1 Introduction
In 2020, more than 400,000 new cases of renal cell carcinoma (RCC) were reported worldwide, which accounts for approximately 3% of all adult cancers [
1], and the majority of the cases were clear cell renal cell carcinomas (ccRCC) [
2]. Over 30% of the RCC patients were diagnosed with metastases, with a 5-year survival rate of approximately 10% [
3].
Recent randomized controlled trials have demonstrated the fundamental role of immunotherapy (IO) plus tyrosine kinase inhibitor (TKI) for metastatic RCC, with superior compared with TKI monotherapy [
4‐
7]. Nonetheless, a significant proportion of patients continue to exhibit inherent resistance to IO + TKI or develop acquired resistance [
6‐
8]. To aid clinical decision-making, novel biomarkers for IO + TKI therapy in RCC are urgently required.
Annexin A2 (ANXA2) is one of the members of the annexin family, which regulates cellular development and signal transduction pathways [
9,
10]. ANXA2 has been reported to be upregulated in multiple cancer types, including RCC [
9,
10]. ANXA2 has also been demonstrated to be involved in tumor angiogenesis [
11] and metastasis [
12]. In RCC, Annexin A2 was reported to be associated with higher tumor grade, metastatic potential and poor prognosis [
13,
14]. Elevated ANXA2 expression was also able to promote RCC cell motility and migration [
15]. Recently, ANXA2 was also found to regulate Hippo signaling pathway, and may be a potential target for pharmaceutical intervention in RCC [
16]. However, the mechanisms underlying ANXA2 function in RCC remain obscure.
In recent years, increasingly more studies have focused on the relationship between ANXA2 and tumor microenvironment [
17]. ANXA2 was found to be essential for T cell to T cell interactions, and the upregulation of ANXA2 leads to decreased T-cell activation in tumor microenvironment [
18]. Annexin A2 also promotes immune escape of hepatocellular carcinoma, by upregulation of immune checkpoint molecules, and decreased expression of effector factors, including perforin, granzyme B (GZMB), interferon-γ, and TNF-α [
17]. Moreover, soluble ANXA2 also showed immunosuppressive properties in RCC [
19], and elevated serum levels of ANXA2 may be important for the suppression of the immune response [
20]. These growing evidences showed the important role of ANXA2 in tumor immune escape. In fact, ANXA2 contributes to therapeutic resistance of immunotherapy [
21]. Accordingly, preclinical studies have been evaluating the anti-tumor efficacy of Annexin A2-targeting therapy [
22], alone or in combination with anti-PD-1 antibodies [
23,
24]. However, these is no study evaluating the prognostic or predictive role of ANXA2 for IO + TKI therapy in RCC.
In the study, we aimed to evaluate the prognostic role of ANXA2 in advanced RCC treated by IO + TKI, in two independent cohorts. We intended to explore the function of ANXA2 in tumor microenvironment of RCC. We also aimed to build a predictive model for IO + TKI therapy benefit in advanced RCC.
2 Materials and methods
2.1 Study cohorts and datasets
This study included two advanced RCC cohorts, the Zhongshan Hospital metastatic renal cell carcinoma (ZS-MRCC) cohort, and the Javelin-101 cohort.
The inclusion criteria of the ZS-MRCC cohort were RCC metastasis, IO + TKI therapy, no other malignancies, and sample availability. Fifty-one patients were initially enrolled, and therapeutic responses were evaluated every 6–8 weeks after treatment initiation according to computed tomography. Therapeutic responses were defined as complete response (CR), partial response (PR), stable disease (SD) or progressive disease (PD) according to the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria [
25]. Patients of CR or PR were defined as responders, and those of SD or PD were non-responders. The primary endpoint of the study was PFS, defined as the time from IO + TKI therapy initiation to disease progression or death. Six patients were excluded due to unavailable sample, or loss of follow-up. Finally, the ZS-MRCC cohort included 45 patients. Baseline characteristics of the ZS-MRCC cohort are displayed in Supplementary Table S1. The correlations between baseline characteristics and therapeutic response are listed in Supplementary Table S2.
The Javelin-101 cohort included 726 patients from the phase III Javelin-101 trial of advanced RCC [
7]. In this cohort, 354 patients received IO + TKI, while 372 patients received TKI monotherapy. The inclusion and exclusion criteria have been previously described [
7]. Patients’ characteristics, genomic data, transcriptomic data and PFS were acquired from the previous studies [
7,
26]. Baseline characteristics of the Javelin-101 cohort are listed in Supplementary Table S3.
This study also included another two datasets for transcriptomic, immunologic and functional studies, the Zhongshan Hospital high risk renal cell carcinoma (ZS-HRRCC) dataset, and the Cancer Genome Atlas Clear Cell Kidney Cancer (TCGA-KIRC) dataset. The ZS-HRRCC dataset initially enrolled 43 high-risk localized RCC patients. Three patients were excluded for sample unavailability or not passing sample quality control. The remaining 40 patients’ samples were used for analysis. The TCGA-KIRC dataset included 530 samples of clear cell RCC [
27]. Patients’ characteristics, survival, genomic and transcriptomic data were all downloaded from the UCSC xena browser (
https://xena.ucsc.edu/) [
27].
The study was approved by the Clinical Research Ethics Committee of Zhongshan Hospital, Fudan University (B2021-119), following the Declaration of Helsinki. Written informed consent was obtained from each participant.
2.2 RNA-sequencing procedures
RNA-sequencing were performed in the ZS-MRCC cohort and the ZS-HRRCC dataset. For total RNA isolation, the MagBeads Total RNA Extraction Kit (Catalog #T02-096) was used according to the manufacturer’s instructions. The RNAClean XP Kit (Cat#A63987, Beckman Coulter, Inc. Kraemer Boulevard Brea, CA, USA) and the RNase-Free DNase Set (Cat#79254, QIAGEN, GmBH, Germany) were used for total RNA purification. Library construction and sequencing was performed by Shanghai Biotechnology Corp. (Shanghai, China). For RNA library preparation and sequencing, VAHTS Universal V6 RNA-sequencing Library Prep Kit for Illumina (Cat#NR604-02, Vazyme, Nanjing, China) and NovaSeq 6000 equipment (Illumina, USA) were utilized. RNA-sequencing data were standardized to read count and Fragments Per Kilobase of exon model per Million mapped fragments (FPKM).
2.3 Hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC)
Hematoxylin and eosin (H&E) staining and immunohistochemistry were performed on formalin-fixed, paraffin-embedded RCC slides in the ZS-HRRCC dataset. Tumor-infiltrating lymphocytes (TILs) were assessed by H&E staining, according to previously described methodology by the International TILs Working Group [
28]. IHC was performed according to previously described procedures [
29]. IHC markers and primary antibodies are listed in Supplementary Table S4. The PANNORAMIC® 250 Flash III DX (3DHISTECH Ltd.) and the CaseViewer application (3DHISTECH Ltd.) were used for image capture and display. IHC analysis was conducted under six randomized fields for each sample, by three independent investigators blinded to patients’ information. The average value was calculated for further analysis.
2.4 Flow cytometry
Peripheral blood samples were collected before surgery, stored in heparinized tubes at 4 °C within two hours before the experiment. RBC Lysis Buffer (Thermo Fisher Scientific) was used to isolate white blood cells. RCC samples were collected and minced after resection. After digestion by collagenase IV (Sigma) and DNase I (Sigma) at 37 °C and passing through a 70-μm strainer, RCC samples were then treated with RBC lysis buffer (Thermo Fisher Scientific) to obtain single cell suspensions. Samples were then treated with Fc receptors blockade, and stained for 30 min at 4 °C with fluorescently labeled membrane marker antibodies. Intracellular proteins were stained with corresponding antibodies after being disposed by Intracellular Fixation and Permeabilization Buffer (Thermo Fisher Scientific). Samples were then stained by the fluorochrome-labeled antibodies and preserved in cell staining buffer. Flow-cytometry analysis were performed by BD LSRFortessa™ X-20 (BD Biosciences) and analyzed by Flowjo v10.0 (Tree Star). Antibodies used for flow cytometry are listed in Supplementary Table S4.
2.5 Random forest model construction
Random forest (RF) is a nonparametric approach for classification and regression [
30], which is suitable for analyzing survival and omics data [
31,
32]. Expression of ANXA2, PDCD1, CD4, GZMB, CD8A, CD274, CTLA4, GZMK were included as parameters for the construction of the random forest score (RF score). The random forest model was constructed by the “randomForestSRC” and “ggRandomForests” packages of the R software.
2.6 Statistical analysis
Chi-square test, Fisher’s exact analysis or Cochran–Mantel–Haenszel test were used for categorical variables. Wilcoxon signed-rank test or Kruskal–Wallis H test were used for continuous variable. Spearman’s correlation test was used for correlation analysis. Kaplan–Meier analysis with log-rank test and Cox regression analysis were used for survival analysis, by the “survival” and “survminer” packages of the R software. The forest plots were plotted by the “forestplot” package, and the waterfall plot was plotted by the “ComplexHeatmap” and “ggplot2” packages of the R software. All statistical analyses were performed by using R software (
https://www.r-project.org/). For all statistical analyses, two-tailed P < 0.05 was regarded as statistically significant.
4 Discussion
The combination therapy of IO + TKI has greatly improved the survival of advanced RCC patients [
4‐
7], but there is still no available biomarker for it. ANXA2 has been found to take part in tumor immune escape [
17‐
20], and therapeutic resistance of immunotherapy [
21]. Accordingly, preclinical studies have been evaluating the anti-tumor efficacy of ANXA2-targeting therapy [
22‐
24]. In the current study, high-ANXA2 indicated resistance and poor PFS of IO + TKI therapy, in two independent RCC cohorts. High-ANXA2 was also related with T cell dysfunction, and terminal exhaustion factors. High-ANXA2, combined with immune parameters, could stratify patients who would benefit from IO + TKI therapy.
Although IO + TKI has been proved to be superior than TKI monotherapy by phase III trials, there are still a great proportion of patients that could not benefit from it [
4‐
7]. However, no biomarker has been applied yet. Biomarker-guided therapy selection between IO and TKI has shown potential benefit [
33], but no biomarker is currently available for the selection between IO + TKI and TKI. In the current study, the low ANXA2 group showed a higher rate of tumor shrinkage (CR or PR, 44.8%) compared with the high ANXA2 group (CR or PR, 18.8%), and a lower rate of therapeutic resistance (PD, 24.1%) compared with the high ANXA2 group (PD, 37.5%), although not reaching statistical significance (Fig.
1G). To better represent therapeutic effect of IO/TKI, PFS was also calculated and set as the primary endpoint. ANXA2 correlated significantly with PFS in both the ZS-MRCC cohort and the Javelin-101 cohort (Fig.
2), demonstrating its prognostic value. The results also supported further investigations of ANXA2’s function in tumor immunity. A multivariate Cox regression model was also built to confirm the prognostic role of ANXA2 (Fig.
2B). The multivariate model included three risk factors: tumor histologic subtype (non-ccRCC or ccRCC), IMDC risk group (favorable, intermediate or poor) and ANXA2 expression (high or low). Among these, tumor histologic subtype and IMDC risk group are well-recognized prognostic factors for metastatic RCC. More importantly, the IMDC risk group is a widely-used risk stratification model for metastatic RCC, which includes six important prognostic factors: < 1 year from time of diagnosis to systemic therapy, Karnofsky performance status < 80%, hemoglobin < lower limit of normal, corrected calcium > upper limit of normal, neutrophils > upper limit of normal, and platelets > upper limit of normal. Accordingly, the multivariate model included most prognostic factors of metastatic RCC, which demonstrated ANXA2 expression as an independent factor for progression-free survival (Fig.
2B). More importantly, the integrated RF score could guide treatment selection between IO + TKI and TKI, as only low-RF score patients could benefit from the combined IO + TKI therapy (Fig.
6B–D). However, the potential application of ANXA2 and RF score still need further external validations. Additionally, since IO + IO has also been recommended for first-line therapy of metastatic RCC, further investigations are also expected for the role of ANXA2 in IO + IO therapy.
Previous studies reported up-regulated ANXA2 expression during RCC progression [
13,
14]. In the current study, ANXA2 expression was also found upregulated in tumor tissues, especially in high-grade or high-stage RCC (Fig.
1A–C). As one of the members of the annexin family, ANXA2 is involved in multiple oncogenic pathways, including cellular development and signal transduction pathways [
9,
10], angiogenesis [
11], and metastasis [
12]. In RCC, ANXA2 could regulate Hippo signaling pathway [
16], and promote migration of tumor cells [
15]. However, the mechanisms underlying ANXA2 function in RCC still remain obscure.
Previous evidences have indicated ANXA2 to be correlated with suppressive tumor microenvironment [
17], especially correlated with decreased T-cell activation [
18]. In the current study, ANXA2 was also found associated with T cell dysfunction, represented by decreased GZMB expression (Fig.
4A–D). In addition, ANXA2 was found correlated with exhaustion markers of T cells, including checkpoints (TIGIT, LAG3, TIM-3, CTLA4) and transcription factors (TOX, TCF1, T-bet, EOMES) (Fig.
4F, G). In hepatocellular carcinoma, ANXA2 has been reported to promotes immune escape, by upregulation of immune checkpoint molecules, and decreased expression of effector factors [
17]. These evidences support the potential relevance between ANXA2 and immune suppression, but the mechanism has not clarified yet.
Soluble ANXA2 also showed immunosuppressive properties in RCC [
19]. In our study, high-ANXA2 was related with increased number of PD-L1 + macrophages (Fig.
5E, F), and upregulated expression of suppressive factor, including IL10, CXCL8, MMP9 and TGF-β (Fig.
5I). The infiltration of suppressive macrophages, and overexpression of suppressive factors, are potential mechanism for immune suppression in high-ANXA2 samples. The immunosuppressive properties of ANXA2 may explain why high-ANXA2 tumors were resistance to IO + TKI. Accordingly, ANXA2 contributes to therapeutic resistance of immunotherapy [
21], and ANXA2-targeting therapy in combination with anti-PD-1 antibodies has been evaluated in preclinical studies [
23,
24]. However, the detailed mechanisms by which ANXA2 regulates T cell exhaustion and immunotherapy resistance need further researches.
The study has several limitations. Firstly, the retrospective design is the major limitation, leading to potential recall and enrollment biases. Secondly, the ZS-MRCC cohort enrolled patients of different histological subtypes. Although multivariate Cox analysis and hierarchical analysis were performed to avoid potential biases, the various histological subtypes may still become an important confounding factor. Besides, the ZS-MRCC cohort also enrolled patients of second-line or third-line IO + TKI therapy, mainly due to the fact that this regimen has only recently been approved for first-line treatment. In addition, the significantly different number of participants in the cohorts may lead to different results. However, despite the limited number of patients in the ZS-MRCC cohort, ANXA2 was still related with poor PFS, with statistical significance. Nevertheless, future validations in prospective trials will be more ideal.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.