Background
For patients with metastatic renal cell carcinoma (mRCC), therapeutic options have expanded significantly these years, since vascular endothelial growth factor (VEGF)-targeted tyrosine kinase inhibitor (TKI) drugs such as sunitinib and pazopanib have been established as first-line therapy [
1]. Several clinical based prognostic models, for example the Heng’s risk criteria, have also achieved remarkable progress in mRCC patient survival prediction [
2]. However, the objective response rates (ORRs) for most first-line TKI drugs were only around 30% [
3], and the lack of validated molecular biomarkers impeded their personalized approach [
4]. This was in contrast to many other tumor types, in which protein expression and mutation were used as basic accesses for drugs response and patient survival prediction [
5,
6].
CC-chemokine receptor 7 (CCR7), a G-protein coupled receptor (GPCR) mostly expressed on immune cells, was initially regarded as an important regulator facilitating leukocytes homing to the lymphatic structures, where its two ligands CC-chemokine ligand 19 (CCL19) and CCL21 are constitutively expressed [
7]. However, in recent years, aberrant high CCR7 expression has also been identified on several tumor types, linking to a potential invasive phenotype [
8]. It has been suggested that CCR7 positive tumor cells could mimic the normal lymphocyte homing function and interact with lymph vessels, leading to subsequent lymph node specific metastasis [
9].
Lymphatic and hematogenous disseminations were two regular metastasis pathways for malignancy. For mRCC, although the most common metastatic site was the lung, possibly via a hematogenous approach, local or distant lymph node involvement at diagnosis was not rare [
10]. Patient receiving TKIs could also develop new lymph node metastasis during the treatment, leading to a progressive disease (PD). Several theories of TKI drug resistance emphasized an increase of tumor cell invasiveness after drug administration [
11]. These processes mostly accompanied with tumor cell migration and matrix metalloproteinase-9 (MMP-9) mediated extracellular matrix degradation [
12], which was similar to the CCR7 mediated lymph vessel intravasation process [
8]. As VEGF targeted therapies mostly focused on blood vessels inhibition, whether they could enhance the possibility of mRCC metastasis through other pathways, such as CCR7 mediated lymph vessel invasion and therapy resistance, was not known.
Thus, here through immunohistochemistry (IHC), we retrospectively evaluated the CCR7 expression in 110 primary tumor specimens of mRCC patients treated with sunitinib and sorafenib. The result suggested a positive correlation of CCR7 expression with patient baseline lymph node metastasis and TKI drugs response. CCR7 high expression could predict a poorer overall survival (OS) and progression free survival (PFS) for mRCC patients after TKIs.
Methods
Patient selection
We initially screened a total of 138 mRCC patients treated with sunitinib or sorafenib between March 2005 and June 2014 at the Department of Urology, Zhongshan Hospital, Fudan University. The inclusion criteria were: pathologically proven RCC patient with metastatic lesion, treated with sunitinib or sorafenib at first without further second-line treatment, had enough Formalin Fixed Paraffin Embedded (FFPE) specimens, and had detailed laboratorial, imaging and survival information. Patients who had former malignant history, received metastasectomy or those with tumor necrosis area >80%, unavailable data were excluded. At last, 28 patients were excluded and 110 patients were selected for the study, in which three were excluded from PFS analysis for incomplete drug response information. This study was approved by the Clinical Research Ethics Committee of Zhongshan Hospital, Fudan University (Shanghai, China) (B2015-030).
Patients’ OS was defined as the time from therapy initiation to the time of death, or was censored at the last follow-up. PFS was calculated from the time of therapy initiation to the time of progression, according to the RECIST 1.1 criteria [
13], or was censored at the last follow-up. All data were collected retrospectively from medical records and electronic databases using uniform database templates, and the last follow-up time was December 2015. According to the 2014 EAU guidelines [
14] and 2012 ISUP consensus [
15], two urologic pathologists (Yuan J. and Jun H.) independently reviewed all the H & E slides of patient samples and confirmed the RCC diagnosis and Fuhrman grade classification. Initial stage at diagnosis was reclassified based on the 2010 AJCC TNM classification [
10]. Heng’s risk model was applied as previously reported [
2].
Tissue microarray and immunohistochemistry
Two representative tumor cores 3 mm in diameter from each sample were selected for tissue micro array (TMA) construction. Anti-CCR7 antibody (ab32527, Abcam, diluted 1/1000) and Dako EnVision Detection System were applied in the immunohistochemistry procedure [
16]. Through western blot in RCC cell lines, the specificity of antibody was confirmed. Negative control was performed without applying primary antibody. Olympus CDD camera, Nikon eclipse Ti-s microscope (×200 magnification) and NIS-Elements F3.2 software were used to record the staining results. Using Image-Pro Plus version 6.0 software (Media Cybernetics Inc., USA), an integrated optical density (IOD) score could be calculated for each scan. Two urologists unaware of the patients’ clinical data evaluated these slides. Each person took three independent shots with the strongest staining for each core, and the IOD mean of each patient’s two cores (six scans) were calculated. Kappa value was calculated for evaluating inter-observer agreement.
Statistical analysis
Univariate analysis was carried out to explore the prognostic and predictive value of continuous CCR7 IOD score. The smooth estimates of hazard ratio (HR) of IOD on patient survival were displayed using R software, “phenoTest” package [
17]. For clinical usage, we dichotomized the IOD into high/low expression through minimum p value method using x-tile software [
18], and because the p values obtained might be overestimated, they were corrected using the formula proposed by Altman and colleagues [
19]. The smooth HR curves after dichotomizing were illustrated through R software, “smoothHR” package [
20]. After this,
χ2 test, Fisher’s exact method and Cochran-Mantel-Haenszel
χ2 test were applied for assessing the relationship between CCR7 expression and patients’ clinicopathological parameters. Kaplan–Meier, univariate and subsequent multivariate analysis were performed, in which 1000 bootstrap resamples was used for reducing overfitting bias. Finally, time dependent receiver operating characteristic (ROC) analysis was done to analyze the adding prognostic value of CCR7 expression to the Heng’s risk model. GraphPad Prism 6 (GraphPad Software Inc., USA), SPSS 21.0 (SPSS Inc., USA), X-tile 3.6.1 (Robert L Camp, USA) and R software 3.1.2 (R Foundation for Statistical Computing, Austria) were used in these procedures. A two-sided
P-
value < 0.05 was regarded as statistically significant.
Discussion
CCR7 was naturally a homeostatic chemokine receptor expressed on various subtypes of immune cells encompassing T cells, B cells, natural killer cells and dendritic cells, enabling them to circulate through the CCL-19/21 positive lymphatic highways [
7]. Its expression on cancer cells was first recognized on hematogenous malignancies, in which the connection between high CCR7 expression and lymphoid organ involvement was discovered [
21]. Subsequently, numerous laboratory studies confirmed this pro-invasion, mostly pro-lymph vessel metastatic function in various cancer types, including breast cancer, melanoma, non-small cell lung cancer, prostate cancer, head-and-neck cancer and gastrointestinal cancer [
9]. Here, our results suggested that CCR7 also expressed on several RCC specimens and associated with various patient baseline characteristics (Table
1, Additional file
1: Figure S1).
The clinical prognostic value of CCR7 has been studied in several other cancer types. In a meta-analysis including 1697 gastric cancer patients, high CCR7 correlated with a worse 5-year overall survival rate [
22]. Studies in melanoma and colorectal cancer also displayed similar results [
23,
24]. In this study, we have found that high CCR7 staining intensity could be used as an adverse prognosticator for mRCC patients’ overall survival. Moreover, a new model integrating CCR7 expression into Heng’s risk criteria performed better than using Heng’s risk alone, and the adding prognostic value mostly came from the intermediate patient groups (Additional file
3: Table S2, Fig.
2a and b). This indicated that a substantial number of patients might switch between risk groups with a consequence for choice of treatment strategy after CCR7 status was considered. ROC analysis for the ccRCC patient group were also carried out, considering the poor performance of CCR7 in non-ccRCC patients (Fig.
2c and d).
For PFS analysis, to our knowledge, this study was the first to report an association between tumoral CCR7 expression and PFS in mRCC patients following TKIs. As the most widely used systemic therapy at present, TKI drugs only reached ORRs for about 30% for mRCC patients, and the theories of drug resistance was complicated [
25]. It is increasingly evident that in some tumors, in which angiogenesis is thwarted genetically or pharmacologically, cancer cells could adapt by migrating more aggressively into normal tissue, based on several pre-existing invasion programmes such as epithelial-mesenchymal transition (EMT) and MMP2/MMP9 secretion [
26,
27], or by switching on several distinctive programmes which were currently unknown [
28]. Since the CCR7 mediated lymphatic specific migration and metastasis was also based on several above mentioned programmes [
8,
29], and as a molecule which could give tumor itself survival signal besides [
30], we hypothesized that CCR7 positive RCC cells might have a potential to migrate into the adjacent lymphatic tissue for survival after TKI drugs administration, subsequently leading to resistance and disease progression. As a result, our study did identify an independent significant adverse predictive effect of high CCR7 expression on mRCC patients’ PFS (Fig.
3b) and its association with poorer best drug response (Table
1, Fig.
3a). Patients’ baseline lymph node involvement status was also correlated (Table
1, Fig.
4a) and four patients who developed PD due to new lymph node metastasis all represented high CCR7 staining for their tumor tissues. All these results above indicated an impact of tumoral CCR7 expression on patients’ PFS and lymphatic involvement status, making this molecule a potential predictor for mRCC patients receiving TKIs.
The major limitations of this pilot study were its retrospective design and relatively small sample size, with patients from a single center and same ethnic background. Although central patient data review and bootstrap validation were performed for minimizing inter-observer and over-fitting bias, further external prospective validations were required, and related basic researches were needed for verifying our hypothesis. Samples from the metastatic site after TKI therapies might be excellent candidates for in-depth investigation. Besides, although we took two cores and six scans from each patient’s sample, intratumoral heterogeneity still might confound the results. Patients in this study received sunitinib or sorafenib as first-line therapy because other agents were not available in China at that time, and other VEGF-based TKI drugs such as pazopanib, axitinib and cabozantinib should also be taken into further consideration.
Acknowledgments
The authors would like to thank Dr. Yuan Ji, Dr. Jun Hou and Ms. Haiying Zeng (Department of Pathology, Zhongshan Hospital of Fudan University) for diagnosis confirmation and technical assistance, respectively.