Risk Assessment in Patients with Acute Myeloid Leukemia and a Normal Karyotype

1416 Vol. 11, 1416–1424, February 15, 2005 Clinical Cancer Research

Marianne Bienz, Madleina Ludwig, mutations. We found that high BAALC expression in normal- Beatrice U. Mueller, Elisabeth Oppliger Leibundgut, karyotype AML with neither FLT3-ITD nor CEBPA mutations (18 of 67) indicates adverse prognosis for both Daniel Ratschiller, Max Solenthaler, Martin F. Fey, DFS and OS (P = 0.0001; e.g., P = 0.0001) compared with the and Thomas Pabst group with low BAALC expression and absent FLT3-ITD Institute of Medical Oncology and the Laboratory for Molecular and CEBPA mutations (18 of 67). Thus, BAALC expression Diagnostics, Department of Hematology, University of Berne, Berne, represents a novel prognostic marker particularly for Switzerland normal-karyotype AML patients with neither FLT3-ITD nor CEBPA mutations. ABSTRACT Conclusions: Assessment of CEBPA mutations, FLT3- Purpose: The recognition of a number of leukemia- ITD, and BAALC expression permits to split normal- specific cytogenetic abnormalities and their role as indepen- karyotype AML into clinically distinct subgroups. dent prognostic factors have provided considerable insights into leukemia pathogenesis and have paved the way to adopt INTRODUCTION risk-adapted treatment. However, f50% of newly diagnosed acute myeloid leukemia (AML) have a normal karyotype. The karyotype of acute myeloid leukemia (AML) assessed There has therefore been much interest in identifying at diagnosis is generally recognized as the single most valuable molecular markers that could help to improve the prognostic prognostic factor in AML (1, 2). However, using conventional stratification of patients with normal-karyotype AML. cytogenetic techniques, karyotype abnormalities are detected in Experimental Design: Consecutive untreated AML only half of all AML cases (1, 2), although the other half are patients (n = 67) from a single institution all with normal commonly described as normal-karyotype AML. Patients with karyotype were analyzed for the presence of mutations in the normal-karyotype AML usually have an intermediate risk with myeloid transcription factor gene CEBPA (for CCAAT/ a 5-year overall survival of between 35% and 45% (1–3), but enhancer binding protein-A), for internal tandem duplica- clinical outcome may vary greatly. In addition, the appropriate tions (ITD) of the tyrosine kinase receptor gene FLT3 (for choice of consolidation in first remission (chemotherapy versus fms-like tyrosine kinase 3), and for expression of the BAALC autologous transplantation versus allogenous transplantation) is gene (for brain and acute leukemia, cytoplasmic). unclear for these patients (1–3). Thus, additional markers with Results: 17.9% of normal-karyotype AML had muta- prognostic significance are needed to identify clinically relevant tions in the CEBPA gene, and 28.4% had FLT3-ITD; 65.7% subgroups among AML patients with a normal karyotype. (44 of 67) had high BAALC expression and 34.3% (23 of 67) Some interesting candidate markers are now becoming had low BAALC expression. Patients with CEBPA mutations available. had a very favorable course of their disease. Median disease- The transcription factor CEBPA (for CCAAT/enhancer free survival (DFS) and overall survival (OS) were 33.5 and binding protein-a; for review, see ref. 4) is expressed in 45.5 months, respectively, compared with 10 (e.g., 12 months myelomonocytic cells and specifically up-regulated during in patients without CEBPA mutations; P = 0.0017; granulocytic differentiation (5). cebpa knockout mice show a P = 0.0007). AML patients with FLT3-ITD had significantly selective block in neutrophil differentiation at the stage of shorter median DFS (P = 0.0328) and OS (P = 0.0148) than myeloblasts that is similar to the maturation arrest seen in human patients without FLT3-ITD. High BAALC expression pre- AML patients (6). Dominant-negative mutations of the CEBPA dicted for a shorter DFS (P = 0.0152) and OS (P = 0.0210) gene have been reported by us and others preferentially in AML compared with AML with low BAALC expression; 53.7% of patients with a normal karyotype and with myeloblastic AML normal-karyotype AML had neither FLT3-ITD nor CEBPA subtypes (AML-M1 and M2; refs. 7–12). Interestingly, prognosis of these AML patients seems to be favorable (9–11). FLT3 (for fms-like tyrosine kinase 3) is a class III tyrosine kinase receptor (for review, see ref. 13). It is involved in Received 8/4/04; revised 11/12/04; accepted 11/16/04. signaling pathways regulating the proliferation of pluripotent Grant support: Swiss National Science Foundation grant 31-66899.01 stem cells and early progenitor cells. Internal FLT3 tandemly (T. Pabst) and SAKK pilot project award 2004. duplicated sequences (ITD) within the JM domain encoded by The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked exons 14 and 15 are the most frequent single mutation described advertisement in accordance with 18 U.S.C. Section 1734 solely to in adult AML with a reported incidence between 13% and 32% indicate this fact. (14–22). Remarkably, FLT3-ITD AML exhibit a high relapse Note: M. Bienz and M. Ludwig contributed equally to this work. risk, decreased disease-free survival (DFS) and overall survival Requests for reprints: Thomas Pabst, Institute of Medical Oncology, University Hospital of Berne, CH-3010 Berne, Switzerland. Phone: 41- (OS; refs. 17, 18). Several groups have found in multivariate 31-632-84-30; Fax: 41-31-382-12-37; E-mail: [email protected]. analysis that in AML FLT3-ITD is the most significant factor D2005 American Association for Cancer Research. predicting an adverse outcome (17, 18).

BAALC (for brain and acute leukemia, cytoplasmic) is a threshold (CT) method. The cycle number difference was recently identified gene on chromosome 8q22.3 with a protein calculated as (DCt = CtBAALC CtPBGD) of each replicate. sequence showing no homology to any other known proteins or The mean value from the duplicate was calculated as A (DCt) A(DCt) functional domains (23). In hematopoietic cells, BAALC =(ADCt) / 2 and expressed as 2 . The BAALC-positive expression is restricted to progenitor cells (23). BAALC leukemic Kasumi cell line and a negative control were expression is found in AML and chronic myelogenous included in each assay. The BAALC and PBGD Ct values leukemias in blast crisis whereas no BAALC expression could were measured in duplicate for each patient. In addition, we be detected in patients with chronic-phase chronic myelogenous determined BAALC mRNA levels in both leukocytes from leukemia (23). In AML patients with normal cytogenetics, high peripheral blood and bone marrow samples from 12 healthy BAALC expression seems to predict a poor prognosis (24). volunteers. However, the expression of the BAALC gene and its potential Immunophenotyping and Cytogenetic Analysis. A use as a prognostic marker in normal-karyotype AML in the panel of monoclonal antibodies against myeloid lineage– absence of CEBPA or FLT3-ITD mutations are unknown. associated antigens, including CD9, CD11b, CD13, CD14, In the present study, we assessed the presence of mutations CD15, CD33, glycophorin, and myeloperoxidase; lymphoid in the CEBPA gene, of FLT3-ITD, and of the expression of the lineage–associated antigens, including CD2, CD3, CD7, CD10, BAALC gene in untreated AML with normal cytogenetics. We CD19, CD22, CD79; and lineage-nonspecific antigens, includ- show that this panel of markers adds important prognostic ing HLA-DR, TdT, CD34, CD45, and CD56 was used to analyze information for this largest subgroup in AML. the leukemic cells. The cutoff for a positive result of a particular marker was set at >20%. All cytogenetic analyses were done at a single reference MATERIALS AND METHODS institution, at the university hospital of Lausanne, Switzerland. Patient Samples. The diagnosis of AML was made using Metaphase chromosomes were banded by conventional banding standard morphology and immunophenotype markers. Leukemic technique and karyotyped according to the International System cells were purified at diagnosis from 67 consecutive patients for Human Cytogenetic Nomenclature. A karyotype was with AML of all subtypes from a single university center. Details considered normal if at least 20 metaphases remained without of the patients are given in Table 1. All patients were treated in evidence of a clonal abnormality. previously published or ongoing protocols (SAKK 30/95 and Statistical Analysis. The primary end point was DFS; the SAKK 30/00; ref. 25). Performance status had to be WHO 0-2, secondary end point was overall survival (OS). DFS was defined and no severe organ dysfunction was allowed. All patients had a as the time from achievement of complete remission to first normal karyotype and were thus considered standard risk. appearance of progression/relapse, or death from any cause. OS Treatment for these standard-risk patients in cycle 1 consisted of was defined as the time from diagnosis to death. Patients alive cytarabine and idarubicin and in cycle 2 of cytarabine and without progression/relapse by the time of analysis were amsacrin. Patients in complete remission after cycle 2 were censored at the time of their last follow-up. Time-to-event randomly assigned to a third cycle of chemotherapy with curves were constructed according to the Kaplan-Meier method etoposide and mitoxantrone or high-dose chemotherapy with and were compared with the log-rank v2 test. Correlation busulfan and cyclophosphamide followed by autologous stem coefficient was specified as Pearson correlation (r). cell transplantation. Allogeneic stem cell transplantation was done in first complete remission if a suitable sibling donor was available and the patient was younger than 55 years. RESULTS Mutational and Expression Analysis. Total cellular CEBPA Mutations Define a Subset of Normal Karyo- RNA was extracted from Ficoll density gradient centrifugation type Acute Myeloid Leukemia with Favorable Prognosis. enriched mononuclear cells using the QIAmp RNA Blood Heterozygous mutations of the CEBPA gene were found in 12 of Mini Kit (Qiagen, Chatsworth, CA). cDNA was synthesized 67 AML with a normal karyotype (17.9%). Eight of these 12 from 2 Ag total RNA applying the Superscript system and AML patients (66.7%) had two or more detectable CEBPA random hexamer primers (Invitrogen, San Diego, CA). mutations, and a total of 25 CEBPA mutations were identified For CEBPA mutational analysis, the entire coding region of (Table 3). the gene was amplified using three (A, B, and C) overlapping Ten of the 12 patients with CEBPA mutations had frame- PCR primer pairs as previously described (7). Sequences of shift mutations with truncation of the protein (Fig. 1). In six the primers used are listed in Table 2. PCR products were patients, in-frame insertions at the COOH-terminal in the basic verified on agarose gel electrophoresis, and sequenced in both leucine zipper domain were seen. A total of 10 point mutations directions using BigDye Terminator-Mix Version 3.1 (ABI, were detected in five patients. In two patients, the point Rotkreuz, Switzerland). Abnormal sequencing results were mutations created novel stop codons. Of particular interest is repeated twice in both directions including repetitions of the 212C > A point mutation in patient 7 (Table 3). This PCR. Analysis of the internal tandem duplication of the FLT3 mutation eliminates the serine at amino acid 21. Phosphorylation gene was done by amplification of the JM domain located in of Ser21 has recently been shown crucial for CEBPA function, exons 14 and 15 and subsequent gel electrophoresis. which in this patient is likely to be abolished (26). BAALC mRNA expression (Table 2) was normalized to Interestingly, patients with CEBPA mutations differed in the simultaneously analyzed PBGD gene. The relative BAALC many aspects from AML patients with a normal karyotype and expression was determined using the comparative cycle a wild-type CEBPA gene status (Table 1). Their leukocyte

Table 1 Presenting characteristics CEBPA CEBPA FLT3 BAALC BAALC All mutation wild type FLT3-ITD wild type high low (N = 67) (n = 12) (n = 55) P (n = 19) (n = 48) P (n = 44) (n = 23) P Median, age, y 49 46 49 0.4622 52 48 0.2954 46 53 0.2238 (range, y) (18-71) (22-58) (18-71) (21-69) (18-71) (18-71) (19-70) Sex 33f /34m 5f /7m 28f /27m 0.5546 8f /11m 25f /23m 0.4615 19f /25m 14f /9m 0.1691 WBC, 109/L, 25.9 16.1 29.1 0.1832 38.6 17.6 0.0298 20.9 38.6 0.0938 median (range) (0.5-195.3) (0.9-179) (0.5-195.3) (0.5-195.3) (0.9-190.6) (0.5-195.3) (1.8-190.6) % Blasts in blood, 72 78 72 0.3967 77 71 0.2836 71 85 0.1214 median (range) (1-98%) (26-94%) (1-98%) (24-98%) (1-95%) (7-98%) (1-95%) LDH units/L, median* 1001 368 1074 0.0150 1292 784 0.0325 863 1103 0.3633 De novo AML (%) 84 92 82 0.4040 95 79 0.1210 80 91 0.2173 Secondary AML (%) 16 8 18 5 21 20 9 MDS/therapy-related (6 / 5) (1 / 0) (5 / 5) (1 / 0) (5 / 5) (4 / 5) (2 / 0) Source PB 39 (58%) 7 (58%) 32 (58%) 0.9923 13 (68%) 26 (57%) 0.3304 29 (66%) 10 (43%) 0.1171 BM 28 (42%) 5 (42%) 23 (42%) 6 (32%) 22 (43%) 15 (34%) 13 (57%) Consolidation CR1y Chemotherapy 36 (63%) 7 (58%) 29 (66%) 10 (67%) 26 (62%) 20 (54%) 16 (80%) Autologous Tx 10 (17%) 3 (25%) 7 (15%) 2 (13%) 8 (19%) 7 (19%) 3 (15%) Allogenous Tx 11 (20%) 2 (17%) 9 (19%) 3 (20%) 8 (19%) 10 (27%) 1 (5%) FAB classification M0 5 1 4 1 4 5 0 M1 17 3 14 4 13 15 2 M2 26 8 18 9 17 17 9 M4 5 0 5 2 3 1 4 M5 10 0 10 3 7 2 8 M6 1 0 1 0 1 1 0 M7 3 0 3 0 3 3 0 Gingiva hyperplasia 3 0 3 1 2 1 2 Lymphadenopathy 8 3 5 1 7 5 3 Hepatomegaly 8 2 6 4 4 5 3 Splenomegaly 12 3 9 5 7 6 6 Lung infiltrates 2 0 2 1 1 1 1 Skin infiltrates 1 0 1 0 1 0 1 NOTE. Presenting characteristics of the patients analysed. Seventeen of the 67 patients were older than 60 years at diagnosis. Source indicates whether blood or bone marrow at diagnosis was analysed. Abbreviations: Tx, transplantation; FAB, French American British classification. *LDH normal <480 units/L. yFifty-seven of 67 patients achieved a first complete remission (CR1) and thus underwent consolidation therapy.

count (WBC) at diagnosis tended to be lower (16.1 versus 29.1 AML patients with a normal karyotype and CEBPA mutations G/L). In addition, the median lactate dehydrogenase (LDH) have a remarkably favorable course of their disease. value at diagnosis was not elevated in patients with CEBPA Internal Tandem Duplications of the FLT3 Gene Define mutations compared with patients without CEBPA mutations a Subset of Normal Karyotype Acute Myeloid Leukemia (368 versus 1074 units/L; P = 0.0150). Remarkably, immuno- with Particularly Adverse Prognosis. Tandemly duplicated phenotyping revealed that in 50% of patients with CEBPA sequences within the JM domain of the FLT3 gene were detected mutations (6 of 12), leukemic cells expressed the lymphoid in 19 of 67 AML with a normal karyotype (28.4%). Patients with marker CD7 whereas none of the patients without CEBPA mutations expressed this marker (Table 4). This has not been Table 2 Sequences of primers and probes used in this study reported before to our knowledge. As previously shown by us and others, CEBPA mutations were limited to the myeloblastic Gene Nucleotide sequence subtypes of AML (M1 and M2) but were absent in the FLT3-F 5V-AGCAATTTAGGTATGAAAGCCAG-3V monocytic (M4 and 5), erythroblastic (M6), and megakaryo- FLT3-R 5V-CCTTCCCAAACTCTAAATTTTCTCT-3V blastic (M7) subtypes (17–22). CEBPA-A-F 5V-TCGCCATGCCGGGAGAACTCTAAC-3V CEBPA-A-R 5V-AGCTGCTTGGCTTCATCCTCCT-3V The clinical course was remarkably different in AML CEBPA-B-F 5V-CCGCTGGTGATCAAGCAGGA-3V patients with or without CEBPA mutations. A complete CEBPA-B-R 5V-CCGGTACTCGTTGCTGTTCT-3V remission after induction chemotherapy was achieved in all 12 CEBPA-C-F 5V-CAAGGCCAAGAAGTCGGTGGACA-3V AML patients with CEBPA mutations but only in 82% of AML CEBPA-C-R 5V-CACGGTCTGGGCAAGCCTCGAGAT-3V V V without mutations. Figure 2C and D illustrates that patients BAALC probe 5-CTCTTTTAGCCTCTGTGG TCTGAAGGCCAT-3 BAALC-F 5V-GCCCTCTGACCCAGAAACAG-3V without CEBPA mutations had a significantly shorter median BAALC-R 5V-CTTTTGCAGGCATTCTCTTAGCA-3V DFS (10 months; P = 0.0017) and OS (12 months; P = 0.0007) PBGD probe 5V-CTCATCTTTGGGCTGTTTTCTTCCGCCT-3V than patients with CEBPA mutations (33.5 months; e.g., PBGD-F 5V-GGCAATGCGGCTGCAA-3V 45.5 months) as summarized in Table 5. We conclude that PBGD-R 5V-GGGTACCCACGCGAATCAC-3V

Table 3 C/EBPA mutations in normal karyotype AML patients FLT3-ITD differed in various aspects from other AML patients Patient Age, y FAB Base pair change Amino acid change with a normal karyotype (Table 1). Leucocytes at diagnosis were higher (38.6 versus 17.6 G/L; P = 0.0298) in patients with FLT3- 1 36.5 M1 744-745GC>TT A199L 1V — M1 1167G>A G339S ITD. Also, the median LDH value at diagnosis was markedly 2 40 M2 563-564insCG Y138fsX160 increased in patients with FLT3-ITD as compared with AML 2V — M2 1094-1095insCTG L315-316ins patients without FLT3-ITD (1,292 versus 784 units/L; P = 0.0325). 3 57.8 M2 327-328insC E59fsX107 No significant differences in the imunophenotype between V 3 — M2 1098-1099insGTC V316-317ins patients with and without FLT3-ITD were observed (Table 4). 4 24 M2 551G>A A134A 4V — M2 742-743insGCCGCCCC P199fsX318 The course of AML in patients with FLT3-ITD was 5 38.5 M2 236-237insGC A29fsX160 unfavorable. Although the rate of complete remissions achieved 6 52.8 M2 395del F82fsX159 after induction chemotherapy was similar in patients with and 6V — M2 1076-1077insAAG K309-310ins without FLT3-ITD (84% versus 85%), the median DFS in 7 47 M1 212C>A S21Q 7V — M1 213del P22fsX159 patients with FLT3-ITD was significantly shorter (8.0 versus 7VV — M1 1088-1089insTCT S313-314ins 12.5 months; P = 0.0328). Also, OS of patients with FLT3-ITD 8 53.1 M1 1083C>T Q312X was decreased as compared with normal-karyotype AML 9 33 M2 672C>G L175V without FLT3-ITD (10.1 versus 15.5 months; P = 0.0148). In V 9 — M2 676C>T A176V summary, the presence of FLT3-ITD in normal-karyotype AML 9VV — M2 678-679GG>TT G177F 9VVV — M2 683C>T L178L seems to confer an adverse clinical course. 9VVVV — M2 688C>A P180H High BAALC Expression in Normal Karyotype Acute 9VVVVV — M2 692C>G Y181X Myeloid Leukemia Is Associated with Unfavorable Prognosis. 10* 22.1 M1 1079-1080insTCT S310-311ins No differences were detected for BAALC mRNA levels between 11 51.5 M2 392-393insT A91fsX107 12 44.8 M2 327-328insC E59fsX107 peripheral blood leukocytes and bone marrow cells from 12V — M2 1098-1099insGTC V316-317ins 12 healthy volunteers. Figure 3A depicts BAALC levels for all NOTE. afsXb indicates a frame-shift mutation of the amino acid at volunteers with the values for peripheral blood on the x-axis and position ‘‘a’’ inducing a stop codon at position ‘‘b’’. for bone marrow on the y-axis. The correlation coefficient Abbreviation: FAB, French American British classification. (Pearson) was r = 0.8507 indicating a strong correlation between *Patient 11 had homozygous insertion. BAALC mRNA levels in blood and bone marrow in a given volunteer. Furthermore, the range of BAALC expression among the 12 volunteers was remarkably small (range, 0.03 and 0.15; median, 0.09). We used the median value of 0.09 of these 12 volunteers as cutoff. Therefore, a value above 0.09 was considered ‘‘high’’ expression, whereas a value below 0.09 qualified for ‘‘low expression’’.

Fig. 1 Diagram of proteins encoded by wild- type and mutant CEBPA alleles. Locations of the amino acids corresponding to transactivation (black bars) domain 1 (79-97) and domain 2 (127-200), and the basic zipper domain (278- 358; black bar). In-frame initiation codons at amino acids 1 and 120 encoding proteins of 42 and 30 kDa. A, proteins encoded by six NH2- terminal mutants. The mutant peptide contains wild-type CEBPA sequence (white bars or black bars) followed by a shift of the reading frame (shaded bars) encoding a novel peptide before termination at a novel stop codon. B, four mutants were located at the COOH-terminal encoding novel stop codons (8 and 9VVVVV)or leading to a frame shift with subsequent truncation by a novel stop codon (2 and 4V). C, these six mutants were in-frame insertions within the basic zipper domain. D, point mutations with changes of the amino acids were seen in four patients.

Table 4 Immunophenotypes Interestingly, patients with high BAALC expression did not CD7 CD11b CD34 CD15 MPO differ significantly from patients with low BAALC expression in terms of leukocyte count at diagnosis and the median LDH value All (N = 67) 10 40 43 30 78 at diagnosis (Table 1). However, we observed that monoblastic CEBPA mutation (n = 12) 50 25 50 33 83 CEBPA wild type (n = 55) 043402976subtypes of AML predominantly had low BAALC expression FLT3-ITD (n = 19) 547321690(80%; 12 of 15 AML-M4 and M5) as previously reported (24). FLT3 wild type (n = 48) 10 38 48 34 75 Low BAALC expression was observed in myeloblastic AML BAALC high (n = 44) 16 30 58 21 72 (M1 and M2) in only 25.6% (11 of 43 patients). In contrast, high BAALC high without FLT3/ 026702257 CEBPA mutation (n = 23) BAALC expression was predominantly observed in myeloblastic BAALC low (n = 23) 061174896subtypes of AML (32 of 43; 74.4%) and undifferentiated AML BAALC low without FLT3/ 0 80 7 53 100 (M0; 5 of 5). All AML with M6 and M7 subtypes (four patients) CEBPA mutation (n = 15) also had high BAALC expression. NOTE. Correlation of immunophenotypes and the CEBPA Interestingly, significant differences were detected in the mutation status, the FLT3-ITD, and BAALC expression. Numbers immunophenotype depending on different BAALC expression. indicate percentages of patients in the respective group expressing a Leukemic cells with low BAALC expression had significantly given marker. The cutoff for a positive result of a particular marker was set at >20%. Significant differences were seen for CD7 higher expression of the CD11b, CD15, and myeloperoxidase expression for patients with and without CEBPA mutations (P = antigens. In addition, low BAALC expression correlated with low 0.0023), and for CD11b (P = 0.019), CD34 (P = 0.001), CD15 (P = CD34 expression (Table 4). 0.044), and myeloperoxidase (P = 0.031) depending on high or low The rate of complete remission achieved after induction expression of BAALC. Differences for these markers were also chemotherapy was not different in patients with high versus low significant between high and low BAALC expression when patients with FLT3-ITD or CEBPA mutations were excluded. No differences BAALC expression (82% versus 91%; P = 0.3008). However, were detected in cells with FLT3 wild type versus ITD. No significant median DFS in patients with high BAALC expression was differences were detected in any groups for the lymphoid markers significantly shorter (8.5 versus 21 months; P = 0.0152). Also, CD2, CD3, CD10, CD19, CD22, CD79, and the myeloid markers overall survival of patients with high BAALC expression was CD9, CD13, CD14, CD33, CD45, CD56, and for HLA-DR and TdT. decreased as compared with patients with low BAALC expression (10 versus 21 months; P = 0.0210). In summary, BAALC expression levels in 67 normal-karyotype AML high BAALC expression in normal-karyotype AML seems to be ranged from 0.004 to 67.2. Twenty-three of the 67 patients associated with shortened DFS and OS. (34.3%) fulfilled the criteria for ‘‘low’’ BAALC expression, In Fig. 4, dot blot representations of BAALC expression whereas 44 patients (65.7%) were classified as ‘‘high’’ BAALC levels are depicted for all patients together (left column; n = 67), expression. Again, we determined whether the BAALC mRNA for patients with FLT3-ITD only (second column; n = 19), for levels in blood and bone marrow correlated for a given patient. patients with CEBPA mutations only (middle column; n = 12), for Indeed, Fig. 3B indicates a strong correlation (r = 0.9501) patients having neither FLT3-ITD nor CEBPA mutations (fourth between BAALC levels in blood and bone marrow in the 29 column; n = 36), and finally for the control group of 12 healthy patients in whom both bone marrow and blood were available at volunteers. We found that patients with FLT3-ITD had a broad diagnosis. We also evaluated whether the percentage of blasts range of BAALC expression. Because high and low BAALC in a given sample correlated with BAALC expression levels. expressing patients with FLT3-ITD did not differ in their course However, no correlation was found between the percentage of (data not shown), we conclude that the presence of FLT3-ITD blasts and BAALC expression (Pearson correlation, r = 0.0907). outweighs the significance of BAALC expression. Interestingly,

Table 5 Effect of CEBPA, FLT3, and BAALC status on clinical outcome CEBPA CEBPA FLT3 BAALC BAALC All mutation wild type FLT3-ITD wild type high low (N = 67) (n = 12) (n = 55) P (n = 19) (n = 48) P (n = 44) (n = 23) P CR1 achieved, % 85 100 82 0.1093 84 85 0.9006 82 91 0.3008 Death in induction, % 9 0 11 0.2305 10 8 0.2177 9 9 0.9571 Death in CR1 (n) 101 01 100 Relapse, % 70 25 80 0.0060 84 65 0.0115 77 57 0.0289 OS median, mos 13 45.5 12 0.0007 10 15.5 0.0148 10 21 0.0210 (range, mos) (0-166) (20-166) (0-113) (0-103) (0-166) (0-103) (0-166) DFS median, mos 10 33.5 10 0.0017 8 12.5 0.0328 8.5 21 0.0152 (range mos) (0-136) (9-136) (0-113) (0-103) (0-136) (0-136) (0-113) DFS at three years, (n) 11 5 6 1 10 6 5 OS at three years, (n) 12 6 6 1 11 7 5 Patients disease-free in follow-up*, n 22 8 14 4 18 11 11 Median, mos 38 60 23 30 43 34 40 Range, mos 4-136 27-136 4-113 4-103 10-136 10-136 4-113 NOTE. No patient died during consolidation therapy randomized to chemotherapy (n = 36) or to autologous transplantation (n = 10), and one patient died during allogeneic transplantation (1 of 11). DFS and OS are given as a median and also at the time point 3 years after diagnosis. *Duration of follow-up is indicated for patients who are still disease free with median value and range.

patients with CEBPA mutations were predominantly seen in the expression group whether clinical outcome in patients with very ‘‘high’’ BAALC expression group. Only three out of 12 patients low expression was more favorable than in patients with ‘‘only’’ with CEBPA mutations relapsed. Remarkably, these three patients low expression. Again, we observed that patients with very low showed the highest BAALC expression among the group of 12 BAALC expression did not differ in their DFS and OS from normal-karyotype AML patients with CEBPA mutations. patients with low BAALC expression (P = 0.753 and P = 0.746, We also analyzed within the high BAALC expression group respectively). These results support the usefulness of our cutoff whether DFS or OS in patients with very high expression indicating that this cutoff indeed seems to separate groups (top 50%) was shorter than in patients with ‘‘only’’ high of prognostic favorable and unfavorable patients. expression (lower 50%). However, patients with very high Low BAALC Expression in Normal Karyotype Acute BAALC expression did not differ in their DFS and OS from Myeloid Leukemia with neither CEBPA nor FLT3 -ITD patients with high BAALC expression (P = 0.497 and P = 0.757, Mutations Is Associated with Favorable Prognosis. We respectively.). Similarly, we studied within the low BAALC hypothesized that determining BAALC expression might be

Fig. 2 DFS (A, C, and E ) and OS (B, D, and F) in normal- karyotype AML patients according to the FLT3-ITD status (A-B), the CEBPA mutation status (C-D), and the BAALC expression (E-F).

Fig. 3 Correlation of BAALC mRNA levels in peripheral blood (PB) leukocytes (x-axis) and bone marrow (BM) cells (y-axis) from 12 healthy volunteers (A). Pearson correlation coefficient r = 0.8507 indicates strong correlation between BAALC mRNA levels in blood and bone marrow. B, BAALC mRNA levels in peripheral blood (x-axis) and bone marrow cells (y-axis) from 29 normal- karyotype AML patients at diagnosis with r = 0.9501 again suggesting a strong correlation.

particularly useful in the subset of AML patients with neither DISCUSSION FLT3-ITD nor CEBPA mutations (36 of 67). The median The prognostic effect of various chromosomal aberrations BAALC expression of patients in this subgroup was about 10- in AML is well established with implications for therapy. In fold higher than the value of our control group. Eighteen of these our well-defined cohort of 67 consecutive normal-karyotype 36 patients (50%) had low BAALC expression, and 18 had high AML patients, we have now identified three independent BAALC expression. molecular prognostic factors that define distinct subgroups. The Most interestingly, the clinical course of these two presence of mutations in the CEBPA gene indicated a favorable subgroups differed dramatically both for DFS (18.4 and 7.4 course of the disease, whereas FLT3-ITD confers a bad months, respectively; P = 0.0001) and for OS (22.8 and 9.1 prognosis. In addition, high expression of BAALC mRNA is months, respectively; P = 0.0001) as depicted in Fig. 5A and B. associated with a significantly worse prognosis. We thus conclude that BAALC expression adds significant We detected CEBPA mutations in one sixth of normal- prognostic information particularly in those AML patients with karyotype AML patients. Others have reported slightly lower a normal karyotype where until now other markers such as percentages with between 4.3% and 11% (7–10, 12). However, FLT3-ITD or CEBPA mutations are lacking. most of these studies have not focussed on normal-karyotype Ultimately, we did a multivariable analysis to investigate AML. In accordance with others, we confirmed that CEBPA whether CEBPA mutations, FLT3-ITD and BAALC expression mutations preferably occur in the FAB classes M1 and M2 (7–12). represent independent prognostic markers in normal-karyotype The clinical effect of CEBPA mutations seems to be distinctly AML. The results of this analysis are summarized in Table 6 favorable. Our AML patients with CEBPA mutations had a indicating that CEBPA mutations, FLT3-ITD, and BAALC median DFS of 33.5 months and an OS of 45.5 months. In parallel expression seem to be strong independent predictors of outcome with accepted practice in other types of good-prognosis AML in normal-karyotype AML. such as acute promyelocytic leukemia or AML with abnormalities

Fig. 4 Dot plots representing individual levels of BAALC expression. Median BAALC values (line). All, BAALC values of all 67 normal- karyotype AML patients; FLT3, group of patients with FLT3-ITD (n = 19); CEBPA, group of patients with CEBPA mutations (n = 12), w/o mut, group of patients with neither FLT3-ITD nor CEBPA mutations (n = 36); normal, BAALC values of 12 healthy volunteers. of chromosome 16, patients with CEBPA mutations and a normal karyotype may enjoy long-lasting first remission without allogeneic stem cell transplantation. Several groups reported a significantly increased frequen- cy of FLT3-ITD in normal-karyotype AML compared with other AML subgroups (15, 17, 20). Our data are in accordance with this literature. The most significant effect of an FLT3-ITD on clinical outcome is its association with increased relapse risk, decreased DFS and OS (13). Several groups reported that an ITD is the most significant factor predicting an adverse outcome in multivariate analysis (17, 18). In our series of normal-karyotype AML, patients with FLT3-ITD had a significantly reduced DFS and OS. Thus, normal-karyotype AML patients with FLT3-ITD seem to represent the unfavorable end of the prognostic spectrum, with patients with CEBPA Fig. 5 DFS (top) and OS (bottom) of normal-karyotype AML patients with neither FLT3-ITD nor CEBPA mutations (n = 36) according to their mutations representing the favorable end. BAALC expression (high, n = 18; low, n = 18). In our series, 53.7% of normal-karyotype AML patients had neither CEBPA mutations nor FLT3-ITD. We now propose to assess BAALC expression to obtain further prognostic informa- expressers. This approach might produce varying levels of tion particularly in those patients. In a recent report investigating median expression depending on the collection of patients 86 de novo AML with a normal karyotype, high BAALC studied. We used a slightly different approach. Bone marrow and expression was found associated with significantly decreased peripheral blood samples from 12 healthy volunteers were OS and DFS (24). In the study cited above, patients were analyzed for BAALC expression. We found a very small range of dichotomized at BAALC’s median expression into low and high expression within this normal control group with no differences

Table 6 Multivariable analysis for overall survival and disease-free survival OS DFS Hazard ratio Hazard ratio (95% confidence interval) P (95% CI) P CEBPA mutation 0.059 (0.012-0.293) 0.0005 0.083 (0.018-0.376) 0.0012 FLT3 wild type 0.329 (0.153-0.710) 0.0046 0.450 (0.220-0.920) 0.0287 high BAALC 3.855 (1.703-8.728) 0.0012 3.857 (1.695-8.776) 0.0013 Age 1.002 (1.000-1.004) 0.0185 1.002 (1.000-1.003) 0.1042 WBC 1.001 (0.993-1.009) 0.8151 1.003 (0.995-1.011) 0.4457 LDH 1.000 (1.000-1.001) 0.3045 1.000 (1.000-1.000) 0.7889 NOTE. Hazard ratios and P values are given for CEBPA mutations versus wild type, FLT3 wild type versus FLT3-ITD, and high versus low BAALC expression, as well as for age, WBC, and LDH.

between peripheral blood and bone marrow samples. The cation of prognostically relevant genotypes and analysis of cooperating median expression value of these controls was used as cutoff. mutations. J Clin Oncol 2004 Feb 15;22:624–33. Patients were then classified into two subgroups with either 12. Snaddon J, Smith ML, Neat M, et al. Mutations of CEBPA in acute myeloid leukaemia FAB Types M1 and M2. Genes Chromosomes BAALC expression above or below this ‘‘normal’’ BAALC Cancer 2003 May;37:72–8. expression level. Most interestingly, normal-karyotype AML 13. Kottaridis P, Gale RE, Linch DC. FLT3 mutations and leukaemia. patients with low BAALC expression had a significantly better Br J Haematol 2003 Aug;122:523–38. clinical outcome than high expressers both for DFS and OS. 14. Nakao M, Yokota S, Iwai T, et al. Internal tandem duplication of the In summary, we identified CEBPA mutations, FLT3-ITD, flt3 gene found in acute myeloid leukemia. Leukemia 1996 Dec;10: and differing levels of BAALC expression as having independent 1911–8. prognostic significance in normal-karyotype AML. We propose 15. Thiede C, Steudel C, Mohr B, et al. Analysis of FLT3 activating mutations in 979 patients with acute myelogenous leukemia: association that molecular assessment of these three factors at diagnosis with FAB subtypes and identification of subgroups with poor prognosis. offers valuable additional prognostic information and may Blood 2002 Jun 15;99:4326–35. thereby markedly affect therapeutic decisions. 16. Kiyoi H, Naoe T, Nakano Y, et al. Prognostic implication of FLT3 and N-RAS gene mutations in acute myeloid leukemia. Blood 1999 May 1;93:3074–80. REFERENCES 17. Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3 1. Grimwade D, Walker H, Oliver F, et al. The importance of diagnostic internal tandem duplication in patients with acute myeloid leukemia cytogenetics on outcome in AML: analysis of 1612 patients entered into (AML) adds important prognostic information to cytogenetic risk group the MRC AML 10 Trial. Blood 1998 Oct 1;92:2322–33. and response to the first cycle of chemotherapy: analysis of 854 patients 2. Suciu S, Mandelli F, de Witte T, et al. Allogeneic compared with from the United Kingdom Medical Research Council AML 10 and 12 autologous stem cell transplantation in the treatment of patients younger trials. Blood 2001 Sept 15;98:1752–9. than 46 years with acute myeloid leukemia (AML) in first complete 18. Frohling S, Schlenk RF, Breitruck J, et al. Prognostic significance remission (CR1): an intention-to-treat analysis of the EORTC/GIMEMA of activating FLT3 mutations in younger adults (16-60 years) with AML-10 trial. Blood 2003 Aug 15;102:1232–40. acute myeloid leukemia and normal cytogenetics: a study of the 3. Slovak ML, Kopecky KJ, Cassileth PA, et al. Karyotypic analysis AML study group Ulm. Blood 2002 Dec 15;100:4372–80. Epub 2002 predicts outcome of preremission and postremission therapy in adult Aug 08. acute myeloid leukaemia: a Southwest Oncology Group/Eastern 19. Schnittger S, Schoch C, Dugas M, et al. Analysis of FLT3 length Cooperative Oncology Group study. Blood 2000 Dec 15;96:4075–83. mutations in 1003 patients with acute myeloid leukemia: correlation to 4. Tenen DG. Disruption of differentiation in human cancer: AML cytogenetics, FAB subtype, and prognosis in the AMLCG study and shows the way. Nat Rev Cancer 2003 Feb;3:89–101. usefulness as a marker for the detection of minimal residual disease. Blood 2002 Jul 1;100:59–66. 5. Radomska HS, Huettner CS, Zhang P, Cheng T, Scadden DT, Tenen DG. CCAAT/enhancer binding protein a is a regulatory switch 20. Stirewalt DL, Kopecky KJ, Meshinchi S, et al. FLT3, RAS, and sufficient for induction of granulocytic development from bipotential TP53 mutations in elderly patients with acute myeloid leukemia. Blood myeloid progenitors. Mol Cell Biol 1998 Jul;18:4301–14. 2001 Jun 1;97:3589–95. 6. Zhang DE, Zhang P, Wang ND, Hetherington CJ, Darlington GJ, 21. Steudel C, Wermke M, Schaich M, et al. Comparative analysis of Tenen DG. Absence of granulocyte colony-stimulating factor signaling MLL partial tandem duplication and FLT3 internal tandem duplication and neutrophil development in CCAAT enhancer binding protein mutations in 956 adult patients with acute myeloid leukemia. Genes a-deficient mice. Proc Natl Acad Sci USA 1997 Jan 21;94:569–74. Chromosomes Cancer 2003 Jul;37:237–51. 7. Pabst T, Mueller BU, Zhang P, et al. Dominant -negative mutations 22. Abu-Duhier FM, Goodeve AC, Wilson GA, et al. FLT3 internal of CEBPA, encoding CCAAT/enhancer binding protein-a (C/EBPa), tandem duplication mutations in adult acute myeloid leukemia define a in acute myeloid leukaemia. Nat Genet 2001 Mar;27:263–70. high-risk group. Br J Haematol 2000 Oct;111:190–5. 8. Gombart AF, Hofmann WK, Kawano S, et al. Mutations in the gene 23. Tanner SM, Austin JL, Leone G, et al. BAALC, the human member encoding the transcription factor CCAAT/enhancer binding protein in of a novel mammalian neuroectoderm gene lineage, is implicated in myelodysplastic syndromes and acute myeloid leukemias. Blood 2002 hematopoiesis and acute leukemia. Proc Natl Acad Sci U S A 2001 Feb 15;99:1332–40. Nov 20;98:13901–6. Epub 2001 Nov 13. 9. Preudhomme C, Sagot C, Boissel N, et al. Favourable prognostic 24. Baldus CD, Tanner SM, Ruppert AS, et al. BAALC expression significance of CEBPA mutations in patients with de novo acute myeloid predicts clinical outcome of de novo acute myeloid leukemia patients leukemia: a study from the Acute Leukemia French Association (ALFA). with normal cytogenetics: a cancer and leukemia group B study. Blood Blood 2002 Oct 15;100:2717–23. 2003 Sept 1;102:1613–8. Epub 2003 May 15. 10. Van Waalwijk B, van Doorn-Khosorvani S, Erpelinck C, et al. 25. Lowenberg B, van Putten W, Theobald M, et al. Effect of priming Biallelic mutations in the CEBPA gene and low CEBPA expression with granulocyte colony-stimulating factor on the outcome of chemo- levels as prognostic markers in intermediate-risk AML. Hematol J therapy for acute myeloid leukemia. N Engl J Med 2003 Aug 21; 2003;4:31–40. 349:743–52. 11. Fro¨hling S, Schlenk RF, Stolze I, et al. CEBPA mutations in younger 26. Ross SE, Radomska HS, Wu B, et al. Phosphorylation of C/EBPa adults with acute myeloid leukemia and normal cytogenetics: identifi- inhibits granulopoiesis. Mol Cell Biol 2004 Jan;24:675–86.

Marianne Bienz, Madleina Ludwig, Beatrice U. Mueller, et al.

Clin Cancer Res 2005;11:1416-1424.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/4/1416

Cited articles This article cites 13 articles, 2 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/4/1416.full#ref-list-1

Citing articles This article has been cited by 37 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/4/1416.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/4/1416. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.