Background
Treatment with tyrosine kinase inhibitors (TKIs) improves survival for patients with metastatic renal cell cancer (mRCC) [
1]. One of the first and most commonly used TKI, sunitinib, increases progression free survival (PFS) with a median of 6 months compared to interferon alpha (IFN-α) [
2]. However, not all patients benefit from treatment with TKIs. Side effects, ranging from mild reversible to chronic toxicity, could be avoided and the costs of the medication could be reduced if predictive biomarkers for TKI treatment were available.
There are several established prognostic factors for mRCC patients and these were updated by Heng in 2009. Hengs criteria include Karnofsky performance status, hemoglobin, calcium, time from diagnosis to treatment, neutrophils and platelet counts [
3]. Predictive factors indicate the sensitivity or resistance to a specific medication. There are currently no established predictive markers for TKI treatment.
Most studies trying to find predictors have analysed the connection between different circulating proteins in serum and response to sunitinib therapy [
4‐
7]. Tissue microarray (TMA) analysis enables direct investigation of protein expression in malignant cells and stroma, but only a few studies have been reported. For hypoxia-inducible factor 1α (HIF-1α), CD31, vascular endothelial growth factor (VEGF) receptors, CA9, Ki67 and platelet-derived growth factor receptor α (pPDGFRα) associations with response to sunitinib therapy have been demonstrated [
8,
9]. In another and larger TMA-based study, tumoral expression of programmed death ligand 1 (PD-L1) or PD-L1 plus tumor infiltrating CD8+ T-cells were correlated to significantly shorter PFS and overall survival (OS) in patients treated with sunitinib or pazopanib [
10]. We have previously reported TMA-studies indicating that cubilin (CUBN) and annexin A1 (ANXA1) expressed in the tumor cells are predictive markers in mRCC patients treated with sunitinib and sorafenib [
11,
12].
In about 60–75% of clear cell renal cell cancers (RCC) the tumor suppressor gene von Hippel-Lindau (VHL) is inactivated leading to accumulation of the hypoxia-inducible factor (HIF) which leads to overexpression of VEGF and PDGF [
13]. Vascular endothelial growth factor and PDGF are growth factors stimulating angiogenesis, tumor spread and tumor growth [
14,
15]. Tumor-associated vessels are larger than normal vessels and leaky which leads to high interstitial fluid pressure (IFP) and swelling in and around tumor tissues. The nutrient and oxygen delivery is poor leading to hypoxia within the tumor which stimulates production of pro-angiogenetic factors and continued development of abnormal vasculature [
16]. Tyrosine kinase inhibitors, such as sunitinib, sorafenib and pazopanib, block VEGF- and PDGF-receptors and thereby inhibit development of new pathological blood vessels [
17].
Epidermal growth factor (EGF), latrophilin and seven transmembrane domain-containing protein 1 (ELTD1), also named as adhesion G protein-coupled receptor L4 (ADGRL4), is thought to play a role in pathological angiogenesis. It belongs to the adhesion G-protein-coupled receptor (GPCR) superfamily [
18]. ELTD1 expression is part of a recently identified gene signature commonly associated with angiogenesis in human tumors, and is frequently upregulated in tumor-associated endothelial cells (EC) in human cancer, including, renal carcinoma [
19,
20]. Higher intratumoral EC ELTD1 expression was significantly correlated with smaller tumor size and better prognosis in mRCC patients [
19]. These findings indicate that ELTD1 expression is regulated during tumor angiogenesis in RCC, and is therefore a putative biomarker for response to anti-angiogenic therapy.
The aim of this study was to explore the potential value of assessing the expression of ELTD1, CD34 and VEGFR2 in tumor vessels to predict benefit of sunitinib treatment in mRCC patients.
Discussion
About 20% of RCC patients suffer from a metastatic disease already at the time of diagnosis [
27]. Prognosis has been very poor for these patients throughout the years. No chemotherapy nor radiotherapy has been successful [
28] and only a minority of the patients respond to cytokine treatment with IFN-α and IL-2 used earlier [
29]. Novel molecular agents such as TKIs have changed the standard of treatment for mRCC and prolonged the survival [
1]. Recently, the combination of the anti PD-1 antibody nivolumab and cytotoxic T-lymphocyte associated protein 4 (CTLA4) ipilimumab was registered as a first line treatment for intermediate and poor-risk mRCC patients based on a positive study comparing this combination with single sunitinib [
30]. However, all patients do not benefit from this specific immunotherapy and many patients develop adverse side effects leading to discontinued therapy. Therefore, TKIs are still considered as one of the cornerstones in the treatment arsenal for mRCC patients. Predictive markers are needed to use TKIs more accurately and select individuals who will benefit from the treatment. No such markers are in clinical use yet.
Some serum proteins have been suggested as predictors for treatment with TKIs in mRCC patients, especially for sunitinib treatment. In a few (
n = 21) sunitinib medicated patients, increased baseline levels of tumor necrosis factor α (TNF-α) and metalloproteinase-9 (MMP-9) were associated with non-responders and reduced time to progression (TTP) and OS [
6]. Another sunitinib study (
n = 85) demonstrated that elevated baseline levels of serum VEGF and neutrophilgelatinase-associated lipocalin (NGAL) correlated to higher relative risk of progression [
7]. Some clinical side-effects related to TKIs, such as development of hypertension and hand-foot skin reaction correlate with better response and improved OS [
31‐
33]. Hence, instead of predicting the benefit they have a role as early evaluators of the treatment.
Putative predictive tumoral proteins can be analysed with TMAs, where all cores are stained at the same time and under same conditions. In one study, patients with elevated tumor expression of HIF-α or expression of VEGFR3 in tumor vessels experienced improved PFS while treated with sunitinib. In the same study a low CA9 score was correlated to a shorter OS [
8]. Tumoral expression of PD-L1 or PD-L1 plus tumor-infiltrating CD8+ T cell counts predicted shorter PFS and OS for patients treated with sunitinib or pazopanib in a TMA of mRCC cases [
10].
The function of ELTD1 in tumor vessels has not been fully elucidated, but a role as a regulator of vascular sprouting has been proposed [
19]. In renal cancers (
n = 157), higher ELTD1 expression on tumor-associated ECs was significantly correlated with improved survival. Surprisingly, silencing ELTD1 inhibited tumor growth in experimental models. Upon in vivo ELTD1 knockdown, microvascular density (MVD) was significantly reduced, hypoxia and apoptosis of ECs increased and cell proliferation was inhibited [
19].
The advantage of using imaging analyses for scoring is that it is more objective and quantitative than manual assessment [
34]. We found that patients with high ELTD1 staining in their primary tumor vasculature had a greater benefit from sunitinib treatment in terms of a prolonged PFS. ELTD1, expressed on ECs, may act as an indicator of activated angiogenesis. As the goal of the sunitinib-treatment is to block new pathological vessels to grow, high ELTD1 expression in tumor vasculature could indicate patients with better likelihood to respond to this treatment, as our results demonstrate. Hence, there is a scientific basis for our finding.
In a subgroup analysis we separated patients having received sunitinib up front or as a second line treatment. The difference in PFS remained significant in the group having received sunitinib in the first line (n = 70) while the same correlation could not be recalled for the fewer mRCC patients (n = 29) who were treated with sunitinib after progression (data not shown). The sensitivity of the tumor to sunitinib therapy might be changed by the primary treatment and our subgroup analysis may indicate ELTD1 being a predictor limited only to the first line therapy. Another possible explanation for these findings is that a true difference also for the second line treated group may not be relived because of the limited number of patients.
In addition to PFS we observed also significantly longer OS in sunitinib treated patients with high ELTD1 vessel staining compared to the patients with low expression of ELTD1. The gain seen in PFS could lead to a prolonged OS. However, OS may also have been affected by other therapeutic regimens. Another established end point in cancer studies is objective response (OR). However, the benefit from TKIs is often not captured by OR but better by evaluating PFS as in the present study.
The level of expression of CD34 in the tumors did not correlate with benefit of sunitinib treatment. Since CD34 is a marker which is homogenously expressed on vascular endothelial cells, we conclude that the number of vessels is not predictive for sunitinib treatment. Furthermore, neither VEGFR2 which is a target for sunitinib was associated with response to the treatment.
We found no correlation between expression of ELTD1 and PFS or OS in sorafenib treated patients. The survival curves in this subanalysis do not show even a trend to separate and furthermore, no significant survival differences were demonstrated for the whole group. This confirms ELTD1 being a pure predictive and not a prognostic marker.
Combining ELTD1 with another biomarker might enhance the predictive value. We have previously shown the value of combining different biomarkers [
11,
12].
There are a couple of limitations in the present study. First, potential heterogeneity between the primary tumors and metastases could not be assessed since there were no available metastatic tissues. Second, though the current study is one of the largest predictive marker studies reported for RCC patients, a higher number would have been desirable to yield more reliable results. The retrospective design does not affect our results as the aim of the study was to investigate a putative predictor, not yet another prognostic marker.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.