Clinical features, early treatment responses, and outcomes of pediatric acute lymphoblastic leukemia in china with or without specific fusion transcripts: A single institutional study of 1,004 patients


  • Conflict of interest: Nothing to report


Acute lymphoblastic leukemia (ALL) with distinct fusion transcripts has unique clinical features. In this study, the incidence, clinical characteristics, early treatment response, and outcomes of 1,004 Chinese pediatric ALLs were analyzed. Patients with TEL-AML1 and E2A-PBX1 fusion genes or other B cell precursor ALLs (BCP-ALL) had favorable clinical features, were sensitive to prednisone, had low minimal residual disease (MRD), and an excellent prognosis, with a 5-year event-free survival (EFS) of 84–92%. T-ALL was associated with a high WBC, increased age, more central nervous system involvement, a poor prednisone response, and high MRD, with a 5-year EFS of 68.4 ± 5.2%. Patients with BCR-ABL and MLL rearrangements usually had adverse clinical presentations and treatment responses, and a dismal prognosis, with 5-year EFS of 27.3 and 57.4%, respectively. We also showed that BCR-ABL and MLL rearrangements, the prednisone response, and MRD were independent prognostic factors. Interestingly, the BCH-2003 protocol resulted in a better outcome for E2A-PBX1+ patients than the CCLG-2008 protocol. Intermediate and late relapses were more common in TEL-AML1+ patients and other BCP-ALLs compared with other subgroups (P = 0.018). Therefore, this study suggests that a fusion gene-specific chemotherapy regimen and/or targeted therapy should be developed to improve further the cure rate of pediatric ALL. Am. J. Hematol. 2012. © 2012 Wiley Periodicals, Inc.


Acute lymphoblastic leukemia (ALL) is the most common malignancy in children. The treatment outcome has been profoundly improved over recent decades, leading to a 5-year event-free survival (EFS) in ∼85% of children and adolescents with ALL [1–7]. This impressive achievement is mainly due to increased intensity of treatment, identification of effective combination chemotherapy, and prophylaxis of central nervous system (CNS) leukemia. The introduction of risk stratification to modulate conventional chemotherapy strategies has significantly improved the management as well [8]. This stratification is based on variables of prognostic importance, such as age, initial leukocytosis, immunophenotype, and cytogenetics. Nevertheless, ∼20% of patients relapse, and the outcome of relapsed ALL remains dismal [9]. Therefore, stringent risk assessment has become an important issue to improve survival of patients at high risk and decrease long-term treatment-related side effects in standard risk patients.

In the last two decades, several molecular methods have been developed to allow identification of leukemia-associated gene abnormalities [10]. These genetic lesions, which are currently detected in ∼50% of patients with ALL, represent unique clinical and biological features, and thus are one of the predictors of clinical outcome [11]. Patients with TEL-AML1 or E2A-PBX1 transcripts have favorable clinical features and excellent outcomes [12], while BCR-ABL or MLL rearrangements are always associated with adverse clinical features and prognosis [13, 14]. However, these observations and conclusions are primarily derived from patients with a Caucasian background, and the situation in the Chinese population has not been thoroughly studied. To date, no systematic study in a large Chinese cohort of pediatric ALL patients has been documented.

In this study, the clinical features and biological characteristics of pediatric ALL patients with or without specific fusion transcripts were summarized. The early response to treatment and long-term outcomes were also analyzed with respect to the individual fusion gene. In addition, the prognostic values of these fusion genes in pediatric ALL were investigated.

Design and Methods

Patients and treatment protocols

The study included 1,004 newly diagnosed pediatric ALL patients, 4 months to 16 years of age (median, 5 years), enrolled in the Hematology Center at Beijing Children's Hospital of Capital Medical University between January 2003 and February 2010; 902 cases were B-lineage ALL and 102 cases were T-lineage ALL. Eighty-two patients did not continue treatment in our center after the initial diagnosis was made. The BCH-2003 (used between January 2003 and March 2008) and CCLG-2008 protocols (used after April 2008) were followed in 651 and 271 pediatric ALL patients, respectively (see Supporting Information Tables I–III for a description of the stratification and treatment regimens of the two protocols). Among these 922 patients, 880 achieved hematologic complete remission (CR) after induction, 14 patients had induction failure, 4 patients died of severe infections during induction, and the remaining 24 patients did not complete induction in our hospital. The BCH-2003 and CCLG-2008 protocols were approved by the Beijing Children's Hospital Institutional Ethics Committee. Informed consents were signed by the guardians of each patient.

Morphologic analysis and determination of immunophenotype and cytogenetic abnormalities

Wright-stained bone marrow (BM) aspirate smears and histochemical staining from all patients were reviewed. Patient BM mononucleated cells were separated by Ficoll-Hypaque density gradient centrifugation. Immunophenotyping was performed following the guideline described previously [15], and a standard panel of monoclonal antibodies (Supporting Information Table IV) was used in all patients. The criteria for ALL diagnosis were based on the European Group for the Immunological Characterization of Leukemias (EGIL) scoring system. Conventional cytogenetics analysis was performed in all cases at the time of diagnosis. Chromosomes were G-banded on BM cells from direct and/or 24-hr unstimulated cultures. Karyotypes were described according to the 2005 International System for Human Cytogenetic Nomenclature criteria [16]. Cytogenetic analysis was considered successful if a clonal chromosomal abnormality was detected or at least 20 metaphases were analyzed. If a clonal aberration was absent and ≥20 metaphases had a normal karyotype, the patient was considered to be cytogenetically normal.

RNA extraction, cDNA synthesis, and fusion transcript analysis

Total RNA was extracted with Trizol (Invitrogen Life Technologies, Carlsbad, CA), according to the manufacturer's instructions. The concentration and quality of RNA was determined by absorbance measurements at 260 and 280 nm. Total RNA (2 μg) from patient samples was reverse transcribed into cDNA using random hexamers (Promega, Madison, WI), according to the manufacturer's instructions.

A multiplex reverse transcription polymerase chain reaction (RT-PCR) system, as described by Pallisgaard et al. [10], was adapted with modifications. The system was able to detect simultaneously the following fusion transcripts of ALL: BCR/ABL p190 (e1a2) and p210 (b2a2, b3a2) isoforms; and E2A/PBX1, E2A/HLF, TEL/AML1, SIL/TAL1, TLS/ERG, and SET/CAN with MLL rearrangements, including MLL/AF4, MLL/ENL, MLL/AF9, and MLL/AF10. More details of the setup and conditions of multiplex nested PCR are stated in Supporting Information Table V. The ubiquitously expressed E2A gene was chosen as an internal control. At the end of PCR amplification, 5 μL of each of 8 PCR products were analyzed on 6% polyacrylamide gel electrophoresis for 45 min at 300 V. The gel was stained with silver nitrate. The possible presence of a specific fusion transcript after the multiplex nested PCR was determined and verified, and the split-out PCR, which was the same as the second PCR with the exception of individual fusion transcript-specific primers, was performed.

Minimal residual disease (MRD) analysis

Rearrangements of immunoglobulin (Ig) and the T-cell receptor (TCR) gene were detected at the time of diagnosis by multiplex PCR according to the BIOMED-2 standardized protocol [17]. Monoclonal PCR products were directly sent to Shanghai Sangon Biological Engineering Technology & Service for DNA sequencing with an ABI PRISM 3730 automated sequencer. After comparison within an IMGT database, an allele-specific oligonucleotide (ASO) based on the sequence of the junctional region was designed and synthesized.

MRD analysis was performed using an ABI 7000 sequence detection system (Applied Biosystems, Foster City, CA). The reaction mixtures contained 1 × QPCR ROX Mix (AB Gene, England), 3.125 pmol of ASO and germline primers, 1.25 pmol of probe, 0.3% bovine serum albumin, and 600 ng of DNA. The reaction was carried out at 95°C for 15 min, followed by 50 cycles at 95°C for 15 sec, and 1 min at 54–66°C. To determine the efficiency of amplification and sensitivity of MRD PCR targets, diagnostic DNA was serially diluted 10-fold in normal MNC DNA (from 10−1 to 10−5), and used in a real-time quantitative polymerase chain reaction (RQ-PCR) to generate a standard curve. The follow-up samples were subjected to RQ-PCR analysis in triplicate [18]. Normal MNC DNA was used as a negative control and amplified in triplicate. Patient-specific IgH, IgK, IgL, Kde, TRB, TRG, and TRD gene rearrangements were used as RQ-PCR targets [19–25]. To check the quantity and quality of DNA in the assay, the N-ras gene was used as a control gene for RQ-PCR analysis [26]. The MRD level was expressed as the normalized value. The RQ-PCR settings and interpretation of the results were performed following the guidelines of the European Study Group on MRD detection in ALL (ESG-MRD-ALL) [27].

At time point 1 (TP1; at the end of induction and day 33 after beginning chemotherapy) and time point 2 (TP2; before consolidation and day 78 after beginning chemotherapy), the MRD was analyzed by 1–2 markers. In this study, 536 samples at TP1 and 490 samples at TP2 were examined for MRD.

Criteria for assessment of early response to treatment

The early response to treatment were estimated by the response to prednisone, remission status of BM (day 33), and the MRD at TP1 and TP2. The poor response to prednisone was defined as the absolute value of peripheral blasts > 1,000/μL. Hematologic CR was defined as normal BM cellularity with <5% undifferentiated cells and normalization of peripheral blood counts. Patients with a MRD < 10−4 at TP1 were considered to have a good response to early treatment. A MRD ≥ 10−2 at TP1 or a MRD ≥ 10−3 at TP2 represented a poor response.

Statistical analysis

December 31, 2011 was chosen as the reference date for the end of data collection for statistical analysis purposes. Comparisons between study groups with or without distinct fusion transcripts and associations between patient pretreatment characteristics and response to treatment were evaluated by nonparametric tests. EFS was defined from the date of diagnosis to the date of relapse, death, or induction failure, whichever came first, or the last contact with patients in continuous hematologic CR. Relapse was defined as the reappearance of leukemic cells in BM (>5% blasts) and/or the reappearance of clinical evidence of the disease. Relapse was divided into early- (<18 months), intermediate- (18–36 months), and late-relapse (≥36 months) [28]. Induction failure was defined as not achieving hematologic CR after remission induction therapy. EFS distribution with or without a distinct fusion transcript was estimated with the Kaplan-Meier procedure; comparisons between groups were performed with the log-rank test. The Cox proportional hazards regression model was used to evaluate the significance of differences in survival among the distinct fusion subgroups and other clinical indicators. All tests were two-sided with a P < 0.05 considered statistically significant. SPSS 16.0 software was used for all statistical analyses. The number of patients and differences between those included or not included in each statistical analysis were described in Supporting Information Table VI.


Incidence and biological characteristics of different fusion transcripts in ALL

There were 372 patients who carried distinct fusion transcripts out of 1,004 children with ALL, giving an incidence rate of 37.05%, and involving TEL-AML1, E2A-PBX1, BCR-ABL, SIL-TAL1, MLL rearrangements, E2A-HLF, TLS-ERG, and SET-NUP214 fusion (see Supporting Information Table VII for a description of the incidence of each fusion transcript). TEL-AML1, BCR-ABL, E2A-PBX1, E2A-HLF, and TLS-ERG were only present in B-lineage ALL, while SIL-TAL1 and SET-NUP214 were only found in T-lineage ALL. Although the MLL rearrangements were noted in both B- and T-lineage ALLs, MLL-AF4, MLL-AF9, and MLL-AF10 were correlated with the B lineage, MLL-AF6 was correlated with the T lineage, and MLL-ENL existed in both B and T lineages. All of the patients with MLL rearrangements in B-lineage ALL were CD10-negative.

Clinical presentation

We divided the patients into six subgroups (TEL-AML1+, E2A-PBX1+, BCR-ABL+, MLL rearrangement+, other B cell precursor ALL [other BCP-ALL], and T-lineage ALL [T-ALL]), and found that patients in each subgroup had distinctive clinical features and biological characteristics at the time of initial diagnosis. The results indicated that patients with TEL-AML1+ or other BCP-ALL usually had favorable clinical features, such as low WBC counts, age between 1 and 10 years, and rare CNS involvement, while one-half of the E2A-PBX1+ patients and most of the patients with BCR-ABL+, MLL rearrangement+, and T-ALL had adverse clinical features at the time of presentation. Although boys outnumbered girls in each fusion subgroup, only MLL rearrangements and T-ALL subgroups showed a gender preference, with a significantly higher proportion of boys (Table I).

Table I. Clinical Presentations, Treatment Responses, and Outcomes of ALL Patients in Different Fusion Subgroups
Patient characteristics and outcomeAll patientsTEL-AML1E2A-PBX1BCR-ABLMLL rearrangementsOther BCP-ALLT-ALLP value
  • a

    The EFS between the BCH-2003 and CCLG-2008 protocols in each fusion subgroups was compared, and it was shown that only the outcome of the E2A-PBX1+ patients treated with BCH-2003 was better than CCLG-2008 (P = 0.043).

  • b

    The P value represents a comparison between early and intermediate/late relapse among different fusion subgroups.

  • c

    There were three patients who relapsed in the bone marrow and extramedullary simultaneously.

Number of patients1,00419965552456299
WBC count
 ≥50 × 109/L7961744831649740<0.001
 <50 × 109/L208251724186559
Age (years)
Prednisone response
Induction of remission
MRD at day 33
MRD at day 78
 Total patients65113639321236765<0.001
 Number of events108832155121
 3-Year EFS85.2 ± 1.4%95.3 ± 1.9%92.2 ± 4.3%a24.7 ± 8.7%58.3 ± 14.2%89.4 ± 1.6%68.2 ± 5.9%
 5-Year EFS82.6 ± 1.5%93.3 ± 2.3%92.2 ± 4.3%a20.6 ± 8.1%58.3 ± 14.2%86.9 ± 3.0%68.2 ± 5.9%
 Total patients2715321177152210.002
 Number of events434563196
 3-Year EFS82.9 ± 2.4%92.3 ± 3.7%73.5 ± 10.4%55.0 ± 14.1%57.1 ± 18.7%86.9 ± 3.0%68.8 ± 10.7%
Relapse time
Relapse site
 Bone marrowc806510740120.108
 Isolated extramedullary19413191

Early treatment response

Prednisone response.

Patients with different types of fusion transcripts had diverse responses to prednisone (P < 0.001; Table I). Greater than 95% of patients in the TEL-AML1+, E2A-PBX1+, and other BCP-ALL subgroups were good responders, while the percentages were <90% in the other subgroups. In addition, we also analyzed the correlation between the response to prednisone and clinical features, and showed that the response to prednisone was related to the WBC count, age, gender, and CNS involvement at the time of initial diagnosis, with P values of <0.001, <0.003, <0.010, and <0.010, respectively. Those having features, such as an elevated WBC count, age > 10 years or < 1 year, male gender, and CNS involvement at the time of diagnosis, were more insensitive to glucocorticoids (Supporting Information Table VIII).

Remission status of BM after induction therapy.

Patients with different types of fusion transcripts responded differently to induction therapy (P = 0.002). The BCR-ABL+ and T-ALLs subgroups had low CR rates, while other subgroups had higher CR rates at the end of induction (Table I).

MRD analysis.

Five hundred thirty-six and four hundred ninety patients were examined for the MRD level at the end of induction and before consolidation, respectively. There was a significant difference in the MRD level among the subgroups (Table I). At TP1, there were 76.9, 94.3, and 63.6% of patients with TEL-AML1, E2A-PBX1, or other BCP-ALLs with low MRD levels (<10−4), respectively, while there were >60% of patients in other subgroups with medium-to-high MRD levels. Furthermore, at TP2 there were still 41.2, 25, and 14.6% of BCR-ABL+, MLL rearrangement+, and T-ALL patients with a high MRD > 10−3, respectively, indicating that the leukemic cells responded poorly to combined chemotherapy. In addition, we showed that the MRD levels at TP1 and TP2 were related to the WBC count and response to prednisone. Those having features, such as an elevated WBC count and a poor response to prednisone, had higher MRD levels at the end of remission induction and before consolidation (Supporting Information Table IX).


The EFS rates varied significantly among the different subgroups (P < 0.001; Fig. 1). The treatment outcomes of patients with TEL-AML1, E2A-PBX1, or other BCP-ALL were excellent. The prognosis of T-ALL was worse than the three aforementioned subgroups. The overall outcome of patients with MLL rearrangements was poor. Interestingly, in 19 patients with MLL rearrangements, 8 of 16 cases with the B cell immunophenotype relapsed, and 3 of 3 cases with the T cell immunophenotype were in CR1, with a follow-up of 33–96 months (P = 0.228). As the tyrosine kinase inhibitor, imatinib, was not included for BCR-ABL+ ALL in the BCH-2003 and CCLG-2008 protocols, the outcome of this subgroup treated by chemotherapy with or without hematopoietic stem cell transplantation (HSCT) was the worst among the various subgroups. Forty BCR-ABL+ patients received HSCT after CR1; however, 6 patients died of complications related to HSCT and 13 patients relapsed after HSCT.

Figure 1.

Event-free Survival (EFS) rate of the subgroups. The 5-year EFS rates were 92.3 ± 2.2%, 85.8 ± 4.7%, 27.3 ± 8.3%, 57.4 ± 11.5%, 84.8 ± 1.8%, and 68.4 ± 5.2% for TEL-AML1+, E2A-PBX1+, BCR-ABL+, MLL rearrangement+, other BCP-ALL, and T-ALL, respectively.

As there were two protocols (BCH-2003 and CCLG-2008) that were applied successively in this study, the treatment effects of these two protocols were also analyzed for each subgroup. The median follow-up of the two protocols was 62 and 27.5 months, respectively. The results showed that the effects of the two protocols were similar, except for the E2A-PBX1 fusion subgroup, which had a better outcome with BCH-2003 compared with CCLG-2008 (P = 0.043), despite a shorter follow-up for CCLG-2008 (Table I). This implied that compared to CCLG-2008, the BCH-2003 protocol using prednisone in induction and high-dose cytarabine in the second delayed intensification may be more suitable for E2A-PBX1+ leukemia.

Among the events leading to therapeutic failure, recurrence was the most common (99 cases, 65.3%), which involved BM and extramedullary relapse. Deaths resulted from chemotherapy-related infection in 23 cases, accounting for 15.1% of the cases. Intermediate and late relapses were more common in recurrent TEL-AML1+ or other BCP-ALL (60 and 59.6%, respectively), while there was a dominant proportion (66.7–100%) of early relapses in the other fusion subgroups (P = 0.020). The site of relapse was similar among the different subgroups. Relapse in the BM was the most common in all subgroups, followed by isolated extramedullary relapse. Only three patients had relapses in the BM and extramedullary site simultaneously (Table I).

Using Cox regression, BCR-ABL fusion, MLL rearrangements, response to prednisone, and MRD at TP1 and TP2 were shown to be independent prognostic factors, with odds ratios of 5.355 (95% confidence interval [CI], 2.430–11.799), 4.171 (95% CI, 1.588–10.957), 2.259 (95% CI, 1.041–4.902), 1.576 (95% CI, 1.015–2.448], and 3.055 (95% CI, 1.386–6.736), and P values of <0.001, <0.004, <0.039, <0.043, and <0.006, respectively. Other clinical features at the time of initial diagnosis, such as the WBC count, gender, age, CNS leukemia, and immunophenotype, were not independent prognostic factors. The independent prognostic significance of TEL-AML fusion, which has been reported in some studies, [29] was not validated in this study. These results again emphasized the prognostic significance of fusion genes and response to early treatment.

Conventional cytogenetic analysis could reveal some types of numerous and structural abnormalities other than translocations. The prognostic significance of high hyperdiploidy (2n > 50) and hypodiploidy have been reported extensively [30, 31]. Because of our unsatisfactory culture conditions, karyotyping succeeded in only 392 patients. No difference in clinical features existed between the patients with and without successful karyotyping (Supporting Information Table VI). The 5-year EFS of high hyperdiploidy patients was 88.4 ± 6.5%, which was minimally higher than others (78.9 ± 2.5%, P = 0.228; Supporting Information Fig. 1).


This study is the first report of fusion genes and clinical features in a large cohort of Chinese pediatric ALL patients. We compared our data with the data reported recently from Europe and North America [1–7], and Japan and Taiwan [32, 33] (Supporting Information Table X). In this study, the incidence of E2A-PBX1 and BCR-ABL fusion was higher than other reports, which may be due to the fact that our hospital is one of the leading centers for childhood leukemia treatment in China, and the patients with these two fusion genes were always recommended to come to our hospital for further treatment because of their adverse clinical features and/or poor outcomes. The incidence of other fusion genes in this study is consistent with the reports in the literature, [1–7] suggesting that the clinical and biological characteristics of ALL in the Chinese population are similar to Caucasians and East Asians.

The prognosis of patients with BCR-ABL fusion was the worst in childhood ALL [34, 35]. Therefore, in protocols BCH-2003 and CCLG-2008, these patients were directly stratified into the high-risk group. In protocol BCH-2003, MLL rearrangements were stratified into the intermediate-risk group, as in some reports the 5-year EFS of this subgroup of patients >1 year of age reached 67% [36]. However, the 5-year EFS of patients with MLL rearrangements were only 58.3%. Thus, in protocol CCLG-2008 we stratified the patients with MLL rearrangements directly into the high-risk group. Although no better outcome was obtained, this made our study more reasonable and comparable to other study groups. Moreover, because of the limited technical conditions in our hospital, some important genetic alterations, such as iAMP21, which were reported associated with poor outcomes recently [37], were not investigated in this study.

The MRD has become one of the most important indicators of stratification in current ALL treatment strategies [38–40]. In April 2005, 2 years after the BCH-2003 protocol started, we began to detect the MRD and validated its prognostic significance; specifically, the MRD level at TP1 and TP2 could distinguish patients with different risks for relapse [18, 41]. Therefore, the MRD was introduced into the stratification system of protocol CCLG-2008. In this study, the result that MRD at TP1 and TP2 were both independent prognostic factors validated the clinical significance of our assay.

It is worth noting that in spite of the best outcome of TEL-AML1+ patients among all the fusion subgroups, the MRD level dropped slower than the E2A-PBX1+ subgroup, with 76.9 and 94.3% of the patients <10−4 at the end of induction, respectively. The two protocols we used were similar to the BFM protocol, and in many studies in which the BFM protocol was used, ∼70 and 94% of the TEL-AML1+ and E2A-PBX1+ patients had a MRD < 10−4 at TP1, respectively [42–44], which approximated our results. In fact, nearly 75% (81/108) of the patients with TEL-AML1 received standard-risk remission induction therapy, while those with E2A-PBX1 were stratified into an intermediate- or high-risk group.

In this study, the outcomes of the two protocols were similar in each subgroup, with the exception of the patients with an E2A-PBX1 fusion, in whom the outcome was significantly better when treated with protocol BCH-2003. Thus, to improve further the outcome of pediatric ALL, it is necessary to apply fusion gene or subtype-specific treatment protocols, or combine the current chemotherapy regimens with new target drugs. Two reports from St. Jude Children's Research Hospital indicated that TEL-AML1+ patients should be treated with intensive asparaginase and high-dose methotrexate [45, 46]. Imatinib has been reported to enhance the 3-year EFS rate of BCR-ABL+ ALL to 80% [47] significantly, which is twice the rate achieved before imatinib was used. Moreover, even within the same fusion subgroup, the patient response to treatment and prognosis still varies greatly, suggesting that more molecular mechanisms have not been identified. Thus, identification of novel targets for therapeutic intervention could lead to tremendous improvement in stratification and targeted therapy of leukemia.

Author Contributions

GC and LZG were the principal investigators and wrote the articles. GC, ZXX, LWJ, CL, ZW, LSG, YZX, and JY performed the laboratory work for this study. WMY recruited the patients. GC participated in the statistical analysis. LZG and WMY coordinated the research. LZG and WMY take primary responsibility for the article.