Reassessment of the prognostic significance of hypodiploidy in pediatric patients with acute lymphoblastic leukemia


  • Susana C. Raimondi Ph.D.,

    Corresponding author
    1. Department of Pathology, St. Jude Children's Research Hospital, Memphis, Tennessee
    2. Department of Pathology, University of Tennessee College of Medicine, Memphis, Tennessee
    • Department of Pathology, St. Jude Children's Research Hospital, 332 N. Lauderdale St., Memphis, TN 38105-2794
    Search for more papers by this author
    • Fax: (901) 495-3100

  • Yinmei Zhou M.Sc.,

    1. Department of Biostatistics, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
  • Susan Mathew Ph.D.,

    1. Department of Pathology, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
  • Sheila A. Shurtleff Ph.D.,

    1. Department of Pathology, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
  • John T. Sandlund M.D.,

    1. Department of Pediatrics, University of Tennessee College of Medicine, Memphis, Tennessee
    2. Department of Hematology-Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
  • Gaston K. Rivera M.D.,

    1. Department of Pediatrics, University of Tennessee College of Medicine, Memphis, Tennessee
    2. Department of Hematology-Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
  • Frederick G. Behm M.D.,

    1. Department of Pathology, St. Jude Children's Research Hospital, Memphis, Tennessee
    2. Department of Pathology, University of Tennessee College of Medicine, Memphis, Tennessee
    Search for more papers by this author
  • Ching-Hon Pui M.D.

    1. Department of Pathology, St. Jude Children's Research Hospital, Memphis, Tennessee
    2. Department of Pediatrics, University of Tennessee College of Medicine, Memphis, Tennessee
    3. Department of Hematology-Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee
    Search for more papers by this author
    • C.-H. Pui is an American Cancer Society–F.M. Kirby Clinical Research Professor.



The purpose of the current study was to evaluate the cytogenetic features of the hypodiploid leukemic cells of pediatric patients with this rare subgroup of acute lymphoblastic leukemia (ALL). In addition, the authors determined whether subdivision of the hypodiploid category served a prognostic purpose for these patients.


The authors evaluated the cytogenetic records of 979 patients with ALL admitted to St. Jude Children's Research Hospital (Memphis, TN) between 1984 and 1999.


Of 67 patients (6.8%) whose leukemic cells contained a modal number (MN) of chromosomes less than or equal to 45 (i.e., hypodiploid leukemic cells), 57 had an MN of 45 and 10 had an MN of less than 45. In 19 patients, cells with an MN of 45 had a whole chromosome missing (42%), which was a sex chromosome in 12 patients (63%). Leukemic cells with an MN of 45 contained dicentric chromosomes (n = 33) formed from chromosome 9p (55%), 12p (18%), or both (21%). The ETV6-CBFA2 fusion was present in 39% of 28 evaluable B-lineage cases with an MN of 45. The event-free survival rate (EFS) for patients with hypodiploid leukemic cells of MN less than 45 (5-year EFS = 20.0% ± 10.3%) was significantly (P < 0.001) lower than that for patients with leukemic cells of MN greater than or equal to 45 (5-year EFS = 74.9% ± 1.6%).


Low hypodiploidy (MN < 45) should be recognized as a high-risk feature in pediatric ALL. Only two hypodiploid groups (MN < 45 and MN = 45) may be necessary in prognostic assessments. Cancer 2003;98:2715–22. © 2003 American Cancer Society.

Blast cell ploidy is a well recognized prognostic factor for pediatric acute lymphoblastic leukemia (ALL).1–6 Patients whose leukemic blast cells have a modal number (MN) of chromosomes equal to 51 or more have the best prognosis, whereas children with near-haploid blast cells (MN < 30) have the worst prognosis. Overall, diploidy or pseudodiploidy is associated with an intermediate prognosis. However, patients with pseudodiploid blast cells and translocations like t(4;11) or t(9;22) (i.e., the Philadelphia chromosome) have a poor prognosis and patients with a cryptic t(12;21) generally have a very favorable outcome.

Hypodiploidy (i.e., ≤ 45 chromosomes) is a very heterogeneous category2, 7, 8 and occurs in 6–7% of patients with childhood ALL. Most patients in the hypodiploid category have leukemic cells with 45 chromosomes (approximately 5% of all patients with ALL), whereas leukemic cells with fewer than 45 chromosomes are rarely observed. Patients with 45 chromosomes in their leukemic cells (i.e., hypodiploidy with an MN = 45) generally fare as well as do those with leukemic cells of other ploidies.8 Patients whose leukemic cells contain fewer than 45 chromosomes (hypodiploidy with an MN < 45) have an inferior outcome when treated on protocols that are effective for ALL of other ploidies,7, 8 a finding suggesting that alternative therapy should be considered for this subtype of ALL.

Several investigators have attempted to divide the hypodiploid category into subgroups for outcome analysis.7–10 Near-haploid ALL is a distinct subgroup of hypodiploid ALL in which the blast cells have 24–34 chromosomes or, more frequently, 26–28 chromosomes.7–28 The overall incidence of near-haploidy is low. This category constitutes approximately 0.5% of patients with ALL. Cases of near-haploidy have received much attention in the literature, despite their rare occurrence, because they are associated with a poor prognosis.8, 10, 25 Other cases of hypodiploidy with 35–44 chromosomes are usually referred to as low hypodiploidy. These cases are also rare and occur in approximately 0.8% of patients with ALL.8, 10

Since our initial reports on hypodiploidy in patients with ALL,7, 10 we have accrued additional cases of this rare category of ALL. (Twenty-seven of the hypodiploid cases presented in the current study were reported in these 2 studies.) Unlike one of the previous studies,10 the current analysis includes only consecutive cases treated at a single institution. We have reassessed the prognostic impact of hypodiploidy by updating the cytogenetic findings from the initial studies, reevaluating these cases together with those accrued since the previous studies, and considering in our analysis cases of this type in the literature.



Between 1984 and 1999, 1011 children were enrolled in the St. Jude Children's Research Hospital (Memphis, TN) Total Therapy protocols T11–T14 for the treatment of newly diagnosed ALL.29, 30 Patients in these studies received risk-directed therapy, i.e., higher-risk patients received more intensive therapy including epipodophyllotoxins, cyclophosphamide, cytarabine, and cranial irradiation. Of the 1011 patients with ALL who were examined, 979 had complete chromosome analysis and known ploidy. These patients became the subjects of the current study. After evaluation, 67 (6.8%) patients were classified as hypodiploid (MN ≤ 45) and 912 (93.2%) were classified as nonhypodiploid (MN ≥ 46), according to the International System for Human Cytogenetic Nomenclature (ISCN 1995).31 The diagnosis of ALL was based on the morphologic criteria of the French–American–British cooperative group and on the lack of myeloperoxidase and nonspecific esterase activity in the blast cells. The diagnosis of B-lineage ALL was based on the expression of B-lineage–associated antigens (CD79a, CD19, CD22, CD10). Informed consent was obtained from all patients or their guardians. The investigations were approved by the institutional review board of St. Jude Children's Research Hospital.

Cytogenetic and Molecular Studies

For cytogenetic studies, the bone marrow samples were prepared by a direct method or unstimulated cultures of bone marrow cells were grown overnight, as described previously.32 Chromosomes were identified by using the banding pattern obtained with trypsin-Wright's stain and classified according to ISCN 1995.31 The ETV6-CBFA2 (formerly TEL-AML1) fusion was identified by reverse transcription-polymerase chain reaction (RT-PCR) as previously reported.33 Some cases were evaluated retrospectively, but after December 1995, all cases were evaluated prospectively.

Statistical Analysis

Analyses were based on patient follow-up records from 1984 to April 28, 2003. Clinical and laboratory features of patients with hypodiploid and nonhypodiploid ALL were compared by using the chi-square and Fisher exact tests, as appropriate. The primary end point evaluated was event-free survival (EFS), which was obtained by analysis only of the records of patients who were enrolled in the St. Jude Total Therapy protocols 11–13B (n = 928).29, 30 The median follow-up period was 11.0 years (range, 2.4–18.6 years). Events were defined as induction failure (lack of response to therapy or death during induction therapy), disease recurrence, death during disease remission, second malignancy, lineage switch, or myelodysplastic syndrome, whichever occurred first. Time from achievement of complete disease remission to the first event was used to calculate the EFS probability for patients who experienced an adverse event. For patients who did not experience an adverse event, the EFS period was defined as the time from achievement of complete disease remission to the last follow-up date. Patients for whom complete disease remission was not achieved were categorized as having treatment failure at Time 0. The EFS curves were estimated by the Kaplan–Meier method34 and the EFS rates were compared between groups by using the log-rank test.35 Within the hypodiploid group, we also compared the presenting features and outcome of patients with 25–29 chromosomes (near-haploidy), 36–44 chromosomes (low hypodiploidy), and 45 chromosomes (high hypodiploidy) in their leukemic blast cells.8, 10


Among 979 children with ALL who had a complete chromosome analysis report, 67 had fewer than 46 chromosomes in their leukemic blast cells (Table 1). For 59 of these 67 patients, their leukemic cells had a B-lineage immunophenotype. The remaining eight patients had ALL of T-cell lineage (Table 1).

Table 1. Chromosomal Abnormalities Observed at the Time of Diagnosis in the Leukemic Cells of 67 Patients with Hypodiploid Acute Lymphoblastic Leukemia
Patient no.DIKaryotypeTEL-AML1IPAge (yrs)Leukocytes × 109/LRD (yrs)Failure typeStatus
  1. MN: modal number; DI: DNA index; IP: immunophenotype; B: B lineage; T: T cell; RD: remission duration; ND: not determined; Ind: induction; HR: hematologic recurrence; TR: testicular recurrence; CNSR: central nervous system recurrence; CR: complete remission; S-AML: secondary acute myeloid leukemia; A: alive; D: dead.

Near-haploid (MN 25–29)
 10.69, 1.18, 1.1225–34[5]/60,XY,+X,+Y,del(3)(p25),+4,+8,+9,+9,+10,+10,+14,+18,+20,+21,+21,+mar[4]/60,idem,+add(17)(q25),−mar[4]/46,XY[8]B5.6106.9TRA
 30.55, 1.1127[1]/34[1]/52,XY,+X,+8,+9,+14,+21,+21[16]NDB3.32.71.5HRD
 40.62, 1.11, 1.2129,XY,+4,+8,+18,+21,+mar[5]/57,XY,+X,+Y,+4,+4,del(5)(p13),+6,del(6)(q21),+8,+8,+14,+18,+21,+21[6]B8.911.91.0HRD
Low hypodiploid (MN 36–44)
 70.77, 1.4637,XX,−2,−3,−4,−7,−9,−12,−15,−16,−20[5]/68,XX,+X,+X,+1,+1,+5,+6,+?7,+8,+10,+11,+13,+15,+17,+18,+19,+19,+20,+21,+21,+22,+22,+22[1]/46,XX[2]NDB11.58.04.4S-AMLD
 80.85, 1.8739,XX,−3,−4,der(5)t(5;11)(q35;q12),−7,−9,add(12)(p11.2),−15,−16,−17,−18,−22,+2mar[6]/39,X,add(X)(q24),add(1)(q24),−3,i(3)(p11),−4,inv(5)(p15.3q13),−7,add(9)(p22),−10,der(12)t(7;12)(q11.2;q24.1),−15,−16,add(17)(q25),−18,add(21)(q22),−22,+mar[3]/68–71,idemx2,−3,i(3)(p11)x2,−7,−11[cp5]B10.297.35.6NoneA
High hypodiploid (MN 45 [whole chromosome loss])
High hypodiploid (MN 45 [with dicentric chromosomes])
High hypodiploid (MN 45 [with nondicentric chromosomes])


In three of four patients with near-haploid cells (MN = 25–29), a subclone had twice as many chromosomes as did the near-haploid line (Patients 1–4, Table 1). In the two banded cases, the structural aberrations were present only in the hyperdiploid clone, suggesting clonal evolution had occurred. The DNA index results were in concordance with the findings of cytogenetic analysis (Table 1).

Low Hypodiploidy

A doubling of the hypodiploid line occurred in three of six patients with hypodiploid cells with 36–44 chromosomes (Patients 6–8, Table 1). One patient (Patient 10) with an MN of 43 had a t(9;22)(q34;q11.2), lacked chromosomes 3, 7, and 9, and had a der(7)t(3;7) in the leukemic cells.

High Hypodiploidy

Fifty-seven patients who had leukemic cells with an MN of 45 (Patients 11–67, Table 1) were subgrouped on the basis of the chromosomal abnormality that instigated the chromosome loss (Fig. 1).

Figure 1.

Distribution of chromosomal abnormalities in patients with a modal number of 45 chromosomes in their leukemic cell.

Loss of a whole chromosome

In 12 of 19 patients (63%) with whole chromosome loss, a sex chromosome was missing. In most instances, the missing sex chromosome appeared to be acquired in the leukemic clone and did not represent a constitutional mosaicism, as established by phytohemagglutinin-stimulated peripheral blood studies. The leukemic cells of 2 of the 19 patients (Patients 11 and 27) had a T-cell immunophenotype. Deletions in the q arm of chromosome 6 were observed in 5 of the 19 (26%) patients. With the exception of a 1.7-year-old girl who had a t(9;11)(p22;q23) in her leukemic clone (Patient 28, Table 1), no other patients in this group had translocations known to affect prognosis in childhood ALL, i.e., t(9;22), t(4;11), t(1;19).36

Dicentric chromosomes

It is interesting to note that in 31 of 33 patients with a dicentric chromosome (Patients 30–62, Table 1), dicentrics involved chromosome 9 (n = 18), chromosome 12 (n = 6), or both, with the formation of a dic(9;12) (n =7; Fig. 1). The other frequently observed dicentrics were dic(9;20) (n = 7) and dic(7;9) (n = 4). In only two patients did the dicentric chromosome formed not involve chromosome 9 or 12 [Patient 57, dic(10;17) and Patient 62, dic(15;22)].

A dicentric chromosome was the only abnormality in approximately half (n = 14) of this subgroup. Three patients who had leukemic cells with a dic(7;9) (Patient 33), a dic(7;12) (Patient 36), or a dic(9;20) (Patient 53) were younger than 2 years at the time of diagnosis of ALL. One infant (Patient 42, younger than 9 months) had a dic(9;12) and the leukemic cells were positive for the ETV6-CBFA2 (TEL-AML1) fusion gene (detected by RT-PCR). Uncommon dicentric formation occurred in all five patients with leukemic cells with a T-cell immunophenotype (Patients 30, 56, 57, 59, 60).

Nondicentric chromosomes

All five patients with nondicentric chromosome formation involved derivative chromosomes that contained only one centromere (Patients 63–67, Table 1). In each patient, partial DNA material was missing from chromosome 4, 7, 8, 10, or 13 (one patient each).

Clinical and Biologic Characteristics of Children with Hypodiploid and Nonhypodiploid Acute Lymphoblastic Leukemia

Age at the time of diagnosis differed significantly among the three hypodiploid groups (P < 0.001). The only infant in the study group had leukemic cells with an MN of 45 and all 6 children who had leukemic cells with an MN of 36–44 were 10 years or older. The median age of the patients in the MN 25–29 group was 7.3 years (range, 3.3–9.3 years). The other factors evaluated (gender, ethnicity, leukocyte count at the time of diagnosis, and ALL lineage and risk group) were not significantly different between the hypodiploid and nonhypodiploid groups or among the hypodiploid subgroups. However, the MN 45 group included seven of the eight patients with T-lineage ALL and 11 patients with B-lineage ALL (of a total of 28 evaluable patients with B-lineage ALL in the MN = 45 group) with the cryptic t(12;21), which resulted in the fusion of the ETV6 and CBFA2 genes. Indeed, the ETV6-CBFA2 (TEL-AML1) chimeric transcript was observed only in leukemic cells that had an MN of 45. It was present in 6 of 13 patients who had whole chromosome loss (all six patients had a missing X chromosome), in 4 of 13 patients who had dicentric chromosomes, and in 1 of 2 patients who had derivative chromosomes resulting from an unbalanced translocation.

For the 928 patients who received treatment on St. Jude's Total Therapy protocols 11–13B, the 5 and 10-year EFS rates ± 1 standard error were 74.4% ± 1.5% and 69.7% ± 2.0%, respectively. The 5-year EFS estimates for patients with 25–29, 36–44, or 45 chromosomes in their leukemic cells are shown in Figure 2. These EFS estimates were significantly different between the group with an MN less than 45 and the group with an MN of 45 (P < 0.001). Figure 3 shows the EFS curves and 5 and 10-year EFS estimates for patients whose blast cells were diploid, pseudodiploid, low hyperdiploid (47–50 chromosomes), or high hyperdiploid (≥ 51 chromosomes) in comparison to the curves for the combined group of patients with an MN less than 45. Notably, patients in the hypodiploid subgroup with an MN of 45 had an EFS outcome that was similar to that of patients in the diploid, low-hyperdiploid (MN = 47–50), and high-hyperdiploid (MN ≥ 51) groups (P = 0.424) and significantly better (P = 0.020) than that of patients in the pseudodiploid group.

Figure 2.

Event-free survival (EFS) and 5-year EFS estimates for patients in the hypodiploid acute lymphoblastic leukemia subgroups. MN: modal number.

Figure 3.

Event-free survival (EFS) and 5- and 10-year EFS estimates by ploidy for patients with acute lymphoblastic leukemia. MN: modal number.

At the time of the current report, all seven patients with a dic(9;12) were alive and free of disease (median follow-up time, 14.3 years; range, 9.1–16.9 years). Among the seven patients with a dic(9;20), two had secondary acute myeloid leukemia (one of whom was alive) and one patient died after a hematologic disease recurrence. The median follow-up time for the patients with the dic(9;20) who were alive was 11.4 years (range, 4.6–18.4 years).


This large, single-institution study of 67 cases of hypodiploid ALL (which represent approximately 7% of nearly 1000 cases of ALL analyzed) confirms that despite improved outcome for most patients with the use of contemporary risk-directed treatment,36 hypodiploidy with an MN less than 45 continues to confer an inferior outcome. Patients with 45 chromosomes in their blast cells fared as well as did those in the other ploidy groups who had good outcomes. Although the distribution of chromosomal abnormalities in the hypodiploid ALL subgroup is heterogeneous, the abnormalities in the near-haploid and low hypodiploid groups (25–44 chromosomes) are distinct from those in the high-hypodiploid (45 chromosomes) group. The type and frequency of the structural abnormalities we identified in patients with an MN less than 45 are similar to those found previously.8, 10 We were unable to perform an exhaustive analysis of the pattern of whole chromosome loss or gain because there were too few such patients in this series.

The high-hypodiploid (MN = 45) subgroup had a more homogeneous pattern of chromosomal abnormalities. For example, a sex chromosome was absent in 50% of patients lacking a dicentric chromosome. Similar to the results of the Children's Cancer Group (CCG) study by Heerema et al.,8 94% of the 33 dicentric chromosomes we identified involved the p arm of chromosome 9 (n = 18), chromosome 12 (n = 6), or both (n = 7). Previous studies have shown that the dic(9;12)(p11;p11.2), dic(9;20)(p11;q11.2), and dic(7;9)(p11.2;p11) are recurrent chromosomal abnormalities in ALL that are mostly associated with a favorable prognosis.37–42 The preferential involvement of some centromeres, such as those of chromosomes 1, 9, and 12 in rearrangements of hematopoietic malignancies, has been suggested to occur as a result of homology of the DNA sequences surrounding the centromeres of the most frequently rearranged chromosomes. This similarity facilitates the recombination of nonhomologous centromeres.43 This mechanism could explain the translocations in the current study that resulted in the dicentric formation for chromosomes 9 and 12. It is interesting to note that in 11 of 28 (39%) patients with B-lineage blast cells with an MN of 45 the chimeric fusion transcript ETV6-CBFA2 was expressed. The rationale for this finding, which differs from those of other studies (ETV6-CBFA2 is found in 25% of ALL cases in general33), is unknown.

The difficulty in determining the MN in some cases may lead to an underestimation of the frequency of near-haploidy. For example, in one of our patients (Patient 3), two metaphase spreads with unbanded chromosomes indicated near-haploidy and 16 metaphases indicated hyperdiploidy. The results of flow cytometric analysis to determine the DNA index suggested that two populations of abnormal cells were present. At subsequent disease recurrence, all abnormal metaphase cells were near-haploid and no hyperdiploid metaphases were observed. Similar discrepancies between findings at the time of diagnosis and findings at disease recurrence were noted by Stark et al.28 Apparent hyperdiploidy (and consequent misdiagnosis of haploidy or near-haploidy) could, in rare instances, arise from doubling of a haploid or a near-haploid clone. We found that abnormal clones had undergone doubling in 6 of 10 patients with an MN less than 45. Similar findings were noted in 4 of 23 patients in the CCG series.8 Therefore, careful evaluation of flow cytometry reports is crucial for the accurate identification of near-haploidy and correct diagnosis of this rare subgroup of ALL.

Of 52 pediatric patients with ALL and near-haploidy reported in the literature,7–28 few had unfavorable features at the time of diagnosis: 31 (62%) were girls and 19 (38%) were boys; age ranged from 1.3–19 years, with 14 patients (27%) being older than 10 years; and the leukocyte count ranged from 1 × 109/L to 354 × 109/L, with only 15 patients (29%) having a leukocyte count higher than 50 × 109/L. None of these patients had central nervous system or other extramedullary involvement. Similarly, most patients with near-haploidy in our series had favorable clinical features at the time of diagnosis, despite their high rate of disease recurrence. The results of our study emphasize the importance of including cytogenetic studies in the evaluation of ALL, especially in view of the particularly poor outcome of low-hypodiploid (MN < 45) cases.

The results of the current study confirm that the MN of chromosomes in the leukemic cells is a powerful prognostic factor for newly diagnosed ALL in children. Therefore, chromosomal abnormalities and the DNA content of malignant leukemic cells should be determined to delineate the prognosis in each case. Particular care should be taken to avoid misdiagnosing hypodiploidy as hyperdiploidy. Hypodiploidy with an MN less than 45 is a significant adverse risk factor, despite treatment with contemporary intensive therapies. Therefore, patients with hypodiploidy and less than 45 chromosomes in their leukemic cells are reasonable candidates for allogeneic hematopoietic stem cell transplantation. In view of the rarity of this genetic subtype, international efforts are needed to determine whether transplantation can improve outcome in these patients.


The authors thank Dr. J. R. Davies for editorial review and T. O'Neill, P. Mardis, E. Entrekin, and P. Dalton for technical expertise.