Incidence of TEL/AML1 Fusion in Children With Relapsed Acute Lymphoblastic Leukemia

THE TEL/AML1 REARRANGEMENT in pediatric acute lymphoblastic leukemia (ALL) is the consequence of t(12;21)(p13;q22).1,2 The gene rearrangement is cryptic by standard cytogenetics but can be detected using reverse transcriptase-polymerase chain reaction (RT-PCR), Southern blot, or fluorescence in situ hybridization (FISH). TEL/AML1 occurs in 20% to 25% of children with B-progenitor ALL and is the most common gene rearrangement yet identified in any pediatric leukemia.1,3-13 The TEL/AML1 fusion frequently occurs in patients previously characterized as low or standard risk. A retrospective analysis of 81 patients treated on Dana-Farber Cancer Institute (DFCI)-ALL Consortium protocols demonstrated that 22 (27%) were TEL/AML1-positive.3 Of these, 11 patients were treated as high risk, although with current risk criteria, 7 patients would be identified as high risk.14 Moreover, 100% of the 22 TEL/AML1-positive patients remained in complete continuous remission (CCR) with no relapses observed at a median follow-up of 8.3 years.3 One TEL/AML1-positive patient died of a brain tumor. Sixteen of 54 TEL/AML1-negative patients relapsed.3 

Other investigators have also reported favorable outcomes forTEL/AML1-positive patients in both retrospective analyses and prospective analyses with relatively short follow-up (Table 1).4,6,7,12,13,15,16However, at least four analyses of patients with relapsed ALL have shown that the frequency of TEL/AML1 positivity at relapse is similar to that at diagnosis (Table2).7,17-19 One study retrospectively identified 9 of 35 (26%) patients who were TEL/AML1-positive at relapse.17 In another retrospective analysis, 32 of 146 (22%) patients enrolled on relapse protocols of the Berlin-Frankfurt-Munster (BFM) group were TEL/AML1-positive at first or second relapse.18 Although the incidence ofTEL/AML1-positive patients at relapse approximates that reported at diagnosis, both studies showed that the duration of initial remissions was longer in TEL/AML1-positive patients.17,18,TEL/AML1-positive relapsed patients also had a significantly higher probability of event-free survival after relapse therapy.18 In a smaller study of 16 relapsed cases of B-progenitor leukemia, 3 patients who experienced late relapses (>30 months) had evidence of TEL/AML1rearrangement.7 It is not clear if these patients were screened for the TEL/AML1 rearrangement at initial diagnosis. In a study that assessed the use of the TEL/AML1 fusion transcript as a marker of minimal residual disease, 2 of 7 patients relapsed with TEL/AML1 rearrangements in their bone marrow.19 

Taken together, these analyses show that the frequency of theTEL/AML1 rearrangement in relapsed patients is similar to that reported at diagnosis. This suggests that TEL/AML1 positivity may not be as favorable a prognostic indicator as previously reported.3,4,6,12,15 16 To clarify the apparent discrepancy between the favorable prognosis of TEL/AML1 in our previous study and reports of high TEL/AML1 frequency at relapse, we screened for the TEL/AML1 fusion using RT-PCR from available bone marrow from relapsed ALL patients initially treated on DFCI ALL Consortium protocols from 1981 through 1995.

Between January 1, 1981 and December 31, 1995, 683 children (<18 years of age) with newly diagnosed ALL were treated on four consecutive DFCI ALL Consortium protocols (81-01, 85-01, 87-01, and 91-01) at three institutions: Boston Children’s Hospital and the Dana-Farber Cancer Institute, International Hospital of Puerto Rico, and, for 81-01 only, Eastern Maine Medical Center. The initial diagnosis was made at the treating institution. Informed consent for sample acquisition and treatment was obtained from parents or guardians at the time of diagnosis. The therapy delivered to patients on the four DFCI ALL Consortium protocols is summarized in Table3. Details of protocols 81-01 and 85-01 have been previously published.20 21 Treatment was based on risk group assignment at the time of diagnosis (Table4).

A total of 125 patients initially treated at these three institutions subsequently relapsed after achieving first clinical remission. Bone marrow samples obtained at diagnosis and relapse were sent to the DFCI immunophenotyping lab and tumor bank. Only those patients with cells obtained in excess of immunophenotyping requirements had samples viably frozen. Samples were selected without knowledge of age at diagnosis, sex, race, immunophenotype, cytogenetics, initial risk stratification, or duration of first clinical remission. When available, paired samples from initial diagnosis and relapse were analyzed.

Forty-seven patients had 64 available cryopreserved bone marrow samples obtained from the DFCI tumor bank. Samples from 8 of these patients failed to yield analyzable RNA. Thirty-nine patients with cryopreserved bone marrow samples yielded RNA of sufficient quality for analysis ofTEL/AML1 gene rearrangement. In summary, a total of 51 samples from these 39 patients had adequate RNA for analysis.

Total RNA was extracted from cryopreserved bone marrow or peripheral blood mononuclear cells using guanidinium/acid phenol extraction (RNA STAT-60; Tel-Test, Friendwood, TX) according to the manufacturer’s instructions. Total RNA (4 μg) was reverse transcribed as previously described.22 An aliquot of cDNA was then used in a control PCR reaction to verify the integrity of the RNA sample using TELspecific primers 458 (5′AGGTCATACTGCATCAGAAC3′) and 750R (5′ATTATTCTCCATGGGAGACA3′) to amplify a 292-bpTEL fragment spanning exons 4 and 5. Forty cycles of PCR (94°C for 1 minute, 56°C for 1 minute, and 72°C for 1 minute) were performed as previously described on an MJ Research thermal cycler (Watertown, MA), and 10 μL of the PCR product was visualized on 2.5% agarose gels stained with ethidium bromide. Samples that were negative for the control TEL expression were excluded from further analysis.

Amplification of the TEL/AML1 fusion was performed in a nested PCR reaction. First-round PCR used TEL primer 937 (5′AACCTCTCTCATCGGGAAGA3′) and AML1 primer 1142R (5′CAGAGTGCCATCTGGAACAT3′). Forty cycles of PCR were performed at an initial denaturing step of 94°C for 5 minutes, followed by cycles of 94°C for 1 minute, 62°C for 1 minute, and 72°C for 1 minute. Four microliters of PCR product was reamplified with second-round PCR using TEL primer 969 and AML1 primer Z3R.TEL primer 969 (5′GAACCACATCATGGTCTCTG3′) and AML1 primer Z3R (5′AACGCCTCGCTCATCTTGCCTG3′) amplify a 174-bp fragment in 40 cycles of PCR (initial step of 94°C for 5 minutes, followed by cycles of 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute). Subsequent analysis of the PCR product (10 μL) was visualized on 1% agarose/3% NuSieve (Pharmacia, Piscataway, NJ) gels stained with ethidium bromide.

All assays were confirmed at least once. The positive control was RNA extracted from the Reh cell line that harbors the TEL/AML1 gene rearrangement.23 24 Negative controls contained PCR reaction mixture without added cDNA. Appropriate positive and negative controls were used in each reaction. Template cDNA was added in a separate lab after the PCR master mix was aliquoted into 96-well reaction plates.

The Fisher’s Exact25 test was used for comparison of categorical data, and the Wilcoxon Exact test was used when the categories were ordered.25 Mantel’s log rank test was used to compare duration of first remission between relapsed patients who were tested and relapsed patients who were not tested.26The survival distribution for time to relapse data was estimated according to Kaplan and Meier.27 

Fifty-one cryopreserved bone marrow samples from 39 patients treated on four consecutive DFCI protocols provided quality RNA for analysis. Paired bone marrow samples were available in 13 patients at initial diagnosis and relapse, in 19 patients at relapse only, and in 7 relapsed patients at initial diagnosis only. Four of the 13 patients for whom bone marrow samples were available at initial diagnosis and at relapse were previously analyzed at diagnosis and have been reported.3 The seven relapsed patients who had bone marrow samples only available at initial diagnosis are reported here but are not included in the group that was tested. For statistical analysis, these 7 patients are included in the 93 patients who were not tested. The presenting characteristics of the remaining 32 patients analyzed are presented in Table 4 and are compared with the presenting characteristics of the 93 relapsed patients who were not tested. No significant differences were found in age at diagnosis, immunophenotype, sex, presence of central nervous system (CNS) leukemia, or presence of t(9;22) between the two groups. Analyzed patients were more likely to have been treated as high risk or very high risk (P = .04). When using current DFCI risk classification criteria (high risk includes the presence of any one of the following: age <1.0 years or ≥10 years, white blood cell count [WBC] ≥50,000/μL, T-cell phenotype, presence of anterior mediastinal mass, any CNS leukemia), this difference is no longer significant (P = .22).14 There was a significantly longer duration of first clinical remission in the 32 patients who were tested compared with the 93 patients who were not tested (log rank,P = .01; Table 4). The median follow-up of the remaining 558 patients who were treated on these protocols and did not relapse was 7.8 years.

Only 1 relapsed patient of 32 (90% confidence interval, 0% to 14%) was found to be TEL/AML1-positive. Additionally, none of the 7 relapsed patients with samples at initial diagnosis only wasTEL/AML1-positive. The relapsed patient harboring theTEL/AML1 fusion was a male diagnosed at 43 months of age with a B-progenitor, CD10⁺ phenotype with a presenting WBC of 40,200/dL and normal cytogenetics. He was stratified and treated as high-risk on DFCI 91-01 because of a WBC greater than 20,000/μL at presentation (Table 3). The patient relapsed in bone marrow and cerebrospinal fluid (CSF) after 23 months of CCR, 2 months before completion of continuation therapy. Reinduction was unsuccessful, and he underwent allogeneic bone marrow transplantation with persistent leukemia. He died of progressive bone marrow and CNS disease with concomitant bacterial sepsis approximately 6 months after initial relapse. A bone marrow sample at initial diagnosis wasTEL/AML1-positive. Peripheral blood mononuclear cells were positive for the TEL/AML1 fusion on day 29 after bone marrow transplantation, 2 weeks before obtaining a bone marrow aspirate demonstrating refractory ALL with 58% lymphoblasts. Although no blood samples were available at initial relapse, the patient had refractory ALL throughout the last 6 months of his life and never achieved a clinical remission.

Comparison of these results with other reported analyses ofTEL/AML1 frequency in relapsed patients is shown in Table 5. Clinical characteristics of the patients reported in other series appear to be similar to those reported here. The median duration of first clinical remission for all relapsed patients and all tested patients appears in concordance with other reports. However, there is a significant difference between the incidence of TEL/AML1 in relapsed patients with B-progenitor ALL treated on DFCI ALL Consortium protocols and the incidence ofTEL/AML1 in relapsed patients with B-progenitor ALL reported by other investigators.

Only 1 of 32 relapsed patients initially treated on DFCI ALL Consortium protocols was TEL/AML1-positive. When considering only patients with a B-progenitor phenotype that did not have the presence of t(9;22), only 1 of 28 (3.6%) patients was TEL/AML1-positive. This low frequency is consistent with our previous study, in which a retrospective analysis of newly diagnosed pediatric ALL patients showed 0 of 22 relapsed TEL/AML1-positive patients with 8.3 years of median follow-up.3 Our results are significantly different when compared with other reported analyses of the incidence ofTEL/AML1 in relapsed patients (Table 5).

There are several potential explanations for the differences between our results and the reports of other investigators. First, there might be a selection bias in our analysis, because samples were tested retrospectively based on availability. It is possible that the incidence of TEL/AML1 rearrangement is lower because a disproportionately higher number of high/very high-risk patients had samples available for analysis (Table 4). However, the clinical features of our study cohort appear similar to those reported by other investigators, including the median length of remission.17,18,TEL/AML1 positivity at relapse has been associated with a longer duration of remission.7,17 18The majority of our study cohort (60%) had a clinical remission that exceeded 36 months. Therefore, it is unlikely that selection bias alone explains the low incidence of TEL/AML1 positivity in our series. To fully address the potential problem of selection bias inherent in this and other retrospective studies, we are prospectively analyzing the prognostic significance of TEL/AML1 in all newly diagnosed patients with ALL treated on the current DFCI ALL Consortium protocol (95-01).

Second, it is notable that half of the TEL/AML1-positive patients identified in our previous report were treated as high risk based on a presenting WBC greater than 20,000/μL. The treatment that these patients received may have been more intense than patients treated on other protocols reporting a higher relapse rate who may use a higher WBC for risk classification. Only those patients meeting other standard risk criteria with a WBC greater than 50,000/μL will be treated as high risk on the current DFCI ALL Consortium protocol. Our prospective analysis should determine whether the change in risk stratification based on WBC will influence the outcome ofTEL/AML1-positive patients. However, it should be noted that the 1 TEL/AML1-positive patient in the currently reported analysis was treated as high risk and still failed to sustain a clinical remission.

Third, it is possible that the high incidence of TEL/AML1 at relapse reflects a therapy-related malignancy that isTEL/AML1-positive; that is, the patients were TEL/AML1negative at initial diagnosis but were TEL/AML1-positive at relapse. Many of the patients in the other reports were only analyzed at relapse and not at initial diagnosis.7,17,18 To date, there has been 1 case report of a 4-year-old patient with a B-progenitor, CD10⁺ ALL whose diagnostic cytogenetics demonstrated del(6q) and no abnormalities of 12p.28 Bone marrow obtained at relapse 7 years after CCR demonstrated absence of the original del(6q) clonal abnormality. Instead, a subtle deletion of 12p on the relapse karyotype led to FISH analysis, which detected the t(12;21) rearrangement.28 Subsequent RT-PCR identified theTEL/AML1 gene rearrangement.1 There was no bone marrow available to determine whether the patient wasTEL/AML1-positive at initial diagnosis.

Although the TEL/AML1 rearrangement has not been shown to occur in therapy related leukemias, it is usually a cryptic translocation and may have yet to be identified. Balanced translocations involvingAML1 with other fusion partners (ETO, EVI1) have been demonstrated in therapy-related acute myeloid leukemias (t-AML), particularly in association with exposure to topoisomerase II inhibitors, including epipodophyllotoxins and anthracyclines.29-33 In fact, the majority of relapsed patients reported by Seeger et al18 and Harbott et al17 were treated initially on BFM or Co-ALL protocols, both of which include therapy with these agents.34-36Additionally, the BFM and CoALL protocols also incorporate the use of alkylating agents that have an association with myelodysplasia and t-AML demonstrating unbalanced translocations.29 34-36Whereas the patients treated on DFCI ALL Consortium protocols received anthracyclines, none has received epipodophyllotoxins or alkylating agents as part of their initial therapy (Table 3).

Finally, it is possible that the low incidence ofTEL/AML1-positive relapses in this study reflects differences in the efficacy of the up-front therapy TEL/AML1-positive patients. Historically, the DFCI and BFM treatment programs have achieved similar outcome results in children with newly diagnosed ALL, but with different treatment strategies.34 The intensive BFM regimens, which include 2 months of induction therapy and a delayed reinduction phase, use a greater number of drugs.34,36 The DFCI regimens, distinguished by early consolidation with intensive asparaginase for all patients and doxorubicin for higher risk patients, have used fewer agents but at higher cumulative dosages.20,21 34 It is possible thatTEL/AML1-positive patients represent a biologically distinct subset of patients whose leukemia is more effectively treated by the agents used more intensively by the DFCI group, such as asparaginase. Ongoing prospective trials may help clarify whether a particular treatment strategy more effectively treats patients withTEL/AML1-positive ALL. Additionally, the development of quantitative minimal residual disease assays may prove crucial to following TEL/AML1-positive patients on different protocols.

At a minimum, these data strongly suggest that the presence ofTEL/AML1 at diagnosis is a favorable prognostic indicator but that treatment specific variables may influence outcome in this potentially curable cohort. Taken together with other published reports, our data would not support consideration of decrements in therapy for TEL/AML1-positive patients at this time. Rather, they emphasize the ongoing need to confirm the prognostic significance of TEL/AML1 prospectively and to develop sensitive quantitative minimal residual disease assays to follow TEL/AML1-positive patients on therapy.

The authors thank Virginia Dalton, Gaylord Garroway, Jennifer Peppe-Bonasera, and Stacey Waters for assistance in obtaining clinical data and members of the Connell-O’Reilly Cell manipulation and Gene transfer laboratories at the Dana-Farber Cancer Institute for their assistance in obtaining samples.