Background
Angiosarcomas account for approximately 1% of adult soft-tissue sarcomas, the latter of which account for 1–2% of adult malignancies. Angiosarcomas are heterogeneous and involve at least, on the one hand, de novo versus radio-induced sarcoma and, on the other hand, skin or scalp locations versus visceral locations [
1]. Regardless, these entities exhibit aggressive behavior, leading to a poor outcome. Localized angiosarcomas are best treated by large en-bloc surgery followed by adjuvant radiotherapy. Locally advanced (not amenable to curative-intent surgery) or metastatic angiosarcomas are treated with palliative chemotherapy aiming to alleviate symptoms and maintain quality of life [
1]. Doxorubicin and weekly paclitaxel are both regarded as a preferred option as the first or second line, and these regimens provide a median overall survival of approximately 8 to 12 months [
2]. Thus, advanced angiosarcoma treatment remains an unmet medical need.
Assessing the activity of anti-angiogenic agents, such as bevacizumab (a humanized monoclonal antibody against vascular endothelial growth factor (VEGF)), in angiosarcoma is important, and preclinical studies have demonstrated a key role for angiogenesis in angiosarcoma proliferation [
3‐
12]. In general, angiosarcomas overexpress VEGF-A as well as multiple VEGF receptors (VEGFRs), including the major pro-angiogenic VEGF-A receptor VEGFR-2 [
3‐
12]. Some studies have reported recurrent activating mutations in angiogenesis signaling genes, especially VEGF receptors [
10], and in other genes encoding proteins associated with regulation of VEGF receptors, such as PTPRB and PLCG1 [
13]. In vitro, blockade of the VEGF pathway inhibits tumor growth by decreasing proliferation and increasing apoptosis in tumor cell [
10].
We previously assessed the activity and safety of weekly paclitaxel and weekly paclitaxel in combination with bevacizumab for advanced angiosarcoma and found no clinical benefit of adding bevacizumab [
14]. The present ancillary analysis aims to identify prognostic and predictive factors in the current trial with a longer follow-up.
Methods
Clinical trial
AngioTax-Plus was a multicenter, randomized, stratified and open-label phase II trial. The stratification factors were as follows: de novo versus radiation-induced angiosarcoma and superficial (skin and soft tissue) versus visceral angiosarcoma. The efficacy of the phase II trial were previously reported [
14]. Patients aged 18 years and older were considered eligible for the study. All patients had histologically proven metastatic or advanced angiosarcoma, as reviewed by the Pathology Committee of the French Sarcoma Group, and were not amenable to curative-intent surgery. Radiation-induced angiosarcomas were eligible if there was no evidence of recurrence of the prior radiotherapy-treated malignancy. Up to two previous lines of systemic chemotherapy were allowed. Tumors were required to be measurable by computed tomography (CT) or magnetic resonance imaging (MRI), as per RECIST 1.1. In both arms, patients received paclitaxel intravenously at the dose of 90 mg/m
2 on days 1, 8 and 15 of a 28-day cycle for 6 cycles. In the experimental arm (Arm B), patients received bevacizumab during chemotherapy cycles at the dose of 10 mg/kg every 2 weeks until intolerance or progression. In the absence of disease progression after the 6 cycles of chemotherapy, bevacizumab was administered as maintenance therapy at the dose of 15 mg/kg every 3 weeks until intolerance or progression.
Study investigations were conducted after approval by the local Ethics Committee (Nord-Ouest IV Commit de Protection des Patients) on 9th March, 2010, and after the approval of the French Health Authority (AFSSAPS) on 12th April, 2010. Informed consent was obtained from each patient. The study was registered in the European Clinical Trials Register (EudraCT N° 2009–017020-59) and in the US Clinical Trials Register (NCT01303497) and was conducted in agreement with the Declaration of Helsinki and International Conference on Harmonization of Good Clinical Practice guidelines.
Biomarker analysis
Blood samples were collected into 5-ml serum-separating tubes (SSTs) on day 1 (D1; baseline) and day 8 (D8). The samples were centrifuged at 3000 rpm for 15 min. The plasma was then transferred to 2 labeled cryotubes and frozen, as soon as possible, at − 80 °C, at each center. The tubes were then transported in containers filled with dry ice to maintain cold chain integrity to the Molecular Pharmacology Laboratory in Paoli-Calmettes Institute, Marseille. Enzyme-linked immunosorbent assay (ELISA) was performed as previously described [
15] to measure the following circulating biomarkers: VEGF-C in pg/ml by Quantikine human VEGF Immunoassay (R&D, Minneapolis, USA); sE-Selectin in ng/ml by Human sE-Selectin Immunoassay (R&D, Minneapolis, USA); thrombospondin in ng/ml by Human TSP-1 immunoassay (Neogen, Lexington, USA via interchim); VEGF in ng/ml by Quantikine human VEGF Immunoassay (R&D, Minneapolis, USA); stem-cell factor (SCF) in ng/ml by Quantikine human SCF Immunoassay (R&D, Minneapolis, USA); fibroblast growth factor (FGF) in pg/ml by Quantikine Human FGF basic Immunoassays (R&D, Minneapolis, USA); and placental growth factor (PlGF) in pg/ml by Quantikine Human PlGF Immunoassay (R&D, Minneapolis, USA). Each sample was analyzed in duplicate, and the average value was used for correlations with clinical outcomes.
Statistical analysis
Biomarker baseline (D1) values were compared between the treatment arms and according to medical history (de novo versus radio-induced sarcoma) using the Wilcoxon and Mann-Whitney tests. In each treatment arm, D8 values were compared with D1 values using the Wilcoxon signed-rank tests for paired data. Differences in biomarker values (D8 – D1) were compared between the treatment arms using linear regression models with adjustment for the D1 values.
The endpoint for the prognostic factor analysis was progression-free survival (PFS), which was defined as the time from randomization to the first progression or death from any cause and analyzed in the entire study population. PFS curves were generated using the Kaplan-Meier method. The impact of covariates on PFS was estimated using Cox models (hazard ratio, HR, and 95% Confidence Interval, 95%CI). The clinical factors investigated were tumor location (superficial versus visceral) and radio-induced sarcoma versus de novo sarcoma. Biomarkers at D1, as well as the variation between D1 and D8, were analyzed as continuous values. The impact of baseline covariates (clinical factors and biomarkers at D1) found to be associated with a p-value< 0.20 in univariate analysis were then evaluated in multivariate Cox regression. The significance level was not corrected for multiple testing and was set at 0.05 in this exploratory analysis. For each biomarker, the impact of the (D8-D1) observed variation on the risk of progression was evaluated using Cox models that were adjusted to the baseline value.
The predictive value of each covariate (clinical factor, biomarker at D1, and D8-D1 variation) was investigated using an interaction term between the treatment arm and covariate in a multivariate Cox model including these parameters. The results are illustrated by forest plots in which the biomarkers were categorized as binary variables using the observed median value as the cut-off, as the results were very similar when considering the continuous value or binary variable.
Discussion
The results of the present study confirmed that adding bevacizumab to weekly paclitaxel did not improve the PFS of advanced angiosarcoma. This study showed that at advanced stages and when treated with weekly paclitaxel, the outcome of radiation-induced angiosarcoma is better than that of de novo angiosarcoma. At baseline, high values of circulating PlGF and low values of VEGF-C were associated with poor outcomes. Treatment with bevacizumab was associated with a significant increase in circulating PlGF and a decrease in circulating VEGF. Finally, an increase in FGF between day 1 and day 8 was associated with a heightened risk of disease progression.
To date, the use of anti-angiogenic antibodies in angiosarcoma patients is somewhat disappointing. The therapeutic role of antiangiogenic antibodies has been assessed in three phase II trials (including the AngioTax-Plus study), and a phase I/II trial has yet to be published. Agulnik et al. conducted a non-randomized phase II trial assessing bevacizumab alone in angiosarcoma patients; the median PFS was 3 months, and the reported best objective response rate was only 8% [
16]. In addition, a phase II trial assessing the peptibody trebananib, which inhibits angiopoietin 1 and 2, failed to demonstrate clinical activity and was closed after the first interim analysis (PFS of 1.8 months and best objective response rate of 0%) [
17]. The phase I/II trial assessing the combination of the multi-kinase inhibitor pazopanib with TRC 105, an endoglobin antibody, showed convincing signs of activity in angiosarcoma patients, with a median PFS of 5.6 months in nine patients and with two durable complete responses [
18]. An ongoing randomized phase III trial is comparing the efficacy of pazopanib alone compared with pazopanib and TRC 105; this trial will formally establish the benefit of adding an endoglobin antibody to the multi-kinase inhibitor
(NCT02979899). Nonetheless, the present study clearly shows that adding bevacizumab did not improve the outcome of advanced angiosarcoma patients, regardless of clinical or biological characteristics (see Additional file
2: Table S1).
When treated with weekly paclitaxel, the outcome of angiosarcoma patients in advanced stages is better for radiation-induced tumors than it is for de novo angiosarcoma. These findings are consistent with data in the literature. Italiano et al. analyzed the outcome of 117 advanced angiosarcoma patients treated with doxorubicin (
n = 42) and weekly paclitaxel (
n = 75) as first-line treatment. These authors reported that the objective response rate was statistically higher in radiation-induced angiosarcoma than in de novo angiosarcoma with doxorubicin (50 versus 15%,
p = 0.03) or with paclitaxel (74 versus 45%,
p = 0.04) [
19]. In another retrospective study comprising 142 patients treated with different chemotherapy regimens, the outcome of radiation-induced angiosarcoma was slightly better than that of de novo angiosarcoma (median overall survival of 14.3 versus 10.3 months), but significance was not reached (
p = 0.32) [
2]. These data suggest that radiation-induced angiosarcoma is highly sensitive to chemotherapy, though this needs to be confirmed by independent studies.
In the present study, we observed some major changes in biomarker values over a short period of time influenced by the addition of the anti-angiogenic agent. A low level of circulating VEGF-C and a high level of circulating PlGF at baseline were found to be associated with a poor outcome. In a previously published phase II trial assessing the therapeutic role of sorafenib, an oral anti-angiogenic tyrosine kinase inhibitor, we observed baseline PlGF to be associated with a poor outcome in advanced angiosarcoma patients [
15]. Moreover, we found a linear correlation between circulating PlGF and time to progression (
p = 0.02). Of note, Sleijfer et al. described that among sarcoma patients receiving pazopanib, those with increased circulating PlGF experienced a worse outcome with a shorter PFS and overall survival [
20]. Nevertheless, the literature data on circulating biomarkers in human angiosarcoma are very spare.
We found that bevacizumab induces a significant decrease in circulating VEGF and an increase in circulating PlGF. To our knowledge, no prior published study has analyzed the impact of bevacizumab on circulating proangiogenic biomarkers in sarcoma patients, especially in angiosarcoma patients. A large body of evidence shows that treatment with bevacizumab induces significant changes in circulating biomarkers in different carcinomas (for example, colo-rectal cancers [
21,
22], hepatocarcinoma [
23], non-small cell lung cancers [
24]), and our observed decrease in circulating VEGF is consistent with bevacizumab’s mechanism of action. Despite the few data on the impact of bevacizumab treatment on circulating PlGF, an increase in PlGF after bevacizumab treatment has been described in the case of non-small cell lung cancer [
24] and colo-rectal cancer [
25]. In addition, Kopetz et al. described that an increase in circulating FGF may be observed before radiological progression in metastatic colorectal cancer patients treated with bevacizumab [
26]. Our findings cannot be further discussed due to the sparse literature data and the notable absence of prior studies assessing the biomarker VEGF network in sarcoma patients treated with bevacizumab.
Regardless, conclusions with respect to the prognostic or predictive values of these changes must be considered with caution owing to the multiple tests performed on this limited sample of patients, the complexity of the circulating pro- and anti-angiogenic factor network, and the fact that negative results can be caused by a lack of power of the tests performed.
Acknowledgements
The authors would like to thank the staff members involved in the trial management: Stéphanie Bacquaert, Caroline Decamps, Emilie Decoupigny, Margaux Labroy, Julien Thery and Marie Vanseymortier. The data management and analysis were conducted by the Centre de Traitement des Données du Cancéropôle Nord-Ouest clinical research platform. We would also like to thank the patient advocacy group “Info Sarcome” and its president Estelle Lecointe.