Dr Francesc SoléLaboratori de Citologia Hematològica, Laboratori de Referència de Catalunya, Unitat d'Hematologia 1973, Hospital del Mar, Passeig Marítim 25–29, 08003 Barcelona, Spain. e-mail: E0037@imas.imim.es
Recently, a consensus International Prognostic Scoring System (IPSS) for predicting outcome and planning therapy in the myelodysplastic syndromes (MDS) has been developed. However, the intermediate-risk cytogenetic subgroup defined by the IPSS includes a miscellaneous number of different single abnormalities for which real prognosis at present is uncertain. The main aims of this study were to evaluate in an independent series the prognostic value of the IPSS and to identify chromosomal abnormalities with a previously unrecognized good or poor prognosis in 640 patients. In univariate analyses, cases with single 1q abnormalities experienced poor survival, whereas those with trisomy 8 had a higher risk of acute leukaemic transformation than the remaining patients (P = 0.004 and P = 0.009 respectively). Patients with single del(12p) had a similar survival to patients with a normal karyotype and showed some trend for a better survival than other cases belonging to the IPSS intermediate-risk cytogenetic subgroup (P = 0.045). Multivariate analyses demonstrated that IPSS cytogenetic prognostic subgroup, proportion of bone marrow blasts and haemoglobin level were the main prognostic factors for survival, and the first two characteristics and platelet count were the best predictors of acute leukaemic transformation risk. A large international co-operative study should be carried out to clarify these findings.
The myelodysplastic syndromes (MDS) constitute a heterogeneous group of haematological disorders characterized by peripheral blood cytopenia(s) in the presence of hypercellular bone marrow with features of ineffective haematopoiesis. MDS have been associated with a high risk of progression to acute myeloid leukaemia (AML) and an overall short survival, death being generally due to the consequences of cytopenias or to progression to AML ( Bennett et al, 1985 ). MDS are classified by the French/American/British (FAB) group, based on the percentage of bone marrow and peripheral blood blasts, the percentage of bone marrow ringed sideroblasts and the level of circulating monocytes, into five subtypes: refractory anaemia (RA); refractory anaemia with ringed sideroblasts (RARS); refractory anaemia with excess of blasts (RAEB); refractory anaemia with excess of blasts in transformation (RAEB-t); and chronic myelomonocytic leukaemia (CMML).
Several studies have demonstrated that the main prognostic factors in MDS are the percentage of bone marrow blasts, chromosomal abnormalities, the number and degree of cytopenias and, to a lesser extent, age ( Greenberg et al, 1997 ). As a result of these studies of prognostic factors, different scoring systems for predicting outcome in individual patients with MDS have been developed ( Mufti et al, 1985 ; Sanz et al, 1989 ; Morel et al, 1993 ). Recently, an international consensus score (International Prognostic Scoring System; IPSS), which has refined the prognostic subgroups defined by cytogenetics, has been published ( Greenberg et al, 1997 ). This scoring system is based on the percentage of bone marrow blasts (< 5%, 5–10%, 11–20% and 21–30%), karyotype [good, normal, −Y, del(5q), del(20q); poor, complex (≥ 3 abnormalities) or chromosome 7 abnormalities; intermediate, other abnormalities], and the degree of cytopenias (0/1 or 2/3). This score is gaining wide acceptance, but its prognostic value in independent series is disputed ( Estey et al, 1997 ).
The main goals of this study were to evaluate by multivariate analysis the prognostic value of the IPSS in an independent and large series of MDS patients and to try to identify chromosomal abnormalities with a previously unrecognized good or poor prognosis.
PATIENTS AND METHODS
The subjects in this retrospective study were 640 consecutive patients who were referred to hospitals belonging to the Spanish Cytogenetic Group between January 1984 and December 1996. Only patients that fulfilled FAB criteria for MDS were included ( Bennett et al, 1985 ). Patients with an ambiguous diagnosis of MDS, those who had previously received chemo/radiotherapy, and those with MDS secondary to a previous malignancy were excluded from analysis. 1 Table I provides a summary of the main clinical data of the patients. Among the 41 patients who were intensively treated, 35 received standard antileukaemic chemotherapy and six received an allogeneic bone marrow transplantation.
Table 1. Table I. Demographic and haematological characteristics and prognostic score of the patients. Hb, haemoglobin level; WBC, white blood cell count; ANC, absolute neutrophil count; Plts, platelet count; BM, bone marrow; IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ); Int-1, intermediate 1; Int-2, intermediate 2; SPS, Spanish Scoring System ( Sanz et al, 1989 ).* Excluding patients intensively treated and CMML patients with WBC counts > 12 × 109/l.
Investigators from the participating institutions completed a standard registration form for each patient detailing the patient's laboratory, clinical and cytogenetic features at presentation and clinical outcome (survival time from diagnosis and time until AML evolution).
Cytogenetic analysis of bone marrow samples was performed at the individual centres. The results were reviewed and collated centrally by the main investigator. Whenever possible, at least 20 metaphases were analysed and ten of them karyotyped. Chromosome identification and karyotype description followed the International System for Chromosome Nomenclature ( ISCN, 1995). A complex karyotype was considered when three or more cytogenetic abnormalities were found. When two or more clones were noted in a patient with two aberrations, the patient was categorized in the complex aberration group, whereas patients with two karyotypically independent clones with a single change were categorized in the two-aberration group.
Cytogenetic and some clinical information concerning 112 patients who were referred from Hospital Central l'Aliança (Barcelona), from Hospital Vall d'Hebron (Barcelona) and from Hospital Universitario La Fe (Valencia) has been previously reported elsewhere ( Soléet al, 1992 , 1998; Greenberg et al, 1997 ).
Analysis of prognostic factors for survival and progression to acute leukaemia
The following characteristics were analysed for establishing their association with survival or acute leukaemic evolution: age, sex, haemoglobin, platelet count, white blood cell (WBC) count, absolute neutrophil count, percentage of blasts in bone marrow (BM) and karyotype. The FAB classification ( Bennett et al, 1985 ), Spanish prognostic score (SPS) ( Sanz et al, 1989 ) and IPSS ( Greenberg et al, 1997 ) were also evaluated.
Chi-squared tests were used to compare proportions. The Kaplan–Meier product limit method was used to estimate the actuarial probability of survival and cumulative risk of leukaemic transformation ( Kaplan & Meier, 1958). Survival was measured from diagnosis to death or last follow-up. All deaths, whether related or not to MDS, were considered as the end point of the follow-up interval. Acute leukaemic transformation was measured from diagnosis to development of acute leukaemia. Patients dying from any cause before developing acute leukaemia were considered as censored data in the date of death for the calculation of risk of leukaemic transformation. Statistical comparisons between different actuarial curves were based on log rank tests ( Mantel, 1966) or, if applicable, the test for trend, as recommended by Peto et al (1977 ). Further multivariate analysis by means of the proportional hazards regression method developed by Cox was used in order to identify the most significant independent prognostic factors related to survival. In the first phase, all prognostic variables, with the exception of FAB, SPS ( Sanz et al, 1989 ) and IPSS ( Greenberg et al, 1997 ), were included in the multivariate regression procedure. In the second phase, those characteristics entering the regression were included along with SPS and IPSS in a new multivariate regression to evaluate whether they added important prognostic information to that already afforded by the scoring systems. For multivariate analyses, only cases with complete data were included. Cases of CMML with > 12 × 109 WBC/l and/or treated with intensive chemotherapy or bone marrow transplantation were excluded from those multivariate analyses in which IPSS was required. Given the multiple comparison problems, the selected P-value for considering differences as statistically significant in all analyses was 0.01. All tests were two-sided. All analyses were performed by using programs from the BMDP statistical data package ( Dixon, 1983).
Characteristics of the patients
The group included 364 males (57%) and 276 females (43%). The median age was 66 years (range 1–94). Twenty patients were under 15 years of age. Of the 640 patients, 184 (29%) were classified as RA, 76 (12%) as RARS, 208 (32%) as RAEB, 79 (12%) as RAEB-t and 93 (15%) as CMML ( 1 Table I). The analytical data concerning haemoglobin, WBC count, platelet count, absolute neutrophil count, bone marrow blasts percentage and the IPSS and SPS scoring systems in the five different subtypes of MDS are shown in 1 Table I. One hundred and fifty-nine patients (25%) developed acute leukaemia during follow-up. The cumulative risk of acute leukaemic transformation was 20% and 38% at 1 and 5 years respectively. At the time of analysis, 320 patients were still alive and 320 had died. Actuarial median survival of the 640 MDS patients was 2.5 years.
Among the 640 patients, 327 (51%) showed clonal karyotypic abnormalities at diagnosis. The frequency of the different chromosomal abnormalities and their relationship to the FAB classification is shown in 2 Table II. As can be appreciated, the most common abnormalities found were del(5q) (78 cases), monosomy 5 (15), monosomy 7/del(7q) (45/24), trisomy 8 (56), involvement of 11q (16), del(12p) (27), involvement of 12q (10), involvement of 13q (11), i(17q) (23), del(20q) (21) and −Y (12). Cytogenetic abnormalities found as a single anomaly were del(5q) in 32 cases, monosomy 7/del(7q) in 24, trisomy 8 in 31, involvement of 11q in six, del(12p) in 13, 13q involvement in four, i(17q) in four, involvement of 1q in four, trisomy 19 in three, del(20q) in eight and loss of Y chromosome in eight.
Table 2. Table II. Chromosomal abnormalities according to FAB subtype. Excluding patients intensively treated and CMML patients with WBC counts > 12 × 109/l. IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ). Good, normal, −Y, del (5q), del (20q); intermediate, +8, other single or double abnormalities; poor, complex (more than three abnormalities) or chromosome 7 abnormalities.
One hundred and ninety patients presented a single chromosome abnormality (FAB subtypes and cytogenetic abnormalities found as a single abnormality other than del(5q), −7/del(7q), +8, del(12p), i(17q), del(20q) and loss of X/Y chromosome are detailed in 3 Table III) and 49 showed two chromosomal defects. In 88 patients, a complex karyotype was observed. In five of these patients, two unrelated clones were observed. In four of these five cases, one of the clones bore a 5q anomaly and the other trisomy 8; in the remaining case, monosomy 7 and trisomy 8 were found in different clones.
Table 3. Table III. FAB subtypes and cytogenetic abnormalities found as a single abnormality other than del(5q), −7/del(7q), +8, 12p, i(17q), del(20q) and loss of X/Y chromosome.
Chromosomal abnormalities and FAB classification
As shown in Table II, 80 patients with RA (43%), 25 with RARS (33%), 128 with RAEB (61%), 55 with RAEB-t (70%) and 39 with CMML (42%) showed a clonal abnormality. In RA, the more frequent cytogenetic abnormalities were del(5q), trisomy 8, del(20q) and −Y; in RARS, the most common were del(5q), trisomy 8 and del(20q); in RAEB, the most common were del(5q), monosomy 7/del(7q), trisomy 8 and del(12p); in RAEB-t, the most common were del(5q), monosomy 7 and trisomy 8; and in CMML the predominant abnormalities were trisomy 8 and del(12p).
In 28 patients, a new cytogenetic analysis was performed after transformation into acute leukaemia. Five of them had acquired a new chromosomal abnormality.
Prognostic factors for survival and leukaemic transformation
The results of the univariate analysis of prognostic factors for survival and risk of leukaemic evolution are summarised in 4 Tables IV and 5 V. Haemoglobin, WBC count, platelet count, number of cytopenias, proportion of bone marrow blasts, FAB subtype and IPSS and SPS scores showed a close association with survival and leukaemic transformation risk. Gender showed a strong association with survival but did not influence leukaemic evolution.
Table 4. Table IV. Clinical variables related to survival and AML evolution. Hb, haemoglobin level; WBC, white blood cell count; ANC; absolute neutrophil count; Plts, platelet count; BM, bone marrow; IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ); Int-1, intermediate 1; Int-2, intermediate 2; SPS, Spanish Scoring System ( Sanz et al, 1989 ).* Excluding patients intensively treated and CMML patients with WBC counts >12 × 109/l.
Table 5. Table V. Cytogenetics variables related to survival and AML evolution. IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ); NR, not reached.* Excluding patients intensively treated and CMML patients with WBC counts >12 × 109/l. Good, normal, −Y, del(5q), del(20q); intermediate, +8, other single or double abnormalities; poor, complex (more than three abnormalities) or chromosome 7 abnormalities.
Cytogenetic findings had a clear impact on survival and risk of leukaemic transformation. Patients with an abnormal karyotype had shorter survival and higher leukaemic transformation risk than those with normal karyotype. The number of chromosomal abnormalities also had a great impact on outcome. The group of patients with a complex karyotype had the worst prognosis, those with no abnormalities had the best prognosis, whereas single or double abnormalities experienced an intermediate outcome. Patients with double abnormalities experienced a somewhat shorter survival and higher acute leukaemic transformation risk than patients with single abnormalities, although we were not able to demonstrate significant differences among them. The cytogenetic prognostic subgroups defined by the IPSS strongly influenced survival and leukaemic transformation risk. Survival and cumulative risk of acute leukaemic transformation curves according to the number of chromosomal aberrations and the IPSS cytogenetic prognostic subgroups are plotted in Figs 1 and 2 respectively. Cases with del(5q), del(11q), del(12p), del(20q) and −Y experienced a somewhat longer survival than the overall series of cases, whereas involvement of 1q, del(3q), monosomy 7/del(7q), trisomy 8, del(13q) and −X showed a somewhat shorter survival (Fig 3). However, none of the chromosomal abnormalities found as single abnormalities, including those universally accepted as portraying good or poor prognosis, showed a statistically significant association with outcome when compared with the remaining cases without that single anomaly, except for monosomy 7/del(7q) (P = 0.01) and involvement of 1q (P = 0.004) and survival and for trisomy 8 (P = 0.009) and acute leukaemic evolution. We tried to ascertain whether any of the chromosomal abnormalities occurring as sole aberrations and without a previously recognized particular prognosis could be segregated from the miscellaneous intermediate-risk IPSS cytogenetic prognostic subgroup. For this purpose, outcome regarding survival and acute leukaemic evolution risk for each specific abnormality was compared with the remaining set of cases within this risk category. None of the specific chromosomal abnormalities demonstrated clear differences in outcome. However, there was some trend for longer survival for the 13 cases with del(12p) (P = 0.045) and shorter survival for the four cases showing involvement of 1q (P = 0.013) compared with the remaining cases in the IPSS intermediate-risk cytogenetic subgroup. Interestingly, eight of the 13 cases with del(12p) had RAEB, three of the 13 had CMML and only two of the 13 had RA, which suggests that the longer median survival of cases with del(12p) was not due to an association with a favourable FAB subtype.
6 Table VI shows the results of the multivariate prognostic analysis for survival in which all demographic and haematological characteristics and cytogenetic information, coded in multiple ways, were entered in the regression modelling procedure. The IPSS cytogenetic prognostic subgroups, haemoglobin level and proportion of bone marrow blasts showed a strong and independent relationship with survival length. When the IPSS and SPS scores were introduced into the regression procedure, the IPSS score was the main prognostic indicator, and none of the remaining characteristics added significant prognostic information.
Table 6. Table VI. Multivariate analysis of survival.* IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ).* Excluding patients intensively treated and CMML patients with WBC counts >12 × 109/l.† Good, normal, −Y, del(5q), del(20q); intermediate, +8, other single or double abnormalities; poor, complex (more than three abnormalities) or chromosome 7 abnormalities.
Table 7. Table VII. Multivariate analysis of leukaemic transformation.* IPSS, International Prognostic Scoring System ( Greenberg et al, 1997 ).* Excluding patients intensively treated and CMML patients with WBC counts >12 × 109/l.† Good, normal, −Y, del(5q), del(20q); intermediate, +8, other single or double abnormalities; poor, complex (more than three abnormalities) or chromosome 7 abnormalities.
As in other studies, the most common cytogenetic abnormalities found in this series included del(5q), monosomy 7/del(7q) and trisomy 8, del(5q) being the most frequent abnormality. Other chromosomal abnormalities frequently found included del(11q), del(12p), involvement of 12q, involvement of 13q, isochromosome 17q, del(20q), trisomy 21, monosomy 21 and loss of sex chromosomes. In MDS, chromosome loss accounts for about half of the chromosomal abnormalities. Among partial chromosome losses, del(5q) is the most common, followed by del(20q), del(11q) and del(7q) ( Third MIC Co-operative Study Group, 1988; Fenaux et al, 1996 Vallespíet al, 1998 ). In this series, the most frequent losses were del(5q), del(12p), del(7q) and del(20q).
None of the cytogenetic abnormalities found in patients with MDS was specific to an FAB subtype. As in other series, certain chromosomal abnormalities showed some degree of association with the FAB subtype ( Yunis et al, 1988 ; Suciu et al, 1990 ; Toyama et al, 1993 ; Parlier et al, 1994 ). Monosomy 7/del(7q) and complex karyotypes were more frequently found in patients with excess of blasts (RAEB and RAEB-t). Deletion 5q as a sole anomaly, trisomy 8, del(20q) and loss of Y chromosome had a higher incidence in patients with RA. In patients with RARS, there was a noticeably high incidence of del(5q), trisomy 8 and del(20q). In CMML, the most frequent aberrations were trisomy 8, del(12p), monosomy 7/del(7q) and -Y.
One of the major aims of the present study was to evaluate in an independent and large series and by means of multivariate analysis the prognostic accuracy of the recently developed IPSS ( Greenberg et al, 1997 ). Our results clearly demonstrate that the IPSS is the most powerful prognostic indicator in MDS patients, both for survival and for risk of leukaemic evolution. In fact, when the IPSS was included in the multivariate analyses, no other characteristic added significant prognostic information.
One possible minor pitfall of the IPSS is the inclusion in the intermediate cytogenetic prognostic subgroup of a miscellaneous number of single chromosomal abnormalities and double abnormalities. Some of the single chromosomal abnormalities might well prove to be of good or poor prognosis when a large number of cases are properly analysed. In the present study, there was some suggestion that certain chromosomal abnormalities could be segregated from the IPSS intermediate-risk cytogenetic subgroup. Patients with single del(12p) experienced a similar survival to patients with a normal karyotype. In fact, single del(12p) showed in univariate analysis some trend for a better prognosis than other cases belonging to the IPSS intermediate-risk cytogenetic subgroup (P = 0.045). Rearrangements of 12p have been reported in about 10% of patients with CMML and in about 5% of patients with RAEB and RAEB-t, usually as deletions at 12p11–p13 ( Wilmoth et al, 1985 ; Berger et al, 1986 ). Interestingly, in this series, eight of the 13 patients with deletion 12p as a sole abnormality had RAEB, three of the patients had CMML, and only two of the patients had RA. These data suggest that the better survival of cases with del(12p) was not conditioned by their association with a good prognosis FAB subtype. In spite of the low number of patients with single involvement of 1q (only four patients), this group had a significantly poorer outcome when compared with the overall series of patients in univariate analysis (P = 0.004), and there was some trend for a statistically significant poorer survival than that observed in the remaining cases of the IPSS intermediate-risk cytogenetic subgroup (P = 0.013). Obviously, these results should be interpreted with caution. The possible prognostic impact of these single abnormalities, del(12p) and 1q involvement has not been previously recognized. In the present series, patients with single trisomy 8 experienced a significantly increased risk of leukaemic transformation (P = 0.009) and a shorter, but not statistically significant, survival than the overall series of MDS patients. This finding indicates that single trisomy 8 might be segregated from the intermediate-risk IPSS cytogenetic category. Obviously, all of these observations require confirmation in future in larger studies before being accepted and used in clinical practice.
To conclude, our results confirm the incidence and relationship with different FAB subtypes of distinct structural and numerical chromosomal abnormalities in primary MDS, demonstrate that the IPSS accurately predicts survival and leukaemic transformation risk in the individual MDS patient, and suggest that some specific chromosomal abnormalities, such as involvement of 1q, del(12p) and trisomy 8, may be segregated from the IPSS intermediate-risk cytogenetic prognostic subgroup. A very large co-operative study should be undertaken to clarify these issues.
The active members of the group for this study were:
Laboratori de Citologia Hematològica, Laboratori de Referència de Catalunya, Hospital del Mar/Hospital de l'Esperança: Dr Francesc Solé, Dra. Blanca Espinet, Dra. Lourdes Florensa, Dra. Elisabet Lloveras, Dra. Soledad Woessner.
Departamento de Anatomía Patológica, Facultad de Medicina, La Laguna, Tenerife: Dr Juan Cruz Cigudosa.
Departamento de Genética, Facultad de Ciencias, Universidad de Navarra, Pamplona: Dra. Ma José Calasanz.
Servicio de Hematología, Hospital Central de Asturias. Oviedo: Dra. Elisa Lun~o.
Servicio de Genética, Hospital Universitario Puerta del Mar, Cádiz: Dr Daniel Armenta-Gil.
Servicio de Hematología, Hospital General Universitario, Valencia: Dr Felix Carbonell.
Servicio de Hematología, Hospital la Fe, Valencia: Dr Guillermo F. Sanz, Dr José Cervera.
Servicio de Genética, Hospital la Fe, Valencia: Dr Félix Prieto, Dra. Lourdes Badia.
Servicio de Hematología, Hospital Clínico Universitario, Valencia: Dr José Angel Martínez-Climent, Dr Javier García-Conde.
Servicio de Hematología, Hospital Universitario de Salamanca: Dr Jesús María Hernández, Dr Juan Luis García, Dr Agustín Ríos, Dr Jesús F San Miguel.
Unidad de Transplante, Hospital General Universitario Gregorio Maran~ón, Madrid: Dr José Luís Díez-Martín, Natividad Polo.
Servicio de Genética. Fundación Jimenez-Diaz, Madrid: Dr Javier Benítez, Dra. Eva Arranz.
Servicio de Genética, Hospital Doce de Octubre, Madrid: Dra. Luisa Marques.
Servicio de Hematología y Hemoterapia, Hospital Universitario Germans Trias i Pujol, Badalona (Barcelona): Dra. Isabel Granada, Dra. Fuensanta Millà, Dr Josep Ma Ribera.