The International Prognostic Scoring System (IPSS) for childhood myelodysplastic syndrome (MDS) and juvenile myelomonocytic leukemia (JMML)


The International Prognostic Scoring System (IPSS) for myelodysplastic syndrome (MDS) is based upon weighted data on bone marrow (BM) blast percentage, cytopenia, and cytogenetics, separating patients into four prognostic groups. We analyzed the value of the IPSS in 142 children with de novo MDS and 166 children with juvenile myelomonocytic leukemia (JMML) enrolled in retro- and prospective studies of the European Working Group on childhood MDS (EWOG-MDS). Survivals in MDS and JMML were analyzed separately. Among the criteria considered by the IPSS score, only BM blasts <5% and platelets >100 × 109/l were significantly associated with a superior survival in MDS. In JMML, better survival was associated with platelets >40 × 109/l, but not with any other IPSS factors including cytogenetics. In conclusion, the IPSS is of limited value in both pediatric MDS and JMML. The results reflect the differences between myelodysplastic and myeloproliferative diseases in children and adults.


Myelodysplastic syndrome (MDS) and juvenile myelomonocytic leukemia (JMML) are rare disorders in childhood, although they have gained increasing attention in recent years.1, 2, 3, 4, 5, 6, 7 Some patients show a prolonged and stable clinical course without treatment, but most cases will eventually progress. Hematopoietic stem cell transplantation (HSCT) is the treatment of choice, but the outcome following HSCT is jeopardized if disease progression has occurred.8, 9

Information on prognostic factors predicting progression or death is important in the planning of therapy. There have been few systematic attempts to define a prognostic score in pediatric MDS. A prognostic scoring system proposed by the British group1 assigned one point each to fetal hemoglobin (HbF) >10%, platelet count <40 × 109/l, and two or more cytogenetic abnormalities (FPC score). A significantly superior survival was found in children with an FPC score of zero. The score has not been applied to other large series of children with MDS mainly because HbF is not routinely evaluated in MDS patients.4, 7

Poor prognosis in JMML has been associated with low platelet count, elevated HbF, and age at diagnosis of 2 years or more,1, 10, 11, 12 confirming two of the factors included in the FPC score.1 In particular, in the EWOG-MDS experience, a platelet count below 33 × 109/l at presentation is the strongest negative prognostic factor.12

The International Prognostic Scoring System (IPSS) was developed for adult MDS and includes weighted data on BM blast percentage, cytopenia, and cytogenetics, dividing patients into four prognostic groups.13 The IPSS has shown strong prognostic value in adults with MDS. The present study evaluates IPSS and other prognostic factors in a group of 308 children with primary MDS or JMML.


We analyzed the data of patients less than 19 years of age with a diagnosis of MDS or JMML,14 who have been included in the retro- or prospective studies of EWOG-MDS.9, 12, 15 We excluded patients with Down syndrome, previous bone marrow failure both congenital (eg, Fanconi anemia, congenital neutropenia, Diamond–Blackfan anemia, Shwachman–Diamond syndrome), and acquired (aplastic anemia) or therapy-related disease (previous chemo- or radiotherapy).

Patients were only included if complete data from the time of diagnosis were available on hemoglobin, WBC with differential count, platelet count, bone marrow differential, and cytogenetics.

Patients without standard cytogenetic studies, but with FISH analysis showing monosomy 7 or trisomy 8 in more than 30% of the cells, were included. Diagnostic cytogenetic examination was accepted if performed within 4 months (120 days) from diagnosis. Patients without cytogenetic results at diagnosis and a normal karyotype more than 4 months from diagnosis were considered to have a normal karyotype if they had received no treatment before the first cytogenetic study. Patients with a normal karyotype at diagnosis and an abnormal karyotype more than 4 months from diagnosis were considered as having normal cytogenetis in the present study. Patients with t(8;21), t(15;17) or inv(16) were considered to have de novo AML and thus excluded from the analysis.

All children had been diagnosed between August 9, 1978 and February 22, 2001. An observation time longer than 4 months from diagnosis was required for inclusion in the study. Analyses used April 7, 2003 as the reference date. The median follow-up for MDS patients who did or did not receive HSCT was 2.7 and 2.6 years, respectively. The median follow-up for JMML patients who did or did not receive HSCT was 2.1 and 0.9 years, respectively.


Separate analyses were performed for MDS and JMML. As most children with MDS and JMML are transplanted within the first year from diagnosis, the end point for all patients was survival irrespective of therapy. The survival analyses did not censor patients receiving HSCT.

MDS patients were classified according to the pediatric modification of the WHO classification14 into three groups: refractory cytopenia (RC), refractory anemia with excess of blasts (RAEB), and RAEB in transformation (RAEB-T).

The IPSS score was calculated according to the criteria proposed by Greenberg et al.13 Cytopenia was defined as hemoglobin (Hb) <10 g/dl, neutrophil count <1.5 × 109/l and platelet count <100 × 109/l and scored as 0 with cytopenia of zero or one lineage or as 0.5 with cytopenia of two or three lineages. Bone marrow blasts were scored as 0 when <5%, as 0.5 when 5–10%, 1.5 when 11–20%, and as 2.0 when 20–30%. Good risk cytogenetic characteristics included normal karyotype, -Y, 5q-, and 20q-. Poor risk cytogenetics included chromosome 7 abnormalities and complex aberrations (>2 abnormalities). The intermediate risk group included all other aberrations. Four risk groups were formed based on the scores (Table 1).

Table 1 The International Prognostic Scoring System (IPSS) for MDS13

The FPC score was calculated as described by the British group1 by assigning one point each for HbF >10%, platelets <40 × 109/l, and two or more cytogenetic abnormalities adding to an FPC score between 0 and 3. The only patient who scored 3 was analyzed in the group with a score of 2.

The Kaplan–Meier method was used to estimate survival rates. Standard errors were calculated using Greenwoods formula. The two-sided log-rank test was used to test the equality of the survivorship functions in different subgroups.16 Quantitative variables were categorized using the cut-point, resulting in the highest risk ratio in univariate Cox regression.17 This classification was repeated for each subgroup. The variable chosen for partition was the one with the largest risk ratio meeting the criteria of a P-value less than 0.01 or 0.05 for initially categorical variables. Partitioning was restricted to splits that resulted in subgroups with a minimum of 10 patients. Spearman's rank correlation and Cramérs V were used to analyze associations between prognostic factors.18 All factors with a P-value less than 0.05 in the univariate analysis were included in a multivariate analysis using the Cox proportional hazard regression model. Statistical analysis was performed using SPSS for Windows 11.0.1 (SPSS Inc, Chicago, IL, USA).


Characteristics of the 142 children with MDS and 166 children with JMML fulfilling the inclusion criteria are presented in Table 2. Monosomy 7 was the most common cytogenetic abnormality in both MDS and JMML (Tables 3 and 4). With the exception of a normal karyotype, the cytogenetic abnormalities considered to be favorable prognostic factors by the IPSS, -Y, 5q-, and 20q- as sole aberrations were absent in this pediatric cohort. Due to these cytogenetic findings and the frequent occurrence of bi- or trilinear cytopenia in children, only a few children were classified as low risk. In contrast to adult MDS, more children were classified in the high-risk group (Table 3).

Table 2 Characteristics of the children with MDS and JMML
Table 3 Distribution of children with MDS and JMML in the four IPSS groups, data from adult MDS13 are included for comparison
Table 4 Survival in children with MDS according to characteristics at diagnosis


HSCT was performed in 99 of the 142 patients (70%); 13 of 43 patients died without HSCT (median time from diagnosis to death 2.6 years) and 41 of 99 patients died following HSCT (median time from diagnosis to death 2.7 years). Survival analyses by the IPSS components showed a significantly better overall survival in patients with BM blasts <5% compared to those with BM blasts 5% (Figure 1). There was no difference in survival among patients with BM blasts 5–10, 11–20 and 21–30% (data not shown). Neither hemoglobin <10 g/dl, neutrophil count <1.5 × 109/l, nor the number of cytopenic lineages was associated with survival, while a platelet count 100 × 109/l was a poor prognostic factor (Table 4, Figure 2).

Figure 1

Survival in children with MDS according to bone marrow blast percentage at diagnosis.

Figure 2

Survival in children with MDS according to platelet count at diagnosis.

The three cytogenetic groups had comparable outcome (Figure 3). Comparing survival in patients with monosomy 7 with those with other abnormalities or normal karyotype, there was no difference between the two groups for patients with 5% BM blasts (Figure 4). For patients with <5% BM blasts, there was a trend towards a poorer outcome for those with monosomy 7 (data not shown). The analysis of outcome by the four IPSS groups showed no events in the small low-risk group and overlapping survival curves in the other three groups (Figure 5). Age and gender were not associated with survival.

Figure 3

Survival in children with MDS according to IPSS cytogenetic groups.

Figure 4

Survival in children with MDS and bone marrow blasts 5% at diagnosis according to karyotype.

Figure 5

Survival in children with MDS according to the four IPSS groups: low, intermediate 1 (Int-1), intermediate 2 (Int-2), and high.

Among the FPC factors, complex karyotype was the only factor significantly associated with a poor outcome (Table 4). The FPC score could be analyzed in 65 patients who experienced 25 events. No significant difference in survival was observed between the groups.

A multivariate analysis of variables influencing the probability of overall survival was performed with HSCT as a time-dependent covariate. All variables with a P-value less than 0.05 in the univariate analysis were included in the Cox regression model. Among these variables, platelet count >100 × 109/l, BM blast count <5%, IPSS good or intermediate risk cytogenetics, and HSCT were independently associated with superior survival.


HSCT was performed in 100 of the 166 patients (60%); 57 of 66 patients died without HSCT (median time from diagnosis to death 0.9 year) and 47 of 100 patients died following HSCT (median time from diagnosis to death 2.1 years). Survival was not influenced by BM blast percentage at diagnosis (Table 5). Similarly, cytopenia showed no significant association with survival except for platelet count. The IPSS platelet cutoff of 100 × 109/l was not significantly associated with survival; however, a platelet count below 40 × 109/l at diagnosis was a poor prognostic factor (Figure 6). The cytogenetic groups according to IPSS and the IPSS groups themselves had comparable outcome (Table 5).

Table 5 Survival in children with JMML according to characteristics at diagnosis
Figure 6

Survival in children with JMML according to platelet count at diagnosis.

Data of 129 patients could be analyzed for the FPC score.1 Survival was significantly higher in patients scoring zero (Figure 7). Thrombocytopenia was the major factor responsible for the prognostic impact of the FPC score. Survival in patients with HbF <10% tended to be superior (P=0.07) compared with patients with HbF 10%. The cytogenetic scoring was not informative because only three patients had a complex karyotype.

Figure 7

Survival in children with JMML according to the FPC score.

Age below 2 years was associated with significantly better survival (Table 5). Gender was not a risk factor for outcome.

A multivariate Cox analysis incorporating age, platelets, HbF, BM blast percentage, and HSCT as a time-dependent covariate was performed. A low platelet count at diagnosis and no HSCT or long time interval from diagnosis to HSCT were associated with poorer survival.


In contrast to the original study on IPSS in adults,13 we included patients who had received HSCT. The study might have been more informative if more nontransplanted patients had been included. However, early HSCT has been established as the treatment of choice in both MDS and JMML, and it is unrealistic to gather a significantly larger group of patients not treated with HSCT.

This study documents a superior survival in MDS patients with BM blasts <5% and in those with a platelet count 100 × 109/l, but none of the other IPSS factors had any significant impact on outcome. The IPSS identified a small low-risk group (7% of the patients) with an excellent prognosis. These are patients with RC, a normal karyotype, and absence of severe cytopenia.9 The three other prognostic groups as defined by the IPSS were not useful in predicting survival in childhood MDS. The distribution among the IPSS groups in children is skewed towards more patients in the high-risk group due to a higher frequency of cytopenia and monosomy 7 compared to adults.13 A similar predominance of high-risk patients is seen in younger adults with MDS.19

Monosomy 7 was the most common cytogenetic abnormality in MDS and the major cause of classifying cytogenetics as unfavorable. Survival in those with unfavorable cytogenetic features was not significantly different from the remaining group of patients in the univariate analysis. The lack of monosomy 7 as an unfavorable feature in MDS has also been observed in previous studies.4, 15, 20 By contrast, -7/7q- in adults has been associated with a poor outcome both after chemotherapy13 and after HSCT.21 However, in children without excess of blasts, monosomy 7 is associated with a shorter time to progression9 and thereby an inferior survival. Patients with advanced MDS have a poorer outcome and in those patients monosomy 7 has no independent prognostic value (Figure 4). In a Japanese study of childhood MDS, the survival differed between the IPSS cytogenetic risk groups.6 The frequency of cytogenetic abnormalities in the Japanese study was lower than in the present study, and RC and advanced MDS were analyzed together. We found a tendency for a superior survival in the IPSS intermediate risk cytogenetic group (Figure 3). The same tendency was reported in MDS children from the UK,7 probably reflecting the different pattern of cytogenetics in children and adults. The favorable cytogenetic aberrations defined by IPSS (-Y, 20q-, and 5q-) were not observed as the sole abnormality in any of the children included in our study. The rarity of these cytogenetic changes in children has been confirmed in other series.4, 6

The IPSS provided no prognostic information in children with JMML. Poor survival in nontransplanted patients has been associated with low platelet count, high HbF, and high BM blast count.1, 10, 11, 12 Cytogenetics including monosomy 7 did not influence the outcome following HSCT. This is in contrast to a Japanese study showing a very poor outcome in a small group of JMML patients with abnormal karyotype.22 The FPC score is more useful in JMML than IPSS.7 However, after multivariate analysis, only a low platelet count was independently associated with a poor survival.

The IPSS was proposed on the basis of data from adult patients. The present study demonstrates major differences between children and adults with MDS. The clinical characteristics are different and prognostic factors predicting survival or progression in adults are of little value in children. The BM blast count and thrombocytopenia were the only IPSS factors being associated with outcome in childhood MDS.

Potentially curative therapy with HSCT is offered to the majority of children with MDS and JMML. HSCT should be performed before progression or other complications have occurred. We identified thrombocytopenia and BM blasts >5% being associated with an inferior survival in MDS. Low platelet count was associated with a poor survival in JMML. HSCT should be performed early in patients with these characteristics.


  1. 1

    Passmore SJ, Hann IM, Stiller CA, Ramani P, Swansbury GJ, Gibbons B et al. Pediatric myelodysplasia: a study of 68 children and a new prognostic scoring system. Blood 1995; 85: 1742–1750.

  2. 2

    Hasle H, Kerndrup G, Jacobsen BB . Childhood myelodysplastic syndrome in Denmark: incidence and predisposing conditions. Leukemia 1995; 9: 1569–1572.

  3. 3

    Bader-Meunier B, Mielot F, Tchernia G, Buisine J, Delsol G, Duchayne E et al. Myelodysplastic syndrome in childhood: report of 49 patients from a French multicenter study. Br J Haematol 1996; 92: 344–350.

  4. 4

    Luna-Fineman S, Shannon KM, Atwater SK, Davis J, Masterson M, Ortega J et al. Myelodysplastic and myeloproliferative disorders of childhood: a study of 167 patients. Blood 1999; 93: 459–466.

  5. 5

    Hasle H, Wadsworth LD, Massing BG, McBride M, Schultz KR . A population-based study of childhood myelodysplastic syndrome in British Columbia, Canada. Br J Haematol 1999; 106: 1027–1032.

  6. 6

    Sasaki H, Manabe A, Kojima S, Tsuchida M, Hayashi Y, Ikuta K et al. Myelodysplastic syndrome in childhood: a retrospective study of 189 patients in Japan. Leukemia 2001; 15: 1713–1720.

  7. 7

    Passmore SJ, Chessells JM, Kempski H, Hann IM, Brownbill PA, Stiller CA . Paediatric MDS and JMML in the UK: a population based study of incidence and survival. Br J Haematol 2003; 121: 758–767.

  8. 8

    Anderson JE, Appelbaum FR, Schoch G, Gooley T, Anasetti C, Bensinger WI et al. Allogeneic marrow transplantation for refractory anemia: a comparison of two preparative regimens and analysis of prognostic factors. Blood 1996; 87: 51–58.

  9. 9

    Kardos G, Baumann I, Passmore SJ, Locatelli F, Hasle H, Schultz KR et al. Refractory anemia in childhood: a retrospective analysis of 67 patients with particular reference to monosomy 7. Blood 2003; 102: 1997–2003.

  10. 10

    Castro-Malaspina H, Schaison G, Passe S, Pasquier A, Berger R, Bayle-Weisgerber C et al. Subacute and chronic myelomonocytic leukemia in children (juvenile CML). Clinical and hematologic observations, and identification of prognostic factors. Cancer 1984; 54: 675–686.

  11. 11

    Owen G, Lewis IJ, Morgan M, Robinson A, Stevens RF . Prognostic factors in juvenile chronic granulocytic leukaemia. Br J Cancer Suppl 1992; 18: S68–S71.

  12. 12

    Niemeyer CM, Aricò M, Basso G, Cantù-Rajnoldi A, Creutzig U, Haas OA, et al., Members of the European Working Group on Myelodysplastic Syndromes in Childhood (EWOG-MDS). Chronic myelomonocytic leukemia in childhood: a retrospective analysis of 110 cases. Blood 1997; 89: 3534–3543.

  13. 13

    Greenberg P, Cox C, Le Beau MM, Fenaux P, Morel P, Sanz G et al. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood 1997; 89: 2079–2088.

  14. 14

    Hasle H, Niemeyer CM, Chessells JM, Baumann I, Bennett JM, Kerndrup G et al. A pediatric approach to the WHO classification of myelodysplastic and myeloproliferative diseases. Leukemia 2003; 17: 277–282.

  15. 15

    Hasle H, Aricò M, Basso G, Biondi A, Cantù-Rajnoldi A, Creutzig U et al. Myelodysplastic syndrome, juvenile myelomonocytic leukemia, and acute myeloid leukemia associated with complete or partial monosomy 7. Leukemia 1999; 13: 376–385.

  16. 16

    Kaplan EL, Meier P . Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–481.

  17. 17

    Cox DR . Regression models and life tables. J R Stat Soc B 1972; 34: 187–220.

  18. 18

    Hosmer DW, Lemeshow S . Applied Survival Analysis. Regression Modeling of Time to Event Data. New York: Wiley, 1999.

  19. 19

    Chang KL, O'Donnell MR, Slovak ML, Dagis AC, Arber DA, Niland JC et al. Primary myelodysplasia occurring in adults under 50 years old: a clinicopathologic study of 52 patients. Leukemia 2002; 16: 623–631.

  20. 20

    Woods WG, Barnard DR, Alonzo TA, Buckley JD, Kobrinsky N, Arthur DC et al. Prospective study of 90 children requiring treatment for juvenile myelomonocytic leukemia or myelodysplastic syndrome: a report from the Children's Cancer Group. J Clin Oncol 2002; 20: 434–440.

  21. 21

    Nevill TJ, Fung HC, Shepherd JD, Horsman DE, Nantel SH, Klingemann HG et al. Cytogenetic abnormalities in primary myelodysplastic syndrome are highly predictive of outcome after allogeneic bone marrow transplantation. Blood 1998; 92: 1910–1917.

  22. 22

    Manabe A, Okamura J, Yumura-Yagi K, Akiyama Y, Sako M, Uchiyama H et al. Allogeneic hematopoietic stem cell transplantation for 27 children with juvenile myelomonocytic leukemia diagnosed based on the criteria of the International JMML Working Group. Leukemia 2002; 16: 645–649.

Download references


This work was supported in part by grants from the Danish Childhood Cancer Foundation (Børnecancerfonden), the German José Carreras Leukemia Foundation, and the German BMBF Competence Network Pediatric Oncology (Project E: Preleukemic Bone Marrow Disorders).

Author information

Correspondence to H Hasle.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Hasle, H., Baumann, I., Bergsträsser, E. et al. The International Prognostic Scoring System (IPSS) for childhood myelodysplastic syndrome (MDS) and juvenile myelomonocytic leukemia (JMML). Leukemia 18, 2008–2014 (2004) doi:10.1038/sj.leu.2403489

Download citation


  • children
  • JMML
  • IPSS
  • MDS
  • prognosis

Further reading