Correlation of neuropilin-1 overexpression to survival in acute myeloid ...

Leukemia (2006) 20, 1950–1954 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu

ORIGINAL ARTICLE Correlation of neuropilin-1 overexpression to survival in acute myeloid leukemia M Kreuter1, K Woelke, R Bieker, C Schliemann, M Steins1, T Buechner, WE Berdel and RM Mesters Department of Medicine/Hematology and Oncology, University of Muenster, Muenster, Germany

Neuropilin-1 (NRP-1), a vascular endothelial growth factors and semaphorin receptor functioning as mediator of angiogenesis and neuronal guidance, is expressed by various solid tumors. The importance of NRP-1 in hematological malignancies such as acute myeloid leukemia (AML) remains to be elucidated. Therefore, we determined NRP-1 expression by immunohistochemical analysis of bone marrow biopsies of patients with newly diagnosed, untreated AML. The expression of NRP-1 was significantly increased in AML patients (n ¼ 76; median 12.9 arbitrary units (a.u.)) as compared with controls (n ¼ 38; median 2.75 a.u.). Survival was significantly poorer in patients with high (4median) versus low (pmedian) NRP-1 expression levels with 5-year overall survival rates of 16.9 versus 49.6% (P ¼ 0.050). In conclusion, our data provide evidence of increased NRP-1 expression in AML with significant correlation to survival. Thus, NRP-1 might constitute a promising target for antileukemic and antiangiogenic treatment strategies in AML. Leukemia (2006) 20, 1950–1954. doi:10.1038/sj.leu.2404384; published online 31 August 2006 Keywords: neuropilin; angiogenesis; acute myeloid leukemia

Introduction In analogy to solid tumors, emerging data suggest the requirement of angiogenesis also for the development and progression of hematopoietic malignancies. Recently, we and others reported increased angiogenesis in the bone marrow of patients with acute myeloid leukemia (AML).1–5 Neuropilins (NRP) are membranous receptors capable of binding two disparate ligands, class 3 semaphorins (Sema) and vascular endothelial growth factors (VEGF), and regulating two diverse systems, neuronal guidance and angiogenesis. NRPs are expressed in many types of cancer cells, in endothelial cells and in additional many types of normal diploid cell types.6 In bone marrow, NRP-1 expression was detected on bone marrow stromal cells suggesting mediating interactions between stroma and primitive hematopoietic cells7 and on bone marrow adipocytes indicating a regulatory role in adipocyte activity.8 Furthermore, it was reported that NRP-1 expression on fetal liver hematopoietic cells may serve as a source of vascular development9 and is involved in the VEGF-dependent differentiation of human bone marrow progenitor cells into endothelial precursor cells.10 In a systemic leukemia mouse model, it was demonstrated that AML progression was induced by VEGF and that a soluble form of NRP-1 significantly inhibited AML progression in vivo.11 However, little is known about the clinical significance of NRP-1 expression in AML patients. Therefore, the goal of our Correspondence: Dr RM Mesters, Department of Medicine/ Hematology and Oncology, University Hospital Muenster, Albert Schweitzer Str. 33, D-48129 Muenster, Germany. E-mail: [email protected] 1 Present address: Clinic for Thoracic Diseases at the University of Heidelberg, Department of Oncology, Heidelberg, Germany. Received 2 June 2006; revised 31 July 2006; accepted 2 August 2006; published online 31 August 2006

study was to investigate the expression of NRP-1 in bone marrow of adult patients with newly diagnosed, untreated AML and to explore its possible correlation to clinical parameters.

Materials and methods

Bone marrow specimen Seventy-six patients with newly diagnosed, untreated AML were chosen for our study. This study population is a cohort of the previously published 62 patients in whom the degree of bone marrow angiogenesis (microvessel density (MVD))1 and 81 patients in whom the overexpression of basic fibroblast growth factor (bFGF) have been reported.3 All AML patients were treated within trials of the German AML Cooperative Group.12 Thirty-eight control patients were examined as an extension from the initial group of 22 controls.1 From all AML and control patients, a bone marrow core biopsy (iliac crest) for histological diagnosis was obtained at presentation. After every core biopsy, a bone marrow aspiration was obtained through a separate puncture for cytological analyses.

Immunohistochemical staining For NRP-1 expression quantification, bone marrow specimens were fixed in paraformaldehyde, decalcified with ethylenediaminetetraacetic acid and embedded in paraffin. Serial sections (4 mm thick) of each sample were processed immunohistochemically for the expression of NRP-1 with rabbit polyclonal antihuman NRP-1 (Zymed No. 34-7300, San Francisco, CA, USA; working dilution 1:100). As described by the manufacturer and previously demonstrated,13 the anti-NRP-1 antibody used in this study is specific for NRP-1. In order to demonstrate specificity, sections of renal tissues served as positive controls with specific staining of glomeruli, afferent arterioles, glomerular tufts and of proximal tubules as described before (Figure 1).13 Controls for immunostaining using non-immune rabbit immunoglobulin G (sc-2027; Santa Cruz Biotechnology, Santa Cruz, CA, USA) in substitution for the specific first antibodies were consistently negative (data not shown). Immunohistochemical staining was performed by the alkaline phosphatase/anti-alkaline phosphatase double bridge technique (Dako-APAAP kit, Dako, Glostrup, Denmark). Briefly, tissue sections were deparaffinized in xylene and rehydrated in a graded ethanol series. Samples were pretreated to promote antigen retrieval in a microwave oven at 450 W twice for 7 min in 10 mM sodium citrate (pH 6.0; Dako). The primary antibodies were applied overnight at 41C. Subsequent steps were performed according to the manufacturer’s instructions. The fast red substrate (Dako) supplemented with 0.1% (w/v) levamisole was used for revelation of phosphatase activity (30 min at room temperature). Sections were counterstained with 0.1% (w/v) erythrocin solution.

Neuropilin-1, a prognostic marker in acute myeloid leukemia M Kreuter et al

1951

Evaluation of NRP-1 expression Immunostaining was simultaneously assessed by two independent experienced investigators using light microscopy. The investigators were blinded to the clinical characteristics of the patients, the results of MVD, growth factor and receptor expression when performing the receptor evaluation. Expression of NRP-1 protein was semiquantitatively assessed as described previously.2,3 Results are reported as arbitrary units (a.u.). In each biopsy sample, NRP-1 expression was evaluated in two to three sections processed in independent immunostainings, and the mean value was calculated. The agreement between the investigators was 98.4%. To ensure the reliability of the quantification assay, marrow slides have been randomly selected during the study and reanalyzed with excellent agreement. A bone marrow as well as renal tissue specimen served as internal control. Sections of these specimens have been repeatedly immunostained for the different antibodies and evaluated by the two investigators at different times.

Statistics The distributions of the time-to-event variables were estimated using the Kaplan–Meier method, and comparisons were based on the log-rank test. Differences in expression of NRP-1 between AML and control group were analyzed by the Mann–Whitney

Figure 1 Immunohistochemical staining of renal tissue with antiNRP-1 antibody, which served as positive controls showing specific staining of glomeruli, afferent arterioles, glomerular tufts and of proximal tubules as described before.13

Table 1

rank-sum test for independent groups. Statistical significance of overall differences between more than two groups was analyzed by the Kruskal–Wallis one-way analysis of variance. Data are presented as individual data plots or as medians, interquartile ranges (low quartile–high quartile (LQ–HQ)) and ranges. Correlations between MVD, bFGF levels, VEGF levels and NRP-1 expression were assessed by the Spearman rank correlation coefficient. P-values reported are two-sided and P-values p0.05 were considered significant. All calculations were performed using the SPSS package (SPSS Inc., Chicago, IL, USA).

Results

NRP-1 expression in newly diagnosed, untreated AML patients The expression of NRP-1 was evaluated by immunohistochemical staining in bone marrow core sections of 76 AML patients at first diagnosis. Patient characteristics are shown in Table 1. The results from these AML patients were compared with the degree of NRP-1 protein expression in bone marrow biopsies obtained from 38 patients with different diseases but with normal bone marrow morphology. In the case of patients with underlying malignancies, the bone marrow was histologically not involved (Table 1). NRP-1 antigen was consistently detected in every bone marrow specimen from AML as well as control patients. Location of the immunostaining was limited to the cell membrane (Figure 2). Both in AML and control specimen, megakaryocytes were stained with anti-NRP-1 antibodies as well. The NRP-1 protein staining scores of the 76 bone marrow specimens from AML patients ranged from 6 to 15 a.u. with a median value of 12.9 a.u. (LQ–HQ: 10.5–14.6), whereas control bone marrow samples (n ¼ 38) showed significantly weaker NRP-1 protein staining (median 2.75, LQ–HQ: 1.3–4.1 a.u.) (Po0.0001; Figure 3). Survival data were available for 69 of 76 patients. The median overall survival for all analyzed patients was 3.6 (range 0.2–8.6) years with a 5-year overall survival rate of 24.9%. Based on the median NRP-1 expression level of the entire AML patient group, we classified patients into two groups each for patients with high (4median) and low (pmedian) NRP-1 expression. Patients with a high NRP-1 expression level (n ¼ 38) had significantly worse overall survival than those (n ¼ 31) with a low NRP-1 expression level (P ¼ 0.050). Estimated 5-year overall and relapse-free survival rates for patients with high NRP-1 expressing AML bone marrows were 16.9 and 20.9% compared to 49.6 and 46.4% for patients with low NRP-1 expression (Figure 4). However,

Patient characteristics

Age (years)a Sex (males/females) FAB distributionb Percentage of leukemic blastsa (bone marrow) Disease

AML patients (n ¼ 76)

Control patients (n ¼ 38)

62 (19–84) 39/37 1 M0, 6 M1, 27 M2, 2 M3, 15 M4, 21 M5, 3 M6, 1 other 75 (30–95) F

63 (18–82) 22/16 F F Five non-Hodgkin’s lymphoma, two Hodgkin’s disease, four solid tumors, 27 non-malignant disorders

Abbreviation: AML, acute myeloid leukemia; FAB, French–American–British classification for AML.27 a Median (range). b French-American classification for AML.27 Leukemia


1952 relapse-free survival differences did not reach statistical significance (P ¼ 0.15). Female patients showed significantly higher NRP-1 immunoreactivity than male patients (P ¼ 0.033). However, there was no correlation with AML subtype, karyotype, age, initial white blood cell count or the degree of leukemic blast infiltration at presentation and after induction chemotherapy. From bone

15 14 13 12

NRP expression (AU)

11

a

10 9 8 7 6 5 4 3 2 1 0 AML

b

Control

Figure 3 Distribution of NRP-1 expression of 76 AML and 38 control patients. Median expression of NRP-1 was 12.9 a.u. for AML patients and 2.75 a.u. for control patients. Data are presented as individual values (open circles) and medians (solid bar).

a 100

Overall survival (%)

90 80 70 60 low NRP

50 40

p = 0.050

30 20

high NRP

10 0 0

c

1

2

3

5 4 Years

6

7

8

9

b 100 Relapse free survival (%)

90 80 70 60 50

low NRP

40

p = 0.146

30

high NRP

20 10 0

Figure 2 Immunohistochemical staining of bone marrow tissue slides with anti-NRP-1 antibody. Sections from AML specimens with high (a) and low (b) NRP-1 expression and from a control patient (c). Original magnification 400. Leukemia

0

1

2

3

4 5 Years

6

7

8

9

Figure 4 Kaplan–Meier estimates of overall (a) and relapse-free survival (b) of AML patients (n ¼ 69) grouped into high and low NRP-1 expression. Overall survival was significantly diminished in patients with high NRP-1 expression.


1953 marrow specimen, adjacent sections to those processed in the present study were immunostained with specific markers of endothelial cells (anti-human thrombomodulin antibodies) for the assessment of MVD and with specific antibodies for the expression of angiogenic factors; results have been reported previously.1–3 No relation was found between NRP-1 expression and MVD (P ¼ 0.59), expression of VEGF (P ¼ 0.18), VEGFreceptor 1 (VEGFR-1) (P ¼ 0.47), VEGFR-2 (P ¼ 0.22) or bFGF (P ¼ 0.15).

Discussion Despite progress in understanding the molecular biology of AML, its treatment remains challenging. Advances in understanding the pathophysiology of AML may lead to improvements in survival of patients with this disease. Current investigations have revealed that angiogenesis plays a significant role in the pathogenesis of AML and in the mechanisms of disease progression. The present investigation clearly documents a significantly increased expression of NRP-1 in the bone marrow of patients with newly diagnosed, untreated AML compared with controls. Until now, NRP-1 expression has been reported in several solid tumors6 but is not well characterized in hematological malignancies. Our data are in line with the report by Schuch et al.11 who described NRP-1 expression on AML cell lines and in leukemic blast isolates from five patients with AML. However, this report is the first demonstrating a significant increase of NRP-1 expression in situ in the bone marrow of a large cohort of patients with newly diagnosed AML. The here described expression of the VEGF receptor NRP-1 is in line with several others reporting an increased expression of angiogenic factors in AML (reviewed by Moehler et al.14). MVD and VEGFR-2 expression have been reported to decline following antileukemic treatment,1,2 whereas bFGF promotes autocrine growth stimulation of myeloid leukemic blasts.3 Furthermore, we demonstrate a significant association of the expression of NRP-1 with survival in a large group of AML patients treated within trials by the German AML Cooperative Group.12 Patients with a high NRP-1 expression had significant worse overall survival than patients with a low NRP-1 expression. Only some other angiogenic factors have already been identified as independent prognostic factors in AML as angiopoietin-2 expression, cellular and plasma VEGF levels and the ratio of soluble VEGF/soluble FMS-like tyrosine kinase 1 (FLt1).5,15–17 Although the prognostic impact of NRP-1 expression in AML was unknown so far, the prognostic value of NRP-1 expression in solid tumors has already been investigated by others. Preserved expression of NRP-1 contributes to a better prognosis in colon cancer,18 whereas overexpression of NRP-1 correlates with poor prognosis in human glioma.19 In non-smallcell lung cancer (NSCLC), co-expression of NRP-1 and NRP-2 is associated with poorer prognosis.20 Nevertheless, the reason why relapse-free survival in our patients was not significantly associated with NRP-1 expression status has to be considered critically. An explanation might be that the sample size in our cohort was too small to detect significant differences. Moreover, other prognostic factors in AML as cytogenetics, subtype, overexpression of growth factors and comorbidity between both groups have to be taken into account. However, the function of NRP-1 overexpression in AML is not well understood. In the angiogenesis pathway, NRP-1 functions as a VEGF co-receptor on endothelial cells enhancing VEGFR-2mediated angiogenic stimuli.21 NRP-1 expression might there-

fore contribute to the angiogenic phenotype in AML. In vivo overexpression of NRP-1 in rat prostate cancer resulted in enlarged tumor growth with increased MVD, dilated blood vessels, decreased numbers of apoptotic tumor cells and upregulation of VEGF expression.22 Derived from this observation, one hypothesis to explain the mechanism of NRP-1 in AML is that NRP1 enhances storage and bioavailability of VEGF within the tumor microenvironment, thereby enhancing tumor angiogenesis in situ. Yet, the lack of correlation between NRP-1 and the expression of VEGF, VEGFR-2 and MVD in our report might argue against this assumption. However, NRP-1 expression on hematopoietic cells in fetal liver was described as a source of vascular development enhancing vasculogenesis and angiogenesis.9 Thus, NRP-1 function might be similar in leukemia and the above named lack of correlation might be attributable to further functions of NRP-1 in AML. Furthermore, NRP-1 expression was also not associated with the expression of VEGF and other VEGFRs in human glioma and in NSCLC.19,20 Also a juxtracrine interaction between NRP-1-expressing blasts and neighboring bone marrow cells has to be considered. Indeed, NRP-1 was already identified as a receptor on bone marrow stromal cells and incubation of bone marrow stromal cells with VEGF led to increased expression of thrombopoietin and FLt3ligand in these cells.7 Recent reports have shown that VEGFdependent autocrine and paracrine loops regulate leukemia survival.23 In accordance with the observation that NRP-1 regulates VEGF-mediated survival in breast carcinoma cells lacking other VEGFR expression,24 NRP-1 could also act as a VEGFR on AML blasts contributing to leukemic blast survival. Additional support of this hypothesis comes from a study by Fons et al.10 who reported that NRP-1 – together with VEGFR-2 – acts as a receptor on human bone marrow progenitor cells mediating VEGF-induced differentiation and proliferation, an effect that hypothetically might be translatable to leukemic stem cells. Besides their function as VEGFRs, NRPs serve as co-receptors with plexins for class 3 Sema. Tordjman et al.7 already described the expression of high levels of Sema3B, Sema3C and Sema3F on hematopoietic progenitor cells. The function of NRPs in tumorigenesis has usually been ascribed to their interactions with VEGF, but mounting evidence suggests that class 3 Sema may also be involved in angiogenesis and tumor growth, but in an inhibitory manner.6 Sema3F, for example, was identified as a potent inhibitor of metastasis.25 Hence, hypothetically, NRP-1 might serve as a receptor for class 3 Sema on leukemic blasts. Through a diminished secretion of inhibitory class 3 Sema, leukemic blast proliferation and/or survival could be increased. To support this hypothesis, a diminished expression of Sema3 in AML patients was already reported by Wang et al.26 Furthermore, VEGF and Sema3A have overlapping binding domains on NRP-1 and thus are competitive inhibitors of one another.6 A predominance of VEGF could thus lead to further enhancement of survival and proliferation of the leukemic blast population via NRP-1. A potential role of anti-NRP-1 treatment in AML has already been investigated in a murine chloroma model.11 Treatment with soluble NRP-1, a naturally occurring protein corresponding to the NRP-1 a and b domains, inhibited tumor angiogenesis and growth; however, besides its possible role as NRP-1 antagonist, soluble NRP-1 also has to be considered as a direct VEGF-inhibitor. In summary, we have demonstrated that AML is associated with an increased expression of NRP-1 in the bone marrow. Furthermore, a significant association between survival and NRP-1 expression was detected. Thus, our results support the Leukemia


1954 hypothesis of an important role for NRP-1 in AML, so that targeting NRP-1 could constitute a novel strategy in the treatment of AML. Future studies should focus on the pathomechanism of NRP-1 in AML.

12

Acknowledgements We are grateful to Mrs Margret Lindermann, Department of Medicine/Hematology and Oncology for her excellent technical support, to Dr Joachim Gerss, Coordinating Centre for Clinical Trials for his statistical advice and to Dr Horst Buerger, GerhardDomagk Institute of Pathology, University of Muenster, Germany. This work was supported by grants from the ‘Deutsche Forschungsgemeinschaft’ (Me 950/3-2) and the program ‘Innovative Medizinische Forschung’ (KR 110303) of the Medical Faculty at the University of Muenster, Germany.

References 1 Padro T, Ruiz S, Bieker R, Bu¨rger H, Steins M, Kienast J et al. Increased angiogenesis in the bone marrow of patients with acute myeloid leukemia. Blood 2000; 95: 2637–2644. 2 Padro T, Bieker R, Ruiz S, Steins M, Bu¨rger H, Bu¨chner T et al. Overexpression of vascular endothelial growth factor (VEGF) and its cellular receptor KDR (VEGFR-2) in the bone marrow of patients with acute myeloid leukemia. Leukemia 2002; 16: 1302–1310. 3 Bieker R, Padro T, Kramer J, Steins M, Kessler T, Retzlaff S et al. Overexpression of basic fibroblast growth factor and autocrine stimulation in acute myeloid leukemia. Cancer Res 2003; 63: 7241–7246. 4 Hussong JW, Rodgers GM, Shami PJ. Evidence of increased angiogenesis in patients with acute myeloid leukemia. Blood 2000; 95: 309–313. 5 Loges S, Heil G, Bruweleit M, Schoder V, Butzal M, Fischer U et al. Analysis of concerted expression of angiogenic growth factors in acute myeloid leukemia: expression of angiopoietin-2 represents an independent prognostic factor for overall survival. J Clin Oncol 2005; 23: 1109–1117. 6 Bielenberg DR, Pettaway CA, Takashima S, Klagsbrun M. Neuropilins in neoplasms: expression, regulation, and function. Exp Cell Res 2006; 312: 584–593. 7 Tordjman R, Ortega N, Coulombel L, Plouet J, Romeo PH, Lemarchandel V. Neuropilin-1 is expressed on bone marrow stromal cells: a novel interaction with hematopoietic cells ? Blood 1999; 94: 2301–2309. 8 Belaid Z, Hubint F, Humblet C, Boniver J, Nusgens B, Defresne MP. Differential expression of vascular endothelial factor and its receptors in hematopoietic and fatty bone marrow: evidence that neuropilin-1 is produced by fat cells. Haematologica 2005; 90: 400–401. 9 Yamada Y, Oike Y, Ogawa H, Ito Y, Fujisawa H, Suda T et al. Neuropilin-1 on hematopoietic cells as a source of vascular development. Blood 2003; 101: 1801–1809. 10 Fons P, Herault JP, Delesque N, Tuyaret J, Bono F, Herbert JM. VEGF-R2 and neuropilin-1 are involved in VEGF-A induced differentiation of human bone marrow progenitor cells. J Cell Physiol 2004; 200: 351–359. 11 Schuch G, Machluf M, Bartsch G, Nomi M, Richard H, Atala A et al. In vivo administration of vascular endothelial growth factor (VEGF) and its antagonist, soluble neuropilin-1, predicts a role of

Leukemia

13 14 15 16

17

18 19 20

21

22 23

24

25

26

27

VEGF in the progression of acute myeloid leukemia in vivo. Blood 2002; 100: 4622–4628. Buchner T, Hiddemann W, Wormann B, Loffler H, Gassmann W, Haferlach T et al. Double induction strategy for acute myeloid leukemia: the effect of high-dose cytarabine with mitoxantrone instead of standard-dose cytarabine with daunorubicin and 6-thioguanine: a randomized trial by the German AML Cooperative Group. Blood 1999; 93: 4116–4124. Harper SJ, Xing CY, Whittle C, Parry R, Gillatt D, Peat D et al. Expression of neuropilin-1 by human glomerular epithelial cells in vitro and in vivo. Clin Science 2001; 101: 439–446. Moehler TM, Ho AD, Goldschmidt H, Barlogie B. Angiogenesis in hematologic malignancies. Crit Rev Oncol Hematol 2003; 45: 227–244. Aref S, El Sherbina M, Goda T, Fouda M, Al Askalany H, Abdalla D. Soluble VEGF/sFLt1 ratio is an independent predictor of AML patient out come. Hematology 2005; 10: 131–134. Aguayo A, Estey EH, Kantarjian HM, Manshouri T, Gidel C, Keating M et al. Cellular vascular endothelial growth factor is a predictor of outcome in patients with acute myeloid leukaemia. Blood 1999; 94: 3717–3721. Aguayo A, Kantarjian HM, Estey EH, Giles FJ, Verstovsek S, Manshouri T et al. Plasma vascular endothelial growth factor levels have prognostic significance in patients with acute myeloid leukemia but not in patients with myelodysplastic syndromes. Cancer 2002; 95: 1923–1930. Kamiya T, Kawakami T, Abe Y, Nishi M, Onoda N, Miyazaki N et al. The preserved expression of neuropilin 1 contributes to a better prognosis in colon cancer. Oncol Rep 2006; 15: 369–373. Osada H, Tokunaga T, Nishi M, Hatanaka H, Abe Y, Tsugu A et al. Overexpression of the neuropilin 1 gene correlated with poor prognosis in human glioma. Anticancer Res 2004; 24: 547–552. Kawakami T, Tokunaga T, Hatanaka H, Kijima H, Yamazaki H, Abe Y et al. Neuropilin 1 and neuropilin 2 co-expression is significantly correlated with increased vascularity and poor prognosis in nonsmall cell lung carcinoma. Cancer 2002; 95: 2196–2201. Soker S, Miao HQ, Nomi M, Takashima S, Klagsbrun M. VEGF165 mediates formation of complexes containing VEGFR-2 and neuropilin-1 that enhance VEGF165-receptor binding. J Cell Biochem 2002; 85: 357–368. Miao HQ, Lee P, Lin H, Soker S, Klagsbrun M. Neuropilin-1 expression by tumor cells promotes tumor angiogenesis and progression. FASEB J 2000; 14: 2532–2539. Dias S, Hattori K, Heissig B, Zhu Z, Wu Y, Witte L et al. Inhibition of both paracrine and autocrine VEGF/VEGFR-2 signaling pathways is essential to induce long-term remission of xenotransplanted human leukemias. Proc Natl Acad Sci USA 2001; 98: 10857–10862. Bachelder RE, Crago A, Chung J, Wendt MA, Shaw LM, Robinson G et al. Vascular endothelial growth factor is an autocrine survival factor for neuropilin-expressing breast carcinoma cells. Cancer Res 2001; 61: 5736–5740. Bielenberg DR, Hida Y, Shimizu A, Kaipainen A, Kreuter M, Kim CC et al. Semaphorin 3F, a chemorepulsant for endothelial cells, induces a poorly vascularized, encapsulated, nonmetastastic tumor phenotype. J Clin Invest 2004; 114: 1260–1271. Wang Z, Li Y, Wu X, Cheng S, Yang L, Wu Y. KDR and Sema3 genes expression in bone marrow stromal cells and hematopoietic cells from leukaemia patients and normal individuals. Hematology 2005; 10: 307–312. Bennet JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, Granlnick HR et al. Proposed revised criteria for the classification of acute myeloid leukemia. A report of the French–American– British Cooperative Group. Ann Intern Med 1985; 103: 620–626.