Gene expression signatures separate B-cell chronic lymphocytic ...

Leukemia (2006) 20, 1774–1782 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu

ORIGINAL ARTICLE Gene expression signatures separate B-cell chronic lymphocytic leukaemia prognostic subgroups defined by ZAP-70 and CD38 expression status A Hu¨ttmann1, L Klein-Hitpass2, J Thomale2, R Deenen2, A Carpinteiro1,3, H Nu¨ckel1, P Ebeling4, A Fu¨hrer1, J Edelmann1, L Sellmann1,2, U Du¨hrsen1 and J Du¨rig1 1

Clinic of Hematology, University Hospital, University of Duisburg-Essen, Essen, Germany; 2Institute of Cell Biology, University Hospital, University of Duisburg-Essen, Essen, Germany; 3Institute of Molecular Biology, University Hospital, University of Duisburg-Essen, Essen, Germany and 4Department of Internal Medicine (Cancer Research), University Hospital, University of Duisburg-Essen, Essen, Germany

B-cell chronic lymphocytic leukaemia (B-CLL) is a heterogenous disease with a highly variable clinical course and analysis of zeta-associated protein 70 (ZAP-70) and CD38 expression on B-CLL cells allowed for identification of patients with good (ZAP-70CD38) and poor (ZAP-70 þ CD38 þ ) prognosis. DNA microarray technology was employed to compare eight ZAP-70 þ CD38 þ with eight ZAP-70CD38 B-CLL cases. The expression of 358 genes differed significantly between the two subgroups, including genes involved in B-cell receptor signaling, angiogenesis and lymphomagenesis. Three of these genes, that is, immune receptor translocation-associated protein 4 (IRTA4)/Fc receptor homologue 2 (FcRH2), angiopoietin 2 (ANGPT2) and Pim2 were selected for further validating studies in a cohort of 94 B-CLL patients. IRTA4/FcRH2 expression as detected by flow cytometry was significantly lower in the poor prognosis subgroup as compared to ZAP70CD38 B-CLL cells. In healthy individuals, IRTA4/FcRH2 protein expression was associated with a CD19 þ CD27 þ memory cell phenotype. ANGPT2 plasma concentrations were twofold higher in the poor prognosis subgroup (Po0.05). Pim2 was significantly overexpressed in poor prognosis cases and Binet stage C. Disease progression may be related to proangiogenic processes and strong Pim2 expression. Leukemia (2006) 20, 1774–1782. doi:10.1038/sj.leu.2404363; published online 17 August 2006 Keywords: B-cell chronic lymphocytic leukaemia (B-CLL); IRTA-4/ FcRH2; Pim2; angiopoietin 2; protein tyrosine kinase ZAP-70; CD38 antigen

Introduction B-cell chronic lymphocytic leukaemia (B-CLL) is a heterogenous disease with a highly variable clinical course.1,2 In a continuing effort to identify patients with poor prognosis and to facilitate the clinical management of B-CLL, several molecular prognostic markers have been identified during the last decade. Landmark studies by Hamblin et al.3 and Damle et al.4 have shown that the survival probability in B-CLL is associated with the absence or presence of somatic mutations in rearranged immunoglobulin heavy-chain variable (IgVH) genes. Currently, the IgVH gene mutation status is considered as one of the most powerful prognostic factors,5 where B-CLL cases with unmutated IgVH genes are characterized by an unfavourable clinical outcome. However, IgVH mutation analysis is time consuming and Correspondence: Dr J Du¨rig, Clinic of Hematology, University Hospital, University of Duisburg-Essen, Hufelandstr. 55, Essen 45122, Germany. E-mail: [email protected] Received 4 March 2006; revised 8 June 2006; accepted 30 June 2006; published online 17 August 2006

technically demanding, thus limiting its use in clinical practice. Several DNA microarray studies comparing the transcriptomes of patient subgroups defined by the presence or absence of IgVH gene mutations6–8 have identified genes that could be used as surrogate markers for the IgVH mutation status. One of these genes, the Syk-ZAP-70 protein tyrosine kinase family 70 kDa zeta-associated protein (ZAP-70) has recently been shown to play a role in B-cell receptor (BCR) signaling in IgVH-unmutated CLL cells.9,10 In addition, ZAP-70 protein expression can be conveniently measured by flow cytometry and a series of studies established ZAP-70 as a reliable prognostic marker,11–13 possibly of even greater predictive value than IgVH mutation analysis.14 Another surrogate marker for the mutated and unmutated IgVH B-CLL subgroups has been identified with the CD38 protein.4 However, although vast numbers of subsequent studies could clearly demonstrate the clinical applicability of CD38 quantitation with some degree of correlation to IgVH mutation status,4,5,13,15–17 at present both parameters are regarded as independent prognostic variables in B-CLL.5,17 CD38 is a type II transmembrane glycoprotein that acts as a complex ectoenzyme and receptor molecule, which leads to enhancement of BCR signaling in B-CLL cells.18,19 Alike ZAP-70 detection, CD38 cell surface expression is amenable to routine flow cytometry and provides a valuable adjunct in the diagnostic work-up of B-CLL patients.5,16 Interestingly, coexpression of CD38 and ZAP-70 in B-CLL cells appears to correlate with a particularly strong BCR signaling capacity.9,10 Under the assumption that sustained BCR signaling in response to a yet unknown (auto)antigen may play an important role in B-CLL pathogenesis, we recently investigated in a cohort of 252 CLL patients whether CD38 can further refine the adverse prognostic relevance of high ZAP-70 expression levels.13 In ZAP-70 þ CD38 þ B-CLL cases, median treatment-free survival was 30 months as compared to 130 months in patients with a ZAP-70CD38 status. Discordant ZAP-70/CD38 resulted in a median treatment-free survival time of 43 months. Thus, ZAP-70 and CD38 expression analyses provided complementary prognostic information identifying three patient subgroups with good, intermediate and poor prognosis.13 Aiming at molecular differences that may help explain the diverse clinical behaviour of B-CLL, the present study investigated cases at the extremes of the disease spectrum. A transcriptome analysis was performed on eight ZAP-70 þ CD38 þ vs eight ZAP-70CD38 B-CLL samples employing the Affymetrix U133A platform. Supervised analysis revealed a panel of differentially expressed genes designated as ‘B-CLL subtype distinction genes’.7 Validation of these results was performed in a larger B-CLL patient cohort on a restricted panel of genes

Gene expression signatures separate B-CLL prognostic subgroups A Hu¨ttmann et al

1775 involved in BCR signaling, B-cell differentiation, angiogenesis and lymphomagenesis.

Patients, materials and methods

Patients Peripheral blood samples from 94 patients with B-CLL were analysed after obtaining informed consent according to our institutional guidelines. The diagnosis of B-CLL required a persistent lymphocytosis of more than 5.0 109/l and a typical CD19 þ , CD20 þ CD5 þ , CD23 þ , Ig light chain (k or l light chain) restricted immunophenotype as revealed by flow cytometry of peripheral blood cells. Peripheral blood mononuclear cells (PBMCs) were isolated by Lymphoprep (Invitrogen, Karlsruhe, Germany) density-gradient centrifugation and cryopreserved until further analysis. Patient selection was based on the availability of frozen freshly isolated total RNA stored in our CLL cell bank and the previously determined expression status of ZAP-70 and CD38 (see below). Clinical and laboratory data of the study population are shown in Supplementary Table 1.

Gene expression profiling Mononuclear cells were isolated from fresh peripheral blood samples as described. CD19-positive leukaemic B cells were enriched employing anti-CD19 magnetic microbeads (MidiMacs, Miltenyi Biotec, Bergisch Gladbach, Germany), resulting in a purity of 493% as determined by flow cytometry (Table 1). Gene expression profiling was performed on total RNA extracted from 1 to 2 108 CD19 þ cells (RNeasy Midi Kit, Qiagen, Hilden; Germany), using the Affymetrix U133A microarrays and GeneChip platform (Affymetrix, Santa Clara, CA, USA), as described recently.13,20 Expression values were determined using the Affymetrix MAS 5.0 software. Comparative analysis of gene expression profiles was performed in 16 patients (eight ZAP-70 þ CD38 þ vs eight ZAP-70CD38 Table 1

M M M F F M M M

71 68 75 73 60 74 66 67

Mean (7s.d.) CLL-9 CLL-10 CLL-11 CLL-12 CLL-13 CLL-14 CLL-15 CLL-16 Mean (7s.d.)

Real-time reverse transcriptase-polymerase chain reaction

Total RNA from 1 to 2 108 freshly isolated PBMC was extracted, purified (RNeasy Midi Kit) and quantified spectrophotometrically as described previously.20,22 The RNA isolation procedure included an on-column digestion step of residual genomic DNA using DNase I as recommended by the manufacturer in all samples analysed. First strand complementary DNA (cDNA) was synthesized from 1 mg of RNA using oligo(dT) primers employing a commercially available kit (RT-PCR Amplimers, Becton Dickinson, Heidelberg, Germany) according to the manufacturer’s instructions, yielding a final cDNA volume of 100 ml. Real-time polymerase chain reaction (PCR) was performed with the ABI Prism 7900HT Sequence Detector (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions as described previously.20,22 PCR was carried out in a 20 ml reaction volume using 2 ml of cDNA with a Pim2-specific assay (Assays-on-demand Gene

Clinical and laboratory data of patients whose leukaemic cells were subjected to comparative microarray gene expression profiling

Patient’s Sex Age ID (years) CLL-1 CLL-2 CLL-3 CLL-4 CLL-5 CLL-6 CLL-7 CLL-8

patients) whose clinical characteristics are shown in Table 1. For multiclass supervised comparative analysis of the two patient groups, we utilized the significance analysis of microarrays (SAM) method, which uses a modified t-test with samplelabeled permutations to evaluate statistical significance.21 Differentially regulated genes were identified by comparing the mean signal intensities of eight ZAP-70 þ CD38 þ vs eight ZAP-70CD38 patient samples at a false discovery rate of qp10% and defining a cutoff fold change of 72.0. Only genes called ‘present’ by the Affymetrix algorithm in at least 30% of either of the two groups were considered for SAM analysis. As CD19 purity and ZAP-70 expression levels showed some degree of variation (Table 1), SAM analysis was also performed after excluding the lowest purity sample (i.e. CLL-12) and the samples closest to the ZAP-70 expression-defining margin (i.e. CLL-4 and CLL-15). Results are reported in Figure 1 and Supplementary Table 2.

F M M M M M M M

CD19 purity (%) 98.8 99.9 97.9 98.1 99.0 98.0 97.2 97.7

Light chain

Diagnosis

Sample

Year

Binet

Year

k k k k k k l k

1990 1993 1989 1994 2001 1995 1997 2000

A A A B A A A A

2003 2003 2003 2003 2003 2003 2003 2003

k k k k l l l k

2002 1996 1996 2002 1994 1998 2000 1997

A A A A A B A A

2003 2003 2003 2003 2003 2003 2004 2004

Expression (%)

Risk group

FISH

Treatment

Binet CD19+/38+ ZAP-70 C B C A B A A A

0.4 5.0 0.0 0.6 1.0 2.8 1.0 0.7

6.0 2.6 6.0 16.0 1.5 0.4 8.2 1.3

ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38 ZAP-70/CD38

del(13q) Trisomy 12 del(13q) del(13q) Normal del(13q) del(13q) ND

Yes No Yes Yes No Yes No No

21.0 63.7 93.0 25.0 95.0 33.7 49.0 74.0

82.0 64.0 83.0 65.0 96.0 77.0 22.0 58.0

ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+ ZAP-70+/CD38+

ND Normal del(13q),del(17p) del(13q),del(11q) del(13q) del(13q),del(11q) del(13q) del(13q)

No Yes Yes No Yes Yes Yes Yes

56.8 (729.3)

68.4 (722.4)

98.3 (70.9) 47 68 50 72 72 58 47 81

99.3 98.7 99.0 93.3 98.5 99.1 99.8 98.6 98.3 (72.1)

B C C C C C

Abbreviations: CLL, chronic lymphocytic leukaemia; FISH, fluorescence in situ hybridisation; F, female; M, male; ND, not determined. Leukemia


1776 Al l samples IRTA4 ANGPT2 Pim2

417

426

34

111

CLL-12 excluded IRTA4 ANGPT2 Pim2 48

250 31

8 36

325 CLL-4 and CLL-15 excluded IRTA4 Pim2 Figure 1 Overlap of differentially regulated probe sets. SAM analysis of all 16 B-CLL cases listed in Table 1 showed differential regulation of 426 probe sets. These include 345 probe sets (encoding 292 genes) upregulated and 81 probe sets (encoding 66 genes) downregulated in ZAP-70CD38 B-CLL cases (top, red circle to the left). When the sample with the lowest purity was excluded, 417 probe sets were differentially regulated (top, blue circle to the right). Out of these 417 probe sets, 361 were intersecting with the ‘all samples’ analysis. When cases with borderline ZAP-70 expression were excluded from the analysis (bottom, green circle), 325 probe sets were differentially regulated. The intersection between the three analyses leaves 250 differentially regulated probe sets encoding 210 genes at the core of the analysis. These 250 probe sets include all genes discussed in the present work, except for ANGPT2, which was found upregulated in both analyses at the top, but not in the analysis at the bottom where CLL-4 and CLL-15 were excluded. A detailed analysis of intersecting probe sets can be taken from Supplementary Table 2.

expression system, Hs00179139_m1, Applied Biosystems). Target-specific primers were designed from sequences of exons 3 and 4 of the Pim2 gene. For normalization, the expression level of the housekeeping gene glycerolaldehyde-3-phosphate dehydrogenase (GAPDH) was measured as an endogenous control (Hs 99999905_m1, Applied Biosystems). For quantification of each PCR result, we calculated the DCt value between the target gene (Pim2) and its endogenous control (GAPDH). The mean DCt values of PBMC samples of five normal donors were then used as calibrators for relative quantification.

ZAP-70 expression Cytoplasmic ZAP-70 expression was determined by flow cytometry as recently described.12,13 In brief, permeabilized cells were stained with anti-ZAP-70 (clone 2F3.2, Upstate Biotechnology, Waltham, MA, USA) as primary antibody and goat-anti-mouse immunoglobulin fluorescein isothiocyanate (FITC; Dako, Glostrup, Denmark) as secondary antibody. CD3-phycoerythrin (PE), CD56-PE (Becton Dickinson) and CD19-peridinin chlorophyll protein cytochrome 5.5 (Dako) were used for staining of T, natural killer and B lymphocytes, respectively. After appropriate lymphocyte gating, cytoplasmic ZAP-70 expression was determined in CD19 þ B-CLL cells as described.12,13 Data acquisition and analysis were performed using CellQuest software (Becton Dickinson).

clonal antibodies using a standard three-colour flow cytometry approach: CD45 (FITC)/CD14 (PE)/CD20-peridinin chlorophyll (PerCP); CD4 (FITC)/CD8 (PE)/CD3 (PerCP); l light-chain (FITC)/ CD19 (PE)/CD5-phycoerythrin-cyanin (PeCy5); k light-chain (FITC)/CD19 (PE)/CD5 (PeCy5); IgM (FITC)/CD23 (PE)/CD19 (PECy5); CD10 (FITC)/CD38 (PE, clone HB7)/CD19 (PECy5); CD23 (FITC)/CD38 (PE, clone HB7). To minimize potential contamination with coexisting normal B cells, only B-CLL cases in which 90% of CD19 þ CD20 þ cells coexpressed CD5 were included in the study.

Detection of genomic aberrations by fluorescence in situ hybridization Prognostically relevant anomalies of chromosomal regions 11q, 13q, and 17p, and of chromosome 12 were assessed by fluorescence in situ hybridization, as described previously.13

Evaluation of immune receptor translocation-associated protein 4/Fc receptor homologue 2 expression on CLL cells by flow cytometry Immune receptor translocation-associated protein 4 (IRTA4)/Fc receptor homologue 2 (FcRH2) (also known as SPAP1) expression on the surface of B-CLL cells was determined by threecolour flow cytometry employing a combined direct/indirect staining technique using antibodies at pretested optimal concentrations. In brief, freshly thawed CLL–PBMC stored in dimethyl sulfoxide were stained with biotinylated anti-FcRH2/ IRTA4 (R&D Systems, Minneapolis, MN, USA) or an irrelevant IgG control antibody (Dako) as primary antibody and streptavidin-PE (R&D Systems) as secondary antibody in combination with directly labeled anti-CD5-FITC (Becton Dickinson) antiCD19-PeCy5 (Dako). The relative density of IRTA4/FcRH2 on lymphocyte-gated CD19 þ /CD5 þ cells was quantified by determining the mean fluorescent intensity (MFI) for PE and subtracting the corresponding MFI values of the control antibody. Comparative experiments were performed on normal donor-derived PBMC (n ¼ 3) stained with biotinylated antiIRTA4/FcRH2 (R&D Systems) followed by streptavidin-FITC (Dako) in combination with directly labeled anti-CD27-PE (R&D Systems) and anti-CD19PeCy5 (Dako). In these experiments, the relative density of IRTA4/FcRH2 on lymphocyte-gated CD19 þ CD27 naive B cells was compared to that of CD19 þ CD27 þ memory B cells by determining the MFI for FITC and subtracting the corresponding MFI values of the control antibody.

Detection of proangiogenic factors in the blood plasma of CLL patients Plasma was prepared from freshly collected heparinized peripheral blood samples and stored at 801C until analysis. Plasma levels of vascular endothelial growth factor (VEGF) and angiopoietin 2 (ANGPT2) were quantified using Quantikine solid phase enzyme-linked immunosorbent assays (ELISAs; R&D Systems).

Cell cultures CD38 expression Fresh heparinized peripheral blood samples were prepared for flow cytometry by ammonium chloride erythrocyte lysis. As described previously,13 the immunophenotype was characterized using the following panel of fluorochrome-labeled monoLeukemia

CD19-positive B cells from the peripheral blood of five patients with B-CLL were enriched employing anti-CD19 magnetic microbeads (MidiMacs, Miltenyi), resulting in a purity of CD19 þ B cells of 495% as determined by flow cytometry. Purified B-CLL cells were cultured at a concentration of 1.5 106/ml in Iscove’s modified Dulbecco’s medium (Gibco


1777 Laboratories, Grand Island, NY, USA) with 10% (v/v) foetal calf serum (Greiner, Limburg, Germany) supplemented with 1000 U/ ml penicillin and 100 U/ml gentamycin (both from Sigma, Deisenhofen, Germany). After 3 days incubation in a 5% CO2 and air incubator at 371C, cell culture supernatants were frozen at 801C until analysis for ANGPT2, as described above.

Statistical analysis Comparison of clinical or laboratory parameters between patient subgroups was performed using the non-parametric Mann–Whitney U-test or Kruskal–Wallis test for continuous variables and the w2 test for categorical data. Differences were regarded significant at Po0.05. Retrieval of Gene-Ontology (GO) annotations and statistical analysis for significant (Po0.05) overrepresentation of individual annotations in the analysed list of genes was performed using GOstat (http://gostat.wehi. edu.au).

Results

Comparative gene expression analysis of ZAP70 þ CD38 þ vs ZAP-70CD38 CLL cells Applying the filter and evaluation strategies detailed above SAM analysis revealed that a total of 292 genes (defined by 345 probe sets) were significantly upregulated in ZAP-70 þ CD38 þ as compared to the ZAP-70CD38 B-CLL group. Conversely, 66 genes (defined by 81 probe sets) were significantly downregulated in the ZAP-70 þ CD38 þ B-CLL samples (Supplementary Table 3). The raw experimental data can be accessed through the internet (http://www.ncbi.nlm.nih.gov/geo/). Excluding the sample with the lowest CD19 purity or the samples closest to the ZAP-70 expression margin, left 210 genes (defined by 250 probe sets) at the core of the analysis (Figure 1). Of note, a substantial number of these 210 B-CLL subtype distinction genes including lipoprotein lipase (LPL), CRY1, ADAM29, TCF7, NRIP1, AKAP12 (gravin), DMD (dystrophin), TGFBR3, FGFR1, FUT8 and PTK2 have been described previously to be differentially expressed in microarray studies comparing B-CLL cases with the absence or presence of mutations in their IgVH genes.6–8 To further elucidate, the biological processes and molecular pathways underlying the diverse clinical behaviour of the two patient subgroups the gene identifiers were linked to the GO database.23 Pathways in the GO ‘biological process’ and ‘molecular function’ categories that showed significant enrichment for genes differentially expressed in ZAP-70 þ CD38 þ vs ZAP-70CD38 B-CLL samples are listed in Supplementary Table 4. Selection of genes for confirmatory studies was based on (i) novelty, (ii) their potential functional relevance for the disease process and (iii) availability of reagents (antibodies/ELISA; fluorochrome-labeled DNA probes). As the number of CLL samples subjected to microarray analysis was small, we proceeded to validate three of the B-CLL subtype distinction genes (Supplementary Table 3), that is, IRTA4/FcRH2, ANGPT2 and Pim2 in a cohort of 94-well characterised B-CLL patients (52 ZAP-70CD38 and 42 ZAP-70 þ CD38 þ samples, Supplementary Table 1) stored in our CLL cell bank.

Immune receptor translocation-associated protein 4/Fc receptor homologue 2 IRTA4/FcRH2 belongs to a recently recognized family of immunoglobulin-like type I membrane proteins that have

homology to Fcg receptors and platelet endothelial cell adhesion molecule 124 and have both activating25 and inhibitory25,26 signaling potential. Comparing the transcriptomes of ZAP-70CD38 vs ZAP-70 þ CD38 þ samples, we found IRTA4/FcRH2 to be significantly overexpressed in the ZAP-70CD38 subgroup. This finding was confirmed at the protein level in a series of 26 patients by flow cytometric analysis of IRTA4/FcRH2 surface expression on gated CD19 þ CD5 þ CLL cells (Figure 2a). Interestingly, IRTA4/FcRH2 protein expression on normal donor-derived PB B-cells was significantly higher on CD19 þ CD27 þ memory as compared to CD19 þ CD27 naive B cells (Figure 3a).

Angiopoietin 2 To investigate the relation between ANGPT2 mRNA levels in B-CLL cells and protein secretion, we determined the concentration of ANGPT2 in plasma samples of 56 B-CLL patients (Figure 4) by ELISA. In line with the microarray experiments, we found a twofold increase of ANGPT2 plasma levels in the ZAP-70 þ CD38 þ as compared to the ZAP-70CD38 patient subgroup. Furthermore, ANGPT2 plasma concentrations detected in Binet B and C patients exceeded those of Binet stage A patients suggesting that ANGPT2 protein secretion may be correlated to disease progression. In a cell culture experiment where five immunomagnetically purified B-CLL cases were studied, leucemic cells proved capable of secreting ANGPT2 into the culture supernatant in three cases (mean value 180 pg/ ml, range 70–255 pg/ml), thereby providing circumstantial evidence that the CLL cells are indeed the probable source of excess ANGPT2 protein production observed in the ZAP70 þ CD38 þ patient cohort (Supplementary Table 5). As ANGPT2 may act synergistically with VEGF as a proangiogenetic stimulus in vivo we also measured the plasma concentrations of VEGF in 38 B-CLL individuals. As depicted in Figure 4, ZAP-70 þ CD38 þ patients exhibited significantly higher VEGF plasma concentrations as compared to the ZAP-70CD38 subgroup. In summary, these findings suggest a proangiogeneic phenotype for the ZAP-70 þ CD38 þ patient subgroup.

Pim2 The serine/threonine kinase Pim2 is a transcriptionally regulated oncogenic kinase that promotes cell survival by activating nuclear factor-kB (NF-kB)-dependent gene expression in response to a wide variety of proliferative signals.27–29 Deregulation of Pim2 expression has been documented in several human malignancies including leukaemia, lymphoma and also B-CLL.30–32 Pim2 mRNA expression was measured by real-time quantitative reverse transcriptase-PCR (RQ-PCR) in 94 patients using PBMC obtained from healthy individuals as a reference with an expression level of one. Overall, Pim2 was found to be expressed at 12-fold higher levels in B-CLL samples than in normal PBMC with a distribution range from 0.9 to 56.2 (data not shown). As inferred from the microarray experiments, Pim2 mRNA concentrations in ZAP-70 þ CD38 þ B-CLL samples were slightly higher than in the ZAP-70CD38 samples (Figure 5, left part of panel a). However, this difference did not reach statistical significance if unfractionated PBMCs were used. Of note, there was a highly significant correlation between disease stage and Pim2 expression with Binet stage C exhibiting 1.8-fold higher Pim2 levels than Binet stage A samples (Figure 5, right part of panel a). We next compared Pim2 expression levels in unfractionated PBMCs, the CD3 þ and CD19 þ subset of normal donor and B-CLL samples (Figure 5b). Taken together, a high Leukemia


1778

a 200

n = 14

n = 12

b 25

x1000 r = 0. 48

20 150 U133A relative signal intensity

IRTA4/FcRH2 FACS signal intensity (geometric mean)

p = 0.0003

100

15

10

50 5

0

0 CD38 neg. ZAP70 neg.

CD38 pos. ZAP70 pos.

0

50

100

150

IRTA4/FcRH2 FACS signal intensity (geometric mean)

Figure 2 IRTA4/FcRH2 expression in B-CLL. (a) Flow cytometric analysis of B-CLL samples belonging to the prognostic favourable ZAP70CD38 (J) and unfavourable ZAP-70 þ CD38 þ (K) subgroup. The expression levels of individual B-CLL samples are plotted as geometric mean. (b) Correlation between relative signal intensity in the microarray experiment and fluorescence-activated cell sorting signal intensity for the ZAP-70CD38 (J) and ZAP-70 þ CD38 þ (K) subgroup.

Figure 3 Flow cytometric analysis of FcRH2/IRTA4 expression. (a) CD19-positive B cells of a healthy individual were stained with a FITCconjugated isotype control (isotype). Gating on CD19-positive/CD27-negative (naive) and CD19/CD27 double positive (memory) PBMCs indicates that acquisition of a memory cell phenotype is associated with a gain in FcRH2/IRTA4 expression levels. (b) CD19/CD5 double-positive B cells of B-CLL cases representing the poor ZAP-70 þ CD38 þ and good ZAP-70CD38 prognosis subgroups were stained with a PE-conjugated isotype control (isotype). Staining with PE-FcRH2/IRTA4 antibody revealed higher expression levels in samples from ZAP-70CD38 (good prognosis) as compared to ZAP-70 þ CD38 þ (poor prognosis) cases. The seemingly lower signal intensity of IRTA4/FcRH2 in the healthy individual (a) is explained by use of the FITC-coupled secondary antibody. The PE conjugate employed for evaluation of FcRH2/IRTA4 expression in the diseased samples (b) provides approximately one log higher signal intensities.

proportion of non-CD19 þ cells may dilute Pim2 expression levels in unfractionated B-CLL samples with a low burden of leukaemic cells. If, in contrast to PBMCs, highly purified B-CLL Leukemia

samples from the favourable and unfavourable prognostic subgroups were tested for Pim2 expression (Figure 5c), a significant difference with higher levels in the unfavourable


1779 p = 0.00769

p = 0.00001 15000

Angiopoietin2

5000

n = 27 0

n = 29

n = 27

p = 0.03486

n = 12

n = 13

p = n.s.

400

VEGF

Plasma concentration (pg/ml)

10000

300 200 100 n = 16

n = 22

n = 14

n = 10

n = 10

0 CD38 neg. ZAP70 neg.

CD38 pos. ZAP70 pos.

Binet A

Binet B

Binet C

Figure 4 Plasma concentrations of the proangiogenic factors ANGPT2 and VEGF in B-CLL cases. ELISA analysis of B-CLL plasma samples collected from the prognostic favourable ZAP-70CD38 (J) and unfavourable ZAP-70 þ CD38 þ (K) subgroups (left part of the panel) reveals statistically significant differences in ANGPT2 and VEGF concentrations (Mann–Whitney U-Test). Analysis by clinical stage (right part of the panel) revealed a statistically significant difference in ANGPT2 plasma concentrations for Binet A vs Binet B vs Binet C in the Kruskal–Wallis test (P ¼ 0.00769). No difference was observed for VEGF plasma concentrations.

subgroup became apparent. This finding suggests that Pim2 expression in B-CLL cells may be related to both disease stage and B-CLL subtype as defined by ZAP-70 and CD38 expression.

Discussion Recent work has shown that measurement of ZAP-70 and CD38 expression on B-CLL cells by flow cytometry allowed for the separation of three patient subgroups with good (ZAP70CD38), intermediate (discordant expression of the two markers) and poor prognosis (ZAP-70 þ CD38 þ ).13,33 Following up on these observations, the present study investigated the transcriptomes of B-CLL cases at the extremes of the disease spectrum in order to identify molecular signatures underpinning the diverse clinical phenotypes. To this end eight ZAP70 þ CD38 þ vs eight ZAP-70CD38 B-CLL samples were analysed employing the Affymetrix U133A platform. Supervised analysis revealed a panel of 358 B-CLL subtype distinction genes, the majority of which (292 genes) was found to be overexpressed in the prognostically unfavourable ZAP70 þ CD38 þ subgroup. Notably and reassuringly, despite the small sample size this list of differentially expressed genes showed a substantial overlap with results from previously published microarray studies comparing B-CLL cases with the absence or presence of mutations in their IgVH genes.6–8 Among these distinction genes LPL, ADAM29, SPG20, NRIP1, AKAP12 and SEP10 have been attributed prognostic potential.8,34,35 In a recent retrospective RQ-PCR study measuring mRNA transcript levels of LPL and ADAM29 in the leukaemic cells of 133 patients with B-CLL,22 we (i) confirmed their value as independent prognostic variables and (ii) detected highly significantly different expression levels of both markers in

ZAP-70 þ CD38 þ as compared to ZAP-70CD38 B-CLL samples validating the microarray results presented herein. Comparative analysis of LPL and ADAM29 mRNA concentrations in patients for whom both microarray and RQ-PCR results were available, yielded concordant results (data not shown). Distinct from previously published microarray studies,6–8 the current study employed a patient selection approach based on ZAP-70 and CD38 quantitation. Both of these molecules can enhance BCR signaling,9,10,18 thereby conveying an important prosurvival stimulus to the leukaemic cells which may partially explain the more aggressive clinical course of ZAP-70 þ CD38 þ B-CLL.1,9,10,18 Remarkably, the microarray experiments described herein revealed relative overexpression of additional BCR pathway components such as CD5, IGHD, IGL, IGLJ3 and IGLC2. These findings are in accordance with a recent flow cytometry study showing higher IgM surface levels on IgVH unmutated as compared to mutated B-CLL cells.36 Furthermore, the present microarray analysis showed that FcRH2/IRTA4 was significantly downmodulated in ZAP-70 þ CD38 þ B-CLL, results which were subsequently confirmed at the protein level using flow cytometry in a series of 26 B-CLL patients. FcRH2/IRTA4 belongs to a recently recognized family of Ig domain containing type I membrane proteins, which exhibit sequence similarity to Fc receptors and are predominantly expressed in the B-cell compartment.25,37–39 It is currently unknown which ligands bind to this receptor family. The cytoplasmic segment of FcRH2/ IRTA4 contains a pair of typical immunoreceptor tyrosine-based inhibition motifs (ITIM).26 These ITIM can be phosphorylated in response to receptor ligation and thereby recruit terminating SH2 domain containing phosphatases including SHP-1. As an anchor molecule of SHP-1, FcRH2/IRTA4 is believed to play an inhibitory role in BCR signaling.26 The published in situ hybridization data suggest that FcRH2/IRTA4 RNA is Leukemia


1780

a

p = 0.50333

60

p = 3.5x10e-19

Pim2 expression ratio (CLL / normal donor)

50

40

30

20

10

n = 52

0

CD38 neg. ZAP70 neg.

b

4.0

PBMC

CD3+

n = 42 CD38 pos. ZAP70 pos. CD19+

n = 48 Binet A

c

Binet B

10

n = 21 Binet C

p = 0.0229

9

3.5

8 Pim2 expression ratio (CLL / normal donor)

3.0 ∆CT (Pim2 - GAPDH)

n = 17

2.5 2.0 1.5 1.0 0.5

7 6 5 4 3 2

0.0

1

-0.5

0

Normal donor B-CLL

n=3 n=3

n = 24

n = 21

CD38 neg. CD38 pos. ZAP70 neg. ZAP70 pos.

Figure 5 Quantitative analysis of Pim2 mRNA expression in B-CLL cases. (a) Quantitative RT-PCR of B-CLL samples belonging to the prognostic favourable ZAP-70CD38 (J) and unfavourable ZAP-70 þ CD38 þ (K) subgroups (left part of panel a) does not reveal a difference in Pim2 expression when unpurified mononuclear cells (PBMCs) were used as a template. Analysis of unpurified PBMCs by clinical stage showed a highly significant difference in the Kruskal–Wallis test for Binet A vs Binet B vs Binet C (right part of panel a, P ¼ 3.5 1019). (b) Quantitative RT-PCR of normal donor (&) and B-CLL (’) PBMCs, CD3 þ cells and CD19 þ cells. Low Pim2 expression levels (i.e. high DCT values) were observed in unfractionated PBMCs isolated from normal donors, whereas Pim2 expression was several fold higher (i.e. low DCT values) in unfractionated PBMCs isolated from B-CLL patients. Pim2 expression was detected at approximately even levels in the CD3 þ fraction. Minor differences were observed in the CD19 þ fraction with a trend towards higher Pim2 expression in the diseased samples. (c) Analysing Pim2 expression in 45 CD19 purified B-CLL samples (CD19/CD5 coexpression available for 29 samples: 98.7%, s.d.71.3) belonging to the favourable (J) and unfavourable (K) subgroups revealed a statistically significant difference with higher levels in the unfavourable subgroup. As derived from panel b, Pim2 expression levels were low in normal donor PBMCs with a considerable proportion of ‘contaminating’ non-CD19 þ cells. If the proportion of ‘contaminating’ non-CD19 þ cells decreases, as it is usually the case for B-CLL progression into stage Binet C, Pim2 expression levels rise (right part of panel a). Thus, analysis of unfractionated cells from various clinical stages as depicted in the left part of panel a did not reveal any differences. However, analysing highly purified B-CLL samples (c) showed higher Pim2 expression in the unfavourable subgroup.

Leukemia


predominantly expressed in naive B cells located in the mantle zone of the lymph follicle.39 By contrast employing multiparameter flow cytometry, we showed that the FcRH2/IRTA4 protein is stronger expressed on normal peripheral blood-derived CD19 þ CD27 þ memory as compared to CD19 þ CD27 naive B cells. The reasons for these discrepancies are not clear but are likely technical in nature as different cell populations and methods of detection were used in the two studies. In all, our finding of differential FcRH2/IRTA4 expression in the two B-CLL subtypes is compatible with the concept that indolent ZAP-70CD38 B-CLLs are more closely related to memory B cells than their ZAP-70 þ CD38 þ counterparts. Also, relative overexpression of the inhibitory-type FcRH2/IRTA4 receptor may render ZAP-70CD38 B-CLL cells less susceptible to antigenic stimulation, resulting in an anergic phenotype as compared to ZAP-70 þ CD38 þ leukaemic cells. Gene ontology analysis of B-CLL subtype distinction genes23 revealed significant enrichment of angiogenesis-related genes including ANGPT2 which was found to be overexpressed in the ZAP-70 þ CD38 þ CLL subgroup when all patients as listed in Table 1 were analysed. To further validate these results, we measured ANGPT2 concentrations in the plasma of 56 CLL patients. In accordance with the microarray experiments, we found a twofold increase of ANGPT2 plasma levels in the ZAP70 þ CD38 þ as compared to the ZAP-70CD38 patient subgroup. Furthermore, ANGPT2 plasma concentrations detected in Binet B and C patients exceeded those of Binet stage A patients suggesting that ANGPT2 protein secretion also may be correlated with disease progression. This observation is relevant in light of experiments showing that VEGF and basic fibroblast growth factor secreted by B-CLL cells induced bone marrow neoangiogenesis,40–42 thereby adding to the maintenance of the leukaemic clone. As ANGPT2 acts in concert with VEGF as a proangiogenic stimulus,43–45 we also measured VEGF plasma concentrations in our study population. Confirmatory of previously published studies,40 VEGF plasma levels were found to be significantly increased in the ZAP-70 þ CD38 þ as compared to the ZAP-70CD38 patient cohort. In another example of a non-solid malignancy, work by Loges et al.46 hinted towards ANGPT2 mRNA expression levels in the leukaemic blasts of patients with newly diagnosed acute myeloid leukaemia as independent prognostic factor in this disease. This finding together with our observation of concordance between ANGPT2 and ZAP-70/CD38 expression status raises the possibility that ANGPT2 could also serve as a novel prognostic marker in B-CLL. Potentially important, our analysis identified the serine/ threonine kinase Pim2 as another B-CLL subtype distinction gene that was found to be overexpressed in the ZAP70 þ CD38 þ B-CLL cases. In line with data published by Cohen et al.,31 our RQ-PCR analysis in PBMCs from 94 CLL patients (and in 45 highly purified CLL samples) revealed significantly higher Pim2 expression levels in B-CLL cells as compared to normal donor-derived PBMCs. Pim2 mRNA concentrations in ZAP-70 þ CD38 þ B-CLL cells were significantly higher as compared to ZAP-70CD38 samples when purified cells were tested. Also, we found a strong and statistically significant positive correlation of Pim2 expression with the Binet stage. This indicates that Pim2 may be upregulated in the more aggressive subgroup as well as in the course of the disease and therefore is a causative player in disease progression. Our results are confirmatory of findings in a series of 48 non-Hodgkin’s lymphoma and B-CLL cases where a correlation of Pim2 expression levels in B-CLL cells with lymphocyte doubling time and Binet stage was reported and expand these to a larger

patient cohort.31 Pim2 functions as an inhibitor of apoptosis and is transcriptionally regulated by a variety of proliferative signals, promoting cell survival by inducing NF-kB-dependent gene expression.27 Acting synergistically with CMYC and NMYC in generating T- and B-cell lymphomas, the precise role of Pim2 in lymphomagenesis remains to be elucidated.28–32 Thus, based on the data initially reported by Cohen et al.31 and the results presented herein Pim2 could serve as a promising target for the development of a novel treatment strategy in B-CLL.

1781

Acknowledgements We thank Anja Fu¨hrer and Ute Schmu¨cker for excellent technical assistance and Brigitte Fischer for help with compiling patient data. This work is dedicated to Professor G Brittinger on the occasion of his 75th birthday. This work was supported by a ‘Deutsche Krebshilfe’ grant provided to JD and AC.

References 1 Chiorazzi N, Rai KR, Ferrarini M. Chronic lymphocytic leukemia. N Engl J Med 2005; 352: 804–815. 2 Keating MJ. Chronic lymphocytic leukemia. Semin Oncol 1999; 26 (Suppl 14): 107–114. 3 Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood 1999; 94: 1848– 1854. 4 Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL et al. Ig V gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood 1999; 94: 1840–1847. 5 Shanafelt TD, Geyer SM, Kay NE. Prognosis at diagnosis: integrating molecular biologic insights into clinical practice for patients with CLL. Blood 2004; 103: 1202–1210. 6 Klein U, Tu Y, Stolovitzky GA, Mattioli M, Cattoretti G, Husson H et al. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med 2001; 194: 1625–1638. 7 Rosenwald A, Alizadeh AA, Widhopf G, Simon R, Davis RE, Yu X et al. Relation of gene expression phenotype to immunoglobulin mutation genotype in B cell chronic lymphocytic leukemia. J Exp Med 2001; 194: 1639–1647. 8 Vasconcelos Y, De Vos J, Vallat L, Reme T, Lalanne AI, Wanherdrick K et al. Gene expression profiling of chronic lymphocytic leukemia can discriminate cases with stable disease and mutated Ig genes from those with progressive disease and unmutated Ig genes. Leukemia 2005; 19: 2002–2005. 9 Chen L, Widhopf G, Huynh L, Rassenti L, Rai KR, Weiss A et al. Expression of ZAP-70 is associated with increased B-cell receptor signaling in chronic lymphocytic leukemia. Blood 2002; 100: 4609–4614. 10 Chen L, Apgar J, Huynh L, Dicker F, Giago-McGahan T, Rassenti L et al. ZAP-70 directly enhances IgM signaling in chronic lymphocytic leukemia. Blood 2005; 105: 2036–2041. 11 Crespo M, Bosch F, Villamor N, Bellosillo B, Colomer D, Rozman M et al. ZAP-70 expression as a surrogate for immunoglobulinvariable-region mutations in chronic lymphocytic leukemia. N Engl J Med 2003; 348: 1764–1775. 12 Du¨rig J, Nu¨ckel H, Cremer M, Fu¨hrer A, Halfmeyer K, Fandrey J et al. ZAP-70 expression is a prognostic factor in chronic lymphocytic leukemia. Leukemia 2003; 17: 2426–2434. 13 Schroers R, Griesinger F, Tru¨mper L, Haase D, Kulle B, KleinHitpass L et al. Combined analysis of ZAP-70 and CD38 expression as a predictor of disease progression in B-cell chronic lymphocytic leukemia. Leukemia 2005; 19: 750–758. 14 Rassenti LZ, Huynh L, Toy TL, Chen L, Keating MJ, Gribben JG et al. ZAP-70 compared with immunoglobulin heavy-chain gene mutation status as a predictor of disease progression in chronic lymphocytic leukemia. N Engl J Med 2004; 351: 893–901. Leukemia


1782 15 Del Poeta G, Maurillo L, Venditti A, Buccisano F, Epiceno AM, Capelli G et al. Clinical significance of CD38 expression in chronic lymphocytic leukemia. Blood 2001; 98: 2633–2639. 16 Du¨rig J, Naschar M, Schmu¨cker U, Renzing-Kohler K, Ho¨lter T, Hu¨ttmann A et al. CD38 expression is an important prognostic marker in chronic lymphocytic leukaemia. Leukemia 2002; 16: 30–35. 17 Hamblin TJ, Orchard JA, Gardiner A, Oscier DG, Davis Z, Stevenson FK. Immunoglobulin V genes and CD38 expression in CLL. Blood 2000; 95: 2455–2457. 18 Deaglio S, Capobianco A, Bergui L, Du¨rig J, Morabito F, Du¨hrsen U et al. CD38 is a signaling molecule in B-cell chronic lymphocytic leukemia cells. Blood 2003; 102: 2146–2155. 19 Deaglio S, Vaisitti T, Bergui L, Bonello L, Horenstein AL, Tamagnone L et al. CD38 and CD100 lead a network of surface receptors relaying positive signals for B-CLL growth and survival. Blood 2005; 105: 3042–3050. 20 Du¨rig J, Nu¨ckel H, Hu¨ttmann A, Kruse E, Ho¨lter T, Halfmeyer K et al. Expression of ribosomal and translation-associated genes is correlated with a favorable clinical course in chronic lymphocytic leukemia. Blood 2003; 101: 2748–2755. 21 Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001; 98: 5116–5121. 22 Nu¨ckel H, Hu¨ttmann A, Klein-Hitpass L, Schroers R, Fu¨hrer A, Sellmann L et al. LPL expression is a novel prognostic factor in BCLL. Leukemia Lymphoma 2006; 47: 1053–1061. 23 Beissbarth T, Speed TP. GOstat: find statistically overrepresented Gene Ontologies within a group of genes. Bioinformatics 2004; 20: 1464–1465. 24 Guselnikov SV, Ershova SA, Mechetina LV, Najakshin AM, Volkova OY, Alabyev BY et al. A family of highly diverse human and mouse genes structurally links leukocyte FcR, gp42 and PECAM-1. Immunogenetics 2002; 54: 87–95. 25 Leu CM, Davis RS, Gartland LA, Fine WD, Cooper MD. FcRH1: an activation coreceptor on human B cells. Blood 2005; 105: 1121– 1126. 26 Xu MJ, Zhao R, Zhao ZJ. Molecular cloning and characterization of SPAP1, an inhibitory receptor. Biochem Biophys Res Commun 2001; 280: 768–775. 27 Fox CJ, Hammerman PS, Cinalli RM, Master SR, Chodosh LA, Thompson CB. The serine/threonine kinase Pim-2 is a transcriptionally regulated apoptotic inhibitor. Genes Dev 2003; 17: 1841–1854. 28 Hammerman PS, Fox CJ, Cinalli RM, Xu A, Wagner JD, Lindsten T et al. Lymphocyte transformation by Pim-2 is dependent on nuclear factor-kappaB activation. Cancer Res 2004; 64: 8341–8348. 29 Hammerman PS, Fox CJ, Birnbaum MJ, Thompson CB. Pim and Akt oncogenes are independent regulators of hematopoietic cell growth and survival. Blood 2005; 105: 4477–4483. 30 Allen JD, Verhoeven E, Domen J, van d V, Berns A. Pim-2 transgene induces lymphoid tumors, exhibiting potent synergy with c-myc. Oncogene 1997; 15: 1133–1141. 31 Cohen AM, Grinblat B, Bessler H, Kristt D, Kremer A, Schwartz A et al. Increased expression of the hPim-2 gene in human chronic lymphocytic leukemia and non-Hodgkin lymphoma. Leukemia Lymphoma 2004; 45: 951–955.

32 Mizuki M, Schwable J, Steur C, Choudhary C, Agrawal S, Sargin B et al. Suppression of myeloid transcription factors and induction of STAT response genes by AML-specific Flt3 mutations. Blood 2003; 101: 3164–3173. 33 Del GI, Morilla A, Osuji N, Matutes E, Morilla R, Burford A et al. Zeta-chain associated protein 70 and CD38 combined predict the time to first treatment in patients with chronic lymphocytic leukemia. Cancer 2005; 104: 2124–2132. 34 Oppezzo P, Vasconcelos Y, Settegrana C, Jeannel D, Vuillier F, Legarff-Tavernier M et al. The LPL/ADAM29 expression ratio is a novel prognosis indicator in chronic lymphocytic leukemia. Blood 2005; 106: 650–657. 35 van’t Veer MB, Brooijmans AM, Langerak AW, Verhaaf B, Goudswaard CS, Graveland WJ et al. The predictive value of lipoprotein lipase for survival in chronic lymphocytic leukemia. Haematologica 2006; 91: 56–63. 36 Porakishvili N, Kulikova N, Jewell AP, Youinou PY, Yong K, Nathwani A et al. Differential expression of CD 180 and IgM by B-cell chronic lymphocytic leukaemia cells using mutated and unmutated immunoglobulin VH genes. Br J Haematol 2005; 131: 313–319. 37 Davis RS, Wang YH, Kubagawa H, Cooper MD. Identification of a family of Fc receptor homologs with preferential B cell expression. Proc Natl Acad Sci USA 2001; 98: 9772–9777. 38 Ehrhardt GR, Davis RS, Hsu JT, Leu CM, Ehrhardt A, Cooper MD. The inhibitory potential of Fc receptor homolog 4 on memory B cells. Proc Natl Acad Sci USA 2003; 100: 13489–13494. 39 Miller I, Hatzivassiliou G, Cattoretti G, Mendelsohn C, DallaFavera R. IRTAs: a new family of immunoglobulin like receptors differentially expressed in B cells. Blood 2002; 99: 2662–2669. 40 Kay NE, Bone ND, Tschumper RC, Howell KH, Geyer SM, Dewald GW et al. B-CLL cells are capable of synthesis and secretion of both pro- and anti-angiogenic molecules. Leukemia 2002; 16: 911–919. 41 Kay NE. The angiogenic status of B-CLL B cells: role of the VEGF receptors. Leukemia Res 2004; 28: 221–222. 42 Lee YK, Shanafelt TD, Bone ND, Strege AK, Jelinek DF, Kay NE. VEGF receptors on chronic lymphocytic leukemia (CLL) B cells interact with STAT 1 and 3: implication for apoptosis resistance. Leukemia 2005; 19: 513–523. 43 Holash J, Maisonpierre PC, Compton D, Boland P, Alexander CR, Zagzag D et al. Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF. Science 1999; 284: 1994– 1998. 44 Lobov IB, Brooks PC, Lang RA. Angiopoietin-2 displays VEGFdependent modulation of capillary structure and endothelial cell survival in vivo. Proc Natl Acad Sci USA 2002; 99: 11205–11210. 45 Vajkoczy P, Farhadi M, Gaumann A, Heidenreich R, Erber R, Wunder A et al. Microtumor growth initiates angiogenic sprouting with simultaneous expression of VEGF, VEGF receptor-2, and angiopoietin-2. J Clin Invest 2002; 109: 777–785. 46 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.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia