The impact of intensive antileukaemic treatment strategies on prognosis of myelodysplastic syndrome patients aged less than 61 years according to International Prognostic Scoring System risk groups

Authors


M. Oosterveld, Department of Hematology, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands. E-mail: M.Oosterveld@haemat.umcn.nl

Abstract

Summary. The present study applied the International Prognostic Scoring System (IPSS) to 306 consecutive myelodysplastic syndrome (MDS) patients diagnosed between August 1977 and September 2000 at the University Medical Centre Nijmegen. The aim was to investigate whether the IPSS could be used as a prognostic tool in MDS patients aged less than 61 years who were treated with acute myeloid leukaemia (AML)-like chemotherapy with or without transplantation, and whether the scoring system discriminated between the subgroups of patients who benefit from intensive treatment strategies. The patients were retrospectively assigned to the IPSS risk categories and compared with the IPSS workshop patients. Eighty-three of 159 patients aged < 61 years, classified as intermediate 1, intermediate 2 and high risk according to the IPSS, received intensive treatment consisting of chemotherapy only (n = 30), chemotherapy followed by either autologous stem cell transplantation (n = 7) or allogeneic stem cell transplantation (n = 46). After intensive treatment, the median survival was 2·6 years for the intermediate 1 risk group (n = 33), 3·4 years for the intermediate 2 risk group (n = 27) and 0·9 years for the high-risk group (n = 23). We conclude that the IPSS is an improved scoring system for patients receiving supportive care. Nevertheless, the scoring system does not seem to be the best method for predicting outcome after intensive antileukaemic treatment. In particular, intermediate 2 risk patients may benefit from intensive treatment.

The myelodysplastic syndromes (MDS) are clonal stem cell disorders characterized by peripheral blood cytopenias in the presence of hypercellular bone marrow with features of ineffective haematopoiesis. Since 1982, MDS have been classified according to French–American–British (FAB) criteria, based on the number of blasts in bone marrow and peripheral blood and the percentage of ringed sideroblasts in the bone marrow and monocytes in the blood (Bennett et al, 1982). Five subgroups have been defined: refractory anaemia (RA), refractory anaemia with ringed sideroblasts (RARS), refractory anaemia with excess of blasts (RAEB), refractory anaemia with excess of blasts in transformation (RAEBt) and chronic myelomonocytic leukaemia (CMML). Although this classification has shown its usefulness in predicting the outcome of patients, considerable variation still exists within the different subgroups regarding clinical features and survival. Numerous scoring systems have been proposed to predict the prognosis of individual patients. (Sanz et al, 1989; Morel et al, 1993) In 1997, an international workshop combined the data from seven previously reported studies (Mufti et al, 1985; Oscier, 1987; Sanz et al, 1989; Aul et al, 1992; Greenberg et al, 1993; Morel et al, 1993; Toyama et al, 1993) to generate an International Prognostic Scoring System (IPSS) (Greenberg et al, 1997a). The IPSS determined four risk groups for survival and acute myeloid leukaemia (AML) evolution (low risk, intermediate 1 risk, intermediate 2 risk and high risk), based on karyotype, number of blasts in the bone marrow and number of cytopenias (Table I). Age was an additional prognostic factor for survival, but not for AML evolution. The IPSS seemed to be an improved classification system for predicting the natural history in MDS (Maes et al, 1999). However, the IPSS was based on data from patients treated with transfusions, biological response modifiers and low-dose oral chemotherapy. Patients treated with intensive chemotherapy and/or stem cell transplantation and patients with therapy-related MDS were excluded from the analysis.

Table I.  The International Prognostic Scoring System (IPSS).
Prognostic variableScore value
00·51·01·52·0
  • Scores for risk groups are as follows: Low: 0; Intermediate 1: 0·5–1·0; Intermediate 2: 1·5–2·0; and High: ≥ 2·5.

  • *

    Good: normal, –Y, del(5q), del(20q); Poor: complex (≥ three abnormalities), chromosome 7 abnormalities; Intermediate: other abnormalities.

  • † Hb < 10 g/dl, neutrophil count < 1·8  ×   109 /l, platelet count < 100 × 109/l.

BM blasts (%)55–1011–2021–30
Karyotype*GoodIntermediatePoor  
Cytopenias0/12/3   

The present study is a retrospective analysis of 306 untreated MDS patients who have been seen in the Nijmegen University Hospital, many of whom subsequently received intensive antileukaemic treatment consisting of AML-like chemotherapy with or without autologous stem cell transplantation, allogeneic transplantation with or without preceding chemotherapy. Classified according to the IPSS criteria, the outcome of our patients was compared with that of the IPSS workshop patients.

The principle aim was to investigate whether the IPSS predicts the outcome of patients aged less than 61 years treated with intensive therapy, and whether the scoring system discriminates between subgroups of patients who may benefit from these intensive treatment strategies.

Patients and methods

Patients.  Between August 1977 and September 2000, 380 newly diagnosed MDS patients were referred to the University Medical Centre Nijmegen. This group consisted of patients referred by their family practitioners or physicians from general hospitals in south-east Netherlands. We reviewed the data of all MDS patients. In 74 patients (19%), sufficient data were not available, mainly because of lack of cytogenetic data, and these patients were excluded from the present analysis. In 302 out of the remaining 306 patients, the karyotype was known. Four patients with unknown karyotype belonged to the high-risk group of the IPSS based on the number of blasts in the bone marrow (21–30%) and the number of cytopenias (2/3). Therefore, these patients were not withdrawn from the analysis. In all, 306 patients were retrospectively assigned to the various IPSS risk categories.

Treatment.  All patients received supportive care. The majority of elderly patients (aged 61 years or over) received supportive care without intensive treatment. Only nine elderly patients (7%) were treated with intensive chemotherapy followed by autologous stem cell transplantation in one patient.

Generally, the accepted age limit for allogeneic stem cell transplantation varies between 50 and 60 years. In Nijmegen, patients aged < 61 years are eligible for allogeneic stem cell transplantation, in the absence of other medical contraindications. Low-risk IPSS patients were not candidates for intensive antileukaemic treatment, in view of their relatively good prognosis. In contrast, patients aged < 61 years, belonging to the intermediate 1, intermediate 2 and high-risk categories of the IPSS could be considered as candidates for intensive treatment strategies including allogeneic stem cell transplantation. We divided the patients into two groups, diagnosed before and after 1 January 1992. Before 1992, the decision to start intensive treatment was individually based. In those years, 13 patients with a human leucocyte antigen-identical donor underwent allogeneic stem cell transplantation: five patients with RA received a transplant without preceding chemotherapy, and eight RAEB/RAEBt patients underwent AML-like chemotherapy followed by an allogeneic transplantation. Six additional patients received chemotherapy but no transplantation at the time of progression of their disease. In January 1992, the department changed the treatment policy and, from 1992 onwards, all high-risk MDS patients were offered intensive antileukaemic treatment, if they met the eligibility criteria of successive European Organization for the Research and Treatment of Cancer (EORTC) protocols. Patients were candidates for intensive treatment if they were aged 16–60 years and had untreated (i) RAEBt, (ii) RAEB with > 10% blasts cells in the bone marrow, (iii) other forms of MDS with multiple chromosomal abnormalities and/or profound cytopenias defined as: neutrophil count < 0·5 × 109/l and/or platelet count < 20 × 109/l, or (iv) CMML with > 5% blasts cells in the bone marrow or with a neutrophil count > 16 × 109/l or a monocyte count > 2·6 × 109/l in the blood.

After 1992, 64 patients were offered intensive antileukaemic treatment. Twenty-two patients diagnosed after 1992 did not receive intensive treatment for the following reasons: treatment refused (n = 3), psychiatric disorder (n = 1), language barrier and platelet refractoriness (n = 1), concomitant disease (n = 2), death before start of treatment due to infection (n = 2), cardiomyopathy (n = 1), not eligible for the protocol (n = 4), doctor's decision (n = 5), unknown (n = 3).

Statistical analysis.  Chi-square tests were used to compare differences between groups. Kaplan–Meier survival analyses were used to estimate median survival time and to plot survival curves. To compare survival times in different strata, the log-rank test was used (Cox & Oakes, 1984).

The prognostic effect of IPSS, age, diagnosis period, treatment, bone marrow blasts, cytopenias and cytogenetics on survival time was analysed in multivariate analyses using proportional hazards models. In this multivariate analysis, we also evaluated the interaction between IPSS and age (< 61 years versus > 60 years). A P-value < 0·05 was considered statistically significant.

Results

The Nijmegen patients were younger than the IPSS workshop patients, with a median age of 58 years versus 69 years (Table II). Eighteen per cent of patients developed MDS after prior treatment with chemotherapy or radiotherapy for a non-related disease (therapy-related MDS). The Nijmegen patients presented with more cytopenias and more severe cytopenias resulting in median haemoglobin level, neutrophil and platelet counts of 9·2 g/dl, 1·4 × 109/l and 75 × 109/l respectively. The corresponding blood counts for the IPSS workshop patients were 9·7 g/dl, 2·0 × 109/l and 132 × 109/l. Median survival was 5·1 years in the low-risk group, 2·4 years for the intermediate 1 risk group, 1·2 years for the intermediate 2 risk group and 0·8 years for the high-risk group (P < 0·0001) (Fig 1). Median survival in the low-risk and intermediate 2 risk group was comparable in the Nijmegen and the IPSS workshop patients. However, the median survival in the intermediate 1 risk group was lower in the Nijmegen patients, and median survival in the high-risk group was better for the Nijmegen patients (Table III).

Table II.  Characteristics of all Nijmegen MDS patients compared with IPSS workshop patients and of Nijmegen patients aged less than 61 years at diagnosis (low risk excluded), according to applied treatment.
VariableNijmegen
n (%)
IPSS
workshop
n (%)
Nijmegen < 61 years P-value*
No intensive
treatment
n (%)
Intensive
treatment
n (%)
  • *

    P-value for the comparison between intensive treatment and no intensive treatment in the Nijmegen patients.

  • Median age (years).

No. of patients306 (100)816 (100)76 (100)83 (100) 
Gender
 Male198 (65)491 (60)43 (57)54 (65)0·27
 Female108 (35)325 (40)33 (43)29 (35) 
Age (years)
  ≤  60176 (58)205 (25)5345< 0·001
 > 60130 (42)611 (75)   
MDS
 Primary250 (82)816 (100)56 (74)72 (87)0·04
 Therapy-related56 (18)020 (26)11 (13) 
FAB classification
 RA94 (31)294 (36)29 (38)18 (23)< 0·001
 RARS37 (12)125 (15)8 (11)5 (6) 
 RAEB96 (31)208 (26)26 (34)24 (29) 
 RAEBt64 (21)61 (8)9 (12)34 (41) 
 CMML11 (4)126 (15)4 (5)2 (2) 
 Unknown4 (1)    
BM blasts
 < 5%136 (44)483 (59)37 (49)22 (27)0·03
 5–10%81 (27)183 (22)20 (26)26 (31) 
 11–20%60 (20)114 (14)13 (17)24 (29) 
 21–30%29 (10)36 (5)6 (8)11 (13) 
Cytopenias
 0/1114 (37)474 (58)27 (36)24 (29)0·37
 2/3192 (63)342 (42)49 (64)59 (71) 
Cytogenetic subgroup
 Good139 (46)570 (70)22 (29)36 (44)0·15
 Intermediate83 (27)112 (14)26 (34)22 (26) 
 Poor80 (26)134 (16)28 (37)24 (29) 
 Unknown4 (1)  1 (1) 
IPSS
 Low37 (12)267 (33)   
 Intermediate 1124 (41)314 (38)38 (50)33 (40)0·17
 Intermediate 288 (29)176 (22)26 (34)27 (33) 
 High57 (19)59 (7)12 (16)23 (28) 
Intensive treatment
 Yes93 (30)0083 (100) 
 No213 (70)816 (100)76 (100)0 
Diagnosis
 Before (1992)163 (47)816 (100)54 (71)19 (23)< 0·001
 After 1992143 (53)022 (29)64 (77) 
Figure 1.

Survival from diagnosis in all patients according to IPSS risk group (n = 306).

Table III.  Survival of Nijmegen patients and IPSS workshop patients (Greenberg et al, 1997a) per IPSS risk group in all patients and in patients aged less than 61 years.
IPSSNijmegen
all
(n = 306)
n (%)
Median
survival
(years)
Greenberg
all
(n = 816)
n (%)
Median
survival
(years)
Nijmegen
< 61 years
(n = 176)
n (%)
Median
survival
(years)
Greenberg
< 61 years
(n = 205)
n (%)
Median
survival
(years)
Low37 (12)5·1267 (33)5·717 (10)11·160 (29)11·8
Int 1124 (41)2·4314 (39)3·571 (40)2·587 (42)5·2
Int 288 (29)1·2176 (22)1·253 (30)1·449 (24)1·8
High57 (19)0·859 (7)0·435 (20)0·99 (4)0·3

Intensive antileukaemic therapy with or without stem cell transplantation was administered to patients younger than 61 years. Therefore, the outcome of patients aged < 61 years was compared with the same age category of the IPSS workshop patients. The differences in median survival between the Nijmegen patients and the IPSS workshop patients in the younger patient groups were similar to the differences found in the overall groups (Table III).

Effect of intensive treatment in patients aged < 61 years

There were 176 patients who were aged < 61 years. Patients with intermediate 1, intermediate 2 and high-risk IPSS scores could be considered as candidates for intensive antileukaemic treatment. Seventeen patients belonged to the low-risk group of the IPSS. Of the remaining 159 patients, classified as IPSS intermediate 1, intermediate 2 and high-risk categories, 83 patients (52%) received intensive treatment, and 76 patients (48%) did not. Thirty patients received chemotherapy only, seven patients underwent chemotherapy followed by autologous transplantation and, in 46 patients, allogeneic stem cell transplantation was performed (21 patients with RARS without preceding chemotherapy). Four patients received a transplant from a voluntary unrelated donor (VUD).

Intensive chemotherapy in patients < 61 years consisted of cytarabine only (n = 3), cytarabine and anthracycline (n = 7), cytarabine, amsacrine and etoposide (n = 2), cytarabine, anthracycline and vincristine (n = 3) or cytarabine, anthracycline and etoposide (n = 47).

Seventy-six patients did not receive intensive treatment. Patient characteristics for both groups are shown in Table II. The median age of the patients who received intensive treatment was 45 years versus 53 years for patients who received supportive care only. The estimated 4-year survival of the 83 patients who underwent intensive treatment was 36·3% versus 23·0% for the 76 patients who received supportive care only (P = 0·02).

In the intensive treatment group, median survival from diagnosis was 2·6 years for the intermediate 1 risk group, 3·4 years for the intermediate 2 risk group and 0·9 years for the high-risk group. The estimated 4-year survival rates were 46%, 40% and 19% respectively (Table IV). In the intermediate 1 risk group, six patients received chemotherapy only, three patients received chemotherapy followed by autologous transplantation, and 24 patients underwent allogeneic stem cell transplantation. In the intermediate 2 risk group, 12 patients received chemotherapy only, three patients received chemotherapy followed by autologous transplantation, and 12 patients underwent allogeneic stem cell transplantation. These numbers were 12, 1 and 10, respectively, in the high-risk group. The median time between diagnosis and start of treatment was 0·2 years (range 0–5·7 years). At the start of treatment, 15 patients had progressed to a higher IPSS risk category than at initial diagnosis. Hence, at the start of intensive treatment, 25 patients were classified in the intermediate 1 risk group, 23 patients in the intermediate 2 risk group and 35 patients in the high-risk group. Survival from start of treatment for the different IPSS risk categories at start of intensive treatment is shown in Fig 2.

Table IV.  Survival of patients aged less than 61 years according to IPSS risk group at diagnosis (low risk excluded) for patients with and without intensive treatment (n = 159).
IPSSNo intensive
treatment
(n = 76)
Median
survival
(years)
4-year
survival
(%)
Intensive
treatment
(n = 83)
Median
survival
(years)
4-year
survival
(%)
Intermediate 1382·237332·646
Intermediate 2260·811273·440
High120·70230·919
Figure 2.

Survival from start of treatment in patients aged less than 61 years after intensive antileukaemic therapy according to IPSS risk group at start of intensive treatment (n = 83).

The survival from diagnosis in patients who received no intensive treatment is shown in Table IV.

An additional analysis was performed, which excluded the secondary MDS patients, as this group of patients was not included in the IPSS workshop. The median survival according to IPSS risk groups of both intensively treated patients (n = 72) and non-intensively treated patients (n = 56) was not different, if the analysis was restricted to primary MDS patients (data not shown).

Effect of allogeneic stem cell transplantation in patients aged < 61 years

Forty-six patients underwent allogeneic stem cell transplantation. The median survival of these 46 patients was 5·7 years compared with 0·9 years for the 37 patients who received chemotherapy with or without autologous stem cell transplantation (P = 0·0005). Of the 46 patients receiving an allogeneic transplantation, 24 belonged to the intermediate 1 risk group, 12 to the intermediate 2 risk group and 10 to the high-risk group. The median survival after allogeneic stem cell transplantation was 5·7 years for the intermediate 1 risk group, 10·3 years for the intermediate 2 risk group and 1·1 years for the high-risk group (P = 0·28). The estimated 4-year survival rates were 50% for the intermediate 1 risk group, 67% for the intermediate 2 risk group and 30% for the high-risk group.

Effect of the year of diagnosis on treatment in patients aged < 61 years

A total of 159 patients were assigned to the intermediate-1, intermediate-2 and high-risk groups of the IPSS. Eighty-three patients who were younger than 61 years received intensive treatment. Nineteen patients diagnosed before 1992 received intensive treatment and 64 patients thereafter. The median survival was 1·4 years versus 1·8 years respectively (P = 0·5).

Seventy-six patients received supportive care only. The median survival of 54 patients diagnosed before 1992 was 1·2 years versus 2·2 years for the 22 patients diagnosed after 1992 (P = 0·7).

Influence of cytogenetic abnormalities

The karyotype was known for 302 out of 306 patients.

A total of 139 patients (46%) showed good prognostic cytogenetic features according to the IPSS (normal cytogenetics, 5q-, 20q- or –Y). Poor prognostic features (three or more abnormalities, chromosome 7 abnormalities) were observed in 80 patients (26%) and intermediate prognostic features (any other cytogenetic abnormality) in 83 patients (27%). The median survival of patients with a good karyotype was 2·8 years, with a poor karyotype 0·8 years and with an intermediate karyotype 2·0 years (P < 0·0001).

According to the Keating classification (Keating et al, 1988) as used for AML, no patients had favourable cytogenetic abnormalities [inv (16), t(8,21) or t(15,17)]. Intermediate prognostic cytogenetic abnormalities (normal, –Y) occurred in 149 patients (49%), and unfavourable prognostic cytogenetic abnormalities (chromosome 5 or 7 abnormalities, 11q, other abnormalities) were found in 153 patients (61%). The median survival for the intermediate prognostic group according to this classification was 2·7 years versus 1·2 years for the unfavourable prognostic group.

Multivariate analyses

As the IPSS was designed to predict the survival of patients treated with supportive care or low-intensity regimens, we analysed the prognostic effect of the different IPSS categories and age on survival in 213 patients who received supportive care only. The results of the multivariate Cox's proportional hazards model are shown in Table V. The classification into the four IPSS categories appeared to be highly prognostic for the duration of survival. The estimated hazard ratio for the high-risk group compared with the intermediate 1 risk group was 2·84 [95% confidence interval (CI), 1·80–4·49; P < 0·0001]. The estimated hazard ratio of older versus younger patients was 1·01 (95% CI 0·99–1·02; P = 0·18).

Table V.  Results of Cox's proportional hazards model for overall survival in patients who received no intensive treatment (n = 213).
VariableHazard ratio*95% CIP-value
  • CI, confidence interval.

  • *

    A value > 1 indicates that the outcome is worse for that category in comparison with the baseline.

IPSS risk group
 Low0·400·23–0·690·0009
 Intermediate 11·00  
 Intermediate 21·651·14–2·400·008
 High2·841·80–4·49< 0·0001
Age (years)
 < 601·00  
  ≥  601·010·99–1·020·18

Secondly, the impact of intensive treatment was tested in 159 younger patients, belonging to the intermediate 1, intermediate 2 and high-risk categories of the IPSS. The results are shown in Table VI. The IPSS risk groups appeared to be of significant prognostic value. The comparison between patients who received intensive treatment and those who received supportive care only yielded a hazard ratio of 0·65 (95% CI 0·45–0·94). This indicated that the rate of death was 35% lower in patients receiving intensive antileukaemic treatment (P = 0·02).

Table VI.  Results of Cox's proportional hazards model for survival of intermediate 1, intermediate 2 and high-risk patients aged less than 61 years (n = 159).
VariableHazard ratio*95% CIP-value
  • CI, confidence interval.

  • *

    A value > 1 indicates that the outcome is worse for that category in comparison with the baseline.

IPSS risk group
 Intermediate 11·00  
 Intermediate 21·460·96–2·230·08
 High2·221·38–3·580·001
Intensive treatment
 Yes0·650·45–0·940·02
 No1·00  
Diagnosis
 Before 19921·170·80–1·700·41
 After 19921·00  

Ultimately, we analysed the prognostic effect of the IPSS in 83 intensively treated patients aged < 61 years (excluding the low-risk group). The estimated hazard ratio for the high-risk group compared with the intermediate 1 risk group was 1·81 (95% CI 0·16–21·16; P = 0·63). The estimated hazard ratio for the intermediate 2 risk group compared with the intermediate 1 risk group was 0·83 (95% CI 0·22–3·13; P = 0·78). The percentage of bone marrow blasts cells was not predictive for survival. There was a trend for better survival in patients with fewer cytopenias and without poor prognostic cytogenetic features, although this did not reach statistical significance. Outcome after allogeneic stem cell transplantation was significantly better than after chemotherapy with or without autologous stem cell transplantation (P = 0·006).

Discussion

Since 1982, the FAB classification has been used to classify MDS patients and to predict outcome in terms of survival and risk of AML development. The FAB classification is based solely on morphological criteria. It has been increasingly acknowledged that biological and molecular variables are important for diagnosing and stratifying MDS patients.

The development of the International Prognostic Scoring System (IPSS) in 1997, based on cytogenetic criteria, number of blasts in the bone marrow and number of cytopenias, was a major step forward towards a risk-adapted treatment strategy for an individual patient. We applied the IPSS to 306 MDS patients from a single university centre. The Nijmegen patients formed a rather heterogeneous group compared with the IPSS workshop patients (i.e. different age distribution, including secondary MDS patients and patients receiving intensive treatment). To investigate the effectiveness of the IPSS in a large group of unselected patients, we applied the IPSS to all Nijmegen patients. Despite the different nature of the Nijmegen patients, the IPSS stratified our patients effectively into the four different categories. The number of reports applying the IPSS is still limited (Estey et al, 1997a; Lee et al, 1999; Maes et al, 1999; Jaiyesimi et al, 2000; Sperr et al, 2001). An overview of the reported results is given in Table VII. The patients from the M. D. Anderson Cancer Center (Estey et al, 1997a) were younger, more often had RAEBt, two or three cytopenias and prognostically poor cytogenetic abnormalities. A shorter survival for each of the IPSS risk categories was found. In the Leuven analysis (Maes et al, 1999), patients were diagnosed during a different period. This analysis included only patients who underwent bone marrow biopsy. Out of 184 patients, 30 (16%) received high-dose chemotherapy. Outcome of low- and high-risk patients was better (6·5 and 0·7 years) and outcome of intermediate 1 risk patients (2·6 years) was worse compared with the IPSS workshop patients. The median age of the patients in the Korean study was 53 years (Lee et al, 1999). This study included patients treated with low-dose cytarabine, androgen and all-trans retinoic acid. Jaiyesimi et al (2000) applied the IPSS to 74 patients diagnosed between 1990 and 1997. Compared with the workshop patients, their patients were older and presented with more cytopenias. The median survival for the intermediate 1 (3·4 years) and high-risk (0·5 years) groups was comparable, but an improved survival for the intermediate 2 risk group (4·1 years) was found. However, patient numbers in the low-risk, intermediate 2 risk and high-risk groups were small.

Table VII.  Application of the International Prognostic Scoring System in the literature regarding clinical outcome of MDS patients.
 IPSSEstey et al (1997a)Maes et al (1999)Lee et al (1999)Jaiyesimi et al (2000)Sperr et al (2001)Present
study
  • NR, not reached.

  • *

    Percentage of patients in the different IPSS risk groups.

No. of patients8162191849179102306
Median age (years)69656453747158
Time period1970–92Before 19911980–971989–971990–971989–991977–2000
Median survival, years (%)*
 Low5·7 (33)2·1 (13)6·5 (22)3·8 (4)NR (15)7·5 (23)5·1 (12)
 Intermediate 13·5 (38)1·2 (41)2·6 (46)3·6 (47)3·4 (49)1·7 (41)2·4 (41)
 Intermediate 21·2 (22)0·7 (30)1·3 (25)0·8 (24)4·1 (19)1·9 (20)1·2 (29)
 High0·4 (7)0·4 (16)0·7 (7)0·7 (24)0·5 (16)1·0 (17)0·8 (19)

The observed differences in median survival within the four risk categories raises several questions. Do the IPSS workshop patients represent patients diagnosed now? Does the selection of patients play a role in the reported results? The IPSS workshop obtained clinical data from 816 primary MDS patients from seven previously reported studies. Together, these studies included over 1600 patients between 1970 and 1992. In about 50% of reported patients, insufficient data were available, and hence selection may have occurred. Compared with the IPSS workshop patients, the Nijmegen patients presented with more adverse prognostic features. As Nijmegen is a referral centre for MDS patients in the south-eastern part of The Netherlands, this may in part explain the observed differences. As in the series of Estey et al (1997a), median blood counts in Nijmegen patients were lower than in the IPSS workshop patients. The workshop authors (Greenberg et al, 1997b) confirmed a shorter survival in patients with more profound cytopenias.

Another point to consider is whether the natural history of MDS has changed over time, and whether supportive care has improved in recent years. All IPSS workshop patients were diagnosed before 1992. In the Nijmegen patients, a trend towards better survival was observed for both intensively treated and non-intensively treated patients diagnosed after 1992. However, this was not statistically significant.

In the Nijmegen cohort, 213 out of 306 patients received supportive care only. In these patients, the IPSS was highly predictive for survival. Multivariate analysis showed a sevenfold risk of death for the IPSS high-risk patients compared with the low risk patients.

The key question is whether the IPSS could also predict the outcome of MDS patients aged < 61 years treated with intensive therapies, including stem cell transplantation. Therefore, we evaluated survival in 83 intensively treated patients. In this study, the median survival of intermediate 2 risk patients was better than that of intermediate 1 risk patients. This suggests that patients belonging to the intermediate 2 risk group, in particular, benefited from intensive treatment strategies. In multivariate analysis, the different IPSS risk groups were not predictive of survival. Hence, this analysis does not support the value of the IPSS in predicting survival after intensive antileukaemic treatment.

Median survival after intensive antileukaemic treatment was significantly better than median survival after supportive care only. Opponents of intensive treatment in MDS often argue that the reported results are biased by selection of patients with a relatively good prognosis. The present analysis establishes that survival of non-intensively treated patients did not deteriorate with the course of time. This supports the fact that we did not select patients with a relatively good prognosis for intensive treatment. Nevertheless, this analysis was not designed to compare the outcome of intensively treated patients with that of patients who received supportive care only. An important pitfall of observational studies is that selection bias can never be ruled out completely.

Another point to consider is whether the applied treatment was effective. Treatment included chemotherapy only, chemotherapy followed by autologous transplantation and allogeneic transplantation with or without preceding chemotherapy. We demonstrated that chemotherapy only and chemotherapy followed by autologous stem cell transplantation were less effective than allogeneic stem cell transplantation. One can argue that the imbalance of the different treatments in the various IPSS risk groups may have influenced the outcome. However, in the more homogeneous group of 46 patients who underwent allogeneic stem cell transplantation, the conclusions remained unchanged.

Finally, it is still unclear which patients benefit most from intensive treatment strategies. The present study suggests that survival in the intermediate 2 risk group improved after intensive therapy. The estimated 4-year survival rate in the intermediate 2 risk group was 40% for patients treated with intensive therapy versus 11% for patients receiving supportive care only. Survival in the high-risk group of the IPSS was rather disappointing. After intensive antileukaemic therapy, the 4-year survival rate of the high-risk patients was 19%, whereas none of the patients receiving supportive care only achieved prolonged survival. The literature contains only a few reports on the value of the IPSS in intensively treated patients. Appelbaum & Anderson (1998) applied the IPSS to 251 patients who underwent allogeneic bone marrow transplantation. They concluded that the IPSS was useful in predicting relapse and disease-free survival after allogeneic transplantation. The reported 5-year disease-free survival rates were 60% for low-risk and intermediate 1 risk patients, 36% for intermediate 2 risk patients and 28% for high-risk patients. In our subanalysis of 46 patients who received allogeneic stem cell transplantation, the estimated 4-year survival rates were 50% for the intermediate 1 risk patients, 67% for the intermediate 2 risk patients and 30% for the high-risk patients.

Cytogenetic abnormalities occur in about 50% of MDS patients (Yunis et al, 1988; Morel et al, 1993; Vallespi et al, 1998; Balduini et al, 1999). Two large studies demonstrated that karyotype has independent prognostic value (Morel et al, 1993; Toyama et al, 1993). Estey et al (1997b) reported that cytogenetic risk groups were highly predictive for outcome after intensive chemotherapy. A study from Vancouver (Nevill et al, 1998) showed that IPSS cytogenetic risk groups have impact on survival after allogeneic bone marrow transplantation. The Spanish group (Sole et al, 2000) confirmed the prognostic value of the IPSS in 640 patients. Their main criticism was that the intermediate cytogenetic prognostic risk group is a mixture of single- and double-chromosome abnormalities and that some single abnormalities might well prove to be of good or poor prognosis when a larger number of cases are analysed.

Although the Nijmegen patients showed worse prognostic features at diagnosis than the IPSS workshop patients, we conclude that the IPSS is an effective scoring system for MDS patients treated with supportive care only. However, the scoring system did not seem to be the best method for predicting outcome after intensive antileukaemic treatment. In particular, the intermediate 2 risk patients seemed to benefit from intensive antileukaemic treatment. Use of the IPSS is recommended for planning treatment for an individual patient. A scoring system based on the number of cytopenias and cytogenetic risk group might predict outcome after intensive therapy better than a scoring system that includes the percentage of bone marrow blasts.

Acknowledgments

We like to thank Dr Eli Estey for the critical reading of the manuscript. This work has been supported by a stipendium from the Jaap Steenbergen Foundation.

Ancillary