Document e5v1RBYLRNqzpRmZeozDOdYme

314 Bone Marrow Dysplasia in Patients wiu, Newly Diagnosed Acute Myelogenous Leukemia Does Not Correlate with History of Myelodysplasia or with Remission Rate and Survival Karen K. Ballen, M.D.,* D. Gary Gilliland, M.D., Ph.D.,* Leslie A. Kalish, D.Sc.,t and Lawrence N . Shulman, M.D.* Background. The records and initial bone marrow studies of 106 patients with newly diagnosed acute myelogenous leukemia (AML) were analyzed retrospectively to determine whether bone marrow dysplasia was predictive of a previous myelodysplastic disorder or correlated with remission rate and survival. Methods. Bone marrow aspirates and biopsy specimens were reviewed in a blinded fashion; dysplasia was assessed in as objective a manner as possible by numerically scoring nine specific findings: erythrocyte multinuclearity, nuclear fragmentation, megaloblastic characteristics, leukocyte granulation abnormalities and nuclear malformations, Pelger-Huet cells, and megakaryocytic dysplasia (mononuclear megakaryocytes, micromegakaryocytes, and megakaryocytes with multiple distinct nuclei). Results. Dysplasia of the megakaryocytic line was seen in 34% of patients; 70% of the patients had erythrocyte dysplasia; and 68% had leukocyte dysplasia. PelgerHuet cells were seen in bone marrow of 35% of the patients. Overall dysplasia score and specific dysplastic findings such as Pelger-Huet cells did not correlate with a known history of myelodysplasia (P= 0.47),cytogenetic abnormalities (P= 0.35),prior chemotherapy treatment with or without alkylating agents (P = 1.00), previous malignant disorders such as polycythemia vera (P = 1.00),remission rate (P = 0.93),or survival (P= 0.42). Multivariate analysis confirmed known independent Presented in part at the American Society of Hematology meetings, Denver, Colorado, December 1991. From the *Hematology-Oncology Division, Brigham and Women's Hospital, Harvard Medical School, and the tDivision of Biostatistics, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts. Address for reprints: Karen K. Ballen, M.D., Hematology-Oncology Division, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115. Accepted for publication August 16, 1993. risk factors for remission in this patient population. cluding age ( P = 0.04),history of prior chemotherapy(p 0.04),abnormalities in chromosomes 5,7,or 8 (P= om, and type of antileukemia therapy (P< 0.001). Conclusions. Bone marrow dysplasia is commga i patients with AML and does not correlate with a &ta\ of myelodysplasia or predict outcome when patients treated with standard intensive AML therapy. C w , 1994;73:314-21. Key words: acute myelogenous leukemia, myelodyspb sia, bone marrow. Acute myelogenous leukemia (AML)is a heterogenm clonal disorder characterized by early myeloid precm. sors that have largely, but not entirely, lost their abili? to differentiate into mature progeny.'-' If some degm of maturation occurs, dysplastic erythrocytes, le&. cytes, and megakaryocytes may be p r e ~ e n t . ~ Patients with AML who have a previous history ~t myelodysplastic syndrome, a previous hematologicdisorder such as polycythemia vera, a history of therap with alkylating agents, or certain specific chromosomd abnormalities have lower remission and survival rate than do patients with de novo AML.5-9Morpholo@ examination of the bone marrow for dysplasia has b. used to determine whether patients with AML haw pre-existing myelodysplasticsyndrome, and somestud. ies have suggested that dysplasia can predict remission For example, in one study 45% of patients with sevm dysplasia and 73% of patients with minimal dysplsb had disease enter remission.' We reviewed 106 consecutive patients with new]! diagnosed AML to determine whether bone marr0\\ dysplasia correlated with clinical patient charactedbe at diagnosis (history of myelodysplasia, previous h e m Marrow Dysplasia at Diagnosis of AML/Ballen ef al. 315 I W c disorder, previous treatment with alkylating rgents, specific chromosomal abnormalities known to k associated with myelodysplasia) and with attain' ment of complete remission and survival. An objective merical dysplasia score was assigned based on a review of bone marrow aspirates and biopsy Ipecimens. Statistical analyses were performed to ex- pbre the independent contributions of patient and bone marrow characteristics in predicting remission Md survival. Materials and Methods Wedical records of 106 consecutive patients with bone m o w examinations available who were seen at the @&am and Women's Hospital between 1982 and 1989 with newly diagnosed AML were reviewed. Age, rc~. French-American-British (FAB) classification, Lrrryotype, blood counts before and at diagnosis, documented history of myelodysplasia, prior malignant disorder, and prior therapy with chemotherapy agents were recorded. A history of myelodysplasia was noted if the patient had a documented bone marrow diagnosis d myelodysplasia more than 3 months before the diagnoJis of AML. Malignant disorders included solid tu- -, the terminal phase of polycythemia vera and mycloid metaplasia, and diseases associated with therapy- dated AML such as Hodgkin's disease. Treatments, duding chemotherapy regimen, and disease remis&m and patient survival rates were analyzed. Patients Who eventually had a bone marrow transplant were aduded. &mp Marrow Morphology Review h e manow aspirates and biopsy specimens were re*ed on all 106 patients by three authors (K.K.B., D.c-G., L.N.S.) in a blinded fashion, without knowlsdgie of patient name or history. In most patients (79), aspirates and biopsy specimens were reviewed. five patients, only a bone marrow biopsy specimen available. In 22 patients, only the bone marrow 'Jpimte was reviewed. Evaluation was based on bone -OW aspirates and biopsy specimen, when avail@* Some characteristics, such as hypogranulation, 'VWe more easily ascertained from the aspirate, other characteristics, such as percent cellular9 8 m d be obtained only from biopsy specimen. n e percentage cellularity, presence or absence of rods, percentage myeloblasts, mye1oid:erythroid w#and presence or absence of bone marrow fibrosis !. evaluated. Specific dysplastic characteristics were e.f Megakaryocytes were scored individually for the presence of micromegakaryocytes, mononuclear megakaryocytes, and megakaryocytes with multiple small, distinct nuclei. A micromegakaryocyte was defined as less than one-third normal size, approximately less than 25 pm in diameter. Erythrocytes were examined for multinuclearity, nuclear fragmentation, and megaloblastic changes. Leukocytes were graded for granulationabnormalities,nuclear segmentation abnormalities, and Pelger-Huet cells." Granulation abnormalities were defined as a lack of granules in the cytoplasm. Nuclear segmentation abnormalities included hypolobular and pseudo-Pelger-Huet abnormalities. Individual characteristicsof leukocyte and erythrocyte dysplasia (Table 1)were graded from 1to 4, with 1 representing no dysplasia; 2, mild dysplasia; 3, moderate dysplasia; and 4, marked dysplasia. Because many of the bone marrow specimens did not have evaluable megakaryocytes or had few megakaryocytes to review, megakaryocyticdysplastic features were scored as mild or marked. Three to four low-power (X10)and 10highpower (X100) fields were examined for each specimen. A characteristicwas scored as mild if less than 10% of a cell line was abnormal; moderate, if 10-30% of the cells of that lineage exhibited this specific characteristic;and marked, if more than 30% of the cell line showed the particular dysplastic characteristic. For example, the presence of a rare Pelger-Huet cell earned a score of mild Pelger-Huet dysplasia; if 20% of the erythrocytes were megaloblastic, this was scored as moderate megaloblastic changes; and if 50% of the megakaryocytes were mononuclear, the bone marrow was considered to have marked mononuclear megakaryocytic dysplasia. Each cell type was given a dysplasia score, and all scores were added for an overall dysplasia score. Cell types that could not be evaluated, because insufficient numbers (fewer than 25) of these cells were present, were marked as not evaluable and assigned the average score. Nonevaluable patients were excluded from some of the statistical analyses, as discussed in Results. Complete remission was defined as less than 5% blasts in a normocellular or hypercellular bone marrow with normal peripheral blood counts. Survival times were measured from the date of diagnosis. Statistical Analysis The Wilcoxon midrank test was used to test for association between a dichotomous outcome (e.& remission) and ordered categorical variates (e.g., none, mild, moderate, marked dysplasia), and Fisher's exact test was used to test for association with unordered categorical covariates.",12 The logistic regression model was used to assess simultaneously the association of more than one covariate with remi~sion.'S~urvival probabilities 316 CANCER Ianuary 15, 1994, Volume 73, No. 2 Table 1. Bone Marrow Dysplasia Scoring and Remission Rate (N = 106) Remission No. rate(%) P value - 1Remimim No. rate (%) Micromegakaryocytes 0.74 Overall erythrocyte dysplasia Not evaluable 61 50 Not evaluable 27 56 Absent 18 55 None 9 56 Mild increase 18 56 Mild 36 53 Marked increase 9 56 Moderate 20 55 Mononuclear megakaryocytes 0.92 Marked 14 57 Not evaluable 61 56 Leukocyte granulation Absent Mild increase 15 53 22 55 abnormalities Not evaluable 30 53 0.Y Marked increase 8 50 Absent 12 42 Multiple nuclei megakaryocytes 0.58 Mild increase 37 59 Not evaluable 62 55 Moderate increase 22 64 Absent 20 55 Marked increase 5 20 Mild increase 14 57 Leukocyte nuclear segmentation Marked increase 10 60 abnormahties 0.94 Overall megakaryocyte dysplasia 0.71 Not evaluable 30 53 Not evaluable 65 54 Absent 13 46 None 7 57 Mild increase 34 59 Mild 18 61 Moderate increase 24 63 Moderate 9 44 Marked increase 5 20 Marked 7 57 Pelger cells 052 Erythrocyte multinuclearity 0.95 Not evaluable 28 54 Not evaluable 26 54 Absent 43 58 Absent 33 55 Mild increase 21 52 Mild increase 19 53 Moderate increase 11 55 Moderate increase 16 63 Marked increase 3 33 Marked increase 12 50 Overall leukocyte dysplasia 0.71 Erythrocyte nuclear fragmentation 0.98 Not evaluable 29 55 Not evaluable 26 54 None 9 33 - Absent 25 60 Mild 41 59 Mild increase 21 48 Moderate 22 64 Moderate increase 18 50 Marked 5 20 Marked increase 16 63 Overall impression of dysplasia 0.93 Erythrocyte megaloblasts 0.46 Not evaluable 16 56 I t Not evaluable Absent 26 54 12 50 None Mild 6 33 42 60 -. Mild increase Moderate increase 27 56 29 48 Moderate Marked 34 53 8 50 Marked increase 12 75 were estimated by the method of Kaplan and Meier.14 The log-rank test, log-rank test for trend, and proportional hazards model were used to test for associations with s~rvival.'~-'' values less than 0.05 were considered significant. Results Clinical characteristics of the 106 patients in the study are shown in Table 2. The average age was 57 years, and 51% of the patients were men. FAB types M2 and M4 were the most common. Corresponding remission rates and tests for association with remission and SUI- viva1 are shown in Table 3. Thirty-eight percent of pa- tients had a known cytogeneticabnormality; a b n o d ties of chromosomes 5, 7, and 8 were associated with a worse prognosis. Twenty-five percent of patients had 1 history of myelodysplasia; as anticipated, these p a t h 6 had lower disease remission rates. Patients who had a history of a prior malignant disorder (22%) and those who had received prior chemotherapy (11%) also had i i I 1 Marrow Dysplasia at Diagnosis of AML/Ballen et al. 317 ' rJla 2. Patient Characteristics patient characteristics Overall group Age (F) < 21 21-40 41-60 61-80 > 80 sex Female Male FAB dassificahon M1 M2 M3 M4 M5 M6 M7 No. 106 3 20 25 55 3 52 54 17 38 5 28 8 6 4 32 2 5 1 Abnormal chromosome 5 6 Abnormal chromosome 7 9 Abnormal chromosome 8 8 Other abnormal chromosomes 9 unknown 34 k t o q of myelodysplasia None 79 Refractory anemia 8 Refractory anemia-excess blasts 7 Refactory anemia-transformation 6 Sideroblastic anemia 5 CMML 1 plior hematologic diagnosis None 83 Lymphoma 3 h a s t cancer 1 Pdycythemia 2 Myelofibrosis 6 Hodgkin's 1 other 10 Chemotherapy None 94 Allrylating agent 7 N W k ylating 5 -on chemotherapy None 10 Hyboxyurea 11 W R and ARA-C 2 and 5' 4 DNR and ARA-C 3 and 7. 43 DNR and ARA-C 3 and 7*/HIDAC 11 D m and ARA-C 3 and 7'/6TG 2 ----o"-&th-etor-w-n-------_------- 23 2 p/%;MDAC: ara-C, 4 g/m /d for 3 d; 6TG: 6-thioguanine, 200 mg/ FAB:bch-American-British; M M L chronic myelomecrocytic leu- to days of administration of daunorubidn and Ara-C. rrSpeC- i Table 3. Association of Patient Characteristics With Remission and Survival Remission N (%I All cases 106 Age (Yr) 5 40 41-60 > 60 23 25 58 Sex Female Male FAB classification M1 M2 M3 M4 M5 M6 M7 Chromosomes Unknown Normal Abnormal 5, 7, 8 Other abnormalities 52 54 17 38 5 28 8 6 4 34 32 23 17 History of myelodysplasia No YeS Prior malignant diagnosis No Yes Prior chemotherapy No Yes Induction chemotherapy Aggressive Nonaggressive or none Unknown' 79 27 83 23 94 12 56 48 2 55 91 64 36 58 52 41 47 80 64 63 50 75 50 63 35 76 63 30 60 35 61 8 86 21 0 Remission P value - < 0.001 0.55 0.54 0.05 0.003 0.04 0.001 < 0.001 Survival P value 0.001 0.53 0.93 0.09 0.21 0.02 0.001 < 0.001 FAB French-American-British. Excluded from analysis. lower remission rates. Blood counts at the time of diagnosis and before diagnosis did not influence remission rate (data not shown). The type of induction chemotherapy was the strongest predictor of remission rate. Patients who received intensive chemotherapy (defined as daunorubicin 45 mg/m2/day for 3 days and cytosine arabinoside 100200$mg/m2/dayfor 7 days, with or without high-dose cytosine arabinoside or 6-thioguanine) were more likely to have disease enter remission (P < 0.001). 318 CANCER january 15, 1994, Volume 73, No. 2 Table 1outlines the dysplastic findings on the bone marrow aspirates and biopsy specimens, with corresponding remission rates and the test for correlations between degree of dysplasia and remission rate. Nonevaluable patients were excluded from these tests, although formal tests comparing evaluable with nonevaluable patients did not change the conclusions. Thirty-two percent of patients had megakaryocytic dysplasia (15% moderate or marked) manifest by micromegakaryocytes, mononuclear megakaryocytes, or megakaryocytes with multiple distinct nuclei. Remission rate did not correlate with the degree of megakaryocytic dysplasia. Sixty-sixpercent of patients had erythrocyte dysplasia (32% moderate or marked) demonstrated by multinuclearity, nuclear fragmentation, and megaloblastic changes. There was no correlation between remission rate and degree of erythrocyte dysplasia. Sixty-four percent of patients had leukocyte dysplasia (25% moderate or marked) with granulation abnormalities, nuclear malformations, and Pelger-Huet cells. The remission rate was not correlated with the degree of leukocyte dysplasia. Pelger-Huet cells were found in the bone marrow of 33% of patients; the presence of Pelger-Huet cells did not correlate with remission rate. Dysplasia did not correlate with remission rate when measured by overall impression (P = 0.93) or by dysplasia score (P = 0.91). The percentage of myeloblasts, presence of reticulin, mye1oid:erythroid ratio, and presence of Auer rods also did not affect the remission rate. The results are unchanged whether patients had both bone marrow aspirates and biopsy specimens reviewed or only one study examined. In multivariate analysis, four factors were independent predictors of remission. Prior chemotherapy (P = 0.04), older age (P = 0.04), abnormalities of chromosomes 5, 7, or 8 (P = 0.02), and nonintensive induction chemotherapy (P < 0.001) were predictors of lower remission rates. Although prior malignant disorder and a history of myelodysplasia correlated inversely with remission rate by univariate analysis, they were not significant in multivariate analysis because of their correlations with older age and nonintensive chemotherapy, respectively. The median survival time was 9 months from the date of diagnosis. Forty-six percent of patients were alive at 1year, 25% at 2 years, and 18% at 3 years. Bone marrow dysplasia at presentation did not correlate with survival. Each measure of dysplasia in Table 1 was tested for association between degree of dysplasia and survival. Results of all tests were nonsignificant, as were results of comparisons of evaluable and nonevaluable patients. There was no correlation between dysplasia and survival when measured by overall impres- sion (P = 0.42) or by an example, Figure total dysplasia score 1 shows survival of (P = 0 . 1 patients ~.)AI overall impression of dysplasia scored as hvil) versus moderate/marked (P= 0.54).Type of antileuh mic therapy, prior chemotherapy, and patient age we, the most important significant predictive factorsform, viva1 (Table 3). At 3 years, 32% of patients intensive induction treatment (including patients wyfk subsequently underwent bone marrow transplantah were alive, compared with none in the group kea& with less than intensive therapy. Morphologic features of dysplasia did not combt, with abnormal karyotype (P = 0.25), prior chemothcr apy (P= 1.00),prior (P = 0.55), or with malignant disorder (P = 1.00), a history of myelodysplasia ( pam 0.47), as outlined in Table 4. Discussion This study attempted to determine if dysplasia th bone marrow of patients with AML had prognosticnK nificance or correlated with a history of myelodyspla sia, previous alkylating agent therapy, or prior hematt, logic disorder. This analysis was important becam AML arising after one of these disorders has a pahm larly dire prognosis. In addition, in a patient with AML it may not always be clear whether a previous my& dysplastic syndrome was Realizing that dysplasia is considered a subjdve finding, we attempted to make our study as objectiveas possible by rigorously quantifying cellular characttics. Objective criteria for dysplasia were established fm each hematopoietic lineage. Nine morphologic charar- teristics were graded, and a numerical dysplasia score was obtained for each characteristic, each cell line, and the bone marrow as a whole. The morphology re* was blinded for name, age, and history of the patient We found that some characteristics, such as erythrocysr multinuclearity and Pelger-Huet cells, were easier b quantify than others, such as leukocyte granulationab normalities. All three observers had to agree before a cell was marked as dysplastic. Dysplasia was a common finding, with the bow marrow of 32% of patients demonstrating m e g a k q e ' cytic dysplasia, 66% showing erythrocyte dysplask 64% exhibiting leukocyte dysplasia, and 33% ha- Pelger-Huet cells. This high incidence of dysplasia m. ceeded the percent of patients known to have disease arising from a myelodysplastic syndrome (25%))a+ malignant disorder (22%),or Aylating agent therapy (79bL A high number (25%) of our patients with AML had a history of myelodysplastic syndrome that be documented by bone marrow examination than 3 months before the diagnosis of AML. This ht%h Marrow Dysplasia at Diagnosis of AMLIBallen et al. Survival 319 -oo--0-----4--oo------------------0---- -1. Survival by dysplasia *heorics (P = 0.54). -: none or dysplasia; - - -:moderate or dysplasia. 0 12 24 36 48 60 - 72 Months after Diagnosis D splasia Ovcrnl~.hpresslonALIYE DEAD TO4T6AL ME1D3I.0AN --------------.NMoodn-eM-mardked 9 37 13 27 40 8.0 :*Wenceof myelodysplasia may be attributed to our *mal-based patient population. Our seriesrepresents ` w t i v e patients with newly diagnosed AML with mMls available, and thus includes patients with dt M V O and secondary AML without any selection. ?`hedegree of bone marrow dysplasia at time of diag- W s of AML was not different in the de novo group in the secondary AML group, despite a significant w m c e in prognosis between these two groups. Factors that were significant for predicting remis- uan atein our patients were age, chromosomalabnor- pnor chemotherapy, and type of inductioncheT e a p y , all in accordance with previous studies in &&ahre.5,6.18-2 1 results differ from some published studies on VhOlOgy in patients with AML.'2,'3 Curtis et %that the percentage of bone marrow myelo- *h a Was inversely related to survival. Brito-Babapulle studied 160 patients with AML. Twenty-four of drge Patients had evidence of trilineage dysplasia and u s i o n rate of 50%,compared with a remission rate of 86% for patients without dysplasia. However, no significant difference in overall survival was found. Other studies have reported no relationship between remission rate and morphologic features.'"'' Fenaux et al.29reviewed 214 patients, and 30% had trilineage dysplasia. There was no differencein the disease remission rate between patients with and those without dysplasia. A review of morphology by Swirsky et al.30found that only Auer rods and eosinophilia were important positive predictors of remission rate, whereas a low platelet count was an adverse factor. Patient age and performance status were the only historic factors found to correlate with survival. Auer rods were found to be positive predictors of survival by Mertelsman et Auer rods were found in the leukemic cells of only 17 of our patients, making statistical analysis difficult. A study of 332 patients with de novo AML found that only 11% of patients had trilineage dysplasia. Granulocytic dysplasia was associated with a lower remission rate (57%versus 72%),but erythrocyticandmegakaryocytic dysplasia had no effect on remission rate.3' 320 CANCER January 15, 1994, Volume 73, No. 2 Table 4. Correlation of Bone Marrow Dysplasia With Patient Characteristics Marrow dysplasia* N (%) P value All evaluablet Age (yr) 5 40 41-60 > 60 Chromosomes Unknown Normal Abnormal 5, 7, 8 Other abnormality Myelodysplasia history No Yes Prior malignant disorder No Yes Prior chemotherapy No 90 18 23 49 31 28 19 12 68 22 71 19 81 47 56 44 45 42 43 58 50 44 55 46 47 47 AA 0.57 0.35 0.47 1.oo 1.00 * Percentageof bone marrow specimensscored as moderate or marked dysplasia. t Sixteen patients were srored asnot evaluableand excluded from this analysis. The etiology of dysplastic cells in patients with AML is unclear. If the leukemic cell is an early progenitor cell common to granulocytes, erythrocytes, and megakaryocytes, all of these lineages will be involved.' Fialkow et al.,3using patients who are heterozygous for X-chromosome-linked glucose-6-phosphate dehydrogenase, showed that AML develops clonally and is heterogeneous. In some patients, the leukemic stem cell manifests granulocytic, megakaryocytic, and erythrocytic differentiation,whereas in other patients, only the granulocytic line is involved. Using recombinant DNA probes, Fearon et al.33demonstrated that in some patients leukemic blast cells can differentiate to form mature cells. We speculate that the dysplastic findings we observed represent incomplete maturation of the leukemic clone. Our study demonstrates that the presence of specific and trilineage dysplastic findings in the bone marrow of patients with newly diagnosed AML is common, not predictive of a history of myelodysplasia, and is not prognostically significant. Because patients with a previous history of myelodysplasia sometimes are treated differently and less intensively than are patients with presumed de novo AML, dysplastic findings in the bone marrow of patients should not be used to deter- mine whether there may have been a p r e v i a : $ dysplastic syndrome, which might deny the propriate intensive chemotherapy. -4, References 1. Vellenga E, Griffin JD. The biology of acute myelomia. Semin Oncol 1987; 14:365-71. hc 2. Griffin ID, Lowenberg B. Clonogenic cells in acute leukemia. Blood 1986; 68:1185-95. pdabl*+ 3. Fialkow PJ, Singer JW, Raskind WH, Adamson JW-,1 Bemstein ID, et al. Clonal development, stem-ceu diff Q tion, and clinical remissionsin acute n o n l y m p h w JaAa~*,J Y ~ N Engl] Med 1987; 317:468-73. 4. Keinanen M, Griffin JD, Bloomfield CD, Machnicki1, ,&h,- pelle A. Clonal chromosomal abnormalities shocell-lineage involvement in acute myeloid l e u Med 1988; 318:1153-8. k e-Nb4 f 5. Cadman EC, Capizzi RL, BertinoJR. Acute nonlymphacpLb kemia. Cancer 1977; 40:1280-96. 6. Layton DM, Mufti GJ. Myelodysplastic syndromes: tory, evolution, and relation to acute myeloid- 1 L 1986; 53~423-36. 7. Bloomfield CD, de la Chapelle A. Chromosomeabn-. acute nonlymphocytic leukemia: clinical and biologic cance. Semin Oncol 1987; 14:372-83. 8. Rowley JD, Alimena G, Garson OM, Hagemeister A, F, Prigogina EL. A collaborativestudy of the relatio&pd& morphologic type of acute nonlymphocytic leukemia wi(h , tient age and karyotype. Blood 1982; 59:1013-22. 9. Islam A, Catovsky D, Goldman JM, Galton DAG. Bone biopsy changes in acute myeloid leukemia: I. observa- h fore chemotherapy. Histopathology 1985; 9:939-57. 10. Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton Gralnick HR, et al. Proposals for the classificationof themy& dysplastic syndrome. Br Haemafol 1982; 51:189-99. i 111. Lehmann EL. Nonparametrics: statistical methods ba& Q ranks. San Francisco: Holden Day, 1975:5-23. 12. Mehta CR, Patel NR. A network algorithm for perfarrq Fisher's exact test in rxc contingency tables. 1 Am Std Arc 1983; 78:427-34. 13. Hosmer DW, Lemeshow S. Applied logistic regresslar. pk York: John Wiley and Sons, 1989:25-34. 14. Kaplan EL, Meier P. Nonparametric estimation from mcunph observations. \Am Stat Assoc 1958; 53:457-81. 15. Pet0 R, Peto J. Asymptotically efficientrank invariant test pm dures (with discussion).J R St-' -oc A 1972; 135:185-206. 16. Tarone RE. Tests for trend in litt table analysis. BiornetrikP 197 62:679-82. 17. Cox DR. Regression models and life tables (with discussion).IK Sfat SOC B 1972; 34~187-220. 18. Rai KR, Holland JF, Glidewell OJ, Weinberg V, B " m . Obrecht JP, et al. Treatment of acute myelocyfic leukemk study of cancer and leukemia Group B. Blood 1981;58:1203-12 a19. Keating MJ, Smith TL, Gehan EA, McCredi KB, *Spitzer G, et al. Factors related to length of complete in adult acute leukemia. Cancer 1980; 45:2017-29. 20. Foon KA, Gale RP. Controversiesin the therapy of acutemy& enous leukemia. Am 1Med 1982; 72:963-79. 21. Peterson BA, Bloomfield CD. Treatment of acute nodymphDcr tic leukemia in elderly patients. Cancer 1977; 40647-52. 22. SandbergAA. The chromosomes in human leukemia. matol 1986; 23:201-17. 23. Kantarjian HM, Keating MJ, Walters RS, McCredie KB, i i .f Marrow Dysplasia at Diagnosis of AML/Ballen et al. 321 @The characteristicsand outcomesof patients with late relapse mtemyelogenous leukemia. ]Clin Oncol 1988; 6:232-8. a Q& JE, Messner HA, Hasselback R,Elhakim TM, McCulloch a contributions of host- and disease-related attributes to the opffome of patients with acute myelogenous leukemia. J Clin -1 1984; 2:253-9. 5 kjto-Babapulle F, Catovsky D, Galton DAG. Clinical and labo- features of de novo acute myeloid leukaemia with trilinregcmyelodysplasia. Br ] Hemat 1987; 66:445-50. IaonRA, Le Beau MM, Vardiman JW, Testa JR, Golomb HM, IlawleyJD.The predictive value of initial cytogenetic studies in 148 adults with acute nonlymphocytic leukemia: a 12 year rtudy.Cancer Genet Cytogenet 1983; 5219-36. 3.Cuson OM.Acute non-lymphocytic leukaemia. Clin Hematol 1980; 939-51. J W - n n W, Wormann B, Gohde W, Buchner T. DNA aneuploidies in adult F- 'nts with acute myeloid leukemia. Cancer 1986; 572146-52. 29. Fenaux P, Preudhomme C, Lai JL,Morel P, Beuscart R, Bauters F.Cytogenetics and their prognostic value in de novo acute my- eloid leukaemia:a report on 283 cases. BrlHernatolI987; 73:61-7. 30. Swinky DM, de Bastos M, Parish SE, Rees JKH, Hayhoe FGJ. Features affecting outcome during remission induction of acute myeloid leukemia in 619 adult patients. Br Hematol 1986; 64:435-53. 31. Mertelsmann R, Thaler HT, To L, Gee TS, McKenzie S, Schauer P, et al. Morphological classification, response to therapy, and survival in 263 adult patients with acute nonlymphoblasticleukemia. Blood 1980; 56:773-81. 32. Matsuo T, Goasquen J, b y L, Bennet Jh4.Trilineagedysplasia in acute myeloid leukemia (AML): a study of 332 patients from an ECOG trial. Blood 1990; 76(Suppl 1):300. 33. Fearon ER, Burke PJ, Schiffer CA, Zehnbauer BA, Vogelstein B. Differentiation of leukemia cells to polymorphonuclear leukocytes in patients with acute nonlymphocytic leukemia. N Engl] Med 1986; 315:lS-24.