Tumor endothelial ELTD1 as a predictive marker for treatment of renal cancer patients with sunitinib

The local Research Ethics Committee in Uppsala granted approval for the study (2009/139) and patients still alive gave their written informed consent. The TMA cohort and treatment characteristics have been described previously [11, 12, 21] (Table 1) [12]. In short, the TMA consists of 139 cases of RCC diagnosed 2006–2010. All the patients underwent nephrectomy at diagnosis and later on were treated with anti-cancer therapy for mRCC. A total of 99 patients were treated with sunitinib in the first or second line setting, while 53 received treatment with sorafenib. A flow chart presenting enrollment of the sunitinib treated patients is demonstrated in Fig. 1.

Treatment characteristics for renal cancer patients treated for metastatic disease with sunitinib in the first- and second-line setting [12].

Patient and tumor characteristics in our cohort have been described previously (Table 2) [12].

Clinical characteristics of renal cancer patients treated for metastatic disease with sunitinib in the first- and second-line setting [12].

TMA, immunohistochemistry and slide scanning were essentially performed in accordance to standards used in the Human Protein Atlas (www.​proteinatlas.​org) and the procedure has been described previously [11, 12, 21‐23]. The TMA in our study consists of core biopsies from the primary renal tumor. The antibody concentrations were tested according to experience from testing and validating antibodies within the Human Protein Atlas Projects using a test TMA containing normal and tumor tissues. The proteins of interest are known to display a vascular expression pattern and to be enriched in many tumor vessels, providing internal positive and negative controls within the TMA.

The staining and immunohistochemical procedure has been previously in detail described [11, 12, 21]. In this study we used the primary rabbit polyclonal antibody towards ELTD1 (HPA025229, Atlas Antibodies, Stockholm, Sweden), CD34 (CAB000018, Dako Cat#7165, Agilent (Formerly DakoCytomation) and VEGFR2 (CAB004028, Cell Signaling Technology Cat#2479, Cell Singaling Technology, Inc).

High-resolution images from each individual TMA core underwent colour deconvolution based on the H-DAB setting of the Fiji software to separate the DAB from the Hematoxylin staining [24]. The ELTD1-, CD34- and VEGFR2-positive area percentage of each TMA core was measured from the generated DAB images by a CellProfiler pipeline based on color-thresholding [25].

Tissue microarrays representing tumor cores from 99 patients diagnosed with mRCC that had undergone nephrectomy and were treated with sunitinib in first or second line therapy were stained using ELTD1-specific antibodies as described in materials and methods. ELTD1 expression was exclusively noted in tumor vessels, while other stromal cells and the tumor cells were uniformly negative (Fig. 2). A varying proportion of vessels within each core expressed ELTD1, while the intensity of staining was fairly similar in positive vessels. The area of ELTD1-expressing tumor vessels in each core was determined through digital analysis of scanned images as described above, and the percentage of ELTD1 positive area as compared to the total tumor area was calculated. To determine if vascular expression of ELTD1 could predict response to sunitinib therapy, patients were classified into two groups according to the median expression of ELTD1. The median expression was 1.1 (%) and the expression range 0.3–5.6 (%). Patients with < 1.1 (%) expression were categorized as ELTD1 low and patients with ≥1.1 (%) as ELTD1 high. The primary end-point of the study was PFS, calculated as the time from treatment initiation to the time of clinical and/or radiological progression, treatment termination due to toxicity, death or end of follow up and the second end-point was OS, calculated from the diagnosis of mRCC, in regard with ELTD1 expression. Using the cut-off value for staining described above, 39/77 (50%) cases were ELTD1 high.

Patients with a higher area percentage of ELTD1-expressing tumor vessels had a significantly better PFS (p = 0.017, Fig. 3a). We observed that the median sunitinib treatment period for patients with a high ELTD1-positive vessel area was 8 months (range 1–31 months) as compared to 5,5 months for patients with a low ELTD1-positive vessel area (range 0.5–34). Patients with a high ELTD1-positive vessel area had a significantly better OS (p = 0.03, Fig. 3b). The ELTD1 high group had a median OS of 31 months (range 2–108 months) while the ELTD1 low group had a median OS of 24 months (range 1–84 months). To determine if tumor vessel expression of ELTD1 generally predicted response to TKI-therapy, we analysed TMA cores from patients treated with sorafenib in the first or second line setting (n = 53). ELTD1-positive area fraction was not associated with either PFS or OS (p = 0.67/0.79) in these patients (Fig. 3c and d).

To determine if the area fraction of ELTD1-positive vessels simply reflected vessel density, the expression of pan-endothelial markers CD34 and VEGFR2 were assessed. CD 34 is widely expressed in both normal and tumor vessels and VEGFR2 is the most prominent receptor for VEGF-induced blood vasculature development and also one of the targets for sunitinib [26]. CD34 and VEGFR2 expression were uniformly expressed in tumor vessels and not present in other cell types within the tumor microenvironment (Fig. 2). The area fraction of CD34 and VEGFR2 within each core was calculated. Patients were categorized into CD34 and VEGFR2 high versus CD34 and VEGFR2 low groups according to the median expression (0.7 and 0.9%, respectively). There were no significant correlation between either CD34-positive area fraction or VEGFR2-positive area fraction and PFS/OS (p = 0.15/0.77 for PFS and 0.62/0.85 for OS) for sunitinib treated patients (Fig. 4a and b).