Analysis of angiogenic and stromal biomarkers in a large malignant mesothelioma cohort

Puey Ling Chia a,b,c,*, Prudence Russell d, Khashi Asadi e, Bibhusal Thapa b,c, Val Gebski f, Carmel Murone b, Marzena Walkiewicz b, Ulf Eriksson g, Andrew M. Scott b,c,h,i, Thomas John a,b,h

A B S T R A C T

The malignant mesothelioma vascular endothelial growth factor platelet-derived growth factor fibroblast growth factor receptor microvessel density Background: Malignant mesothelioma (MM) is an aggressive malignancy of the pleura and other mesothelial membranes. Agents targeting vascular endothelial growth factor (VEGF) such as bevacizumab; and multi-kinase inhibitors such as nintedanib [angiokinase inhibitor of VEGF, platelet-derived growth factor (PDGF) receptor and fibroblast growth factor receptor (FGFR)] have recently demonstrated efficacy in MM.
Methods: Tissue microarrays (TMAs) were created from formalin-fixed, paraffin-embedded tissue samples obtained from 326 patients with MM who were treated surgically. PDGF-CC, FGFR-1, VEGF and CD31 expression were analysed by immunohistochemical (IHC) staining. The H-score method assigned a score of 0–300 to each sample, based on the percentage of cells stained at different intensities. CD31 was evaluated via Chalkley’s method to evaluate microvessel density. We evaluated the association between expression of the biomarkers, clinicopathological factors and outcomes, in patients with MM.
Results: CD31 high (≥5) was seen in only 31/302 (10.3%) irrespective of histology. PDGF-CC high (≥150) was seen in 203 /310 (65%) of all samples. VEGF high (≥80) was seen in 219/322 (68%) of all MM with 143/209 (68%) of epithelioid histology. FGFR-1 high (≥40) was seen in 127/310 (41%) of all MM. There was no association of VEGF and FGFR-1 IHC with survival nor differences between histological subtypes.
There was a non-significant trend towards poorer survival in epithelioid tumours with increased PDGF-CC expression (OS 18.5 vs 13.2 months; HR 0.7928; 95% CI 0.5958 to 1.055, P =0.1110). High CD31 score was associated with significantly poorer survival (OS 12 vs 8.6 months; HR 0.48; 95% CI 0.2873 to 0.7941, P =0.0044). Of the 31 patients with high CD31 scores; 23/31 (74%) were also high for PDGF-CC and 20/31 (64%) with high VEGF scores. CD31 was found to be an independent prognostic factor in multivariate analysis (HR 1.540; 95% CI 1.018 to 2.330; p =0.041).
Conclusions: High CD31 was an independent poor prognostic factor and high PDGF-CC expression was associated with poor survival in MM. Abrogating these pathways may have prognostic implications.

1. Introduction

Angiogenesis is a complex process involving the interaction between factor-1α, and the subsequent activation of VEGF, along with many tumour cells, growth factors and cells within the tumour other growth factors including platelet derived growth factors, endothelial growth factors and fibroblast growth factor.1–3 The interaction between these cells within the TME and the supporting growth factors induce pro-inflammatory mediators which supports proliferation and survival of malignant cells, stromal remodelling and propagates angiogenesis-related factors.4–7
VEGF is a growth factor that is linked to angiogenesis and stimulates neovascularization of tumours. In malignant mesothelioma (MM), elevated levels of VEGF and its receptor have been detected by immunohistochemistry8,9 and correlated with the density of microvessels, increased tumour necrosis and poorer survival.10
A wide range of VEGFR tyrosine kinase inhibitors and monoclonal antibodies against VEGF (such as bevacizumab) have been trialled in mesothelioma. The Mesothelioma Avastin Cisplatin Pemetrexed Study (MAPS) study by the French Cooperative Thoracic Intergroup (IFCT) showed the addition of bevacizumab to pemetrexed plus cisplatin significantly improved overall survival (OS) from 16.1 months with doublet chemotherapy alone to 18.8 months with the combination of chemotherapy and bevacizumab.11 The combination of chemotherapy with bevacizumab now exists as a new standard of care for the management of MM. Given that steadily rising health care costs are not sustainable, the use of bevacizumab in the management of malignant mesothelioma for a potential 2.7 months improvement in OS are often limited by the cost-benefit calculations.
The FGF pathway is involved in signal transduction pathways that regulate processes such as cell proliferation, differentiation and angiogenesis as well as playing a fundamental role in carcinogenesis.12–14 FGF inhibitors are currently being investigated as a potential therapeutic agent for MM and include oral FGF inhibitors such as AZD4547 and BGJ398.15,16 Schelch et al also investigated the role of FGFR1 inhibition using genetic constructs or tyrosine kinase inhibitors and demonstrated suppression of mesothelioma cell growth in vitro and in vivo; as well as increasing the sensitivity to cisplatin treatment and irradiation.17 FGF/FGFR-targeting strategies are currently explored in many clinical studies to evaluate its efficacy in the management of mesothelioma.
Platelet-derived growth factors (PDGF) are receptor tyrosine kinases that regulate diverse cellular functions, including cell proliferation, transformation, migration and embryonic development. The PDGF/ PDGFR pathway has been implicated in mesothelioma carcinogenesis with both autocrine and paracrine mechanisms of proliferation.18–24 PDGFs consists of four different polypeptide chains (PDGF-A, -B, -C and -D) which form disulphide-linked homodimers (PDGF-AA, -BB, -CC and -DD) or heterodimers (only one known heterodimer PDGF-AB).18 PDGF-CC has been reported to induce angiogenic activity indirectly, via upregulation of VEGF19,20 and directly, via activation of PDGF-Rα and Rα − Rβ receptors.21 PDGF-CC has recently been shown to be a negative prognostic factor in triple negative breast cancer.22 Given PDGF-CC’s role in angiogenesis and paracrine functions, this could be an ideal biomarker to investigate in mesothelioma.
There are several small molecule tyrosine kinase inhibitors (TKIs) that target several receptor tyrosine kinases including VEGF that have been studied in MM and are summarized in our previous review article.23 Positive data from the MAPS study has definitely lead to further interest in other anti-angiogenic agents being trialled in MM. In particular, orally available angiokinase inhibitors such as nintedanib were of significant interest over the last few years. Nintedanib is an orally available, triple angiokinase inhibitor that inhibits the VEGF, PDGF, and FGF receptors. The addition of nintedanib to pemetrexed plus cisplatin resulted in 4 months PFS improvement in the phase II LUME-Meso study.24 Unfortunately, the encouraging efficacy signal observed for nintedanib in Phase II was not confirmed in a larger Phase III trial.25 Previous studies of antiangiogenic treatment in MM with the pan-VEGFR inhibitor cediranib26 and the multikinase inhibitor sorafenib27 (inhibits VEGFRs and PDGFRs) also showed only limited activity.
While predictive biomarker studies are crucial to determine the subset of patients who may benefit from the addition of an anti- angiogenic agent, to date none have been forthcoming. CD31 (also known as platelet endothelial cell adhesion molecule-1) is an endothelial marker used to assess microvessel density with preclinical data showing significantly lower microvessel areas were present in tumors treated with nintedanib in xenograft models.28 The complementary LUME-Meso biomarkers study had reviewed the plasma levels of 58 angiogenic factors (Human AngiogenesisMAP® panel, Myriad RBM) including single-nucleotide polymorphisms (SNPs) in genes for mesothelin (MSLN), VEGFR1 (FLT1) and VEGFR3 (FLT4) and assessed microvessel density via CD31 immunohistochemistry assessment of archival biopsy samples. Although there were potential signals for greater PFS and OS benefits in patients with low plasma endoglin and major homozygous VEGFR1/3 genotypes from the phase II study; no biomarkers showed any conclusive significant association with nintedanib efficacy.29 A phase II trial evaluating vatalanib (a small molecule TKI that inhibits VEGFR1-3, with an inhibitory action also against c-KIT and PDGFRα) in previously untreated patients also demonstrated no correlation between baseline VEGF or PDGF levels and response, PFS or survival.30
In the setting of these new anti-angiogenic therapeutic agents which are of interest in MM, we sought to evaluate VEGF, PDGF, FGFR, and CD31 by immunochemistry in MM to determine their frequency of expression and assess their associated prognostic implications.

2. Materials and methods

2.1. Establishment of mesothelioma tissue microarrays with clinical annotation

Tissue microarrays (TMAs) were created from formalin-fixed, paraffin-embedded (FFPE) tissue samples obtained from 329 patients who underwent surgical resection or biopsy for MM between 1988 and 2014 as previously described, and included triplicates from each tumor sample.31

2.2. Immunohistochemistry

The expression of PDGF-CC, VEGF, FGFR-1 and CD31 were analysed by immunohistochemical (IHC) staining. The slides were deparaffinized with xylene then transferred through graded ethanol to H2O. Endogenous peroxidase activity was blocked by 10-minute incubation in a 3% hydrogen peroxide solution. Antigen retrieval was performed via various techniques for different markers according to protocol for the commercially available antibodies. For PDGF-CC IHC, the antigen retrieval was performed by immersing slides in a bath of citrate buffer (pH6.0) and boiling in a 100 degrees water bath for 30 minutes. After antigen retrieval, the slides were blocked its relevant primary and secondary antibody according to protocol. The primary antibodies used were PDGF-CC (produced in-house) at 1.5ug/ml, VEGF (VG1; Dako) at 5ug/ml and CD31 (Dako) ready-to-use solution and FGFR-1 (Cell Signalling) at 0.6ug/ml. The tissues were incubated with various antibodies for at least 1 hour or overnight at 4 ◦C. Novolink Kit Poly-HRP secondary antibody was used. The intensity of immunostaining for VEGF, PDGF- CC, FGFR-1 and CD31 was scored visually by one investigator and pathologist (PR or KA) for validation.
The H-score method assigned a score of 0–300 to each sample, based on the percentage of cells stained at different intensities viewed at various magnifications by a pathologist and an investigator. A discriminatory threshold was set for each IHC stain (usually the median score) and the samples were classified as low (below median score) or high expression (above median score) for each of stain.

2.3. Microvessel quantification by Chalkley’s method

Assessment of angiogenesis was performed using a Chalkley eyepiece graticule as previously described.32–36 Each section was examined under low power to identify 3 intratumoural microvessel “hot spots”. At 200x magnification, a 25-point Chalkley eyepiece graticule was applied to each hotspot and oriented to coincide the dots on the graticule to as many microvessel profiles as possible. The mean of three separate counts was the final Chalkey score as previously described.36,37 The Chalkley’s graticule covers an area of 0.196mm2. All Chalkley counts were performed by one observer without knowledge of the clinical data or outcome. The data were later verified by an independent pathologist (KA) for consensus scoring. CD31 was evaluated via Chalkley’s method (objective method of measuring vascularity) to evaluate microvessel density.29,38

3. Statistical Analysis

Demographic and IHC outcomes were summarized using mean, standard deviation, range if continuous, or using percentages if categorical. Comparisons of categorized IHC outcomes between patient’s histology groups were carried out using the Fisher’s exact test. Overall survival was estimated using the Kaplan-Meier curves generated using GraphPad Prism 6.0. All tests were two-sided and p-values of 0.05 or less were considered statistically significant. Statistical analysis was carried out using Prism Graph-Pad Software program and a stepwise, multivariate Cox model analysis was performed using SPSS software.

4. Results

4.1. Patient characteristics

A total of 329 MM patient samples were collected, with annotated clinical data present only for 326 patients, which formed the cohort reported in this study. The histological subtypes comprised of epithelioid (70%); biphasic (22.8%); sarcomatoid (4.6%) and desmoplastic (<1%). The cohort is shown in Table 1, and could be considered representative of standard demographics for this disease, with older median age and survival being the poorest for non-epithelioid histology.

4.2. Immunohistochemical expression of VEGF, PDGF-CC, FGFR-1 expression

PDGF-CC IHC was considered to be positive if greater or equal to 5% of the cells were stained. Almost all mesothelioma tumour samples (98%) showed presence of PDGF-CC with varying intensity. The discriminatory threshold for PDGF-CC IHC was set at the median of 150.
PDGF-CC high IHC (≥150) was seen in 203 /310 (65%) of all samples but was higher in epithelioid histology (129/203 (64%) epithelioid; 45/ 203 (22%) biphasic and 28/203 (13.8%) sarcomatoid; 1/203 indeterminate). There was a non-significant trend towards poorer survival (OS 13.6 mths vs 10.6 mths; HR 0.844; 95% CI 0.675 to 1.055; p = 0.071) in the analysis including all MM subtypes (Fig. 1A). The median OS difference was accentuated in the epithelioid-only histology subgroup (median OS difference of 5.3 months) but this was not statistically significant [13.2 months in PDGF-CC high (PDGF-CChi) vs 18.5 months for PDGF-CC low (PDGF-CClow); HR 0.793; 95% CI 0.596 to 1.055; p = 0.110] (Fig. 1B). PDGF-CC expression in stromal tissues was present but less than tumour expression with stromal PDGF-CChi in 72/310 (23.2%).
VEGF IHC was considered to be positive if greater or equal to 5% of the cells were stained. 97% of our mesothelioma cohort showed presence of VEGF IHC staining. The cases with negative VEGF IHC were mainly sarcomatoid or biphasic subtypes (80%).
The discriminatory threshold for VEGF IHC was set at 80 (which was the median score). VEGF high ≥80(VEGFhi) was seen in 219/322 (68%) of all MM and was also higher in epithelioid histology [143/209 (68%)].
There was no association of VEGF IHC with survival nor differences between histological subtypes (12 months in VEGF high vs 11.5 months in VEGF low (VEGFlow); HR 0.975; 95% CI 0.780 to 1.218; p = 0.820) (Fig. 2).
FGFR-1 was considered to be positive if greater or equal to 5% of the cells were stained. 65% (194/300) of our mesothelioma cohort showed presence of VEGF IHC staining. Using the median IHC score, the threshold was set as 30. 40% (120/300) are noted to have high FGFR-1 IHC within our cohort. There was no association of VEGF IHC with survival nor differences between histological subtypes (11.1 months for FGFR-1 high vs 11.8mths for FGFR-1 low; HR 0.938; 95% CI 0.145 to 1.732 p = 0.984) (Fig. 3).

4.3. Chalkley estimates and prognostic outcomes

The median Chalkley count was 2 (range 1-9). There was no difference in the median Chalkley count between epithelioid and non- epithelioid histology types. Representative examples of CD31 IHC staining in mesothelioma and microvessel density evaluated via the Chalkley’s method are shown in Fig. 4A (i, ii and iii). High CD31 was defined as 5 or more (above 3rd quartile) and low CD31 count was defined as less than 5.
High CD31 score ≥5 (CD31hi) was associated with significantly poorer survival (8.6 months in high CD31 vs 12 months in low CD31 <5 (CD31low); HR 0.48; 95% CI 0.287 to 0.794, p = 0.004) (Fig. 4C). Of the 31 patients with high CD31 scores; 23/31 (74%) were also high for PDGF-C and 20/31 (64%) with high VEGF scores (Table 2).
A total of 269 cases were included in the multivariate analysis and those with missing data were excluded. CD31 was found to be an independent prognostic marker in the multivariate analysis (HR 1.540; 95% CI 1.018 to 2.330; p = 0.041) (Table 3). The other significant prognostic variables identified in a multivariate Cox regression analysis were neutrophil-lymphocyte ratio, epithelioid tumour histology subtype and advanced stage disease which are all known prognostic factors in MM.
A summary of the biomarkers (VEGF, PDGF-CC, FGFR-1 and CD31), the threshold definition for high vs low cases and their associations with survival outcomes have been summarised in Table 4.

5. Discussion

Our data confirm the role of high microvessel density as a poor prognostic marker in MM. High CD31 score was an independent prognostic marker in multivariate analysis ((HR 1.540; 95% CI 1.018 to 2.330; p = 0.041). Increased PDGF-CC expression in the epithelioid subgroup was associated with poorer prognosis but did not achieve statistical significance (HR 0.793; 95% CI 0.596 to 1.055; p = 0.110). High VEGF and FGFR expression were present in 68% and 40% of the mesothelioma cohort respectively, but no correlation between prognosis and immune expression of VEGF or FGFR was present.
The interaction between mesothelioma cells and angiogenic factors such as VEGF/ VEGFR, PDGF/ PDGFR and FGFR have been known to be of significance in the pathogenesis and progression of MM.23,39 Our data builds on this knowledge but also suggests that targeting these pathways may also have prognostic impact. Despite promising in vitro and in vivo data investigating a variety of receptor tyrosine kinase inhibitors in MM, there has been a lack of significant clinical efficacy for many of these agents. There is a need for research into prognostic and predictive biomarkers to better understand the impact they have on MM patients and better select for patients that are most likely to respond to anti-angiogenesis directed therapies.
As MM is a rare disease, research into prognostic angiogenic biomarkers in MM are less well studied with low numbers (typically less than 100 cases) in most studies. Our data uses a large repository of tissues and clinicopathological data for a more in-depth analysis and clinicopathological correlation. Angiogenesis, assessed by variable methods have been reported by several groups to be a poor prognostic factor in MM.9,35,40,41 Our data from a large cohort are consistent with the literature but we also found those with high microvessel density also have a higher proportion of cases with associated high angiogenic and stromal markers, particularly PDGF-CC. We identified high CD31 (via Chalkey’s methods) to be an independent prognostic marker in both univariate and multivariate analysis (HR 1.540; 95% CI 1.018 to 2.330; p = 0.041), where histological subtype was included in the model. Previous research had shown a trend towards better treatment response in lung cancer patients with a higher undifferentiated-vessel microvessel density.42 This may be as a result of better drug penetrance due to a higher tumour microvessel density. However, responses to treatment were not assessed in our study.
The incidence of positive VEGF IHC in the literature range from 25% (n = 4/16) to 97.3% (n = 36/37) in MM.43,44 In our dataset 97% of samples were positive for VEGF staining (defined as positive if >5% of cells stained) which is similar to that reported by Aoe et al.43 We also did not find any significant correlation between prognosis and expression of VEGF which is consistent with the reported literature.43,44 FGF-1 expression was reported to occur in 50% (8 out of 16) by Kumar-Singh et al44 which was similar to our cohort which were noted to be 40% (120/300) were positive for FGF-1.
To our knowledge the frequency of PDGF-CC has not been reported previously in MM and our data demonstrate that most cases were not only positive but showed strong expression (PDGF-CC high IHC ≥ 150) in 65% of the cohort. Increased PDGF-CC expression was associated with a trend towards poorer survival (13.2 months in PDGF-CC high vs 18.5 months for PDGF-CC low; HR 0.793; 95% CI 0.596 to 1.055; p = 0.110). The significance of PDGF-CC protein overexpression is still unknown and may warrant further investigation of this target in MM. In breast cancer, transcriptional upregulation of estrogen receptor in triple negative disease are related to a paracrine signalling mechanism from stromal cancer associated fibroblasts (with blockade of PDGF-CC induced PDGFRα activation).22,45 A novel neutralising antibody to PDGF-CC has already been developed and early data demonstrated that targeting PDGF-CC in experimental mouse models resulted in conversion of basal-like breast tumours into an estrogen receptor-positive state that conferred sensitivity to endocrine therapy.22 MM is similar to basal breast cancers as there are no targetable receptors and the mainstay of treatment remains that of chemotherapy. This concept of converting a non-targetable cancer to one with more therapeutic options is interesting especially given the dearth of options available to MM patients. Of interest, we observed that a higher proportion of patients with PDGF-CChi were of epithelioid subtypes rather than sarcomatoid histology. PDGF-CChi in the epithelioid subgroup was associated with poorer prognosis hence further studies to explore novel antibodies to PDGF-CC in this subgroup may be warranted. The expression of PDGF-CC in MM stromal tissues was reviewed, and we observed that expression was present but less than tumour expression; with PDGFChi in 23.2% (stromal tissue only) compared to 65% (tumour plus stromal tissue). Such agents may hold promise in MM, given the abundant expression of PDGF-CC in this tumour type.
While our data provide comprehensive clinicopathological correlation in a large dataset, there are limitations which should be acknowledged. The data presented are based mainly on the statistical analysis of immunohistochemistry results of the selected angiogenic and stromal markers in archival MM biopsy samples. Analysis of corresponding serum levels of these proangiogenic markers similar to the LUME-Meso study would be ideal for further correlative analysis but are not available for this MM cohort. Given the FGF401 exploratory nature of the study, we did not correct statistical significance level for multiple testing. The variations with regards to tissue processing, reference standards for protein quantification by IHC, as well as intra- and inter-observer variations are also important limitations to highlight. Tumour-rich samples were selected for TMA inclusion hence the study of PDGF-CC in stromal tissue may not reflect the true extent of stromal expression in MM.
Given their location and patterns of infiltration, mesotheliomas recruit vascular and stromal factors to survive and proliferate. Our data shed further light on key factors involved in these processes with some, such as CD31 being independently prognostic while others such as PDGF-CC appear imminently targetable. In a disease with limited treatment options, these data may help in defining pathways worth abrogation and ultimately in identifying therapeutic strategies for this devastating disease.

References

[1] H.E. Ryan, M. Poloni, W. McNulty, et al., Hypoxia-inducible factor-1alpha is a positive factor in solid tumor growth, Cancer Res. 60 (15) (2000) 4010–4015.
[2] W. Risau, H. Drexler, V. Mironov, et al., Platelet-derived growth factor is angiogenic in vivo, Growth Factors. 7 (4) (1992) 261–266.
[3] J. Folkman, M. Klagsbrun, Angiogenic factors, Science. 235 (4787) (1987) 42–447.
[4] J.W. Pollard, Tumour-educated macrophages promote tumour progression and metastasis, Nat Rev Cancer. 4 (1) (2004) 71–78.
[5] A. Sica, Role of tumour-associated macrophages in cancer-related inflammation, Exp Oncol. 32 (3) (2010) 153–158.
[6] Z.M. David Dong, A.C. Aplin, R.F. Nicosia, Regulation of angiogenesis by macrophages, dendritic cells, and circulating myelomonocytic cells, Curr Pharm Des. 15 (4) (2009) 365–379.
[7] C. Porta, P. Larghi, M. Rimoldi, et al., Cellular and molecular pathways linking inflammation and cancer, Immunobiology. 214 (9-10) (2009) 761–777.
[8] F. Demirag, E. Unsal, A. Yilmaz, et al., Prognostic significance of vascular endothelial growth factor, tumor necrosis, and mitotic activity index in malignant pleural mesothelioma, Chest. 128 (5) (2005) 3382–3387.
[9] Y. Ohta, V. Shridhar, R.K. Bright, et al., VEGF and VEGF type C play an important role in angiogenesis and lymphangiogenesis in human malignant mesothelioma tumours, Br J Cancer. 81 (1) (1999) 54–61.
[10] J.G. Edwards, D.E. Swinson, J.L. Jones, et al., Tumor necrosis correlates with angiogenesis and is a predictor of poor prognosis in malignant mesothelioma, Chest. 124 (5) (2003) 1916–1923.
[11] G. Zalcman, J. Mazieres, J. Margery, et al., Bevacizumab for newly diagnosed pleural mesothelioma in the Mesothelioma Avastin Cisplatin Pemetrexed Study (MAPS): a randomised, controlled, open-label, phase 3 trial, Lancet. (2015).
[12] A. Beenken, M. Mohammadi, The FGF family: biology, pathophysiology and therapy, Nat Rev Drug Discov. 8 (3) (2009) 235–253.
[13] D.M. Ornitz, N. Itoh, The Fibroblast Growth Factor signaling pathway, Wiley Interdiscip Rev Dev Biol. 4 (3) (2015) 215–266.
[14] N. Turner, R. Grose, Fibroblast growth factor signalling: from development to cancer, Nat Rev Cancer. 10 (2) (2010) 116–129.
[15] W. Lam, A. Nowak, F. Chen, et al., MA23.01 Phase II Trial of an Oral FGFR Inhibitor AZD4547 as Second or Third Line Therapy in Malignant Pleural Mesothelioma: Final Results of FRAME Study, Journal of Thoracic Oncology. 14 (10) (2019) S342.
[16] J.M. Quispel-Janssen, J. Badhai, L. Schunselaar, et al., Comprehensive Pharmacogenomic Profiling of Malignant Pleural Mesothelioma Identifies a Subgroup Sensitive to FGFR Inhibition, Clin Cancer Res. 24 (1) (2018) 84–94.
[17] K. Schelch, M.A. Hoda, T. Klikovits, et al., Fibroblast growth factor receptor inhibition is active against mesothelioma and synergizes with radio- and chemotherapy, Am J Respir Crit Care Med. 190 (7) (2014) 763–772.
[18] L. Fredriksson, H. Li, U. Eriksson, The PDGF family: four gene products form five dimeric isoforms, Cytokine Growth Factor Rev. 15 (4) (2004) 197–204.
[19] X. Li, A. Ponten, K. Aase, et al., PDGF-C is a new protease-activated ligand for the PDGF alpha-receptor, Nat Cell Biol. 2 (5) (2000) 302–309.
[20] X. Li, U. Eriksson, Novel PDGF family members: PDGF-C and PDGF-D, Cytokine Growth Factor Rev. 14 (2) (2003) 91–98.
[21] R. Cao, E. Brakenhielm, X. Li, et al., Angiogenesis stimulated by PDGF-CC, a novel member in the PDGF family, involves activation of PDGFR-alphaalpha and -alphabeta receptors, FASEB J. 16 (12) (2002) 1575–1583.
[22] P. Roswall, M. Bocci, M. Bartoschek, et al., Microenvironmental control of breast cancer subtype elicited through paracrine platelet-derived growth factor-CC signaling, Nat Med. 24 (4) (2018) 463–473.
[23] P.L. Chia, P.A. Russell, A.M. Scott, et al., Targeting the vasculature: anti-angiogenic agents for malignant mesothelioma, Expert Rev Anticancer Ther. 16 (12) (2016) 1235–1245.
[24] F. Grosso, N. Steele, S. Novello, et al., Nintedanib Plus Pemetrexed/Cisplatin in Patients With Malignant Pleural Mesothelioma: Phase II Results From the Randomized, Placebo-Controlled LUME-Meso Trial, J Clin Oncol. 35 (31) (2017) 3591–3600.
[25] G.V. Scagliotti, R. Gaafar, A.K. Nowak, et al., Nintedanib in combination with pemetrexed and cisplatin for chemotherapy-naive patients with advanced malignant pleural mesothelioma (LUME-Meso): a double-blind, randomised, placebo-controlled phase 3 trial, Lancet Respir Med. 7 (7) (2019) 569–580.
[26] L.L. Garland, K. Chansky, A.J. Wozniak, et al., Phase II study of cediranib in patients with malignant pleural mesothelioma: SWOG S0509, J Thorac Oncol. 6 (11) (2011) 1938–1945.
[27] S. Dubey, P.A. Janne, L. Krug, et al., A phase II study of sorafenib in malignant mesothelioma: results of Cancer and Leukemia Group B 30307, J Thorac Oncol. 5 (10) (2010) 1655–1661.
[28] V. Laszlo, Z. Valko, I. Kovacs, et al., Nintedanib Is Active in Malignant Pleural Mesothelioma Cell Models and Inhibits Angiogenesis and Tumor Growth In Vivo, Clin Cancer Res. 24 (15) (2018) 3729–3740.
[29] N. Pavlakis, F. Grosso, N. Steele, et al., 213O Nintedanib + pemetrexed/cisplatin in malignant pleural mesothelioma (MPM): Phase II biomarker data from the LUME Meso study, Journal of Thoracic Oncology. 13 (4) (2018) S128.
[30] T. Jahan, L. Gu, R. Kratzke, et al., Vatalanib in malignant mesothelioma: a phase II trial by the Cancer and Leukemia Group B (CALGB 30107), Lung Cancer. 76 (3) (2012) 393–396.
[31] B. Thapa, M. Walkiewicz, C. Murone, et al., Calretinin but not caveolin-1 correlates with tumour histology and survival in malignant mesothelioma, Pathology. 48 (7) (2016) 660–665.
[32] S.B. Fox, Microscopic assessment of angiogenesis in tumors, Methods Mol Med. 46 (2001) 29–46.
[33] S.B. Fox, R.D. Leek, K. Smith, et al., Tumor angiogenesis in node-negative breast carcinomas–relationship with epidermal growth factor receptor, estrogen receptor, and survival, Breast Cancer Res Treat. 29 (1) (1994) 109–116.
[34] S.B. Fox, R.D. Leek, M.P. Weekes, et al., Quantitation and prognostic value of breast cancer angiogenesis: comparison of microvessel density, Chalkley count, and computer image analysis, J Pathol. 177 (3) (1995) 275–283.
[35] J.G. Edwards, G. Cox, A. Andi, et al., Angiogenesis is an independent prognostic factor in malignant mesothelioma, Br J Cancer. 85 (6) (2001) 863–868.
[36] S. Hansen, D.A. Grabau, F.B. Sorensen, et al., The prognostic value of angiogenesis by Chalkley counting in a confirmatory study design on 836 breast cancer patients, Clin Cancer Res. 6 (1) (2000) 139–146.
[37] B.V. Offersen, F.B. Sorensen, M. Yilmaz, et al., Chalkley Estimates of Angiogenesis in Early Breast Cancer – Relevance to Prognosis, Acta Oncol. 41 (7-8) (2002) 695–703.
[38] E. Pulford, A. Hocking, K. Griggs, et al., Vasculogenic mimicry in malignant mesothelioma: an experimental and immunohistochemical analysis, Pathology. 48 (7) (2016) 650–659.
[39] J. Remon, P. Lianes, S. Martinez, et al., Malignant mesothelioma: new insights into a rare disease, Cancer treatment reviews 39 (6) (2013) 584–591.
[40] S. Kumar-Singh, P.B. Vermeulen, J. Weyler, et al., Evaluation of tumour angiogenesis as a prognostic marker in malignant mesothelioma, J Pathol. 182 (2) (1997) 211–216.
[41] A. Yasumitsu, C. Tabata, R. Tabata, et al., Clinical significance of serum vascular endothelial growth factor in malignant pleural mesothelioma, J Thorac Oncol. 5 (4) (2010) 479–483.
[42] Y.Y. Zhao, C. Xue, W. Jiang, et al., Predictive value of intratumoral microvascular density in patients with advanced non-small cell lung cancer receiving chemotherapy plus bevacizumab, J Thorac Oncol. 7 (1) (2012) 71–75.
[43] K. Aoe, A. Hiraki, T. Tanaka, et al., Expression of vascular endothelial growth factor in malignant mesothelioma, Anticancer Res. 26 (6C) (2006) 4833–4836.
[44] S. Kumar-Singh, J. Weyler, M.J. Martin, et al., Angiogenic cytokines in mesothelioma: a study of VEGF, FGF-1 and -2, and TGF beta expression, J Pathol. 189 (1) (1999) 72–78.
[45] S. Jansson, K. Aaltonen, P.O. Bendahl, et al., The PDGF pathway in breast cancer is linked to tumour aggressiveness, triple-negative subtype and early recurrence, Breast Cancer Res Treat. 169 (2) (2018) 231–241.