we extended the cohort to include patients initially treated on Dana-Farber Cancer Institute (DFCI) ALL Consortium protocol 91-01 (1991-1995). With these inclusions, the median follow-up for the relapsed patients analyzed was 38 months. Since publication, an additional patient enrolled on DFCI ALL Consortium 87-01 who did not have an analyzable sample in the recent report (Loh et al1-2) has relapsed a second time and isTEL/AML1-positive by RT-PCR and FISH. Thus, there are 2 relapsed patients out of 33 with evidence of the TEL/AML1fusion. Although the numbers are small, they are still statistically significantly different than those reported by Seeger et al1-3 and Harbott et al.1-4
Thus, there is a correlation at our institution between high relapse-free survival and low relapse frequency ofTEL/AML1-positive patients. In addition, other centers have also reported favorable prognosis associated with the TEL/AML1gene rearrangement.1-5-1-8 Therefore, it remains plausible that there is a true difference in outcome in the subgroup ofTEL/AML1-positive patients between centers. We do not believe it is superfluous or premature to discuss differences in the efficacy of upfront therapy as possible explanations for the different outcomes. It has already been shown that certain subtypes of ALL respond better to specific therapies. One classic example is mature B-cell leukemia, which is best treated using intensive, short regimens containing, among other agents, cyclophosphamide and cytarabine. Taken together, the data on the prognostic significance of TEL/AML1 may be guiding clinical investigators towards more specific therapy for these patients.
We agree that it is improbable that the higher incidence ofTEL/AML1-positive leukemia at relapse reflects a therapy-related malignancy, although it is a possibility that deserves mention given the different chemotherapeutic regimens these patients have received.
The additional data provided by Seeger et al corroborate their prior retrospective findings with prospective monitoring of relapsed patients treated on BFM or Co-ALL protocols. Their data adds further impetus to the need for prospective evaluation of newly diagnosed patients. Prospective analysis of the prognostic significance of theTEL/AML1 gene rearrangement is ongoing on the current DFCI ALL Consortium protocol (95-01). Of 223 patients with analyzable diagnostic peripheral blood or bone marrow samples, 53 of 223 (23%) areTEL/AML1-positive. From this experience, we are confident that the prognostic significance of TEL/AML1-positive patients treated on DFCI ALL Consortium protocols will be firmly established in the future. This effort, in conjunction with similar prospective analysis on BFM or Co-ALL protocols, should firmly establish the prognostic significance of the TEL/AML1 gene rearrangement in these centers.
Supported in part by the Howard Hughes Medical Institute and Grant No. CA68484 from the National Institutes of Health. D.G.G. is the Stephen Birnbaum Scholar of the Leukemia Society of America and an investigator in the Howard Hughes Medical Institute.
The molecular biology study is supported by a grant from the Deutsche José Carreras Siftung e.V., Munich, Germany, and the ALL-REZ BFM studies are supported by grants from the Deutsche Krebshilfe, Bonn, Germany.