Evaluation of Endoglin (CD105) expression in pediatric rhabdomyosarcoma

Rhabdomyosarcoma (RMS) is the most common type of soft tissue sarcoma (STS) in children and young adults, accounting for up to 5% of all childhood cancers and for about 40% of pediatric STS [1]. Embryonal (ERMS) and alveolar (ARMS) RMS are the two major histologic subtypes. ARMS is associated with PAX3/7-FOXO1 gene fusions and with a poor prognosis, often being metastatic at diagnosis [2]. Although during the last three decades, multimodal treatment strategies have substantially improved the prognosis of localized RMS, for metastatic disease the prognosis remains dismal [3]. Therefore, new targets and tailored therapies directed against the metastatic process are needed for these patients. The formation of new blood vessels is a requirement for tumor growth and metastatic spread and many regulators of tumor angiogenesis have been identified in different types of cancer [4]. Studies on inhibitors of angiogenesis have shown antitumor activity in pediatric sarcoma models, including RMS, mostly in combination with other drugs [5‐7], and several trials showed promising results for selected clinical indications [8‐10]. The quantification of tumor vasculature is a useful indicator of angiogenesis, by helping patients stratification prior to anti-angiogenic therapy and monitoring patient response. One often-quantified aspect of tumor vasculature is the Intratumoral Microvessel Density (IMVD). IMVD is commonly used as a surrogate marker to quantify angiogenic activity and is usually assessed by pan-endothelial markers, such as CD34 and CD31 [10‐13]. However, these markers are not tumor endothelial-specific, as they are also expressed on pre-existing/mature vasculature and on large vessels [14, 15]. Recent studies have shown that IMVD assessed by detection of Endoglin (CD105) is more specifically associated with tumor neovascularization [16‐20] and represents a significant prognostic marker in several tumors [19‐24]. CD105 is a transforming growth factor β (TGF-β) transmembrane co-receptor required for angiogenesis [25] and is highly expressed on the surface of actively proliferating microvascular endothelial cells, forming immature, highly permeable tumor neovessels [26]. In line with its supportive role in tumor neoangiogenesis, CD105 is up-regulated by hypoxia [27‐29]. The expression of CD105 has been reported on the tumor vasculature of several sarcomas, including Kaposi sarcoma, angiosarcoma, leiomyosarcoma, chondrosarcoma and gastrointestinal stromal tumor and correlated with worse survival for some of these tumors [21, 30‐33]. In this study, we aimed to investigate if CD105 was expressed in pediatric RMS and assess the neovascularization by using the angiogenic ratio IMVD-CD105 to IMVD-CD31. For this purpose, we evaluated the immunohistochemical expression of CD105, CD31 and VEGF in a retrospective series of pediatric patients with RMS. In order to define the proliferation fraction of the endothelium we compared the CD105 microvessels count with CD31 immunoexpression obtaining the CD105/CD31 expression ratio. In the cases where the CD105/CD31 expression ratio is higher, the angiogenesis is increased because CD105 marks the neoformed vessels [26] whereas CD31 is also expressed in mature vessels [18]. This ratio has been reported to have a prognostic value and be a potential predictor of response to anti-VEGF therapy [34‐36].

Neo-angiogenesis has long been implicated in generating a microenvironment suitable for tumor growth and metastatic spread [44]. Several pro-angiogenic factors have been described and among them VEGF appears to play a central role in the activation of angiogenesis in various cancer [45, 46]. Several efforts have been made to develop therapies focused on the inhibition of the VEGF signaling pathway also in RMS [47, 48]. However these drugs led to transient responses and the complementary/dual inhibition of non-VEGF angiogenic pathways might represent a way to improve anti-angiogenic therapy. A phase I study using an anti-endoglin monoclonal antibody (TRC105) in combination with bevacizumab in adults with advanced cancers showed good tolerance and clinical activity in a VEGF inhibitor-refractory population [49]. A trial testing TRC105 in combination with pazopanib in patients with STS (≥12 years old) is currently ongoing [50]. In this context, methods which enable to quantify tumor angiogenesis are useful surrogate markers of angiogenic activity and response to therapy, and might help stratify patients with RMS for treatment. The IMVD is the most commonly used parameter to quantify intra-tumoral neovascularization and is measured by pan-endothelial markers, such as CD31. CD105 presents a higher specificity for new developing vessels and recent studies have shown that IMVD as determined by this marker has a higher prognostic impact than CD31 in several tumors [21‐23]. In particular, IMVD ratio of CD105/CD31 expression, had been used to specifically assess neovascularization showing a prognostic value [34‐36]. In the present study, we analyzed for the first time, the CD105 immunoexpression in pediatric RMS and quantified the presence of proliferating endothelial cells by using the CD105/CD31 expression ratio. CD105 was detected in small tumor capillary-like vessels, whereas CD31 presented a more diffused expression in endothelial cells. The significant positive correlation found between the IMVD measured by the two markers is coherent with the association between CD105 expression and other endothelial markers, such as CD31 and CD34, already described in other tumors [51, 52]. We also evaluated whether a correlation between this neoangiogenic ratio and clinic-pathological variables exists. Several prognostic factors, such as the age at diagnosis, primary tumor size, primary site, histology, post-surgical stage and presence or absence of distant metastases are currently used for risk-adapted treatment approaches in clinical trials of RMS patients [53]. Using these parameters, in the univariable survival analysis, we found that the advanced disease, classified according to the COG risk stratification, was a significant predictor of worse OS and EFS. In line with previously reported works, our study confirms that metastatic disease is the main prognostic factor in RMS [3]. The ROC curve for the OS indicated a cutoff point of 1.3, which was used to separate patients with good and poor prognosis. A value of the CD105/CD31 expression ratio < 1.3 was associated with a significantly better patients’ OS. These findings suggest that neovascularization could be an indicator of prognosis in patients with RMS and are supported by the correlation described between the CD105/CD31 expression ratio and aggressive phenotypes in other tumors [35]. Indeed, it has been already reported that IMVD quantified by CD105 correlate with poor survival in patients with breast carcinoma, non-small cell lung cancer and hepatocellular carcinoma [13, 54, 55]. Interestingly, when the histotype was specifically considered, we found that the ERMS correlated significantly with neo-angiogenesis. Kuda et al., previously described that IMVD, assessed by CD31, was higher in ERMS than ARMS [56]. We speculate that this association could be related to the different growth rate displayed by these two RMS histotypes. Indeed, although angiogenesis is a key process activated during cancer invasion and metastasis, highly aggressive histotypes are also able to support their growth through a process known as vasculogenic mimicry (VM) [57]. The generation of non-endothelialized vessel-like channels allows the perfusion of a variety of tumors, enabling them to aggressively proliferate and metastasize [58]. The VM channels are not lined by endothelial cells, but by tumor cells instead, and therefore are not stained by endothelial markers, including CD31 [59]. A higher incidence of VM has been described in tumors presenting necrosis, as well as in ARMS, and has been associated with poor prognosis [60, 61]. The faster growth of ARMS compared to ERMS may explain the different pattern of neovessels in the two variants. No statistically significant differences in CD105/CD31 expression ratio were encountered with respect to age, tumor size, primary tumor location, COG risk groups and VEGF. Despite VEGF overexpression has been reported to be associated with prognosis in RMS patients [42], data regarding the correlation amongst IMVD, VEGF expression and prognosis has shown conflicting results in several tumors including STS and RMS [39, 56, 62‐64].

In conclusion, this small proof-of-concept study suggests that CD105 is expressed in endothelial cells of pediatric RMS and that CD105/CD31 expression ratio might be useful to measure the proportion of proliferating endothelial cells in this tumor. Despite the small cohort of patients studied, these data indicate that a high value of CD105/CD31 expression ratio could be related with a “pro-angiogenic” RMS subset of patients with low OS.

If further studies confirm these results in larger cohorts of patients, CD105 may also represent a potential therapeutic target as part of combined therapy in RMS. In particular, an inter-institutional cooperative study would be advisable considering the low frequency of this tumor in the pediatric population. This type of large study could also be a tool to elucidate if the CD105/CD31 expression ratio may be useful for patient’s stratification and/or evaluate response to therapy.