SEARCH

SEARCH BY CITATION

Keywords:

  • Ewing sarcoma;
  • chimeric transcript;
  • prediction;
  • recurrence

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

Tumors in the Ewing family (EFTs) are the second most common bone tumors in children and adolescents. Despite aggressive chemotherapy, one-third of patients with localized tumor still may develop recurrences. This implies that not all tumor cells are eradicated and that the patients may have a level of residual disease. EFTs are characterized by specific chromosomal translocations that result in chimeric transcripts that can be detected with reverse transcriptase-polymerase chain reaction (RT-PCR) analysis.

METHODS

The authors report the prognostic potential of the positive chimeric transcript (EWS/FLI1) in bone marrow (BM) and/or peripheral blood (PBL) in 26 patients with EFT during a long follow-up period (median, 61 months).

RESULTS

At diagnosis, 43% of patients had positive RT-PCR BM results, with no correlation to tumor progression (P = 0.3). During follow-up, 58% of patients had positive RT-PCR results in their last sample analyzed (BM and/or PBL). A highly significant correlation between the presence of the chimeric transcript and disease progression was detected (P = 0.0028). In a multivariate analysis, the percentage of tumor necrosis (P = 0.007) and RT-PCR results during follow-up (P = 0.02) remained significant prognostic markers. In 10 of 11 patients who developed disease progression, BM and/or PBL samples were positive for the chimeric transcript before evidence of overt clinical recurrence.

CONCLUSIONS

Occult tumor cells in BM and/or PBL samples during long follow-up are strong predictors of recurrent disease in patients with nonmetastatic EFTs. Cancer 2004;100:1053–8. © 2004 American Cancer Society.

The Ewing family of tumors (EFTs), consisting of Ewing sarcoma of bone, extraskeletal Ewing sarcoma, peripheral primitive neuroectodermal tumors, and Askin tumor, are the second most common bone tumors in children and adolescents.1 Clinically, approximately 25% of patients have metastases at diagnosis, which is a major adverse prognostic factor. Despite aggressive multimodal therapy, one-third of patients with a localized tumor still may develop a recurrence, even after 5 years or later.1 This implies that not all tumor cells were eradicated in these patients and that the patients may have occult disease or submicroscopic levels of residual disease that are below the remission threshold.

EFT tumors all are characterized by the same chromosomal translocations that result in the fusion of the EWS gene on chromosome 22q12 with different ETS-related genes, including FLI1, ERG, ETV1, E1AF, and FEV.2–6 The majority of tumors consist of the EWS-FLI1 (90–95%) and EWS-ERG (5–10%) chimeric transcripts. It has been shown that similar clinical phenotypes are associated with the different gene fusion.7 It is well established that in malignant diseases with specific translocations in which the genes involved have been cloned, the reverse transcriptase–polymerase chain reaction (RT-PCR) is a sensitive, reliable, and rapid assay for diagnosis and accurate staging of the disease.8 Furthermore, when applied to bone marrow (BM) and/or peripheral blood (PBL), it can be used for monitoring response to treatment, to detect minimal residual disease (MRD), and for the early diagnosis of metastasis in patients with EFTs. RT-PCR has been applied in the tumor at the time of diagnosis2, 9 and in samples of BM/PBL for the detection of circulating tumor cells.10–13

Fagnou et al.14 showed an association at diagnosis of a positive chimeric transcript only in BM with a significant unfavorable outcome. In a recent study, Schleiermacher et al.15 reported a significantly poorer outcome in patients with localized EFT who had occult tumor cells in BM and/or PBL at diagnosis. The potential prognostic value of the positive chimeric transcript in PBL during therapy and consolidation has been demonstrated,16 but those researchers concluded that a longer follow-up was needed to predict outcomes.

Because the EWS/FLI1 transcripts arise from several different breakpoints,17 many investigators have searched for an association between the transcript type and prognosis. The EWS/FLI1 type I, resulting from the fusion of EWS exon 7 and FLI1 exon 6, has been associated with a favorable prognosis, independent of other clinical prognostic factors.18, 19 We studied the potential prognostic value of the presence or absence of occult tumor cells in BM and/or PBL during a median follow-up of 61 months in 26 patients with nonmetastatic EFT.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patients

Twenty-six children who were diagnosed with localized EFTs participated in this study. The study was approved after it was reviewed by the Ethics Committee of the Rabin Medical Center and the Ministry of Health and after obtaining informed consent from the parents. The median patient age was 14 years (range, 0.3–27 years). All but two tumors were osseous primaries. The primary tumor site was the pelvis in 10 patients (38%), the limbs in 12 patients (46%), and other sites in 4 patients. Fifteen patients (58%) were male. Clinical metastases developed in 11 patients (42%) and were local in the primary site in 4 patients, in the lung in 4 patients, in bone in 2 patients, and local in the ileum and distal to the cranium in 1 patient. All tumors harbored an EWS-FLI1 chimeric transcript.

Samples

PBL and/or BM samples were available for molecular staging at diagnosis, during therapy, and during follow-up. All samples were obtained before recurrence, and the patients exhibited no evidence of disease. Altogether, 109 samples were analyzed; 21 at diagnosis and 88 during follow-up. The median follow-up was 61 months (range, 7–165 months).

Therapy consisted of 4 courses of intravenous (IV) vincristine, actinomycin D, cyclophosphamide, and doxorubicin (IV-VACA) with ifosfamide and radiation (4000 centigrays [cGy] before surgery and 2000 cGy after surgery) followed by 2 courses of IV-VAC (without doxorubicin) for 18 months. Responses to therapy were assessed postsurgery after 4–6 months of therapy by histopathologic response and percentage of tumor necrosis.

RT-PCR

Total RNA was isolated from PBL and/or BM samples with TRI-reagent (MRC, Cincinnati, OH). RT-PCR was performed with the Access RT-PCR System (Promega, Madison, WI), according to the manufacturer's instructions, with 1 μg total RNA, 2.5 μM magnesium, random hexamers, and previously described primers 22.8 and 22.42, 7 for the normal EWS gene and primer 22.8 with FLI119 for the EWS-FLI1 fusion transcript. Nested PCR was performed with primer 22.3 with FLI3 after the cDNA was diluted 1:50. Thirty-five cycles of nested PCR were performed for each sample. Annealing and extension temperatures for the EWS–FLIA transcript were 68 °C and 72 °C, respectively. Each RT-PCR was performed at least twice. Sequence analysis using the Thermosequenase kit (Amersham Biosciences, Piscataway, NJ) was performed on the majority of the cDNA samples for chimeric transcript validation.

Statistical Analysis

Progression-free survival was determined using Kaplan–Meier analysis for age and response to therapy (< 90% necrosis and > 90% necrosis), and RT-PCR analysis was used on BM samples at diagnosis and on the last sample available from BM only, PBL only, and a combination of BM and/or PBL. Chi-square analysis was used to correlate various clinical parameters with the results of RT-PCR analysis. Multivariate analysis using Cox regression was performed for the parameters age, percent tumor necrosis, and RT-PCR results.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

RT-PCR results were positive in BM samples at diagnosis in 6 of 14 of patients (43%); however, there was no correlation between the presence or absence of the chimeric transcripts and progression-free survival, as determined by Kaplan–Meier analysis (P = 0.39). Only 7 PBL samples were available at diagnosis, and 3 samples (43%) were positive. There was no preferentially EWS-FLI1 tumor subtype 1 or 2 that gave rise to submicroscopic disease.

During follow-up, 32 BM samples and 56 PBL samples were analyzed over 7–165 months (median, 61 months). BM and PBL samples were obtained simultaneously at 16 time points. Their RT-PCR results were compatible in 12 instances (75%). Among the four sets of incompatible results, the BM sample was positive and the PBL sample negative in three cases, and in one instance, the BM sample was negative and the PBL sample was positive. Only one of the patients with a positive BM sample developed a recurrence.

For statistical analysis, the last sample from each patient was analyzed (37 samples; 14 BM samples and 23 PBL samples). Eight of 14 BM samples (57%) had positive RT-PCR results, and these positive results were correlated significantly with disease progression (P = 0.02) (Fig. 1). Eleven of 23 PBL samples (48%) had positive RT-PCR results, and these results also were correlated significantly with disease progression (P = 0.04) (Fig. 2). BM and PBL analyses could be combined, because each separately correlated significantly with disease progression. Fifteen of 26 patients (58%) had positive RT-PCR results in their last sample analyzed (either BM and/or PBL). A highly significant correlation between the presence of the chimeric transcripts and disease progression was detected (P = 0.0028) (Fig. 3).

thumbnail image

Figure 1. Progression-free survival of patients with nonmetastatic Ewing family tumors according to positive (+) or negative (−) reverse transcriptase–polymerase chain reaction results in bone marrow (BM) samples (P = 0.02).

Download figure to PowerPoint

thumbnail image

Figure 2. Progression-free survival of patients with nonmetastatic Ewing family tumors according to positive (+) or negative (−) reverse transcriptase–polymerase chain reaction results in peripheral blood (PBL) samples (P = 0.04).

Download figure to PowerPoint

thumbnail image

Figure 3. Progression-free survival of patients with nonmetastatic Ewing family tumors according to positive (+) or negative (−) reverse transcriptase–polymerase chain reaction results in bone marrow (BM) and/or peripheral blood (PBL) samples (P = 0.002).

Download figure to PowerPoint

When summarizing all RT-PCR results for each patient, four distinct groups could be defined. The first group (Group 1) included 13 patients who had consistently positive RT-PCR results in all samples analyzed, and 69% of those patients developed recurrent disease. The second group (Group 2) consisted of samples from four patients, all with negative RT-PCR results, and none developed recurrences. In the third group (Group 3), RT-PCR results changed from negative to positive in two patients, one of whom developed a recurrence. Finally, in the fourth group (Group 4), RT-PCR results changed from positive to negative in seven patients, and only one developed a recurrence. A significant correlation between RT-PCR results and disease progression was determined (P = 0.0052) when combining the 4 groups into 2 groups: In Group A (Groups 1 and 3), which consisted of patients with circulating tumor cells, 66% of patients developed recurrences, and in Group B (Groups 2 and 4), which consisted of patients with no detectable occult tumor cells, only 1 patient developed a recurrence (Table 1). There was no significant difference in the primary site and treatment response between the two groups (Table 1). In the group with > 90% necrosis, patients still developed recurrences, but the recurrence rate (56%) was remarkably higher in Group A (5 of 8 patients) compared with Group B (1 of 8 patients; 13%).

Table 1. Longitudinal Reverse Transcriptase–Polymerase Chain Reaction Results and Outcomea
VariableGroup AbGroup BcP value
  • a

    For all samples (median follow-up, 61 months).

  • b

    Group A: Samples from patients with positive reverse transcriptase–polymerase chain reaction (RT-PCR) results or with RT-PCR results that changed from negative to positive.

  • c

    Group B: Samples from patients with negative RT-PCR results or with RT-PCR results that changed from positive to negative.

Primary site   
 Pelvis73 
 Other880.4
Necrosis   
 ≥ 90%88 
 < 90%510.3
Progression   
 Recurrence101 
 Remission5100.0052

Univariate and multivariate analyses were performed for the clinical parameters and for RT-PCR results at diagnosis and during follow-up (Table 2). In the univariate analysis, age, percent necrosis, and RT-PCR results during follow-up were identified as highly significant prognostic factors. Age < 12 years at diagnosis, necrosis < 90%, and positive RT-PCR results were correlated with adverse prognosis (P = 0.036, 0.0015, and 0.0028, respectively).

Table 2. Univariate and Multivariate Analyses of the Clinical Parameters and Reverse Transcriptase–Polymerase Chain Reaction Results
ParameterNo. of patientsUnivariate analysisMultivariate analysis
RecurrencesP valueHR (95% CI)P value
  1. HR: hazard ratio; 95% CI: 95% confidence interval; RT-PCR: reverse transcriptase–polymerase chain reaction; NS: not significant.

Age (yrs)     
 < 12960.036NS 
 ≥ 12175   
Necrosis     
 ≥ 90%1760.00159.12 (1.82–45.66)0.007
 < 90%54   
RT-PCR follow-up results   
 Positive15100.002811.52 (1.39–95.59)0.024
 Negative111   
Primary site     
 Pelvis106NS  
 Other165   
Gender     
 Male157NS  
 Female114   
RT-PCR results at diagnosis   
 Positive61NS  
 Negative83   

Using multivariate analysis, the percentage of necrosis (P = 0.007) and the RT-PCR results remained significant prognostic factors (P = 0.02). However, the multivariate analysis data should be interpreted with caution due to the sample size.

Samples of BM and or PBL before patients developed clinical recurrences were available for testing from 11 patients who had disease progression. Ten samples (6 BM samples and 6 PBL samples) were positive for the chimeric transcript before evidence of overt clinical recurrence. The time lag between positive RT-PCR results and disease progression is summarized in Table 3. All samples that were obtained at the time of clinical recurrence had positive RT-PCR results and were not included in this study.

Table 3. Molecular Progression as a Predictor of Clinical Progression
Patient no.No. of mosSite
Positive RT-PCR resultsClinical progression
  1. RT-PCR: reverse transcriptase–polymerase chain reaction; BM: bone marrow; PBL: peripheral blood

141 (BM and PBL)51Lung
525 (PBL)29Local
724 (PBL)41Bone
1137 (PBL)61Local and cranium
13 8 (BM)15Lung
1415 (BM)17Local
15 7 (BM and PBL)12Local
1635 (BM)37Local
2030 (BM)31Bone
23 2 (PBL)5Lung

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

One of the major obstacles to successful treatment of patients with EFT is ‘late’ tumor progression despite intensive combination therapy. Therefore, it is of utmost importance to assess the presence of occult micrometastases at diagnosis, during therapy, and during follow-up in addition to assessing their impact on disease progression and outcome.

In patients with EFT, several studies have focused on the identification of BM micrometastases and circulating tumor cells in PBL at diagnosis. In previously published reports, tumor cells were detected in 19–40% of nonmetastatic patients at diagnosis, and their correlation with outcome was controversial. Fagnou et al.14 reported a significant association between the presence of chimeric transcripts in BM (42 patients) and an unfavorable outcome after a median follow-up of 12 months. That group has continued their study, adding 110 patients and a median follow-up period of 27 months; in this extended study, they have demonstrated a significantly poorer outcome in patients harboring occult tumor cells in BM and/or PBL samples at diagnosis. It should be noted that this significant association was observed when recurrences developed up to 24 months after diagnosis.15 Conversely, Zoubek et al.,13 with a median follow-up of 30 months, could not detect any statistically significant difference in outcome related to positive or negative RT-PCR results in patients with localized EFT at diagnosis. Because EFTs are characterized by late recurrences—after 5 years or more—those authors suggested that observation of at least 5 years is required for the evaluation of the prognostic significance of RT-PCR results at diagnosis.13 We also identified chimeric transcripts in BM samples at a similar rate of 43% in nonmetastatic patients at diagnosis. However, after a very long follow-up (median, 61 months), no significant correlation with outcome was detected. Seven of 17 patients with BM and/or PBL samples at diagnosis developed recurrences. Three patients who had early recurrences (at 5 months, 12 months, and 15 months) had positive RT-PCR results, whereas the patients with late recurrences (41 months, 51 months, 61 months, and 61 months) had negative RT-PCR results. The difference between our results and those of Schleiermacher et al. may be attributed to their substantially larger numbers of patients, their shorter follow-up (median, 27 months), their large number of early recurrences,15 or successful clearing of positive cells by intensive chemotherapy. In the three patients who developed recurrent tumors early in the current study, circulating tumor cells in BM and/or PBL samples were detected at diagnosis. This is in concordance with the results of Schleiermacher et al.,15 who found that patients harboring circulating tumor cells at diagnosis are at risk of early recurrence.

There are only a limited number of reports on the detection of circulating cells in BM/PBL during follow-up. In the largest study performed by de Alava et al.16 on PBL samples from 28 patients, chimeric transcripts were detected in 56% of patients with a median follow-up of 26 months; their presence was not correlated significantly with outcome.

We identified chimeric transcripts in 58% of our patients and found a highly significant correlation between them in BM/PBL samples and progression-free survival during follow-up. The current investigation, to the best of our knowledge, is the first long-term follow-up study monitoring MRD in BM and/or PBL samples from patients with EFT. Furthermore, we identified molecular recurrences before clinical recurrences in 91% of patients, similar to data from de Alava et al.,16 who found that 4 patients had positive RT-PCR samples prior to clinical recurrence.

Using univariate analysis, we found that the significant clinical parameters for recurrence were age (P = 0.03), response to therapy (percentage of necrosis after chemoradiotherapy; P = 0.0015), and RT-PCR results during follow-up (P = 0.0028). In the multivariate analysis, response to therapy and RT-PCR results remained highly significant (P = 0.07 and 0.024, respectively) (Table 2). Thus, the detection of occult circulating tumor cells is a predictor of disease progression. Moreover, 4 of 5 good responders (> 90% necrosis) who were positive for the chimeric transcript during follow-up developed recurrences. However, in 4 patients who had positive RT-PCR samples through follow-up (range, 61–75 months), there was no evidence of disease. These false-positive results may be explained by the low numbers of malignant circulating cells in the BM/PBL that are detected by the over sensitivity of the nested RT-PCR technique. We anticipated that the small numbers of tumor cells could be eliminated by immunologic or other mechanisms and thus do not have significant clinical relevance. Quantification of the chimeric transcripts during follow-up may shed light on the impact of the amount of the circulating tumor cells in the BM and/or PBL on outcome.

The clinical importance of MRD has been demonstrated in another childhood neoplasm, acute lymphoblastic leukemia (ALL). It was found to be of clinical significance at consecutive time points, and it served as an indicator of the effectiveness of treatment. Levels of MRD distinguished patients with good or poor prognoses, and treatment was modified according to the risk stratification based on the MRD results.20, 21 Negative MRD results were associated with low recurrence rates (3–15% at 3 years), whereas high recurrence rates were found in patients with positive MRD results (39–86% at 3 years). A significant correlation was established with quantification at precise time points in patients with childhood ALL.

The prognostic role of tumor cells in BM and/or PBL at diagnosis and follow-up have been addressed in patients with EFT by several investigators. Recently, a significant association between BM micrometastases/circulating tumor cells in PBL and early recurrence was observed. The long follow-up period in the current study enabled us to determine the significance of occult tumor cells in BM and/or PBL and their correlation with the late recurrences commonly observed in patients with EFT. However, we could not determine the critical time points for prediction of disease recurrence, because the samples in the current study were not obtained at precise time points. We currently are applying real-time PCR for quantifying MRD and its correlation with clinical outcome.

We conclude that RT-PCR analysis is a valid test for MRD assessment during therapy and follow-up in patients with EFT and may be used for predicting tumor progression, identifying high-risk patients, and in the application of risk-adapted therapy. Based on our data, it is suggested that serial monitoring with RT-PCR is recommended for the prediction of disease recurrence. A prospective longitudinal study could determine the critical time points early in the course of therapy, offering the possibility of risk stratification and optimal treatment approaches. Quantitative RT-PCR may determine predictive prognostic levels of circulating tumor cells that are critical for treatment stratification.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  • 1
    Ginsberg JP, Woo SY, Johnson ME, et al. Ewing sarcoma family of tumors: Ewing's sarcoma of bone and soft tissue and the peripheral primitive neuroectodermal tumors. In: PizzoPA, PoplackDG, editors. Principles and practice of pediatric oncology ( 4th edition). Philadelphia: Lippincott-Raven, 2002: 9731016.
  • 2
    Delattre O, Zucman J, Melot T, et al. The Ewing family of tumors—a subgroup of small round cell tumors defined by specific chimeric transcripts. N Engl J Med. 1994; 331: 294299.
  • 3
    Sorensen PH, Lessnick SL, Lopez-Terrada D, Liu XF, Triche TJ, Denny CT. A second Ewing's sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family transcription factor, ERG. Nat Genet. 1994; 6: 146151.
  • 4
    Jeon IS, Davis JN, Braun BS, et al. A variant Ewing's sarcoma translocation (7;22) fuses the EWS gene to the ETS gene ETV1. Oncogene. 1995; 10: 12291234.
  • 5
    Kaneko Y, Yoshida K, Handa M, et al. Fusion of an ETS-family gene, E1AF, to EWS by t(17;22)(q12;q12) chromosome translocation in an undifferentiated sarcoma of infancy. Genes Chromosomes Cancer. 1996; 15: 115121.
  • 6
    Peter M, Couturier J, Pacquement H, et al. A new member of the ETS family fused to EWS in Ewing tumors. Oncogene. 1997; 14: 11591164.
  • 7
    Ginzberg JP, de Alava E, Ladanyi M, et al. EWS-FLI1 and EWS-ERG gene fusions are associated with similar clinical phenotypes in Ewing's sarcoma. J Clin Oncol. 1999; 17: 18091814.
  • 8
    Naito N, Kawai A, Ouchida M, et al. A reverse transcriptase polymerase chain reaction assay in the diagnosis of soft tissue sarcomas. Cancer. 2000; 89: 19921998.
  • 9
    Zoubek A, Pfleiderer C, Salzer-Kuntschik M, et al. Variability of EWS chimeric transcripts in Ewing tumours: a comparison of clinical and molecular data. Br J Cancer. 1994; 70: 908913.
  • 10
    Pfleiderer C, Zoubek A, Gruber B, et al. Detection of tumour cells in peripheral blood and bone marrow from Ewing tumour patients by RT-PCR. Int J Cancer. 1995; 64: 135139.
  • 11
    Peter M, Magdelenat H, Michon J, et al. Sensitive detection of occult Ewing's cells by the reverse transcriptase-polymerase chain reaction. Br J Cancer. 1995; 72: 96100.
  • 12
    West DC, Grier HE, Swallow MM, Demetri GD, Granowetter L, Sklar J. Detection of circulating tumor cells in patients with Ewing's sarcoma and peripheral primitive neuroectodermal tumor. J Clin Oncol. 1997; 15: 583588.
  • 13
    Zoubek A, Ladenstein R, Windhager R, et al. Predictive potential of testing for bone marrow involvement in Ewing tumor patients by RT-PCR: a preliminary evaluation. Int J Cancer. 1998; 79: 5660.
  • 14
    Fagnou C, Michon J, Peter M, et al. Presence of tumor cells in bone marrow but not in blood is associated with adverse prognosis in patients with Ewing's tumor. J Clin Oncol. 1998; 16: 17071711.
  • 15
    Schleiermacher G, Peter M, Oberlin O, et al. Increased risk of systemic relapses associated with bone marrow micrometastasis and circulating tumor cells in localized Ewing tumor. J Clin Oncol. 2003; 21: 8591.
  • 16
    de Alava E, Lozano MD, Patino A, Sierrasesumaga L, Pardo-Mindan FJ. Ewing family tumors: potential prognostic value of reverse-transcriptase polymerase chain reaction detection of minimal residual disease in peripheral blood samples. Diagn Mol Pathol. 1998; 7: 152157.
  • 17
    Zucman J, Melot T, Desmaze C, et al. Combinatorial generation of variable fusion proteins in the Ewing family of tumours. EMBO J. 1993; 12: 44814487.
  • 18
    Zoubek A, Dockhorn-Dworniczak B, Delattre O, et al. Does expression of different EWS chimeric transcripts define clinically distinct risk groups of Ewing tumor patients? J Clin Oncol. 1996; 14: 12451251.
  • 19
    de Alava E, Kawai A, Healy JH, et al. EWS-FLI1 fusion transcript structure is an independent determinant of prognosis in Ewing's sarcoma. J Clin Oncol. 1998; 16: 12481255.
  • 20
    van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al. Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet. 1998; 352: 17311738.
  • 21
    Biondi A, Valsecchi MG, Seriu T, et al. Molecular detection of minimal residual disease is a strong predictive factor of relapse in childhood B-lineage acute lymphoblastic leukemia with medium risk features. A case control study of the International BFM Study Group. Leukemia. 2000; 14: 19391943.