Analysis of Concerted Expression of Angiogenic Growth Factors in Acute Myeloid Leukemia: Expression of Angiopoietin-2 Represents an Independ
http://www.100md.com
▲還散笫雖悝◎
the Medizinische Klinik II, University Hospital Hamburg-Eppendorf
Institute of Medical Biometry and Epidemiology, University Hospital Hamburg-Eppendorf, Hamburg
Department of Oncology/Hematology, Medical University Hannover, Hannover, Germany
ABSTRACT
PATIENTS AND METHODS: We investigated the expression of VEGF-A, VEGF-C, angiopoietin-1 (Ang1), angiopoietin-2 (Ang2), and the receptor Tie2 by quantitative polymerase chain reaction in a cohort of 90 patients younger than 61 years with de novo AML entered into the German AML S邦ddeutsche Hmoblastose Gruppe Hannover 95 trial. Uni- and multivariate analyses were performed using clinical and gene expression variables.
RESULTS: Univariate analysis of overall survival indicated the following variables as prognostic factors: good response on a day-15 bone marrow examination after initiation of induction chemotherapy, karyotype, and high Ang2 expression. In multivariate analysis, only bad response and log Ang2 expression remained of statistical significance, with a hazard ratio of 3.51 (95% CI, 1.91 to 6.47) and 0.75 (95% CI, 0.61 to 0.91), respectively. Subgroup analysis suggested that the prognostic impact of Ang2 expression was especially evident in cohorts with low VEGF-C and Ang1 mRNA levels.
CONCLUSION: These results show that expression of Ang2 represents an independent prognostic factor in AML. Additional research into interactions of angiogenic cytokines in the pathogenesis of bone marrow angiogenesis in AML is warranted.
INTRODUCTION
In AML, bone marrow neoangiogenesis is promoted by growth factors released by leukemic blasts. Our group and others demonstrated that constitutive secretion of VEGF-A by AML blasts occurred in 71% of analyzed patients.17,18 Aguayo et al19 showed that VEGF-A expression represents a negative prognostic factor in patients with AML. Furthermore, we reported the expression of VEGF-C and its receptor Flt-4 by leukemic cells in 61% and 36% of AML patients, respectively.20 Dias et al21 recently found that VEGF-C released from the bone marrow endothelium induced proliferation, promoted survival, and protected leukemic cells from chemotherapy-induced apoptosis.
Expression of angiopoietins and their receptor Tie2 by AML cells was studied by two groups. Muller et al22 described RNA expression of Ang1 and Tie2 in 11 of 17 patients with acute or chronic myeloid leukemia. Watarai et al23 reported a significantly higher expression of Ang2 by CD7+ AML blasts compared with those with the translocation t(8,21), suggesting a cell- and karyotype-specific expression.
Given that neoangiogenesis depends on the interplay of different members of the VEGF and angiopoietin family, it is mandatory to determine their expression in the same pretreatment samples. In this study, we investigated the concerted expression of VEGF-A, VEGF-C, Ang1, Ang2, and Tie2 by quantitative polymerase chain reaction (PCR) in 90 patients younger than 61 years of age with de novo AML treated according to a uniform chemotherapy protocol. Uni- and multivariate analyses were performed to identify the prognostic impact of the expression of angiogenic factors on the long-term prognosis of AML patients.
PATIENTS AND METHODS
The study was approved by the ethics committees at each study site and was conducted in accordance with German drug development regulations and the Declaration of Helsinki.
Isolation of PBMCs and Preparation of AC133+ Cells From Leukapheresis Products
Mononuclear cells from four healthy donors were isolated using Ficoll-Hypaque density gradient centrifugation. One milliliter of a leukapheresis product from granulocyte colony-stimulating factor-primed patients with nonmyeloid malignancies was obtained. Patients had given prior informed consent according to German laws. Selection of AC133+ cells was performed using the Auto-MACS system (Miltenyi Biotech, Bergisch Gladbach, Germany). Purity of isolated cells was at least 95%.
Isolation of RNA and Synthesis of cDNA
Total cellular RNA from 1 x 107 patient cells (2 x 106 normal donor cells) was extracted using the Trizol method (Invitrogen, Carlsbad, CA) and reverse transcribed using MuLV-RT (Invitrogen). cDNA was stored at 每20∼C.
Primers for Real-Time PCR
All primers were designed with the Primer 3 Software (Whitehead Institute for Biomedical Research, Boston, MA).25 The PCR product of the glycerinaldehyde-3-phosphate dehydrogenase (GAPDH) spanned intron H of the GAPDH gene; thus, the larger genomic fragment could be detected by melting-curve analysis. Contamination with genomic DNA was not detected in any of the analyzed samples.
The VEGF primer was constructed to amplify the three splice variants of VEGF (VEGF165, VEGF121, and VEGF18926). The Ang1 primers were designed to amplify all known splice variants of Ang1.27 All primer sets had an efficiency of at least 1.81 determined by log dilutions of plasmid standards.
Real-Time Quantitative PCR
Real-time quantitative PCR (qPCR) was carried out on the Light Cycler (Roche, Basel, Switzerland) using the FAST Start DNAMaster SYBR Green Kit (Roche). The relative amount of expressed cDNA was calculated from a standard curve obtained by using log dilutions of plasmids containing the gene of interest. Plasmids were constructed by cloning of amplification products into the pCRII Vector using the TA-Cloning Kit (Invitrogen). All recombinant DNA work was done in an S1 facility after approval according to German law.
Results of two independent analyses for each gene and sample or plasmid dilution were averaged. The calculated amount of the target genes was normalized to the housekeeping gene GAPDH. All data are presented as ratio of the target gene/GAPDH. Primer sequences and PCR conditions are available on request.
Cell Lines and Ang2 Protein Determination
Leukemic cell lines TF-1, Mo7e, HL60, U937, and K562 (DKMZ, Braunschweig, Germany) were seeded at a density of 1 million cells/mL. After 3 days of culture, supernatants were harvested for determination of Ang2 protein by enzyme-linked immunosorbent assay (ELISA; R&D Systems, Wiesbaden, Germany). Cells were pelleted, cDNA was prepared, and qPCR was performed for Ang2 and GAPDH as described above.
Statistical Methods
Data are presented as median and range for continuous variables and as count and percentage for categoric variables. Groups of patients with and without remission were compared using Fisher's exact test for categorized variables and using Mann-Whitney U test for continuous variables. For comparison of gene expression of leukemic blasts with normal AC133+ and PBMCs, the Kruskal-Wallis test was used.
As primary end points for survival analysis, we defined overall survival (starting from time of diagnosis) and relapse-free survival after remission (including only patients with remission, starting from remission time and defining death or relapse as an event). To investigate the relationship between gene expression variables, Spearman's {rho} was computed.
Univariate survival analysis included Kaplan-Meier curves and log-rank tests. In addition, an multiple Cox proportional hazards regression analysis was performed (including construction of Martingale residual plots and verification of the proportional hazards assumption).28,29 All statistical analyses were done with SPSS 11.0 (SPSS Inc, Chicago, IL) and S-Plus 6.0 (Insightful Corp, Seattle, WA).
RESULTS
Samples from 90 AML patients, AC133+ cells from peripheral blood from seven different leukapheresis products, and PBMCs from four healthy donors were analyzed by real-time qPCR. We quantified the expression of VEGF-A, VEGF-C, Ang1, Ang2, and Tie2 as a ratio with the housekeeping gene GAPDH (Fig 1; Table 2). Median levels of VEGF-A, VEGF-C, Ang1, and Ang2 were higher in leukemic blasts than in normal peripheral-blood AC133+ cells. With the exception of VEGF-C, expression of analyzed angiogenic factors was low in normal PBMCs.
To investigate the relation between Ang2 mRNA levels and protein secretion, we determined concomitantly the expression of Ang2 in five human leukemic cell lines (TF-1, Mo7e, HL60, U937, and K562) by qPCR and by ELISA of cell culture supernatants. A correlation of 0.8 between mRNA and protein expression was found.
Because of the wide range of individual data points, we used logarithms of values shifted by 1 for further analysis and graphical display. mRNA specific for VEGF-A was detected in 90%, mRNA specific for VEGF-C was detected in 73%, mRNA specific for Ang1 was detected in 96%, mRNA specific for Ang2 was detected in 98%, and mRNA specific for Tie2 was detected in 60% of patients.
We performed univariate analyses to correlate gene expression with confounding factors age, sex, karyotype, and response on day 15 after induction chemotherapy with long-term outcome. Patients with less than 5% blasts in hypoplastic bone marrows 15 days after initiation of chemotherapy without evidence of extramedullary disease were considered as good responders. The karyotype of patients was initially categorized into three risk groups: t(8;21), inv(16); normal karyotype; and trisomy 8, abnormalities of chromosome 5, 7, 11, or complex abnormalities. Although observed CR rates varied among these groups with 90.9%, 73.5%, and 60.7%, respectively, we could not detect a statistically significant difference. Because in our cohort no significant survival difference was detected between patients with core-binding factor leukemias (n = 11) and those with normal karyotype (n = 50), both groups were combined for additional analysis. Comparable outcomes for patients with CBF leukemias and those with normal karyotype were also found in the entire Hannover 95 study cohort.24 For additional analysis, patients were divided into three groups of approximately equal size (VEGF-A < 3, 3 to 20, and > 20; Ang1 < 0.5, 0.5 to 1.5, and > 1.5; Ang2 < 2, 2 to 15, and >15). Given that only 60% and 73% of patients expressed VEGF-C and Tie2 respectively, patients were analyzed in two groups: zero and more than zero.
First we analyzed whether a correlation existed between gene expression levels of individual angiogenic factors and blast count, age, sex, karyotype, French-American-British classification type, and good or bad response to induction chemotherapy; no association was detected. In particular, no relevant correlation between absolute or relative blast count in peripheral blood and expression of angiogenic factors could be detected (absolute value of Spearman's r < 0.3 for all angiogenic factors and particularly r < 0.1 for Ang2). In addition, for Ang2, mean blast counts were not different among our specified groups.
Univariate analysis was then performed to identify factors that could predict the achievement of a CR. In our patient group, the only variable that correlated with obtaining a CR was good response to induction chemotherapy assessed on day 15 (Fisher's exact test, P < .001). Although the age of the entire group of patients was younger than 61 years at diagnosis, there was a trend that younger patients had higher CR rates (P = .059). No impact of the variables sex (P = .64), karyotype (P = .1), VEGF-A (P = .59), VEGF-C (P = .42), Ang1 (P = .08), Ang2 (P = .28), and Tie2 (P = 1.0) was found.
Univariate analysis of factors associated with OS showed a better prognosis for patients with good response to induction chemotherapy (log-rank P < .001), favorable karyotype (P = .07), and high Ang2 expression (P = .005). We could not detect a prognostic impact on survival for the factors sex (P = .52), age (P = .69), VEGF-A (P = .72), VEGF-C (P = .32), Ang1 (P = .88), and Tie2 (P = .43). Results of univariate analysis are listed in Table 3. Figure 2 shows the Kaplan-Meier curves for survival stratified for good response and for Ang2. It should be noted that patients were grouped into cohorts of approximately equal size to enable us to illustrate the results by Kaplan-Meier curves. Because Cox regression and Martingale plots revealed a linear relation between log Ang2 and risk (described in this section), no clinically relevant cut-offs could be identified and arbitrary cut-offs were chosen.
We used the Cox proportional hazards model based on all 90 patients to perform multivariate analyses of the variables with a P value of less than .1 (karyotype, good or bad response, and Ang2 expression) on overall survival (stepwise procedure). In this analysis, log Ang2 was used as a continuous predictor variable. We could identify Ang2 expression and good versus bad response as independent prognostic factors for survival (Table 4). Response on day 15 was closely related to karyotype, as can be deduced by cross-table analysis (P = .005). In our cohort among 64 patients with good response, 83% possessed a favorable karyotype, whereas among the patients with bad response only 50% had favorable cytogenetics. Therefore, karyotype lost its prognostic significance in multivariate analysis in our cohort.
The Martingale residual plot after inclusion of the variable response indicates that the effect of Ang2 on survival could be adequately modeled by a linear term on the log scale (compare Loess smoothing in Fig 3).
The resulting hazard ratio for bad versus good response was 3.51 (95% CI, 1.91 to 6.47; P < .001) and the hazard ratio for log Ang2 was 0.75 (95% CI, 0.61 to 0.91; P = .005). Karyotype did not show an additional independent effect on survival in multivariate analysis (P = .19). Results of the Cox regression therefore coincide with results of univariate analysis (compare with Fig 2).
The Martingale residual plot might also suggest the presence of four outliers. However, because the omission of these four patients (all of them had a poor response and did not die during time of investigation, probably due to allogeneic bone marrow transplantation in three of them) did not change coefficients and SEs in a noteworthy way, we chose to continue using the model including all 90 patients.
In addition, we investigated the adequacy of the proportional hazards assumption of the final model by fitting a model including a time-dependent covariate for each variable. P values of .48 and .49 for these coefficients suggested no relevant time dependency.
Furthermore, data were reanalyzed after censoring patients with allogeneic bone marrow transplantation at the time point of stem-cell infusion. No substantial difference in results was found. In univariate analysis karyotype seemed to be of more pronounced prognostic value (P = .0039). However, in multivariate analysis there was again no additional significant effect of karyotype (P = .19), which is consistent with the original model without censoring of patients who received bone marrow transplantation.
To further illustrate the impact of good and bad response and level of Ang2 expression on overall survival, Kaplan-Meier plots of patients with good and bad response stratified for Ang2 expression in three groups are shown in Figure 4.
To investigate possible interactions between the expression of various angiogenic growth factors on patient survival, subgroup analysis was performed. Patients were divided into groups with expression of VEGF-A, VEGF-C, and Ang1 below or above the median. Survival estimates were calculated for each subgroup according to the level of Ang2 expression. Kaplan-Meier survival curves for all subgroups are shown in Figure 5. The prognostic relevance of Ang2 expression was most pronounced in the subgroups with low VEGF-C and Ang1 mRNA levels. In the cohorts with high VEGF-C, VEGF-A, and Ang1, Ang2 expression lost its impact on OS, suggesting that the protective effect of Ang2 is diminished by simultaneous expression of proangiogenic factors.
Statistical analysis of RFS was nonyielding, given that 26 patients did not achieve a complete remission. Because of low patient numbers and consequently lack of statistical power, no additional analysis was performed for RFS.
DISCUSSION
Although RFS of the entire group was slightly better than those of analyzed patients, no significant difference between analysis and nonanalysis groups could be found. One limitation of the study is that no plasma samples of the study patients were available; thus, analysis was restricted to investigation of mRNA expression of angiogenic factors. Therefore, exact data on protein concentrations in blood are unknown. However, given that no post-transcriptional regulation of these factors has been described, concerted mRNA and protein expression of these factors can be supposed. Furthermore, mRNA and protein expression of Ang2 by qPCR and ELISA from five human leukemic cell lines was determined. A positive correlation was found, supporting the validity of the use of cDNA for our investigation.
Multivariate analysis revealed that expression of Ang2 and a good response on a day-15 postinduction chemotherapy bone marrow sample were the only independent predictors of OS in our patient cohort. The prognostic impact of persistent leukemia after induction chemotherapy has been confirmed by several groups.35每37 Because bad response to induction chemotherapy is correlated to adverse karyotypes, both groups of patients overlap to a high degree. Therefore, it was not possible to detect an independent prognostic value of karyotype in our cohort. Because of the small number of patients with specific chromosome aberrations, patients were grouped together into only two prognostic cohorts. It cannot be excluded by our analysis that individual translocations may result in a specific risk profile.
The role of Ang2 in angiogenesis is still controversial. In initial studies, Ang2 has been shown to block the effect of Ang1 on endothelial cells in vitro. Furthermore, Ang2 competes with Ang1 for binding to their common receptor Tie2, and acts as a competitive inhibitor of Ang1. Transgenic overexpression of Ang2 resulted in insufficient vascular remodeling, as in Ang1 or Tie2 knockout mice, indicating that Ang2 could act as an antagonist of Ang1 in vivo.13 Furthermore, Lin et al38 demonstrated that neutralization of the effects of Ang1 by soluble Tie2 blocked angiogenesis and tumor growth in a mouse model. Recently, the role of Ang2 as a pure inhibitor of angiogenesis has been challenged. Emphasis has now been put on the concerted action of Ang2 and VEGF-A on endothelial cells. In several experimental models, Ang2 (in the absence of VEGF-A) led to endothelial cell apoptosis and vessel regression. In the presence of VEGF-A, Ang2 promoted endothelial cell proliferation and migration, thereby acting as a proangiogenic agent.14每16,39每41 The dependence of the effects of Ang2 by VEGF-A is also suggested by our results. Although Ang2 expression was an independent predictor of OS for the entire cohort, subgroup analysis indicated that the prognosis of patients with VEGF-A (and especially VEGF-C and Ang1) expression below the median was most dependent on the level of Ang2 production. In particular, patients with high Ang2 and low VEGF-C expression had an excellent long-term prognosis. On the other hand, in patients with high VEGF-C levels, prognosis was much less influenced by Ang2. Unfortunately, in our study no corresponding bone marrow biopsies were available; therefore, we could not directly correlate bone marrow angiogenesis with expression of VEGF-A, VEGF-C, and Ang2.
Recently, several smaller studies have been published in which Ang2 expression was found in a variety of solid tumors.32,42,43 Ang2 could be detected in tumor cells and endothelial cells of invading tumor vessels in subgroups of patients. In the largest of these studies, expression of VEGF-A, Ang1, and Ang2 was analyzed in 236 patients with localized non-small-cell lung cancer. This investigation clearly showed that concerted expression of VEGF-A and Ang2 resulted in increased microvessel density and adverse long-term prognosis, supporting the model of cooperative effects of both types of growth factors.43
On the basis of the concept of cooperation between both classes of endothelial-specific growth factors, additional clinical and experimental research should be done to gain further insights into the dependence of neoangiogenesis on the interaction of Ang2 and VEGF-A or VEGF-C expression. If substantiated, therapeutic strategies directed against the action of VEGF-A and VEGF-C with simultaneous administration of Ang2 may be explored in experimental models to investigate enhanced antiangiogenic and antileukemic efficacy.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank the members of the AML SHG Hannover study group for providing samples.
NOTES
Supported by a grant from Werner-Otto Stiftung Hamburg, Germany (S.L.), and a grant from Eppendorfer: Leukmiehilfe Hamburg and Roggenbuck Stiftung, Hamburg, Germany (U.M.G.).
Presented in part at the annual meeting of the American Society of Hematology in Philadelphia, PA, December 6-10, 2002.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Folkman J: Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1:27每31, 1995
2. Risau W: Mechanisms of angiogenesis. Nature 386:671每674, 1997
3. Fiedler W, Staib P, Kuse R, et al: Role of angiogenesis inhibitors in acute myeloid leukemia. Cancer J 7:S129每S133, 2001 (suppl 3)
4. Aigner A, Butscheid M, Kunkel P, et al: An FGF-binding protein (FGF-BP) exerts its biological function by parallel paracrine stimulation of tumor cell and endothelial cell proliferation through FGF-2 release. Int J Cancer 92:510每517, 2001
5. Berger W, Setinek U, Mohr T, et al: Evidence for a role of FGF-2 and FGF receptors in the proliferation of non-small cell lung cancer cells. Int J Cancer 83:415每423, 1999
6. Strutz F, Zeisberg M, Renziehausen A, et al: TGF-beta 1 induces proliferation in human renal fibroblasts via induction of basic fibroblast growth factor (FGF-2). Kidney Int 59:579每592, 2001
7. Mustonen T, Alitalo K: Endothelial receptor tyrosine kinases involved in angiogenesis. J Cell Biol 129:895每898, 1995
8. Davis S, Aldrich TH, Jones PF, et al: Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by secretion-trap expression cloning. Cell 87:1161每1169, 1996
9. Davis S, Yancopoulos GD: The angiopoietins: Yin and yang in angiogenesis. Curr Top Microbiol Immunol 237:173每185, 1999
10. Valenzuela DM, Griffiths JA, Rojas J, et al: Angiopoietins 3 and 4: Diverging gene counterparts in mice and humans. Proc Natl Acad Sci U S A 96:1904每1909, 1999
11. Sato TN, Tozawa Y, Deutsch U, et al: Distinct roles of the receptor tyrosine kinases Tie-1 and Tie-2 in blood vessel formation. Nature 376:70每74, 1995
12. Korpelainen EI, Alitalo K: Signaling angiogenesis and lymphangiogenesis. Curr Opin Cell Biol 10:159每164, 1998
13. Maisonpierre PC, Suri C, Jones PF, et al: Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis. Science 277:55每60, 1997
14. Lobov IB, Brooks PC, Lang RA: From the cover: Angiopoietin-2 displays VEGF-dependent modulation of capillary structure and endothelial cell survival in vivo. Proc Natl Acad Sci U S A 99:11205每11210, 2002
15. Holash J, Maisonpierre PC, Compton D, et al: Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF. Science 284:1994每1998, 1999
16. Visconti RP, Richardson CD, Sato TN: Orchestration of angiogenesis and arteriovenous contribution by angiopoietins and vascular endothelial growth factor (VEGF). Proc Natl Acad Sci U S A 99:8219每8224, 2002
17. Fiedler W, Graeven U, Ergun S, et al: Vascular endothelial growth factor, a possible paracrine growth factor in human acute myeloid leukemia. Blood 89:1870每1875, 1997
18. Bellamy WT, Richter L, Frutiger Y, et al: Expression of vascular endothelial growth factor and its receptors in hematopoietic malignancies. Cancer Res 59:728每733, 1999
19. Aguayo A, Estey E, Kantarjian H, et al: Cellular vascular endothelial growth factor is a predictor of outcome in patients with acute myeloid leukemia. Blood 94:3717每3721, 1999
20. Fiedler W, Graeven U, Ergun S, et al: Expression of FLT4 and its ligand VEGF-C in acute myeloid leukemia. Leukemia 11:1234每1237, 1997
21. Dias S, Choy M, Alitalo K, et al: Vascular endothelial growth factor (VEGF)-C signaling through FLT-4 (VEGFR-3) mediates leukemic cell proliferation, survival, and resistance to chemotherapy. Blood 99:2179每2184, 2002
22. Muller A, Lange K, Gaiser T, et al: Expression of angiopoietin-1 and its receptor TEK in hematopoietic cells from patients with myeloid leukemia. Leuk Res 26:163每168, 2002
23. Watarai M, Miwa H, Shikami M, et al: Expression of endothelial cell-associated molecules in AML cells. Leukemia 16:112每119, 2002
24. Heil G, Krauter J, Raghavachar A, et al: Risk-adapted induction and consolidation therapy in adults with de novo AML aged
25. Rozen S, Skaletsky H: Primer3 on the WWW for general users and biologist programmers, in Krawetz S, Misener S (eds): Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa, NJ, Humana Press, 2000, pp 365每386
26. Tischer E, Mitchell R, Hartman T, et al: The human gene for vascular endothelial growth factor: Multiple protein forms are encoded through alternative exon splicing. J Biol Chem 266:11947每11954, 1991
27. Huang YQ, Li JJ, Karpatkin S: Identification of a family of alternatively spliced mRNA species of angiopoietin-1. Blood 95:1993每1999, 2000
28. Therneau T, Grambsch P, Fleming T: Martingale-based residuals for survival models. Biometrika 77:147每160, 1990
29. Grambsch PM: Goodness-of-fit and diagnostics for proportional hazards regression models. Cancer Treat Res 75:95每112, 1995
30. Aguayo A, Kantarjian H, Manshouri T, et al: Angiogenesis in acute and chronic leukemias and myelodysplastic syndromes. Blood 96:2240每2245, 2000
31. Albitar M: Angiogenesis in acute myeloid leukemia and myelodysplastic syndrome. Acta Haematol 106:170每176, 2001
32. Lee JH, Koh JT, Shin BA, et al: Comparative study of angiostatic and anti-invasive gene expressions as prognostic factors in gastric cancer. Int J Oncol 18:355每361, 2001
33. Verstovsek S, Kantarjian H, Manshouri T, et al: Prognostic significance of cellular vascular endothelial growth factor expression in chronic phase chronic myeloid leukemia. Blood 99:2265每2267, 2002
34. de Bont ES, Rosati S, Jacobs S, et al: Increased bone marrow vascularization in patients with acute myeloid leukaemia: A possible role for vascular endothelial growth factor. Br J Haematol 113:296每304, 2001
35. Kern W, Haferlach T, Schoch C, et al: Early blast clearance by remission induction therapy is a major independent prognostic factor for both achievement of complete remission and long-term outcome in acute myeloid leukemia: Data from the German AML Cooperative Group (AMLCG) 1992 Trial. Blood 101:64每70, 2003
36. Liso V, Albano F, Pastore D, et al: Bone marrow aspirate on the 14th day of induction treatment as a prognostic tool in de novo adult acute myeloid leukemia. Haematologica 85:1285每1290, 2000
37. Wheatley K, Burnett AK, Goldstone AH, et al: A simple, robust, validated and highly predictive index for the determination of risk-directed therapy in acute myeloid leukaemia derived from the MRC AML 10 trial: United Kingdom Medical Research Council's Adult and Childhood Leukaemia Working Parties. Br J Haematol 107:69每79, 1999
38. Lin P, Polverini P, Dewhirst M, et al: Inhibition of tumor angiogenesis using a soluble receptor establishes a role for Tie2 in pathologic vascular growth. J Clin Invest 100:2072每2078, 1997
39. Vajkoczy P, Farhadi M, Gaumann A, et al: Microtumor growth initiates angiogenic sprouting with simultaneous expression of VEGF, VEGF receptor-2, and angiopoietin-2. J Clin Invest 109:777每785, 2002
40. Holash J, Wiegand SJ, Yancopoulos GD: New model of tumor angiogenesis: Dynamic balance between vessel regression and growth mediated by angiopoietins and VEGF. Oncogene 18:5356每5362, 1999
41. Asahara T, Chen D, Takahashi T, et al: Tie2 receptor ligands, angiopoietin-1 and angiopoietin-2, modulate VEGF-induced postnatal neovascularization. Circ Res 83:233每240, 1998
42. Sfiligoi C, de Luca A, Cascone I, et al: Angiopoietin-2 expression in breast cancer correlates with lymph node invasion and short survival. Int J Cancer 103:466每474, 2003
43. Tanaka F, Ishikawa S, Yanagihara K, et al: Expression of angiopoietins and its clinical significance in non-small cell lung cancer. Cancer Res 62:7124每7129, 2002(Sonja Loges, Gerhard Heil)
Institute of Medical Biometry and Epidemiology, University Hospital Hamburg-Eppendorf, Hamburg
Department of Oncology/Hematology, Medical University Hannover, Hannover, Germany
ABSTRACT
PATIENTS AND METHODS: We investigated the expression of VEGF-A, VEGF-C, angiopoietin-1 (Ang1), angiopoietin-2 (Ang2), and the receptor Tie2 by quantitative polymerase chain reaction in a cohort of 90 patients younger than 61 years with de novo AML entered into the German AML S邦ddeutsche Hmoblastose Gruppe Hannover 95 trial. Uni- and multivariate analyses were performed using clinical and gene expression variables.
RESULTS: Univariate analysis of overall survival indicated the following variables as prognostic factors: good response on a day-15 bone marrow examination after initiation of induction chemotherapy, karyotype, and high Ang2 expression. In multivariate analysis, only bad response and log Ang2 expression remained of statistical significance, with a hazard ratio of 3.51 (95% CI, 1.91 to 6.47) and 0.75 (95% CI, 0.61 to 0.91), respectively. Subgroup analysis suggested that the prognostic impact of Ang2 expression was especially evident in cohorts with low VEGF-C and Ang1 mRNA levels.
CONCLUSION: These results show that expression of Ang2 represents an independent prognostic factor in AML. Additional research into interactions of angiogenic cytokines in the pathogenesis of bone marrow angiogenesis in AML is warranted.
INTRODUCTION
In AML, bone marrow neoangiogenesis is promoted by growth factors released by leukemic blasts. Our group and others demonstrated that constitutive secretion of VEGF-A by AML blasts occurred in 71% of analyzed patients.17,18 Aguayo et al19 showed that VEGF-A expression represents a negative prognostic factor in patients with AML. Furthermore, we reported the expression of VEGF-C and its receptor Flt-4 by leukemic cells in 61% and 36% of AML patients, respectively.20 Dias et al21 recently found that VEGF-C released from the bone marrow endothelium induced proliferation, promoted survival, and protected leukemic cells from chemotherapy-induced apoptosis.
Expression of angiopoietins and their receptor Tie2 by AML cells was studied by two groups. Muller et al22 described RNA expression of Ang1 and Tie2 in 11 of 17 patients with acute or chronic myeloid leukemia. Watarai et al23 reported a significantly higher expression of Ang2 by CD7+ AML blasts compared with those with the translocation t(8,21), suggesting a cell- and karyotype-specific expression.
Given that neoangiogenesis depends on the interplay of different members of the VEGF and angiopoietin family, it is mandatory to determine their expression in the same pretreatment samples. In this study, we investigated the concerted expression of VEGF-A, VEGF-C, Ang1, Ang2, and Tie2 by quantitative polymerase chain reaction (PCR) in 90 patients younger than 61 years of age with de novo AML treated according to a uniform chemotherapy protocol. Uni- and multivariate analyses were performed to identify the prognostic impact of the expression of angiogenic factors on the long-term prognosis of AML patients.
PATIENTS AND METHODS
The study was approved by the ethics committees at each study site and was conducted in accordance with German drug development regulations and the Declaration of Helsinki.
Isolation of PBMCs and Preparation of AC133+ Cells From Leukapheresis Products
Mononuclear cells from four healthy donors were isolated using Ficoll-Hypaque density gradient centrifugation. One milliliter of a leukapheresis product from granulocyte colony-stimulating factor-primed patients with nonmyeloid malignancies was obtained. Patients had given prior informed consent according to German laws. Selection of AC133+ cells was performed using the Auto-MACS system (Miltenyi Biotech, Bergisch Gladbach, Germany). Purity of isolated cells was at least 95%.
Isolation of RNA and Synthesis of cDNA
Total cellular RNA from 1 x 107 patient cells (2 x 106 normal donor cells) was extracted using the Trizol method (Invitrogen, Carlsbad, CA) and reverse transcribed using MuLV-RT (Invitrogen). cDNA was stored at 每20∼C.
Primers for Real-Time PCR
All primers were designed with the Primer 3 Software (Whitehead Institute for Biomedical Research, Boston, MA).25 The PCR product of the glycerinaldehyde-3-phosphate dehydrogenase (GAPDH) spanned intron H of the GAPDH gene; thus, the larger genomic fragment could be detected by melting-curve analysis. Contamination with genomic DNA was not detected in any of the analyzed samples.
The VEGF primer was constructed to amplify the three splice variants of VEGF (VEGF165, VEGF121, and VEGF18926). The Ang1 primers were designed to amplify all known splice variants of Ang1.27 All primer sets had an efficiency of at least 1.81 determined by log dilutions of plasmid standards.
Real-Time Quantitative PCR
Real-time quantitative PCR (qPCR) was carried out on the Light Cycler (Roche, Basel, Switzerland) using the FAST Start DNAMaster SYBR Green Kit (Roche). The relative amount of expressed cDNA was calculated from a standard curve obtained by using log dilutions of plasmids containing the gene of interest. Plasmids were constructed by cloning of amplification products into the pCRII Vector using the TA-Cloning Kit (Invitrogen). All recombinant DNA work was done in an S1 facility after approval according to German law.
Results of two independent analyses for each gene and sample or plasmid dilution were averaged. The calculated amount of the target genes was normalized to the housekeeping gene GAPDH. All data are presented as ratio of the target gene/GAPDH. Primer sequences and PCR conditions are available on request.
Cell Lines and Ang2 Protein Determination
Leukemic cell lines TF-1, Mo7e, HL60, U937, and K562 (DKMZ, Braunschweig, Germany) were seeded at a density of 1 million cells/mL. After 3 days of culture, supernatants were harvested for determination of Ang2 protein by enzyme-linked immunosorbent assay (ELISA; R&D Systems, Wiesbaden, Germany). Cells were pelleted, cDNA was prepared, and qPCR was performed for Ang2 and GAPDH as described above.
Statistical Methods
Data are presented as median and range for continuous variables and as count and percentage for categoric variables. Groups of patients with and without remission were compared using Fisher's exact test for categorized variables and using Mann-Whitney U test for continuous variables. For comparison of gene expression of leukemic blasts with normal AC133+ and PBMCs, the Kruskal-Wallis test was used.
As primary end points for survival analysis, we defined overall survival (starting from time of diagnosis) and relapse-free survival after remission (including only patients with remission, starting from remission time and defining death or relapse as an event). To investigate the relationship between gene expression variables, Spearman's {rho} was computed.
Univariate survival analysis included Kaplan-Meier curves and log-rank tests. In addition, an multiple Cox proportional hazards regression analysis was performed (including construction of Martingale residual plots and verification of the proportional hazards assumption).28,29 All statistical analyses were done with SPSS 11.0 (SPSS Inc, Chicago, IL) and S-Plus 6.0 (Insightful Corp, Seattle, WA).
RESULTS
Samples from 90 AML patients, AC133+ cells from peripheral blood from seven different leukapheresis products, and PBMCs from four healthy donors were analyzed by real-time qPCR. We quantified the expression of VEGF-A, VEGF-C, Ang1, Ang2, and Tie2 as a ratio with the housekeeping gene GAPDH (Fig 1; Table 2). Median levels of VEGF-A, VEGF-C, Ang1, and Ang2 were higher in leukemic blasts than in normal peripheral-blood AC133+ cells. With the exception of VEGF-C, expression of analyzed angiogenic factors was low in normal PBMCs.
To investigate the relation between Ang2 mRNA levels and protein secretion, we determined concomitantly the expression of Ang2 in five human leukemic cell lines (TF-1, Mo7e, HL60, U937, and K562) by qPCR and by ELISA of cell culture supernatants. A correlation of 0.8 between mRNA and protein expression was found.
Because of the wide range of individual data points, we used logarithms of values shifted by 1 for further analysis and graphical display. mRNA specific for VEGF-A was detected in 90%, mRNA specific for VEGF-C was detected in 73%, mRNA specific for Ang1 was detected in 96%, mRNA specific for Ang2 was detected in 98%, and mRNA specific for Tie2 was detected in 60% of patients.
We performed univariate analyses to correlate gene expression with confounding factors age, sex, karyotype, and response on day 15 after induction chemotherapy with long-term outcome. Patients with less than 5% blasts in hypoplastic bone marrows 15 days after initiation of chemotherapy without evidence of extramedullary disease were considered as good responders. The karyotype of patients was initially categorized into three risk groups: t(8;21), inv(16); normal karyotype; and trisomy 8, abnormalities of chromosome 5, 7, 11, or complex abnormalities. Although observed CR rates varied among these groups with 90.9%, 73.5%, and 60.7%, respectively, we could not detect a statistically significant difference. Because in our cohort no significant survival difference was detected between patients with core-binding factor leukemias (n = 11) and those with normal karyotype (n = 50), both groups were combined for additional analysis. Comparable outcomes for patients with CBF leukemias and those with normal karyotype were also found in the entire Hannover 95 study cohort.24 For additional analysis, patients were divided into three groups of approximately equal size (VEGF-A < 3, 3 to 20, and > 20; Ang1 < 0.5, 0.5 to 1.5, and > 1.5; Ang2 < 2, 2 to 15, and >15). Given that only 60% and 73% of patients expressed VEGF-C and Tie2 respectively, patients were analyzed in two groups: zero and more than zero.
First we analyzed whether a correlation existed between gene expression levels of individual angiogenic factors and blast count, age, sex, karyotype, French-American-British classification type, and good or bad response to induction chemotherapy; no association was detected. In particular, no relevant correlation between absolute or relative blast count in peripheral blood and expression of angiogenic factors could be detected (absolute value of Spearman's r < 0.3 for all angiogenic factors and particularly r < 0.1 for Ang2). In addition, for Ang2, mean blast counts were not different among our specified groups.
Univariate analysis was then performed to identify factors that could predict the achievement of a CR. In our patient group, the only variable that correlated with obtaining a CR was good response to induction chemotherapy assessed on day 15 (Fisher's exact test, P < .001). Although the age of the entire group of patients was younger than 61 years at diagnosis, there was a trend that younger patients had higher CR rates (P = .059). No impact of the variables sex (P = .64), karyotype (P = .1), VEGF-A (P = .59), VEGF-C (P = .42), Ang1 (P = .08), Ang2 (P = .28), and Tie2 (P = 1.0) was found.
Univariate analysis of factors associated with OS showed a better prognosis for patients with good response to induction chemotherapy (log-rank P < .001), favorable karyotype (P = .07), and high Ang2 expression (P = .005). We could not detect a prognostic impact on survival for the factors sex (P = .52), age (P = .69), VEGF-A (P = .72), VEGF-C (P = .32), Ang1 (P = .88), and Tie2 (P = .43). Results of univariate analysis are listed in Table 3. Figure 2 shows the Kaplan-Meier curves for survival stratified for good response and for Ang2. It should be noted that patients were grouped into cohorts of approximately equal size to enable us to illustrate the results by Kaplan-Meier curves. Because Cox regression and Martingale plots revealed a linear relation between log Ang2 and risk (described in this section), no clinically relevant cut-offs could be identified and arbitrary cut-offs were chosen.
We used the Cox proportional hazards model based on all 90 patients to perform multivariate analyses of the variables with a P value of less than .1 (karyotype, good or bad response, and Ang2 expression) on overall survival (stepwise procedure). In this analysis, log Ang2 was used as a continuous predictor variable. We could identify Ang2 expression and good versus bad response as independent prognostic factors for survival (Table 4). Response on day 15 was closely related to karyotype, as can be deduced by cross-table analysis (P = .005). In our cohort among 64 patients with good response, 83% possessed a favorable karyotype, whereas among the patients with bad response only 50% had favorable cytogenetics. Therefore, karyotype lost its prognostic significance in multivariate analysis in our cohort.
The Martingale residual plot after inclusion of the variable response indicates that the effect of Ang2 on survival could be adequately modeled by a linear term on the log scale (compare Loess smoothing in Fig 3).
The resulting hazard ratio for bad versus good response was 3.51 (95% CI, 1.91 to 6.47; P < .001) and the hazard ratio for log Ang2 was 0.75 (95% CI, 0.61 to 0.91; P = .005). Karyotype did not show an additional independent effect on survival in multivariate analysis (P = .19). Results of the Cox regression therefore coincide with results of univariate analysis (compare with Fig 2).
The Martingale residual plot might also suggest the presence of four outliers. However, because the omission of these four patients (all of them had a poor response and did not die during time of investigation, probably due to allogeneic bone marrow transplantation in three of them) did not change coefficients and SEs in a noteworthy way, we chose to continue using the model including all 90 patients.
In addition, we investigated the adequacy of the proportional hazards assumption of the final model by fitting a model including a time-dependent covariate for each variable. P values of .48 and .49 for these coefficients suggested no relevant time dependency.
Furthermore, data were reanalyzed after censoring patients with allogeneic bone marrow transplantation at the time point of stem-cell infusion. No substantial difference in results was found. In univariate analysis karyotype seemed to be of more pronounced prognostic value (P = .0039). However, in multivariate analysis there was again no additional significant effect of karyotype (P = .19), which is consistent with the original model without censoring of patients who received bone marrow transplantation.
To further illustrate the impact of good and bad response and level of Ang2 expression on overall survival, Kaplan-Meier plots of patients with good and bad response stratified for Ang2 expression in three groups are shown in Figure 4.
To investigate possible interactions between the expression of various angiogenic growth factors on patient survival, subgroup analysis was performed. Patients were divided into groups with expression of VEGF-A, VEGF-C, and Ang1 below or above the median. Survival estimates were calculated for each subgroup according to the level of Ang2 expression. Kaplan-Meier survival curves for all subgroups are shown in Figure 5. The prognostic relevance of Ang2 expression was most pronounced in the subgroups with low VEGF-C and Ang1 mRNA levels. In the cohorts with high VEGF-C, VEGF-A, and Ang1, Ang2 expression lost its impact on OS, suggesting that the protective effect of Ang2 is diminished by simultaneous expression of proangiogenic factors.
Statistical analysis of RFS was nonyielding, given that 26 patients did not achieve a complete remission. Because of low patient numbers and consequently lack of statistical power, no additional analysis was performed for RFS.
DISCUSSION
Although RFS of the entire group was slightly better than those of analyzed patients, no significant difference between analysis and nonanalysis groups could be found. One limitation of the study is that no plasma samples of the study patients were available; thus, analysis was restricted to investigation of mRNA expression of angiogenic factors. Therefore, exact data on protein concentrations in blood are unknown. However, given that no post-transcriptional regulation of these factors has been described, concerted mRNA and protein expression of these factors can be supposed. Furthermore, mRNA and protein expression of Ang2 by qPCR and ELISA from five human leukemic cell lines was determined. A positive correlation was found, supporting the validity of the use of cDNA for our investigation.
Multivariate analysis revealed that expression of Ang2 and a good response on a day-15 postinduction chemotherapy bone marrow sample were the only independent predictors of OS in our patient cohort. The prognostic impact of persistent leukemia after induction chemotherapy has been confirmed by several groups.35每37 Because bad response to induction chemotherapy is correlated to adverse karyotypes, both groups of patients overlap to a high degree. Therefore, it was not possible to detect an independent prognostic value of karyotype in our cohort. Because of the small number of patients with specific chromosome aberrations, patients were grouped together into only two prognostic cohorts. It cannot be excluded by our analysis that individual translocations may result in a specific risk profile.
The role of Ang2 in angiogenesis is still controversial. In initial studies, Ang2 has been shown to block the effect of Ang1 on endothelial cells in vitro. Furthermore, Ang2 competes with Ang1 for binding to their common receptor Tie2, and acts as a competitive inhibitor of Ang1. Transgenic overexpression of Ang2 resulted in insufficient vascular remodeling, as in Ang1 or Tie2 knockout mice, indicating that Ang2 could act as an antagonist of Ang1 in vivo.13 Furthermore, Lin et al38 demonstrated that neutralization of the effects of Ang1 by soluble Tie2 blocked angiogenesis and tumor growth in a mouse model. Recently, the role of Ang2 as a pure inhibitor of angiogenesis has been challenged. Emphasis has now been put on the concerted action of Ang2 and VEGF-A on endothelial cells. In several experimental models, Ang2 (in the absence of VEGF-A) led to endothelial cell apoptosis and vessel regression. In the presence of VEGF-A, Ang2 promoted endothelial cell proliferation and migration, thereby acting as a proangiogenic agent.14每16,39每41 The dependence of the effects of Ang2 by VEGF-A is also suggested by our results. Although Ang2 expression was an independent predictor of OS for the entire cohort, subgroup analysis indicated that the prognosis of patients with VEGF-A (and especially VEGF-C and Ang1) expression below the median was most dependent on the level of Ang2 production. In particular, patients with high Ang2 and low VEGF-C expression had an excellent long-term prognosis. On the other hand, in patients with high VEGF-C levels, prognosis was much less influenced by Ang2. Unfortunately, in our study no corresponding bone marrow biopsies were available; therefore, we could not directly correlate bone marrow angiogenesis with expression of VEGF-A, VEGF-C, and Ang2.
Recently, several smaller studies have been published in which Ang2 expression was found in a variety of solid tumors.32,42,43 Ang2 could be detected in tumor cells and endothelial cells of invading tumor vessels in subgroups of patients. In the largest of these studies, expression of VEGF-A, Ang1, and Ang2 was analyzed in 236 patients with localized non-small-cell lung cancer. This investigation clearly showed that concerted expression of VEGF-A and Ang2 resulted in increased microvessel density and adverse long-term prognosis, supporting the model of cooperative effects of both types of growth factors.43
On the basis of the concept of cooperation between both classes of endothelial-specific growth factors, additional clinical and experimental research should be done to gain further insights into the dependence of neoangiogenesis on the interaction of Ang2 and VEGF-A or VEGF-C expression. If substantiated, therapeutic strategies directed against the action of VEGF-A and VEGF-C with simultaneous administration of Ang2 may be explored in experimental models to investigate enhanced antiangiogenic and antileukemic efficacy.
Authors' Disclosures of Potential Conflicts of Interest
Acknowledgment
We thank the members of the AML SHG Hannover study group for providing samples.
NOTES
Supported by a grant from Werner-Otto Stiftung Hamburg, Germany (S.L.), and a grant from Eppendorfer: Leukmiehilfe Hamburg and Roggenbuck Stiftung, Hamburg, Germany (U.M.G.).
Presented in part at the annual meeting of the American Society of Hematology in Philadelphia, PA, December 6-10, 2002.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Folkman J: Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1:27每31, 1995
2. Risau W: Mechanisms of angiogenesis. Nature 386:671每674, 1997
3. Fiedler W, Staib P, Kuse R, et al: Role of angiogenesis inhibitors in acute myeloid leukemia. Cancer J 7:S129每S133, 2001 (suppl 3)
4. Aigner A, Butscheid M, Kunkel P, et al: An FGF-binding protein (FGF-BP) exerts its biological function by parallel paracrine stimulation of tumor cell and endothelial cell proliferation through FGF-2 release. Int J Cancer 92:510每517, 2001
5. Berger W, Setinek U, Mohr T, et al: Evidence for a role of FGF-2 and FGF receptors in the proliferation of non-small cell lung cancer cells. Int J Cancer 83:415每423, 1999
6. Strutz F, Zeisberg M, Renziehausen A, et al: TGF-beta 1 induces proliferation in human renal fibroblasts via induction of basic fibroblast growth factor (FGF-2). Kidney Int 59:579每592, 2001
7. Mustonen T, Alitalo K: Endothelial receptor tyrosine kinases involved in angiogenesis. J Cell Biol 129:895每898, 1995
8. Davis S, Aldrich TH, Jones PF, et al: Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by secretion-trap expression cloning. Cell 87:1161每1169, 1996
9. Davis S, Yancopoulos GD: The angiopoietins: Yin and yang in angiogenesis. Curr Top Microbiol Immunol 237:173每185, 1999
10. Valenzuela DM, Griffiths JA, Rojas J, et al: Angiopoietins 3 and 4: Diverging gene counterparts in mice and humans. Proc Natl Acad Sci U S A 96:1904每1909, 1999
11. Sato TN, Tozawa Y, Deutsch U, et al: Distinct roles of the receptor tyrosine kinases Tie-1 and Tie-2 in blood vessel formation. Nature 376:70每74, 1995
12. Korpelainen EI, Alitalo K: Signaling angiogenesis and lymphangiogenesis. Curr Opin Cell Biol 10:159每164, 1998
13. Maisonpierre PC, Suri C, Jones PF, et al: Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis. Science 277:55每60, 1997
14. Lobov IB, Brooks PC, Lang RA: From the cover: Angiopoietin-2 displays VEGF-dependent modulation of capillary structure and endothelial cell survival in vivo. Proc Natl Acad Sci U S A 99:11205每11210, 2002
15. Holash J, Maisonpierre PC, Compton D, et al: Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF. Science 284:1994每1998, 1999
16. Visconti RP, Richardson CD, Sato TN: Orchestration of angiogenesis and arteriovenous contribution by angiopoietins and vascular endothelial growth factor (VEGF). Proc Natl Acad Sci U S A 99:8219每8224, 2002
17. Fiedler W, Graeven U, Ergun S, et al: Vascular endothelial growth factor, a possible paracrine growth factor in human acute myeloid leukemia. Blood 89:1870每1875, 1997
18. Bellamy WT, Richter L, Frutiger Y, et al: Expression of vascular endothelial growth factor and its receptors in hematopoietic malignancies. Cancer Res 59:728每733, 1999
19. Aguayo A, Estey E, Kantarjian H, et al: Cellular vascular endothelial growth factor is a predictor of outcome in patients with acute myeloid leukemia. Blood 94:3717每3721, 1999
20. Fiedler W, Graeven U, Ergun S, et al: Expression of FLT4 and its ligand VEGF-C in acute myeloid leukemia. Leukemia 11:1234每1237, 1997
21. Dias S, Choy M, Alitalo K, et al: Vascular endothelial growth factor (VEGF)-C signaling through FLT-4 (VEGFR-3) mediates leukemic cell proliferation, survival, and resistance to chemotherapy. Blood 99:2179每2184, 2002
22. Muller A, Lange K, Gaiser T, et al: Expression of angiopoietin-1 and its receptor TEK in hematopoietic cells from patients with myeloid leukemia. Leuk Res 26:163每168, 2002
23. Watarai M, Miwa H, Shikami M, et al: Expression of endothelial cell-associated molecules in AML cells. Leukemia 16:112每119, 2002
24. Heil G, Krauter J, Raghavachar A, et al: Risk-adapted induction and consolidation therapy in adults with de novo AML aged
25. Rozen S, Skaletsky H: Primer3 on the WWW for general users and biologist programmers, in Krawetz S, Misener S (eds): Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa, NJ, Humana Press, 2000, pp 365每386
26. Tischer E, Mitchell R, Hartman T, et al: The human gene for vascular endothelial growth factor: Multiple protein forms are encoded through alternative exon splicing. J Biol Chem 266:11947每11954, 1991
27. Huang YQ, Li JJ, Karpatkin S: Identification of a family of alternatively spliced mRNA species of angiopoietin-1. Blood 95:1993每1999, 2000
28. Therneau T, Grambsch P, Fleming T: Martingale-based residuals for survival models. Biometrika 77:147每160, 1990
29. Grambsch PM: Goodness-of-fit and diagnostics for proportional hazards regression models. Cancer Treat Res 75:95每112, 1995
30. Aguayo A, Kantarjian H, Manshouri T, et al: Angiogenesis in acute and chronic leukemias and myelodysplastic syndromes. Blood 96:2240每2245, 2000
31. Albitar M: Angiogenesis in acute myeloid leukemia and myelodysplastic syndrome. Acta Haematol 106:170每176, 2001
32. Lee JH, Koh JT, Shin BA, et al: Comparative study of angiostatic and anti-invasive gene expressions as prognostic factors in gastric cancer. Int J Oncol 18:355每361, 2001
33. Verstovsek S, Kantarjian H, Manshouri T, et al: Prognostic significance of cellular vascular endothelial growth factor expression in chronic phase chronic myeloid leukemia. Blood 99:2265每2267, 2002
34. de Bont ES, Rosati S, Jacobs S, et al: Increased bone marrow vascularization in patients with acute myeloid leukaemia: A possible role for vascular endothelial growth factor. Br J Haematol 113:296每304, 2001
35. Kern W, Haferlach T, Schoch C, et al: Early blast clearance by remission induction therapy is a major independent prognostic factor for both achievement of complete remission and long-term outcome in acute myeloid leukemia: Data from the German AML Cooperative Group (AMLCG) 1992 Trial. Blood 101:64每70, 2003
36. Liso V, Albano F, Pastore D, et al: Bone marrow aspirate on the 14th day of induction treatment as a prognostic tool in de novo adult acute myeloid leukemia. Haematologica 85:1285每1290, 2000
37. Wheatley K, Burnett AK, Goldstone AH, et al: A simple, robust, validated and highly predictive index for the determination of risk-directed therapy in acute myeloid leukaemia derived from the MRC AML 10 trial: United Kingdom Medical Research Council's Adult and Childhood Leukaemia Working Parties. Br J Haematol 107:69每79, 1999
38. Lin P, Polverini P, Dewhirst M, et al: Inhibition of tumor angiogenesis using a soluble receptor establishes a role for Tie2 in pathologic vascular growth. J Clin Invest 100:2072每2078, 1997
39. Vajkoczy P, Farhadi M, Gaumann A, et al: Microtumor growth initiates angiogenic sprouting with simultaneous expression of VEGF, VEGF receptor-2, and angiopoietin-2. J Clin Invest 109:777每785, 2002
40. Holash J, Wiegand SJ, Yancopoulos GD: New model of tumor angiogenesis: Dynamic balance between vessel regression and growth mediated by angiopoietins and VEGF. Oncogene 18:5356每5362, 1999
41. Asahara T, Chen D, Takahashi T, et al: Tie2 receptor ligands, angiopoietin-1 and angiopoietin-2, modulate VEGF-induced postnatal neovascularization. Circ Res 83:233每240, 1998
42. Sfiligoi C, de Luca A, Cascone I, et al: Angiopoietin-2 expression in breast cancer correlates with lymph node invasion and short survival. Int J Cancer 103:466每474, 2003
43. Tanaka F, Ishikawa S, Yanagihara K, et al: Expression of angiopoietins and its clinical significance in non-small cell lung cancer. Cancer Res 62:7124每7129, 2002(Sonja Loges, Gerhard Heil)