Introduction

Renal cell carcinoma (RCC) is a common tumor in both men and women, which originates from renal tubular epithelial cells. Clear cell renal cell carcinoma (ccRCC) is the most common subtype of RCC and accounts for the majority of kidney cancer deaths [1]. Approximately 295,000 new cases of RCC were diagnosed worldwide each year, with approximately 134,000 deaths from RCC [2, 3]. There are several treatments available for local RCC, and the most effective method is surgery, followed by chemotherapy and radiotherapy. RCCs that are not convenient for surgical treatment or distant metastasis has occurred are usually treated with molecular targeted agents or immune checkpoint blockade for systemic therapy [4].

Vascular endothelial growth factor (VEGF) and its tyrosine kinase receptor (VEGFR) and rapamycin (mTOR) targeting mammalian serine-threonine kinase have been used as targets for RCC targeted therapy, and thus, the survival rate of metastatic RCC (mRCC) has improved [5]. Nevertheless, the 3-year overall survival rate of metastatic RCC is no more than 40% [6]. Therefore, new targets and treatments are needed to improve the prognosis of the patients.

Immune checkpoint blockade with antibodies against the PD-L1 pathway has been shown to be able to reduce tumor volume and prolong overall survival RCC [7]. In addition, PD-L1 is the focus of study in the reported immunological checkpoint markers [8].

PD-L1 also known as B7-H1 or CD274 is a surface glycoprotein belonged to B7/CD28 costimulatory factor superfamily. PD-L1 expressed on sorts of tumor cells and immune cells [9]. PD-L1 expression in cancer is considered to be a biomarker for predicting the effect of immune checkpoint blockade therapy [10]. Some reports have shown that PD-L1 is associated with the prognosis and clinicopathological in various cancers [11,12,13,14,15,16,17,18,19,20]. Studies about the prognostic significance of PD-L1 are comparatively few in RCC, and the results are not entirely consistent. So there is an urgent need to analyze the data of PD-L1 in RCC to reach a reasonable conclusion, and detection of PD-L1 expression in RCC after operation may be a great help to verify a high-risk population of RCC patients with poor prognosis.

In this study, we performed an up-to-date meta-analysis of the available evidence in order to evaluate the prognostic value of PD-L1 in RCC.

Materials and methods

Protocol

No protocol had been previously published for this review. Systematic reviews and meta-analyses do not require necessary patients’ consent or ethical approval. We conducted this systematic meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [21].

Literature search

We searched PubMed, Embase, Web of Science, the Cochrane library databases and gray literature for studies published before March 20, 2017. The meta-analysis was carried out in accordance with the PRISMA statement for reporting systematic reviews of studies that evaluate healthcare interventions [22]. The entry terms employed for search included “PD-L1” or “B7-H1” or “CD274” or “B7 homolog 1” or “programmed death ligand 1” and “renal cell cancer” or “renal cell carcinoma” or “renal cell adenocarcinoma” or “kidney tumor” and “survival” or “outcome” or “prognosis” in human, the language of publications was restrained to English language. We contacted the corresponding authors to get additional information if necessary.

Inclusion and exclusion criteria

The inclusion criteria were as follows: (1) all patients were diagnosed with histologically confirmed RCC; (2) the prognostic value of PD-L1 expression for overall survival (OS) and/or cancer-specific survival (CSS) was reported; (3) hazard ratios (HRs) and their 95% confidence intervals (95%CIs) for survival analysis were reported in text or could be computed from given data; (4) the expression of PD-L1 was measured by immunohistochemistry (IHC) method;

The exclusion criteria were as follows: (1) non-English papers; (2) non-human data; (3) review articles, case reports, letters, commentaries or conference abstracts; (4) sample cases fewer than 30; (5) duplicate publication; and (6) insufficient data for risk ratios (RR) and 95% confidence interval (95%CI), or the Kaplan–Meier curve could not be extracted. The most recent study was used for the analysis when the same population was included in different articles.

Data extraction

Two reviewers (WZ and XH) independently retrieved the data from eligible publications, with differences resolved by the third reviewer (JN) when necessary. The quality of the selected articles was assessed according to the Newcastle–Ottawa Scale (NOS) [23]. Author, publication year, country, patient number, cancer type, specimen, detection method, analysis method, the cutoff value, hazard ratios or risk ratio and corresponding 95%CIs of estimates in each comparison, follow-up period as well as positive rates of PD-L1 expression were considered for each included publication. For articles that only provided survival data in a Kaplan–Meier curve, software designed by Jayne F Tierney and Matthew R Sydes was used to digitize and extract the RR and its 95%CI [24].

Statistical analysis

Stata SE12.0 (Stata Corp LP, College Station, TX77845, USA) was used to calculate RR and 95%CI. The correlations between PD-L1 expression and the clinicopathological features of RCC (primary tumor stage, regional lymph node involvement, distant metastases, nuclear grade and histologic tumor necrosis) were assessed by RR and 95%CI. Hazard ratio (HR) with a 95%CI was computed to reveal the correlation between PD-L1 expression and prognosis. Interstudy heterogeneity was evaluated using the Chi-square test and I2 statistic (100% × [(Q − df)/Q]) [25, 26], the value of P < 0.1 and I2 ≥ 50% indicated the existence of significant heterogeneity. A fixed-effects model was used when the value of Pheterogeneity > 0.05 and I2 < 50%, otherwise, a random-effects model was applied. Begg’s funnel plot and Egger’s linear regression test to evaluate the potential for publication bias. Two-tailed P value < 0.05 was considered statistically significant.

Results

Study selection and characteristics

A total of 313 potentially relevant studies were identified through systematic literature searching. After title and/or abstracts screening, 18 articles remained for full-text assessment. Then eight articles were excluded (three not prognosis study, two duplicate study, one paired design and two received therapy). At last, ten studies [27,28,29,30,31,32,33,34,35,36] published from 2004 to 2017 with 1863 patients that met our inclusion criteria were included in the meta-analysis (Fig. 1). All studies were retrospective study design and detected PD-L1 expression using IHC. The sample size ranged from 56 to 306. Four studies were conducted in the USA [27,28,29,30], two conducted in Germany [32, 33], one conducted in Korea [34], France [35], Japan [36], and Brazil [31], respectively. The mean follow-up time ranged from 2.0 to 13 years. None of the patients in the eligible studies received neoadjuvant radiotherapy or chemotherapy before surgery. The characteristics of the ten eligible studies are listed in Table 1. PD-L1 expression levels were measured in tumor tissue. The IHC technique is summarized in Table 2. The IHC technique was varied widely among studies, with a wide range of dilution (1:25–1:200) and sources of primary antibodies coming from different companies. The cutoff used in the selected studies to evaluate PD-L1 positivity was ≥ 5% in seven studies, ≥ 10% in one study, ≥ 1% in one study, ≥ 2% in one study.

Fig. 1
figure 1

Flow diagram of the study selection process

Full size image
Table 1 Characteristics of enrolled studies
Full size table
Table 2 Immunohistochemical technique used in the studies included in the meta-analysis
Full size table

Prognostic value of PD-L1 expression for survival

In this meta-analysis, a total of 1863 cases were included in the final analysis: 1404 were clear cell RCC (ccRCC), and 459 were non-ccRCC. PD-L1 expression was found in 491 cases (26.4%). In the ccRCC population, PD-L1 expression was found in 414 cases (29.5%), while in the non-ccRCC population, 77 cases (16.8%) expressed PD-L1. A random effect model was used for the meta-analysis.

Seven studies evaluated the relationship between PD-L1 expression and OS in ccRCC patients, and four studies performed on non-ccRCC patients. In the total population, the pooled HR was 2.76 (95%CI: 2.28–3.34), indicating that PD-L1 expression has significantly increased the risk of death in RCC patients (Fig. 2). There was no significant heterogeneity exist (Chi2 = 11.22, Pheterogeneity = 0.340, I2 = 10.9%). When the analysis was restricted to ccRCC population, the pooled HR was 2.76 (95%CI: 2.25–3.38), no significant heterogeneity exists (Chi2 = 7.01, Pheterogeneity = 0.320; I2 = 14.4%). In the non-ccRCC population, the pooled HR was 2.77 (95%CI: 1.62–4.72), no significant heterogeneity exists (Chi2 = 4.22, Pheterogeneity = 0.239; I2 = 28.8%). The results illustrate that PD-L1 expression significantly increased the risk of death in both ccRCC and non-ccRCC patients.

Fig. 2
figure 2

Forest plot HR for the correlation between B7-H1 expression and overall survival in RCC patient

Full size image

Evaluation of PD-L1 expression and clinicopathological characteristics

To explore the significance of PD-L1 in pathologic diagnosis, we evaluated the correlation between PD-L1 expression and clinicopathological features. The data of primary tumor stage, lymph node metastasis, distant metastasis, Fuhrman nuclear grade and histologic tumor necrosis were extracted from the studies, and then calculated the pooled OR and 95%CI. As shown in Fig. 3 and Table 3, PD-L1 expression was significantly associated with primary tumor stage (OR = 1.76, 95%CI: 1.39–2.23; I2 = 56.3%), regional lymph node involvement (OR = 2.10, 95%CI: 1.48–2.98; I2 = 14.9%), distant metastases (OR = 2.69, 95%CI: 2.05–3.54; I2 = 0.0%), nuclear grade (OR = 1.72, 95%CI: 1.32–2.23; I2 = 79.4%) and histologic tumor necrosis (OR = 2.25, 95%CI: 1.59–3.18; I2 = 66.1%). In subgroup analysis, PD-L1 expression was significantly associated with primary tumor stage (OR = 1.86, 95%CI: 1.54–2.24; I2 = 27.9%), regional lymph node involvement (OR = 2.14, 95%CI: 1.51–23.05; I2 = 0.0%), distant metastases (OR = 2.72, 95%CI: 2.05–3.62; I2 = 0.0%), nuclear grade (OR = 2.01, 95%CI: 1.48–2.73; I2 = 82.4%), and histologic tumor necrosis (OR = 2.24, 95%CI: 1.53–3.28; I2 = 71.3%) in ccRCC population. Whereas in non-ccRCC population, PD-L1 expression was correlated with primary tumor stage (OR = 1.86, 95%CI: 1.54–2.24; I2 = 27.9%), distant metastases (OR = 2.33, 95%CI: 0.81–6.72; I2 = 77.8%) and histologic tumor necrosis (OR = 2.15, 95%CI: 0.9–5.13); but not associated with regional lymph node involvement (OR = 0.62, 95%CI: 0.02–17.32; I2 = 85.1%) and nuclear grade (OR = 1.26, 95%CI: 0.76–2.08; I2 = 67%).

Fig. 3
figure 3

Association between B7-H1 expression and clinical parameters in RCC. a Primary tumor stage (pT3/4 vs. pT1/2), b regional lymph node involvement (present vs. absent), c distant metastasis (present vs. absent), d nuclear grade (3,4 vs. 1,2), e histologic tumor necrosis (present vs. absent)

Full size image
Table 3 Meta-analysis of B7-H1 expression and clinicopathological features in renal cell carcinoma
Full size table

The relationship between PD-L1 expression and other clinical characteristics of RCC did not evaluate for limited studies.

Publication bias

Begg’s funnel plots and Egger’s test were used to assess the publication bias in this meta-analysis. Both the Begg’s funnel plot test (P = 0.276; Fig. 4) and the Egger’s (P = 0.388) verified that there was no publication bias within the included cohorts. The funnel plots for clinical features also indicated no obvious publication bias (Table 3).

Fig. 4
figure 4

Funnel plots evaluating possible publication bias for a OS, b primary tumor stage, c regional lymph node involvement, d distant metastasis, e nuclear grade, f histologic tumor necrosis

Full size image

Sensitivity analysis

The selected studies were sequentially removed to investigate whether any single study could have an influence on the pooled results. The test suggested that the pooled result did not tend to exhibit alterations when an individual study was excluded (Fig. 5).

Fig. 5
figure 5

Sensitivity analysis for the B7-H1 expression with OS

Full size image

Discussion

PD-L1, a cell-surface glycoprotein belonging to the B7 family of T cell costimulatory molecules, was discovered in 1999 that normally expression in macrophage-lineage cells [37]. Several human cancers [11,12,13,14,15,16,17,18,19,20] which found aberrantly express PD-L1, were associated with clinicopathological and immunological factors. PD-L1 upregulation may allow cancer to escape the host immune system, for PD-L1 inhibited tumor specific T cell-mediated immunity: inducing T cell apoptosis, impairing cytokine production and diminishing the cytotoxicity of activated T cells [38,39,40,41]. Studies about the prognostic significance of PD-L1 are comparatively few in RCC, and the results are not entirely consistent [28, 32,33,34, 42]. Our meta-analysis focus on the relationship between PD-L1 and RCC indicated that RCC patients with PD-L1 expression are at significant risk of tumor progression and overall mortality. The meta-analysis underscores the fact that PD-L1 in independently correlated with poor outcome in RCC.

Our analysis mainly reports the prognostic role of PD-L1 expression in RCC. Studies from different countries are included in the meta-analysis. In the present study, based on results from ten studies with 1863 subjects, the data showed that PD-L1 expression indicated a worse prognosis among RCCs.

The relationship between PD-L1 and clinicopathological features was also evaluated. The pooled OR indicated that PD-L1 expression was significantly associated with advanced tumor stage, distant metastases and the presence of histologic tumor necrosis in both ccRCC and non-ccRCC, and correlated with advanced regional lymph node involvement and higher tumor nuclear grade in ccRCC. These results suggested that PD-L1 detection was feasible for tumor aggressiveness evaluation and tumor staging. PD-L1 could be recommended as a valuable risk factor for RCC diagnosis and prognosis, especially for ccRCC.

Lacovell et al. [43]. have conduct a meta-analysis to estimate the prognostic role of PD-L1 expression in RCC, the result showed that higher expression of PD-L1 predicted poor OS in RCC (HR; 1.81, 95%CI 1.31–2.49; P < 0.001), which was consistent with our result. They also pointed out that PD-L1 evaluated in primary tumors retains its prognostic role in metastatic patients. However, their meta-analysis included six studies (five studies included patients with ccRCC, and one study included patients with non-ccRCC), in which only five studies evaluated PD-L1 expression by immunohistochemistry. We conduct a comprehensive meta-analysis dependent on ten studies (six studies included patients with ccRCC, three studies included patients with non-ccRCC, and one study included patients with both ccRCC and non-ccRCC; all evaluated PD-L1 expression by immunohistochemistry). Our meta-analysis not only evaluated the relationship between PD-L1 expression and OS in both ccRCC and non-ccRCC patients, but also evaluated the relationship between PD-L1 and clinical parameters (primary tumor stage, lymph node involvement, distant metastasis, nuclear grade and histologic tumor necrosis) in ccRCC and non-ccRCC patients. The pooled results showed that PD-L1 expression is correlated with poor prognosis and advanced clinicopathological features in both ccRCC and non-ccRCC patients. The results are more reliably and accurately.

PD-L1 is expressed on a subset of macrophages, and it can be induced by inflammatory cytokines [38, 44,45,46]. Programmed death-1 (PD-1) is expressed on activated T- and B-cells [47], T cell effector functions are attenuated when activated T cells expressing PD-1 encounter PD-L1. Taube et al. [48] reported that tumor cell PD-L1 expression was associated with infiltrating immune cells (including lymphocytes and histiocytes). They also found that PD-L1 expression on tumor and infiltrating immune cells in the tumor microenvironment were association with the presence of B-cells and lymphoid aggregates.

Several anti-PD-L1 antibodies were being developed as immune-oncology therapies and showed good results in clinical trials [49]. The FDA has approved several anti-PD-L1 antibodies, such as nivolumab, pembrolizumab and atezolizumab [50, 51]. The anti-PD-L1 antibodies have demonstrated responses and improved survival for RCC patients in clinical trials [52,53,54]. Callea et al. [55] have investigated the expression of PD-L1 between primary tumors and metastases, 53 primary ccRCCs and 76 corresponding metastases were retrieved. Overall, tumor cell PD-L1 expression was not different in primary tumors and metastases (P = 0.51). PD-L1 expression was associated with higher T stage (P = 0.03) and higher Fuhrman nuclear grade (P < 0.01). Our analysis mainly reported that PD-L1 expression evaluated on primary tumors maintains its prognostic role in RCC patients, especially in ccRCC patients.

There are several limitations that should be acknowledged. First, immunohistochemical was used to detect PD-L1 expression in all included studies in this meta-analysis, but the cutoff criteria to determine the positive or negative expression of PD-L1 and the primary antibodies used for detected PD-L1 expression were inconsistent in different studies, which may potentially contribute to heterogeneity. Therefore, a more unified standard should be defined in the future. Second, the number of patients included in the most eligible studies was relatively small. Therefore, it is needed large-scale studies to conceive more reliable results. Third, relatively few studies were extracted in some subgroup analyses, which might render premature results. Finally, research with positive results is potentially more likely to be submitted and published than work with negative results, which could cause publication bias, although this bias was not detected in the present analysis [56].

Conclusions

Our meta-analysis indicated that PD-L1 expression is a negative prognostic factor in RCC. The results also indicate PD-L1 expression was associated more aggressive clinical features in patients with RCC. More prospective and large-scale studies are needed to clarify our results.