Leukemia (1998) 12, 482–485  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu Refractory with severe dysplasia: clinical significance of morphological features in refractory anemia A Matsuda1, I Jinnai1, F Yagasaki1, S Kusumoto1, M Minamihisamatsu2, S Honda3, I Murohashi1, M Bessho1 and K Hirashima1

1First Department of Internal , Saitama , Saitama; 2Division of Radiobiology and Biodosimetry, National Institute of Radiological Science, Chiba; and 3Department of Radiation Epidemiology, Atomic Disease Institute, Nagasaki University School of Medicine, Nagasaki, Japan

Refractory anemia (RA) in myelodysplastic syndromes (MDS) Patients and methods are very heterogeneous diseases regarding their morphology, clinical features and survival. We proposed the new desig- nations ‘RA with severe dysplasia (RASD)’ and ‘RA with mini- Patients mal dysplasia (RAminiD)’. In our criteria, RASD is considered present if a bone marrow (BM) examination shows Pseudo- A total of 80 consecutive patients with the diagnosis of pri- Pelger–Huet anomalies of mature neutrophils у3% and/or micromegakaryocytes (mMgk) of megakaryocytes у10% in RA mary RA and RA with excess of blasts (RAEB) according to patients. RAminiD is defined as RA cases other than RASD. the FAB criteria were included in the retrospective analysis. After the reclassification of 58 primary RA patients, the group Patients were diagnosed at the Saitama Medical School Hospi- was composed of 45 RAminiD and 13 RASD patients. The blast tal between January 1983 and October 1996. Of the patients, percentage in the BM and the frequency of cytogenetic abnor- 58 were diagnosed as having RA and 22 as having RAEB. malities observed in the RASD patients were intermediate between those in the RAminiD and RAEB patients. The analysis of survival curves revealed differences among the three groups; the RASD patients had lower survival probabilities Morphological analysis than those of the RAminiD group, and significantly higher prob- abilities than those of the RAEB group. (RAminiD vs RASD, P = 0.06; RASD vs RAEB, P = 0.004.) Our data indicate that in Hematological examinations were performed using standard RA patients, RASD is a distinct subset of RA with an unfavor- methods (peripheral blood (PB) and bone marrow (BM) able clinical outcome. Wright–Giemsa stained films). PB and BM differential counts Keywords: myelodysplastic syndromes; refractory anemia; myelo- were performed on 100 and 500 cells, respectively. A mini- dysplasia; survival; Pseudo-Pelger–Huet anomalies; micromega- karyocyte mum of 25 megakaryocytes and 200 mature neutrophils were examined in each patient. The evaluation of dysplasia has been used for Rosati’s criteria: , megalo- Introduction blastoid change and/or nuclear budding and/or multiple nuclei у3% or karyorrhexis у5%; granulocytic dysplasia, cytoplasmic hypogranularity and/or hypo- or hypersegmented Myelodysplastic syndromes (MDS) are acquired clonal stem nuclei у3%; megakaryocytic dysplasia, micromegakaryocytes cell disorders characterized by ineffective hematopoiesis with myelodysplasia1 and are associated with a high risk of pro- or multiple, widely separated or single, non-lobated nuclei у 5 gression to acute leukemias.2 The diagnosis of MDS requires and a fully mature cytoplasm 10%. In addition, two careful light microscopic examination of the peripheral blood especially distinct dysplastic changes were chosen (Pseudo- Pelger–Huet anomalies and mMgk), and their morphological smear, bone marrow aspirate smear and biopsy. abnormalities could be evaluated. In this study, we propose MDS are very heterogeneous diseases regarding their mor- phology, clinical features and survival.3 Refractory anemia the new designations ‘RA with severe dysplasia (RASD)’ and (RA) and RA with ringed sideroblasts (RARS) are generally ‘RA with minimal dysplasia (RAminiD)’. In our criteria, RASD classified as low-risk groups in MDS.4 Rosati et al5 recently was considered present if the BM examination showed Pseudo-Pelger–Huet anomalies of mature neutrophils у3% proposed the entity of refractory cytopenia with multilineage and/or mMgk of megakaryocytes у10% in RA patients. RAmi- dysplasia (RCMD), and reported that the outcome of the RCMD patients was unfavorable. Kuriyama et al6 reported that niD was defined as RA other than RASD. Each case of RA was Pseudo-Pelger–Huet anomalies and micromegakaryocytes reviewed and reclassified according to our criteria and Rosa- (mMgk) are specific in MDS. In the present study, we chose ti’s criteria (RCMD: RA showing trilineage dysplasia in the BM examination; Rosati’s RA: RA showing minimal or absent dys- two distinct dysplastic changes (Pseudo-Pelger–Huet anomal- granulopoiesis and dysmegakaryopoiesis in the BM ies and mMgk), and propose the new designations ‘RA with 5 severe dysplasia (RASD)’ and ‘RA with minimal dysplasia examination). The morphological analysis was separately (RAminiD)’. We investigated the clinical significance of RASD evaluated by two hematologists (AM and IJ). and RAminiD.

Cytogenetic analysis

Cytogenetic analyses were performed with a trypsin-Giemsa banding technique on BM cells from aspirates. Cells from Correspondence: A Matsuda, First Department of , Ͻ Saitama Medical School, 38 Morohongo, Moroyama, Iruma-gun, Sait- short-term unstimulated cultures ( 48 h) were examined. ama, 350-04, Japan; Fax: 81 492 95 8025 Chromosomal abnormalities were identified by the Inter- Received 20 April 1997; accepted 2 December 1997 national System for Human Cytogenetic Nomenclature.7 Myelodysplasia in RA A Matsuda et al 483 Table 1 Clinical and hematological findings in RAminiD, RASD and RAEB patients

Group No. Age Sex WBC ANC Hb (g/dl) PLT BM Disease of (×109/l) (×109/l) (×109/l) progression patients E blast (%) Blast (%) to acute leukemia

RAminiD 45 54.2 ± 18.8 23M 22F 3.36 ± 1.64 1.77 ± 1.27 8.4 ± 2.9 56.6 ± 65.9 39.8 ± 14.8 1.32 ± 0.96 *1   RASD 13 62.5 ± 16.4 7M 6F 3.45 ± 1.54 1.76 ± 1.44 8.3 ± 3.3 63.7 ± 34.0 39.3 ± 16.6  ** 2.77 ± 1.59 2  RAEB 22 65.5 ± 14.3 13M 9F 3.76 ± 2.25 1.70 ± 1.53 7.8 ± 1.7 182.7 ± 289.826.0 ± 11.6 9.15 ± 4.38 11

*P = 0.006; **P = 0.09. RASD, refractory anemia with severe dysplasia; RAminiD, refractory anemia with minimal dysplasia; RAEB, refractory anemia with excess of blasts; M, male; F, female; WBC, white blood cells; ANC, absolute neutrophil count; Hb, hemoglobin; PLT, platelets; E blast, erythyroblast.

Statistical analysis Table 2 Cytogenetic findings in RAminiD, RASD and RAEB patients The Kaplan–Meier method8 was used to estimate the prob- 10 ability of overall survival and leukemic transformation. Sur- Group No. of Chromosomal abnormalities (%) studied vival was measured from the date of diagnosis at our hospital patients or affiliated hospitals until death due to any cause or until the Type A Type B Type C Type A/B/C last patient contact, as of October 1997. Leukemic transform- ation was measured from diagnosis. The analyses of survival RAminiD 34 5 (15%) 1 (3%) 1 (3%) 7 (21%) RASD 11 1 (9%) 1 (9%) 2 (18%) 4 (36%) curve and leukemic transformation curve were done using the RAEB 12 4 (33%) 0 (0%) 5 (42%) 9 (75%) log rank test.9 Statistical comparisons were performed by use of the nonparametric Mann–Whitney test for continuous data. Type A: single chromosome abnormalities other than −7/7q−, +8. Type B: single chromosomal abnormalities of −7/7q−, +8, and two aberrations. Results Type C: complex aberrations (three or more aberrations). RAminiD: refractory anemia with minimal dysplasia, RASD: refrac- The case of each RA patient was reviewed according to Rosa- tory anemia with severe dysplasia, RAEB: refractory anemia with excess of blasts. ti’s criteria and our criteria. After reclassification, the RA group was composed of 45 RAminiD and 13 RASD patients, and 26 Rosati’s RA and 21 RCMD patients. All RA patients could be leukemia. Although this patient was initially classified as hav- classified according to our criteria. However, in 10 of the RA ing RAminiD because her frequency of mMgk was slightly patients, the reclassification according to Rosati’s criteria is lower than10%, we thought that her morphological findings confusing because they have only two-lineage dysplasia, and were similar to those of RASD. She subsequently developed one 5q− syndrome case was also excluded according to Rosa- RASD and thereafter acute leukemia. ti’s criteria. High concordance rate (94.8%) between the two Concerning the causes of death in our patients, the fre- observers and high reproducibility (98.3%) by observer (AM) quency of non-hematological death in the RAminiD group were found regarding our classification. All RAEB patients had was higher than that in the RASD and RAEB groups. The trilineage dysplasia. causes of death in the RASD group were similar to those in The analysis of the age, sex and hematological findings the RAEB group (Table 3). revealed that the RASD patients had higher blast percentage Follow-up periods ranged from 2 to 203 months (median of BM cells than that of the RAminiD patients (P = 0.006), and 31 months). Forty-seven of the 80 patients were still alive 2+ higher erythroblast percentage of BM cells than that of the to 203+ months. These included 38/45 RAminiD patients, RAEB patients (P = 0.09) (Table 1). 7/13 RASD patients, and 2/22 with RAEB. The analysis of sur- In the RAminiD group, the frequency of cytogenetic abnor- vival curves revealed differences among the three groups malities was 7/34 (21%). In contrast, cytogenetic abnormali- (RAminiD vs RASD: P = 0.06, RASD vs RAEB: p = 0.004) ties were found in four (36%) of the 11 RASD patients studied (Figure 2). Patients with RCMD had prognoses intermediate and nine (75%) of the 12 RAEB patients. The frequency of between those of the Rosati’s RA and RAEB groups (Rosati’s cytogenetic abnormalities observed in the RASD group were RA vs RCMD: P = 0.09, RCMD vs RAEB: P = 0.0001) intermediate between those in the RAminiD and RAEB groups. (Figure 3). The types of cytogenetic abnormalities are summarized in Table 2. Complex aberrations (type C in Japanese Score (JS)10 were present in only one RAminiD patient (3%) and two Discussion RASD patients (18%). Type A abnormality in JS was present in only one RASD patient (9%) and five RAminiD patients (15%). Gattermann et al11 suggested that in the setting of RARS, The frequency of disease progression to acute leukemia patients with dysplastic granulocytic and/or megakaryocytic among the patients with RASD was lower than that among the elements were at higher risk of developing acute leukemia patients with RAEB (Table 1). The leukemia-free survival (LFS) than were those who lacked these findings. Rosati et al5 indi- time calculation revealed differences among the three groups cated that RCMD is a distinct subset of MDS, with an unfavor- (RAminiD vs RASD: P = 0.03, RASD vs RAEB: P = 0.04) able clinical outcome. We also believe that the degree of mye- (Figure 1). Only one female RAminiD patient developed acute lodysplasia reflects the prognosis in MDS patients including Myelodysplasia in RA A Matsuda et al 484

Figure 1 Leukemia-free survival in RAminiD, RASD and RAEB Figure 2 Cumulative survival in RAminiD, RASD and RAEB patients. RAminiD vs RASD: P = 0.03; RASD vs RAEB: P = 0.04. patients. RAminiD vs RASD: P = 0.06; RASD vs RAEB: P = 0.004.

Table 3 Causes of death in RAminiD, RASD and RAEB patients

Type Bleeding Infection Acute Others Total leukemia proportional hazards regression multivariate analysis, the independent variables for determining outcome were Pseudo- RAminiD 0 (0%) 3 (43%) 1 (14%) 3 (43%) 7 (100%) Pelger–Huet anomalies (P = 0.009) and Hb (P = 0.04) RASD 1 (17%) 3 (50%) 2 (33%) 0 (0%) 6 (100%) (Table 5). These findings support Aul’s previous results. RAEB 1 (5%) 6 (30%) 11 (55%) 2 (10%) 20 (100%) Our criteria were advantageous compared with Rosati’s cri- teria. Firstly, our criteria were able to identify the patients with RAminiD, refractory anemia with minimal dysplasia; RASD, refrac- RA who were at greater risk for an unfavorable clinical course tory anemia with severe dysplasia; RAEB, refractory anemia with = = excess of blasts. (RAminiD vs RASD: P 0.06, Rosati’s RA vs RCMD: P 0.09). In particular, the prognostic value of our criteria in the first 3 years from the diagnosis (RAminiD vs RASD: P = 0.001) were those with RA. However, individual interpretation of myelody- advantageous compared with Rosati’s criteria (Rosati’s RA vs splasia is variance. In addition, there are low frequencies of RCMD: P = 0.1). We think that these findings are important dysplasia in the BM of patients with liver disease,12 viral infec- for decision-making regarding therapeutic strategies for RA tion13 etc, and myelodysplastic features other than Pseudo- patients. Secondly, Rosati’s criteria are somewhat intricate, Pelger–Huet anomalies, complete degranular neutrophils and while our criteria are simple, more objective, and more spe- mMgk are not able to distinguish MDS from normal subjects.6 cific for dysplasia of MDS. Aul et al14 reported that Pseudo-Pelger–Huet anomalies and In our study, the BM blast percentage, the frequency of mMgk were correlated with the length of survival. Thus, we cytogenetic abnormalities, the length of the LFS and the prog- chose these two distinct dysplastic changes (Pseudo-Pelger– nosis of the patients with RASD were intermediate between Huet anomalies and mMgk) in part because they have spe- those of the RAEB and RAminiD patients. Chromosomal cific, objective and prognostic values. In the present patients, abnormalities associated with good prognosis predominated this univariate analysis using the Kaplan–Meier method and in the RAminiD group, whereas others associated with poor log rank test showed that Pseudo-Pelger–Huet anomalies prognosis predominated in the RASD group. We believe that (P = 0.0001), mMgk (P = 0.07), Hb (P = 0.04) and BM blast RASD is a distinct subset of RA with an unfavorable clinical (P = 0.01) were predictors of poor survival (Table 4). Using outcome. Myelodysplasia in RA A Matsuda et al 485 Table 5 Multivariate analysis of survival in RA patients

Variable P value

Pelger у3% 0.009 Hb р6 g/dl 0.04 mMgk у10% Ͼ0.1 PLT р50 × 109/l Ͼ0.1 ANC р1 × 109/l Ͼ0.1 BM blast Ͼ3% Ͼ0.1

Abbreviations are explained in Table 4.

Acknowledgements

We would like to thank Dr K Fuchigami, Dr T Matsuo and Prof M Tomonaga, Department of , Atomic Dis- ease Institute, Nagasaki University School of Medicine for their valuable advice.

References

1 Goasguen JE, Bennett JM. Classification and morphologic features of the myelodysplastic syndromes. Semin Oncol 1992; 19: 4–13. 2 Ganser A, Hoelzer D. Clinical course of myelodysplastic syn- dromes. Hematol Oncol Clin N Am 1992; 6: 607–617. 3 Koeffler HP. Introduction: myelodysplastic syndromes. Semin Hematol 1996; 33: 87–94. 4 Hofmann WK, Ottmann OG, Ganser A, Hoelzer D. Myelodysplas- tic syndromes: clinical features. Semin Hematol 1996; 33: 177– 185. 5 Rosati S, Mick R, Xu F, Stonys E, Le Beau MM, Larson R, Vardiman JW. Refractory cytopenia with multilineage dysplasia: further characterization of an ‘unclassifiable’ . Leukemia 1996; 10: 20–26. Figure 3 Cumulative survival in Rosati’s RA, RCMD and RAEB 6 Kuriyama K, Tomonaga M, Matsuo T, Ginnai I, Ichimaru M. Diag- nostic significance of detecting pseudo-Pelger–Huet anomalies patients. Rosati’s RA vs RCMD: P = 0.09, RCMD vs RAEB: P = 0.0001. and micromegakaryocytes in myelodysplastic syndrome. Br J Haematol 1986; 63: 665–669. 7 ISCN (1991). Mitelman F (ed). Guideline for Cancer Cytogenetics Nomenclature. Karger: Basel, 1991. Table 4 Univariate analysis of survival in RA patients 8 Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–481. Variable No. patients P value 9 Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966; 50: 163–170. Pelger у3% 6 0.0001 10 Toyama K, Ohyashiki K, Yoshida Y, Abe T, Asano S, Hirai H, Ͻ3% 52 Hirashima K, Hotta T, Kuramoto A, Kuriya S, Miyazaki T, Kakishita Hb р6 g/dl 16 0.04 E, Mizoguchi H, Okada M, Shirakawa S, Takaku F, Tomonaga Ͼ6 g/dl 42 M, Uchino H, Yasunaga K, Nomura T. Clinical implications of mMgk у10% 12 0.07 chromosomal abnormalities in 401 patients with myelodysplastic Ͻ10% 46 syndromes: a multicentric study in Japan. Leukemia 1993; 7: PLT р50 × 109/l 25 Ͼ0.1 499–508. Ͼ50 × 109/l 33 11 Gattermann N, Aul C, Schneider W. Two types of acquired sid- ANC р1 × 109/l 17 Ͼ0.1 eroblastic anemia (AISA). Br J Haematol 1990; 74: 45–52. Ͼ1 × 109/l 41 12 Hadnagy C, Laszlo GA. Acquired dyserythropoiesis in liver dis- BM blast Ͼ3% 11 0.01 ease. Br J Haematol 1991; 78: 283. р3% 47 13 Karcher DS, Frost AR. The bone marrow in human immunodefi- ciency virus (HIV)-related disease. Morphology and clinical corre- Pelger, Pseudo-Pelger–Huet anomalies; Hb, hemoglobin; mMgk, lation. Am J Clin Pathol 1991; 95: 63–71. micromegakaryocytes; PLT, platelets; ANC, absolute neutrophil 14 Aul C, Gatterman N, Germing U, Runde V, Helly A, Schneider count; BM, bone marrow. W. Risk assessment in primary myelodysplastic syndromes: vali- dation of the Dusseldorf score. Leukemia 1994; 8: 1906–1913.