Increased Bone Marrow Microvascular Density in Haematological Malignancies Is Associated with Differential Regulation of Angiogenic Factors

Increased bone marrow microvascular density in haematological malignancies is associated with differential regulation of angiogenic factors

HFS Negaard1,2, N Iversen3, IM Bowitz-Lothe4, PM Sandset1,2, B Steinsvik1, B Østenstad5 and PO Iversen1,2

1Department of Haematology, Ulleva˚l University Hospital Trust, Oslo, Norway; 2Faculty of Medicine, University of Oslo, Oslo, Norway; 3Department of Medical Genetics, Ulleva˚l University Hospital Trust, Oslo, Norway; 4Department of Pathology, Ulleva˚l University Hospital Trust, Oslo, Norway and 5Cancer Centre, Ulleva˚l University Hospital Trust, Oslo, Norway

Antiangiogenic drugs are currently tested in haematological Despite advances in treatment, chronic lymphatic leukaemia malignancies. As these drugs target different angiogenic (CLL), MM and indolent subtypes of NHL are still regarded to be regulators, and as cancers are inherently heterogeneous, a detailed characterization of angiogenesis in individual incurable diseases, and only half of the patients with acute cancers is needed. Hence, we measured bone marrow myeloid leukaemia (AML) and aggressive subtypes of NHL are 8 microvessel density (MVD), plasma concentrations of eight alive after 5 years. Given the role of angiogenesis in angiogenesis-related parameters and the expression in blood haematological neoplasias, several angiogenesis-inhibitors have mononuclear cells of 40 angiogenesis-related mRNAs in 93 been tested in clinical trials, but many have not lived up to patients with haematological neoplasias (acute myeloid expectations.9 As these drugs have different targets, and as leukaemia; chronic lymphatic leukaemia; multiple myeloma (MM); or non-Hodgkin’s lymphoma (NHL)) before start and cancers are heterogeneous, a detailed characterization of after completion of cancer therapy. Compared with healthy angiogenesis in separate cancer diagnoses is needed. individuals, the patients had significantly increased bone The purpose of this prospective cohort study was to determine marrow MVD, especially patients with advanced stage disease. the bone marrow MVD in unselected patients with various A novel finding was that patients with NHL also had increased haematological malignancies. In particular, we wanted to bone marrow MVD. The plasma levels of vascular endothelial investigate whether MVD was increased, especially in NHL, growth factor (VEGF), interleukin (IL)-6 and IL-8 were significantly increased. VEGF levels were highest in those who did and if MVD was associated with disease stage and remission not achieve complete remission after cancer therapy. The status. To further characterize the angiogenic process in mRNA expression of IL-8 was upregulated 15-fold. Our data haematological malignancies, we measured plasma concen- show that patients with haematological malignancies have trations of eight angiogenesis-related markers, with regard to increased bone marrow MVD; hence, supporting the notion that diagnosis, stage and remission, as well as longitudinally during bone marrow angiogenesis plays a role in the pathogenesis treatment. Finally, as various blood cells secrete pro-angiogenic and progression of these cancers. VEGF, IL-6 and IL-8 seem to 10 contribute to the malignant phenotype. molecules, we also studied the mRNA expression of angiogenesis- Leukemia (2009) 23, 162–169; doi:10.1038/leu.2008.255; related genes in peripheral blood mononuclear cells. published online 18 September 2008 Keywords: angiogenesis; lymphoma; myeloma; bone marrow; VEGF Materials and methods

Patients and controls The patient material and study design have previously been Introduction 11 described in detail. Briefly, we studied 93 patients (40 females, 53 males; median age 62 years, range 28–89 years) with AML Emerging evidence suggests that bone marrow microvascular (n ¼ 20), CLL (n ¼ 14), MM (n ¼ 11) and NHL (n ¼ 48), included density (MVD) is increased in leukaemia and multiple myeloma in the ‘Angiogenesis and Haemostasis in Haematological (MM),1–3 and that this may have prognostic value.3,4 However, Neoplasia’ study. Patients were newly diagnosed, apart from whether the bone marrow MVD in non-Hodgkin’s lymphoma 10 patients with relapsed NHL. The malignancies were (NHL) is increased compared with that of healthy individuals, 12–14 diagnosed according to internationally accepted criteria. has not been clarified. From 7 December 2004 to 26 January 2006, patients were Although current understanding of the complex regulation of prospectively and consecutively recruited from Ulleva˚l and angiogenesis emphasizes the importance of vascular endothelial Rikshospitalet University Hospitals, Oslo, Norway. Patients growth factor (VEGF), several other pro-angiogenic factors may were excluded from the study in case of pregnancy, previous be involvedFincluding fibroblast growth factor 2 (FGF2; or or current other form of cancer (except squamous cell basic FGF), interleukin (IL)-6, IL-8, tumour necrosis factor-a carcinoma of the skin and stage 0 cervical cancer), the (TNF-a), angiogenin and tissue factor (TF)5,6Fin addition to promyelocytic subtype of AML, incapability to consent, other antiangiogenic factors, such as TF pathway inhibitor type 1 bone marrow diseases, HIV, other uncontrolled infectious (TFPI-1).7 diseases, use of immunomodulatory drugs or having received chemotherapy within the previous 3 months. Correspondence: Dr HFS Negaard, Department of Haematology, Staging of patients were according to internationally accepted ˚ Ulleval University Hospital Trust, Kirkeveien 166, N-0407 Oslo, classifications.14–16 The following were considered as advanced Norway. E-mail: [email protected] stage disease: Binet stage C CLL, stage 3 MM and Ann Arbor Received 29 May 2008; revised 26 August 2008; accepted 28 August stage IV NHL. Remission status was evaluated at completion of 2008; published online 18 September 2008 cancer therapy according to conventional criteria.14,17–19 After Increased bone marrow MVD in haematological malignancies HFS Negaard et al 163

Figure 1 Bone marrow biopsy stained for CD34 (brown) for microvessel density (MVD) counting. Magniﬁcation at Â 400. Patients with non- Hodgkin’s lymphoma (NHL) with (a) diffuse large B cell subtype, stage II; (b) follicular lymphoma subtype, stage IV. Myeloid blasts that stained for CD34 (arrows point at examples) were not counted as microvessels.

cancer treatment, 44 of the 93 patients (47%) obtained complete sections of tonsils as positive control, and healthy bone marrow remission (CR). For clinical features of consecutive patients with as normal control, were run. NHL, see Supplementary Table 1. Archival bone marrow specimens from 16 patients were used as controls for the bone marrow MVD estimations. None of Estimation of MVD the patients had previous cancer, and neither were any All counting was done in a blinded manner, by the simultaneous diagnosed with cancer after the bone marrow biopsy (minimum assessment of two investigators using a double-headed light follow-up: 1 year). The controls consisted of nine females and microscope (Ernst Leitz, Wetzlar, Germany), as described.20 seven males; with median age of 54 years, range 32–90 years. Briefly, slides were scanned at Â 100 magnification to As reference for the plasma levels and mRNA expressions, we determine three ‘hot spots’ defined as areas with the maximum used samples from 11 healthy individuals (ten males and one number of microvessels. At Â 400 magnification, the field was female; median age 23 years, range 21–29 years). All control set to cover the maximum number of microvessels within the hot individuals were subjected to the same exclusion criteria as the spot, and microvessels were counted in one field (1.13 mm2)in patients. each of the three hot spots. Both investigators agreed on what The protocol was approved by the Norwegian Regional constituted a single microvessel before any vessel was included Committee for Research Ethics in Health Region East. Written in the count. Microvessels were defined as endothelial cells, informed consent was obtained before inclusion. either single or clustered in nests or tubes, clearly separated from adjacent microvessels, with or without a lumen. Great care was taken to omit the counting of stained non-endothelial cells such Bone marrow microvessel staining as myeloid blasts. In Figure 1a stained bone marrow biopsy from Bone marrow biopsy specimens were fixed, decalcified with a patient is shown; arrows point at examples of stained myeloid EDTA and embedded in paraffin. Pretreatment bone marrow blasts that were not counted. We were unable to measure MVD biopsies were evaluated in 72 of the patients (77%). The CD34 in seven biopsies from patients with AML due to large numbers antigen specific for endothelial cells,20 was used as a marker of of stained myeloid blasts. Large vessels or vessels in the vascular endothelial cells. Immunohistochemical staining for periosteum or bone were excluded. MVD is reported as the CD34 was performed by a two-step polymer-based method run median of the three hot spots and expressed as the number of on a Dako Autostainer (Dako, Glostrup, Denmark). Specimens microvessels per mm2. were deparaffinized in xylene and rehydrated in serially graded ethanol. Tissue sections were then pretreated in a microwave oven with a tris/EDTA solution, pH 9.1, for 24 min, followed by Blood sampling and processing a 30 min cooling. Tissue sections were incubated with a 1:50 Blood samples were collected prior to and after end of cancer dilution of the primary monoclonal anti-CD34 (Monoclonal therapy, and stored at À70 1C until assayed. Mouse Anti-Human CD34 Class II Clone QBEnd 10, Dako) for Peripheral blood mononuclear cells (PBMNC) were extracted 30 min at room temperature. The EnVision þ System-HRP from EDTA whole blood samples by density gradient centri- detection kit (Dako) was used for antigen visualization. Sections fugation (Lymphoprep, Axis-Shield, Oslo, Norway) within 2 h of were counterstained with haematoxylin. On each slide, paraffin sampling. They were washed with 0.9% NaCl, lysed with lysis/

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 164 binding buffer (MagNA Pure LC RNA Isolation kit, Roche process. The relative quantification (RQ) of mRNA expressions Applied Sciences, Penzberg, Germany) and stored at À70 1C. were determined by employing the Comparative cycle threshold 22 Later, ribonucleic acid (RNA) was extracted with MagNA Pure (Ct) method, calculated using RQ Manager 1.2 (Applied LC Instrument (Roche Applied Sciences) using the MagNA Pure Biosystems) with a common threshold setting for each gene LC RNA Isolation kit (Roche Applied Sciences). The amount of target. The Ribosomal Protein Large P0 (RPLP0) gene was RNA was quantified using the NanoDrop Spectrophotometre selected as endogenous control as it showed the least scattering (NanoDrop Technologies, Wilmington, DE, USA). in Ct values when running cDNA from the 11 healthy controls All assays were performed examiner-blind. Samples were run and 21 randomly selected patients on a TLDA endogenous in batch by the end of the study with both samples for the same control plate (Applied Biosystems). All targets were normalized individual in each run. to the RPLP0 gene, to calculate the DCt. The normalized results were compared with the normalized results of a calibrator (RNA from the subject among the healthy controls with median value Plasma assays of RPLP0 expression) yielding the DDCt ¼ DCt (sample) ÀDCt VEGF, IL-6, IL-8, FGF2 and TNF-a were assayed in EDTA (calibrator). plasma using the Luminex Multiplex assay with Bio-Plex RNA quality was assessed using an Agilent 2100 Bioanalyzer Cytokine assay reagents (Bio-Rad Laboratories, Hercules, CA, (Agilent Technologies, Santa Clara, CA, USA) with reagents from USA) on a Bio-Plex Luminex xMAP (Bio-Rad Laboratories) as the RNA 6000 Nano LabChip kit (Agilent Technologies), described by the manufacturer. Angiogenin was assayed in calculating the RNA integrity number (RIN). Data were EDTA plasma with Human Angiogenin Flex Set reagents (BD excluded if DRn was (fluorescence intensity minus background Biosciences, San Diego, CA, USA) on a BD FACSArray fluorescence) p1.35, if the amplification curve was non- Bioanalyzer (BD Biosciences). In citrated plasma TF antigen optimal, or if both the RNA integrity number were o4.0 and was assayed with the Imubind Tissue Factor ELISA kit (American the Ct440. Consequently mRNA expression results were Diagnostica, Stamford, CT, USA), TFPI-1 free and total antigen available in 81 (87%) patients. In 43 (46%) patients all 40 gene with Asserachrom Free TFPI and Asserachrom Total TFPI target expressions were available and in 12 (13%) patients 39 (Diagnostica Stago, Asnie`re, France) and TFPI-1 activity with gene target expressions were available. an in-house method.21 In 89 (96%) patients all plasma results were available. Interassay coefficients of variation for individual assays were: Statistical analyses VEGF 5.3%, IL-6 16%, IL-8 9.2%, FGF2 12%, TNF-a 12%, Variables that had normal distribution were reported as means angiogenin 13%, TF 9.9%, total TFPI-1 antigen 2.3%, free and standard deviations. Medians and interquartile intervals TFPI-1 antigen 9.7% and TFPI-1 activity 3.5%. were reported for the following variables whose distribution deviated markedly from the normal distribution: plasma levels of TNF-a, IL-6 and IL-8; and mRNA expression of angiogenic Gene expression analyses growth factor with G patch and FHA domains 1 (AGGF1), Equal amounts (750 ng) of RNA from PBMNC were reversely angiopoietin (ANGPT) 1 and 2, interferon-a (IFNA1), IL-6, transcribed into cDNA using the High-Capacity cDNA Archive platelet-derived growth factor (PDGF) A and B, placental growth Kit (Applied Biosystems, Foster City, CA, USA). Equal amounts factor (PGF) and TNF a-induced protein 2 (TNFAIP2). Compa- (150 ng) of cDNA from each sample were run with TaqMan Low risons between independent groups were performed with the Density Arrays (TLDA; Applied Biosystems) on an ABI PRISM t-test or the Mann–Whitney test, as appropriate. Differences 7900 HT Sequence Detection System RT-PCR (Applied Biosys- between diagnostic groups were evaluated using one-way tems), according to the manufacturer’s instructions. ANOVA or the Kruskal–Wallis test. Comparisons between After scrutinizing relevant reports, we chose to quantify 40 values obtained before and after cancer therapy were evaluated genes because of their possible influence on the angiogenic with the paired samples t-test or Wilcoxon signed rank test.

Table 1 Bone marrow MVD and plasma levels of angiogenic markers before cancer therapy

Patients Controls P-value

Mean/median* s.d./IQR* Mean/median* s.d./IQR*

MVD (microvessels per mm2)3916208o0.001 VEGF (pg/ml) 31.30 18.76 18.50 7.70 o0.001 FGF2 (pg/ml) 14.95 15.16 8.03 6.90 0.139 TNF-a (pg/ml) 4.96* 3.01–9.17* 3.01* 2.73–6.56* 0.062 IL-6 (pg/ml) 4.56* 2.50–10.76* 2.29* 1.23–6.73* 0.034 IL-8 (pg/ml) 7.18* 5.36–10.02* 3.94* 2.90–5.87* 0.001 Angiogenin (ng/ml) 710 228 582 154 0.073 TF (pg/ml) 167.8 174.9 166.1 90.2 0.974 Free TFPI-1 (ng/ml) 17.0 6.9 11.0 2.8 0.005 Total TFPI-1 (ng/ml) 83.8 18.9 75.9 9.2 0.177 TFPI-1 activity (%) 122.9 22.8 110.6 9.6 0.004 Abbreviations: FGF2, ﬁbroblast growth factor 2; IL, interleukin; MVD, microvessel density; TF, tissue factor; TFPI-1, TF pathway inhibitor 1; TNF-a, tumour necrosis factor-a; VEGF, vascular endothelial growth factor. Values presented as mean and standard deviation (s.d.) for markers having normal distribution and evaluated with t-test, and median and interquartile range (IQR) marked with asterix (*) for markers not having normal distribution evaluated with Mann–Whitney test. For MVD n ¼ 65 for patients and n ¼ 16 for controls, for the plasma levels n ¼ 93 for patients and n ¼ 11 for controls.

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 165 Table 2 for separate haematological malignancies). In addition, they had significantly increased plasma levels of VEGF, IL-6, IL-8, as well as elevated levels of both free TFPI-1 antigen and TFPI-1 activity (Table 1). On the other hand, we found no significant differences in plasma levels of FGF2, TNF-a, angiogenin and TF between patients and controls (Table 1, Supplementary Table 2 for separate haematological malignancies). The mRNA expression of IL-8 was upregulated 15-fold, whereas the following mRNAs were downregulated fourfold or more: notch homologue 4 (NOTCH4), VEGFC, ANGPT2, chemokine ligand 2 (CCL2), TFPI2 and TFPI1 (Figure 2, Supplementary Figure 1 for separate haematological malignancies). We have confined the presented data of mRNA expression to these most up/downregulated genes. Owing to the apparent lack of reported data on bone marrow MVD measurements in NHL, we were specifically interested in bone marrow MVD in NHL patients. Before cancer therapy, these patients had higher MVD (mean 38 microvessels per mm2) than healthy controls (mean 20 microvessels per mm2, Po0.001).

Angiogenic markers before cancer therapy according to diagnosis Haematological malignancies represent distinct diseases with specific characteristics. We found no significant statistical Figure 2 mRNA expression of angiogenesis-related genes in peri- differences in bone marrow MVD between the diagnoses or in pheral blood mononuclear cells (PBMNC) from patients with the plasma levels of VEGF (Figures 3a and b). There were haematological malignancies (n ¼ 93) before cancer therapy. Variables significant differences between the diagnoses in plasma levels of having normal distribution are presented as mean (square) and 95% FGF2, TNF-a, IL-6 and IL-8. Patients with CLL had the highest confidence interval (CI), whereas variables not having normal plasma levels of FGF2, whereas patients with NHL had the distribution are presented as median (circle) and 95% CI. All targets were normalized to the endogenous gene ribosomal protein large P0, highest plasma levels of TNF-a, and patients with AML had the and compared with mRNA from PBMNC from a healthy individual. highest plasma levels of IL-6 and IL-8 (Figures 3c–f), and lowest The results were calculated using the Comparative Ct method, plasma levels of total TFPI-1 (mean 72 ng/ml for AML, 87 ng/ml generating relative quantitation (RQ) data, which are presented on a for CLL, 82 ng/ml for MM and 88 ng/ml for NHL; P ¼ 0.009). In log2 scale. Data are sorted according to RQ. AGGF, angiogenic factor contrast to the non-significant differences in bone marrow MVD with G patch and FHA domains; AKT, v-akt murine thymoma viral between the diagnoses; 32 of the 40 mRNA expressions oncogene homologue; ANGPT, angiopoietin; ANGPTL4, ANGPT-like 4; ANG, angiogenin; CCL, chemokine C-Cmotif ligand; EPAS, measured, were statistically significantly different among the endothelial PAS domain protein; EPH, ephrin receptor; FGF, fibroblast diagnoses (data not shown). growth factor; FIGF, c-fos induced growth factor (VEGFD); HIF, hypoxia-inducible factor; IFN, interferon; IL, interleukin; JAG, jagged; NOTCH, notch homologue; PDGF, platelet-derived growth factor; Angiogenic markers before cancer therapy according to PGF, placental growth factor; SERPIN, serpin peptidase inhibitor; SH2D2A, SH2 domain protein 2A; TF, tissue factor; TFPI, TF pathway disease stage inhibitor; TGF, transforming growth factor; TNF, tumour necrosis Patients with advanced stage disease had higher bone marrow factor; TNFAIP, TNF a-induced protein; TNFSF15, TNF superfamily MVD than patients with lower stage disease (mean 45 vs 35 member 15; TP53, tumour protein p53; TYMP, thymidine phospho- microvessels per mm2, P ¼ 0.030). An example illustrating this is rylase (or endothelial cell growth factor); VEGF, vascular endothelial shown in Figure 1. Advanced stage disease patients had higher growth factor. plasma levels of IL-6 (median 7.95 vs 3.44 pg/ml; P ¼ 0.004) and IL-8 (median 7.58 vs 6.46 pg/ml; P ¼ 0.038) before treatment, whereas there were no statistical differences according to stage Correlation between continuous variables was studied with for the other plasma levels measured (data not shown). Patients Pearson’s correlation (r) or Spearman’s rank correlation (rs). All with advanced stage disease had lower mRNA expression of tests were two-tailed and a 5% significance level was applied CCL2 than those with low stage disease (mean 9-fold down- except for the correlation analyses in which a 1% significance regulated vs 2-fold downregulated; P ¼ 0.035). level was applied. Statistical analyses were performed using the Statistical Package for the Social Sciences 14.02 (SPSS Inc., Chicago, IL, USA). Angiogenic markers before cancer therapy according to remission status We found no differences in pretreatment bone marrow MVD Results between patients who did not obtain CR compared with those who did (mean 39 vs 38 microvessels per mm2). Patients who Increased MVD and pro-angiogenic markers before did not obtain CR after cancer therapy had significantly higher cancer therapy pretreatment plasma levels of VEGF (mean 35.11 vs 27.07 pg/ml; Before start of cancer therapy, patients had significantly higher P ¼ 0.038) and FGF2 (mean 18.76 vs 10.71 pg/ml; P ¼ 0.008) bone marrow MVD than controls (Table 1, Supplementary than patients who did obtain CR, whereas the other plasma

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 166

Figure 3 MVD and plasma levels of angiogenic markers before cancer therapy. Comparison among the four haematological malignancies (number given in parenthesis under each group); controls as reference. Variables having normal distribution (MVD, VEGF and FGF2) are presented as means and standard deviations, and variables not having normal distribution (TNF-a, IL-6 and IL-8) are presented as medians and interquartile intervals. P-values: (a) MVD, P ¼ 0.81; (b) VEGF, P ¼ 0.798; (c) FGF2, P ¼ 0.001; (d) TNF-a, P ¼ 0.040: (e) IL-6, P ¼ 0.041; (f) IL-8, P ¼ 0.012. Signiﬁcant results marked with asterix (*). AML, acute myeloid leukaemia; CLL, chronic lymphatic leukaemia; MM, multiple myeloma; NHL, non- Hodgkin’s lymphoma; MVD, microvessel density; VEGF, vascular endothelial growth factor; FGF2, ﬁbroblast growth factor 2; TNF-a, tumour necrosis factor-a; IL, interleukin.

levels were not signiﬁcantly different (data not shown). For their respective mRNA expressions in PBMNC (data not shown). results in the separate haematological malignancies, see There were, however, strong correlations between VEGFA Supplementary Table 3. Additionally, non-CR patients had mRNA expression and both hypoxia-inducible factor 1a (HIF1A) signiﬁcantly lower pretreatment mRNA expression of IL-8 (mean and endothelial PAS domain protein 1 (EPAS1, or HIF2A) mRNA 7-fold vs 30-fold upregulated; P ¼ 0.012) and NOTCH4 (mean expressions (r ¼ 0.7; Po0.001 for both). 47-fold vs 10-fold downregulated; P ¼ 0.029).

Changes in angiogenic markers during cancer treatment Correlation of angiogenic markers before cancer We next compared the levels of angiogenic markers before therapy treatment with those measured after completion of treatment. In 79 (85%) patients, results were available from two or more of There was an increase in plasma levels of total TFPI-1 antigen the parameters, namely bone marrow MVD, all plasma (mean 85.8 ng/ml before cancer therapy to mean 105.9 ng/ml angiogenesis factors or all mRNA angiogenesis factors. Bone after cancer therapy; P ¼ 0.014) and TFPI-1 activity (mean marrow MVD was not significantly correlated to any of the 124.6% before cancer therapy to mean 140.2% after cancer angiogenic markers measured in plasma, nor mRNA expressions therapy; Po0.001). The other plasma levels did not change (data not shown). There were no statistically significant significantly during cancer treatment (data not shown). CCL2 correlations between plasma levels of angiogenic markers and was downregulated fourfold before cancer therapy, whereas

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 167 after cancer therapy it was upregulated 1.2-fold (Po0.001). In been demonstrated in patients with AML,1 CLL,3 and MM.25 addition, the mRNA expression of VEGFC increased during Higher bone marrow MVD in patients with advanced stage treatment from six- to threefold downregulation (P ¼ 0.003). disease has been demonstrated previously in patients with MM.2 We did not find any association between bone marrow MVD and CR, in accordance with some MM studies;26 but in contrast Angiogenic markers after cancer therapy to other studies on MM,4 and CLL.3 After cancer treatment, patients had higher plasma levels of IL-8 Vascular endothelial growth factor is regarded as the most (median 7.12 pg/ml for patients vs 3.94 pg/ml for controls; important pro-angiogenic factor in cancer.5 In this study, P ¼ 0.001) and TFPI-1 activity (mean 140.2% for patients vs patients had higher pretreatment plasma levels of VEGF than 110.6% for controls; Po0.001), whereas the remaining plasma controls, with highest levels among those who did not achieve levels were not significantly different from controls (data not CR. However, elevated pretreatment levels of VEGF does not shown). The relative mRNA expression of IL-8 was strongly directly imply that anti-VEGF drugs are effective, as pretreat- upregulated (mean 20-fold) also after cancer therapy, and PGF ment plasma levels of VEGF did not correlate with response to with median 18-fold was upregulated. The most downregulated the addition of the anti-VEGF monoclonal antibody bevacizu- mRNA after cancer therapy was IL-6 (median 9-fold). mab in metastatic colorectal cancer or advanced pancreatic The patients who did not obtain CR had significantly higher cancer.27 We measured VEGF levels in plasma as 80–90% of the plasma levels than those who did obtain CR of IL-6 (median 6.56 VEGF levels measured in serum originate from ex vivo activated vs 3.50 pg/ml; P ¼ 0.016) and IL-8 (median 9.02 vs 6.44 pg/ml; platelets.28 Increased levels of circulating VEGF have been Po0.001). CCL2 was downregulated in non-CR patients (mean demonstrated previously in patients with the haematological twofold) and upregulated in CR patients (mean twofold; malignancies in this study.29–32 Higher pretreatment plasma P ¼ 0.021). In addition, non-CR patients had significantly lower levels of VEGF in patients who did not obtain CR has been expression of NOTCH4 (mean 19-fold vs 2-fold downregulated; demonstrated previously in AML29 and NHL.32,33 P ¼ 0.002). Interleukin-6 is another activator of angiogenesis5 and an important growth factor for myeloma cells. IL-8 is a pro- angiogenic chemokine, which serves as a growth factor in Discussion several tumour models.5 Levels of IL-6 and IL-8 were elevated in plasma, with highest levels in patients with advanced stage In this study, we have demonstrated that patients with disease. IL-6 and IL-8 plasma levels were higher among non-CR haematological malignancies had higher pretreatment bone patients than CR patients after cancer therapy. We found IL-8 to marrow MVD than controls, and elevated pretreatment plasma be the most upregulated gene. IL-6 levels in plasma have levels of VEGF, IL-6 and IL-8. IL-8 was the most upregulated previously been shown to be elevated in patients with NHL32 gene in PBMNC. There were no differences in bone marrow and CLL.34 Furthermore, elevated IL-6 plasma levels are MVD between the haematological malignant diseases, but the apparently associated to more advanced CLL stage35 as well bone marrow microvascularity seems to be differentially as to poor outcome in both NHL32 and CLL.34,35 In contrast, regulated in the various malignancies studied as we found plasma IL-6 reportedly did not correlate with treatment response significant differences between the diagnoses for the majority of in patients with MM,36 and no significant differences were found angiogenesis-related markers studied. according to MM stage.37 Plasma IL-8 has been reported to be The strengths of this study are that we investigated con- increased in MM,38 as well as increased in advanced stage CLL secutive, unselected patients with various haematological and associated to poorer survival.39 malignancies longitudinally, and with several surrogate markers FGF2 is a heparin-binding growth factor with pro-angiogenic of angiogenesis. Evaluation of MVD is widely used in studies of properties.5 In this study the plasma level of FGF2 was higher angiogenesis. Its major limitation is that it usually involves a before cancer therapy in patients who did not obtain CR, which subjective selection of areas for microvessel enumeration and is in line with previous studies reporting an association between vessel evaluation.23 Although computerized MVD evaluation FGF2 levels and poor prognosis in NHL.33 TNF-a is a has been developed, several limitations are also associated with multifunctional pro-inflammatory cytokine that can induce this technology.23 The grouping of different diagnoses might production of pro-angiogenic factors,5 whereas angiogenin is lead to the loss of important characteristics of separate diseases; known to promote endothelial cell invasiveness.5 However, in and the separate sample sizes were limited as were the control this study plasma levels of TNF-a and angiogenin were not groups. In addition, PBMNC contain different cell types, and the systematically altered, neither according to disease stage nor composition varies with the haematological malignancy accord- remission status. ing to whether malignant cells are present in peripheral blood. Tissue factor is the main trigger of blood coagulation in vivo, This limits the comparison between different diagnoses. There and several reports have found that TF also exerts pro- is increasing evidence for the role of the microenvironment angiogenic effects.6 In somewhat contrast to this we here found in pathogenesis of haematological malignancies. For instance, that TF was not significantly altered in the haematological in vitro data on AML suggest that osteoblasts release IL-8 thus malignancies under investigation. Moreover, we have recently enhancing angiogenesis.24 We were not able to investigate the reported that TF was apparently not responsible for the role of the influence of the microenvironment on angiogenesis hypercoagulable state observed in this cohort of patients.11 in this study. The lack of correlation between plasma levels and The antiangiogenic molecule TFPI-1 inhibits endothelial cell mRNA expression was somewhat expected, because of differ- proliferation.7 We found increased free TFPI-1 antigen and TFPI- ences between mRNA and protein, and different sources 1 activity before cancer therapy, and increased TFPI-1 activity (plasma vs PBMNC). after cancer therapy. Despite a shift in the balance between TF We found that patients had higher bone marrow MVD than and its inhibitor TFPI-1 towards an antiangiogenic phenotype, controls, especially patients with advanced stage disease. Bone we found increased angiogenesis in bone marrows. marrow MVD has, to our knowledge, not been reported for Hypoxia is reportedly an important inducer of increased NHL, whereas increased bone marrow MVD previously has VEGF expression in cancer, and is mediated by the oxygen-

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 168 sensitive transcription factors HIF1A and EPAS1.5 The role of and its correlation with various clinical, histological, and hypoxia as inducer of VEGF and angiogenesis in haematological laboratory parameters. Am J Hematol 2006; 81: 649–656. malignancies is less obvious than in solid tumours, as the 5 Dong X, Han ZC, Yang R. Angiogenesis and antiangiogenic malignant cells are dispersed in blood and bone marrow. Here therapy in hematologic malignancies. Crit Rev Oncol Hematol 2007; 62: 105–118. we demonstrated a strong correlation between VEGFA mRNA 6 He Y, Chang G, Zhan S, Song X, Wang X, Luo Y. Soluble tissue and both HIF1A and EPAS1 expressions, thus corroborating factor has unique angiogenic activities that selectively promote findings in adult T-cell lymphoma.40 migration and differentiation but not proliferation of endothelial The chemokine CCL2 induces endothelial cell migration and cells. Biochem Biophys Res Commun 2008; 370: 489–494. increases the expression of pro-angiogenic factors such as VEGF 7 Hembrough TA, Ruiz JF, Papathanassiu AE, Green SJ, Strickland and IL-8.41 However, the effect of CCL2 on neoplastic cells DK. Tissue factor pathway inhibitor inhibits endothelial cell proliferation via association with the very low density lipoprotein seems to depend on expression levels in the tumour environ- receptor. J Biol Chem 2001; 276: 12241–12248. ment: low expression promotes tumour growth and high 8 Cancer Registry of Norway. Cancer in Norway 2006FCancer 41 expression promotes antitumour activities. We found a incidence, mortality, survival and prevalence in Norway. Cancer downregulated CCL2 expression, especially in patients with Registry of Norway: Oslo, 2007, p 49. advanced stage disease, and after cancer therapy in those failing 9 Negaard H, Dahm A, Sandset PM, Iversen PO, Ostenstad B. to obtain CR. Thus, low expression with subsequent promotion Angiogenesis and hemostasis in hematological neoplasias. Curr Drug Targets 2005; 6: 683–699. of tumour growth seems to have prevailed in these malig- 10 Dirkx AEM, Oude Egbrink MGA, Wagstaff J, Griffioen AW. nancies. Earlier studies in lymph node biopsies of NHL have Monocyte/macrophage infiltration in tumors: modulators of shown that mRNA expression of CCL2 was as infrequent as in angiogenesis. J Leukoc Biol 2006; 80: 1183–1196. reactive lymphoid hyperplasia.42 11 Negaard HFS, Iversen PO, Ostenstad B, Iversen N, Holme PA, NOTCH4 functions as a receptor that regulates cell-fate Sandset PM. Hypercoagulability in patients with haematological determination, and may regulate branching morphology in the neoplasia: No apparent initiation by tissue factor. Thromb developing vascular system.43 We found NOTCH4 to be the Haemost 2008; 99: 1040–1048. 12 WHO. World Health Organization Classification of Tumours: most downregulated gene, both before and after cancer therapy, Pathology and Genetics of Tumours of Haematopoietic and especially in patients who did not obtain CR. Lymphoid Tissues, 3rd edn. IARC Press: Lyon, 2001. An apparent novel finding in this study was the increased 13 Oscier D, Fegan C, Hillmen P, Illidge T, Johnson S, Maguire P et al. bone marrow MVD in NHL. Also of note, we consistently found Guidelines on the diagnosis and management of chronic elevated bone marrow MVD among CLL patients. Of the lymphocytic leukaemia. Br J Haematol 2004; 125: 294–317. circulating molecules investigated, VEGF, IL-6 and IL-8 seem to 14 Smith A, Wisloff F, Samson D. Guidelines on the diagnosis and management of multiple myeloma 2005. Br J Haematol 2005; 132: be of greatest importance to the haematological malignancies 410–451. studied. The lack of correlation between bone marrow MVD 15 Binet JL, Auquier A, Dighiero G, Chastang C, Piguet H, Goasguen J and angiogenic markers in plasma or their mRNA expression et al. A new prognostic classification of chronic lymphocytic can be explained by angiogenesis being the result of a complex leukemia derived from a multivariate survival analysis. Cancer network of cytokines, and that distinct subsets of haematological 1981; 48: 198–206. malignancies differ in their response to cytokines.44 Our results 16 Lister TA, Crowther D, Sutcliffe SB, Glatstein E, Canellos GP, suggest that angiogenesis is differentially regulated by different Young RC et al. Report of a committee convened to discuss the evaluation and staging of patients with Hodgkin’s disease: angiogenesis-related factors in the various haematological Cotswolds meeting. J Clin Oncol 1989; 7: 1630–1636. malignancies, which has been reported earlier in acute 17 Cheson BD, Bennett JM, Kopecky KJ, Buchner T, Willman CL, lymphatic leukaemia and CLL.45 Further investigations into the Estey EH et al. Revised Recommendations of the International complex regulation of angiogenesis in haematological malig- Working Group for diagnosis, standardization of response criteria, nancies are needed. treatment outcomes, and reporting standards for therapeutic trials in acute myeloid leukemia. J Clin Oncol 2003; 21: 4642–4649. 18 Cheson BD, Bennett JM, Grever M, Kay N, Keating MJ, O’Brien S Acknowledgements et al. National Cancer Institute-sponsored Working Group guidelines for chronic lymphocytic leukemia: revised guidelines for diagnosis and treatment. Blood 1996; 87: 4990–4997. The study was financially supported with grants from the 19 Cheson BD, Horning SJ, Coiffier B, Shipp MA, Fisher RI, Connors Norwegian South-Eastern Health Authority Trust (fellowship for JM et al. Report of an international workshop to standardize HFSN), Ulleva˚l University Hospital Trust and the University of response criteria for non-Hodgkin’s lymphomas. NCI Sponsored Oslo, Norway. International Working Group. J Clin Oncol 1999; 17: 1244. 20 Vermeulen PB, Gasparini G, Fox SB, Toi M, Martin L, Mcculloch P et al. Quantification of angiogenesis in solid human tumours: an References international consensus on the methodology and criteria of evaluation. Eur J Cancer 1996; 32: 2474–2484. 1 Padro T, Ruiz S, Bieker R, Burger H, Steins M, Kienast J et al. 21 Bendz B, Andersen TO, Sandset PM. A new sensitive chromogenic Increased angiogenesis in the bone marrow of patients with acute substrate assay of tissue factor pathway inhibitor type 1. Thromb myeloid leukemia. Blood 2000; 95: 2637–2644. Res 2000; 97: 463–472. 2 Niemoller K, Jakob C, Heider U, Zavrski I, Eucker J, Kaufmann O 22 Applied Biosystems. Relative quantification of gene expression. et al. Bone marrow angiogenesis and its correlation with other ABI Prism 7700 Sequence Detection System. User Bulletin No. 2. disease characteristics in multiple myeloma in stage I versus stage 1997. II–III. J Cancer Res Clin Oncol 2003; 129: 234–238. 23 Fox SB, Harris AL. Histological quantitation of tumour angiogenesis. 3 Molica S, Vacca A, Ribatti D, Cuneo A, Cavazzini F, Levato D APMIS 2004; 112: 413–430. et al. Prognostic value of enhanced bone marrow angiogenesis in 24 Bruserud O, Ryningen A, Wergeland L, Glenjen NI, Gjertsen BT. early B-cell chronic lymphocytic leukemia. Blood 2002; 100: Osteoblasts increase proliferation and release of pro-angiogenic 3344–3351. interleukin 8 by native human acute myelogenous leukemia blasts. 4 Bhatti SS, Kumar L, Dinda AK, Dawar R. Prognostic value of bone Haematologica 2004; 89: 391–402. marrow angiogenesis in multiple myeloma: use of light microscopy 25 Rajkumar SV, Mesa RA, Fonseca R, Schroeder G, Plevak MF, as well as computerized image analyzer in the assessment of Dispenzieri A et al. Bone marrow angiogenesis in 400 patients microvessel density and total vascular area in multiple myeloma with monoclonal gammopathy of undetermined significance,

Leukemia Increased bone marrow MVD in haematological malignancies HFS Negaard et al 169 multiple myeloma, and primary amyloidosis. Clin Cancer Res 35 Lai R, O’Brien S, Maushouri T, Rogers A, Kantarjian H, Keating M 2002; 8: 2210–2216. et al. Prognostic value of plasma interleukin-6 levels in 26 Kumar S, Gertz MA, Dispenzieri A, Lacy MQ, Wellik LA, Fonseca patients with chronic lymphocytic leukemia. Cancer 2002; 95: R et al. Prognostic value of bone marrow angiogenesis in patients 1071–1075. with multiple myeloma undergoing high-dose therapy. Bone 36 Rosinol L, Cibeira MT, Segarra M, Cid MC, Filella X, Aymerich M Marrow Transplant 2004; 34: 235–239. et al. Response to thalidomide in multiple myeloma: impact of 27 Jubb AM, Oates AJ, Holden S, Koeppen H. Predicting benefit from angiogenic factors. Cytokine 2004; 26: 145–148. anti-angiogenic agents in malignancy. Nat Rev Cancer 2006; 6: 37 Sezer O, Jakob C, Eucker J, Niemoller K, Gatz F, Wernecke KD 626–635. et al. Serum levels of the angiogenic cytokines basic fibroblast 28 Gunsilius E, Petzer A, Stockhammer G, Nussbaumer W, growth factor (bFGF), vascular endothelial growth factor (VEGF) Schumacher P, Clausen J et al. Thrombocytes are the major source and hepatocyte growth factor (HGF) in multiple myeloma. Eur J for soluble vascular endothelial growth factor in peripheral blood. Haematol 2001; 66: 83–88. Oncology 2000; 58: 169–174. 38 Kuku I, Bayraktar MR, Kaya E, Erkurt MA, Bayraktar N, Cikim K 29 Aguayo A, Kantarjian H, Estey E, Giles FJ, Verstovsek S, Manshouri et al. Serum proinflammatory mediators at different periods of T et al. Plasma vascular endothelial growth factor levels have therapy in patients with multiple myeloma. Mediators Inflamm prognostic significance in patients with acute myeloid leukemia 2005; 2005: 171–174. but not in patients with myelodysplastic syndromes. Cancer 2002; 39 Wierda WG, Johnson MM, Do KA, Manshouri T, Dey A, O’Brien S 95: 1923–1930. et al. Plasma interleukin 8 level predicts for survival in 30 Gora-Tybor J, Blonski JZ, Robak T. Circulating vascular endothelial chronic lymphocytic leukaemia. Br J Haematol 2003; 120: growth factor (VEGF) and its soluble receptors in patients with 452–456. chronic lymphocytic leukemia. Eur Cytokine Netw 2005; 16: 41–46. 40 Tomita M, Semenza GL, Michiels C, Matsuda T, Uchihara JN, 31 Alexandrakis MG, Passam FH, Boula A, Christophoridou A, Okudaira T et al. Activation of hypoxia-inducible factor 1 in Aloizos G, Roussou P et al. Relationship between circulating human T-cell leukaemia virus type 1-infected cell lines and primary serum soluble interleukin-6 receptor and the angiogenic cytokines adult T-cell leukaemia cells. Biochem J 2007; 406: 317–323. basic fibroblast growth factor and vascular endothelial growth 41 Raman D, Baugher PJ, Thu YM, Richmond A. Role of chemokines factor in multiple myeloma. Ann Hematol 2003; 82: 19–23. in tumor growth. Cancer Lett 2007; 256: 137–165. 32 Niitsu N, Okamoto M, Nakamine H, Yoshino T, Tamaru J, 42 Luciani MG, Stoppacciaro A, Peri G, Mantovani A, Ruco LP. The Nakamura S et al. Simultaneous elevation of the serum concentra- monocyte chemotactic protein a (MCP-1) and interleukin 8 (IL-8) tions of vascular endothelial growth factor and interleukin-6 as in Hodgkin’s disease and in solid tumours. Mol Pathol 1998; 51: independent predictors of prognosis in aggressive non-Hodgkin’s 273–276. lymphoma. Eur J Haematol 2002; 68: 91–100. 43 Kerbel RS. Tumor angiogenesis. N Engl J Med 2008; 358: 33 Salven P, Orpana A, Teerenhovi L, Joensuu H. Simultaneous 2039–2049. elevation in the serum concentrations of the angiogenic growth 44 Bruserud O, Ryningen A, Olsnes AM, Stordrange L, Oyan AM, factors VEGF and bFGF is an independent predictor of poor Kalland KH et al. Subclassification of patients with acute prognosis in non-Hodgkin lymphoma: a single-institution study of myelogenous leukemia based on chemokine responsiveness and 200 patients. Blood 2000; 96: 3712–3718. constitutive chemokine release by their leukemic cells. Haemato- 34 Fayad L, Keating MJ, Reuben JM, O’Brien S, Lee BN, Lerner S et al. logica 2007; 92: 332–341. Interleukin-6 and interleukin-10 levels in chronic lymphocytic 45 Aref S, Salama O, Shamaa S, El-Refaie M, Mourkos H. Angiogen- leukemia: correlation with phenotypic characteristics and esis factor pattern differs in acute lymphoblastic leukemia and outcome. Blood 2001; 97: 256–263. chronic lymphocytic leukemia. Hematology 2007; 12: 319–324.

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

Leukemia