Introduction
Cancer is a major cause of death affecting human health worldwide. There are an estimated 19.3 million new cancer cases and almost 10.0 million cancer deaths occurred in 2020 globally [
1]. Since the occurrence and development of tumors may be accompanied by different gene alterations, it is an important problem to find biomarkers for the diagnosis of different tumors. Pan-cancer research is to analyze multiple aspects of a large number of tumors and examine the differences in genes in different tumor types, so as to have a fuller understanding of tumors and find therapeutic and diagnostic targets for a variety of tumors.
Angio-associated migratory cell protein (AAMP) was first discovered by Beckner when they screened cell surface proteins related to cell motility in melanoma cells [
2]. It belonged to the immunoglobulin superfamily and had homologous domains with cell adhesion molecule proteins NCAM, LFA-2, PECAM, etc. The structural characteristics of AAMP protein suggested that it may be involved in cell adhesion and migration. Subsequently, it was found that AAMP was involved in cancer occurrence and development. For example, AAMP could accelerate the adhesion and proliferation of breast cancer cells, and high AAMP expression had a worse prognosis for breast cancer patients [
3,
4]. AAMP interacted with EGFR to enhance the proliferation and drug resistance of non-small cell lung cancer cells [
5]. The interaction between AAMP and CDC42 could accelerate non-small cell lung cancer cells' metastasis [
6]. However, there are few studies on the AAMP gene expression pattern and latent function in pan-cancer.
To study the latency effect of AAMP in pan-cancer, we analyzed the transcription level of AAMP and its relationship with clinical pathology from multiple public databases. Then, we conducted bioinformatics analysis to investigate the biological function and prognostic significance of AAMP in pan-cancer.
Materials and methods
Expression of AAMP in pan-cancer
Tumor Immune Estimation Resource (TIMER) is an online platform for analyzing immune cell infiltration in tumors and gene expression differences between tumors and normal tissues in the TCGA database [
7]. We analyzed AAMP expression in pan-cancer from TIMER. Because there were no normal tissues in some tumors from TIMER, so we downloaded the expression profile data of 33 tumors in TCGA and GTEx to compare AAMP gene expression in pan-cancer. The Wilcoxon test was investigated the difference in AAMP gene expression between tumors and normal tissues. AAMP expression in tumors and its matched normal tissues was studied by paired sample
T test. AAMP expression at the protein level was analyzed using the CPTAC data set from the UALCAN and immunohistochemical image analysis from the HPA database. In addition, the ROC curve was used to estimate the diagnostic significance of AAMP using the pROC package in R. Then, we downloaded RNA sequence and clinical information for liver hepatocellular carcinoma(LIHC) from LIRI-JP in the ICGC database to verify the expression and prognosis of AAMP.
Prognosis analysis in pan-cancer
According to the median value of AAMP, we separated AAMP expression into AAMP high-expression and low-expression groups. Univariate regressive analysis was used to explore the impact of AAMP on the prognosis of 33 kinds of tumors using a survival package and forest plots for visualization. The prognostic indicators included overall survival (OS), disease-specific survival (DSS), and progress-free interval (PFI).
Gene mutation, TMB, and MSI analysis in pan-cancer
Gene mutation is a common mode of epigenetics. cBioPortal is an online website for studying gene mutation analysis in tumors [
8,
9]. Gene alteration contains mutation, structural variant, amplification, and deep deletion. Moreover, tumor mutation burden (TMB) and microsatellite instability (MSI) are two highly effective markers for tumor immunotherapy. Tumors with higher TMB can recruit more neoantigens on the surface of tumor cells, increase the immunogenicity of tumors, and improve the efficacy of immunotherapy [
10]. Studies have shown that tumors with high MSI are highly sensitive to immune checkpoint inhibitor treatment [
11]. Spearman correlation analysis was applied to discuss the correlation between AAMP expression and TMB, MSI.
Association of AAMP expression with clinicopathological characteristics in LIHC
The clinical information of LIHC patients was acquired from TCGA data, including age, sex, T stage, N stage, M stage, clinical stage, and other clinical features. Moreover, the Chi-Square test was used to study the correlation between AAMP and clinical parameters.
Relationship of AAMP expression with the prognosis of LIHC and its nomogram
We used the Kaplan–Meier curve to analyze the effect of AAMP expression on OS, PFI, and DSS in LIHC. Moreover, we explored the independent prognostic factors of AAMP in LIHC by univariate and multivariate regression analysis. The 1- year, 3- year, and 5- year survival rate of AAMP in LIHC was investigated using a time ROC curve. A calibration curve was drawn to evaluate the precision accuracy of the nomogram.
Immune cell infiltration and immune checkpoints analysis in LIHC
We used the ssGSEA algorithm in R packet-GSVA to discuss the correlation of AAMP with 24 kinds of immunocyte infiltration [
12,
13]. Moreover, we used the ESTIMATE package in R to estimate the stromal cells and immune cells in tumor tissue, predict tumor microenvironment (TME) by immune and stromal scores, and analyze the association of AAMP with stromal and immune score in LIHC [
14]. The presence of immune checkpoint inhibitors has made significant progress in immune therapy. We analyzed the difference in expression of eight common immune checkpoints between the high and low AAMP expression groups to predict the effect of immunotherapy. Tumor Immune Dysfunction and Exclusion (TIDE) algorithm is used to predict the response to cancer immunotherapy. The efficacy of immune checkpoint blockade (ICB) is poorer with a higher TIDE score, and the survival time is shorter after receiving ICB treatment [
15,
16]. We predicted immunotherapy response by analyzing TIDE scores of high and low AAMP expression groups in LIHC.
Gene enrichment analysis in LIHC
The AAMP was classified into high- and low-expression groups according to the median expression. We explored the DEGs between high- and low-expression groups in LIHC using the DESeq2 (version 1.26.0) package. Gene Ontology (GO) enrichment and GSEA were performed to explore the relevant pathways involved in AAMP in LIHC using the “Cluster Profiler” package of R language. GO enrichment contains molecular function (MF), cellular component (CC), and biological process (BP). P < 0.05, and FDR < 0.25 are defined as significantly enriched.
Statistical analysis
Wilcoxon test was utilized to analyze the differential expression of two groups. The chi-square test was explored the association between AAMP expression and clinical features. Univariate regression analysis and the Kaplan–Meier curve were used to investigate the effect of AAMP on prognosis. We predicted the impact of AAMP expression in LIHC on 1-, 3-, and 5-year's OS by timeROC package. Spearman correlation analysis was used to research the correlation of AAMP expression with TME and immune checkpoints.
Discussion
AAMP is a member of the immunoglobulin superfamily and is widely distributed in various types of cells. It is essential in transcriptional activation, cell cycle regulation, protein–protein interaction, and signal transduction [
19‐
21]. AAMP is expressed in a variety of cell types, including endothelial cells, aortic smooth muscle cells, dermal fibroblasts, renal proximal tubular cells, glomerular mesangial cells, human breast cancer cells, human melanoma cells and prostate cancer cells [
22‐
24]. Recent studies have shown that AAMP mainly locates in the cytoplasm and membrane of vascular endothelial cells, affecting the angiogenesis, diffusion, migration, and cytoskeleton remodeling processes of endothelial cells [
25,
26]. In addition, it has been reported that AAMP is abnormally up-regulated in metastatic CRC and boosts the occurrence of colorectal cancer by inhibiting SMURF2-mediated RhoA liquefaction and degradation [
27]. Furthermore, AAMP is highly expressed in invasive gastrointestinal and breast carcinoma cells and is a marker of poor prognosis [
4,
28]. In addition, AAMP plays a positive role in angiogenesis and is regulated by Astrocytes in coculture [
29]. These results showed that AAMP is vital in cancer occurrence and progression.
Our study explored AAMP expression, prognostic value, gene alteration, TMB, and MSI in pan-cancer through multiple databases. AAMP had high expression in 18 kinds of tumors, while AAMP had low expression in four tumors. The highly expressed AAMP had inferior OS, DSS, and PFI in ACC and LIHC. We used cBioPortal to study AAMP gene alteration in pan-cancer. In UCEC, AAMP had the highest gene alteration frequency, including mutation 1.89%, amplification 1.32%, and deep deletion 0.19%. TMB and MSI are considered as two biomarkers of response to immunotherapy. Tumors with high TMB and high MSI are sensitive to immunotherapy response [
30]. Therefore, we study the correlation between AAMP and TMB, MSI in pan-cancer. AAMP expression was positively correlated with TMB for ACC, ESCA, LUAD, STAD, LUSC, THYM, and PAAD while negatively correlated for KIRP, THCA, and UCS. Regarding MSI, AAMP had a positive relation to MSI in ESCA, STAD, KIRC, LUSC, LIHC, and UVM, and negative relation in THCA. AAMP correlated with TMB and MSI in ESCA, STAD, and LUSC. It can be speculated that in ESCA, STAD, and LUSC, tumors with high expression of AAMP have better responses to immunotherapy.
We focused on LIHC after screening and discussed the AAMP expression, prognosis, clinical features, and immunity. AAMP was over-expressed at the mRNA expression level from the TCGA database and protein expression level from CPTAC. The immunohistochemical staining in the HPA database also confirmed AAMP's high expression in LIHC. Moreover, AAMP overexpression was correlated with histological grade and pathological stage of LIHC. AAMP had higher expression in the higher histological grade and advanced pathological stage. Kaplan–Meier survival curve demonstrated that AAMP's high expression was related to unfavorable OS, DSS, and PFI. The univariate and multivariate analysis results showed that AAMP was an independent adverse prognostic factor for LIHC patients. Furthermore, ROC curve analysis demonstrated that OS of 1, 3 and 5 years was more than 0.6. We also confirmed the high expression of AAMP in LIHC in another different database. The results showed that 1, 2 and 3 year OS was 0.662, 0.737, and 0.726 by ROC curve analysis. These findings suggested that AAMP expression had an excellent prediction ability in TCGA and ICGC and could predict LIHC prognosis, supporting AAMP expression as a new predictor of survival for LIHC.
TME is essential in tumor development and is closely related to patient outcomes [
31]. In recent years, increasing studies have confirmed that different TME of patients is vital in mediating late metastasis, immune escape, and immunosuppression. We used the ssGSEA algorithm to explore the association between AAMP and immune cell infiltration in LIHC. With the increase of AAMP expression, the number of DC, Cytotoxic cells, pDC, neutrophils, B cells, Th17 cells, Treg, mast cells, eosinophils, iDC, Th1 cells, Tgd, T cells, NK CD56 dim cells decreased, and the number of T helper cells, Th2 cells, and Tcm increased. Therefore, changes in AAMP expression lead to changes in Th1/Th2. The immunosuppressive state will affect the body's anti-tumor immunity and ultimately result in tumors [
32]. Immune and stromal cells, as the non-tumor components of TME, have gradually attracted the attention of researchers due to their essential roles in tumor genesis, metastasis, drug resistance, and prognosis [
33,
34]. We found that the stromal score and immune score of AAMP over-expression were lower than AAMP low-expression, indicating that AAMP affected the occurrence and development of LIHC through immunocyte infiltration. Recently, the treatment of advanced malignant patients has been revolutionized by the introduction of immune checkpoint inhibitors [
35]. Moreover, common checkpoints included CD274, CTLA4, HAVCR2, LAG3, PDCD1, PDCD1LG2, TIGIT, and SIGLEC15. The expression of CD274, CTLA4, HAVCR2, LAG3, PDCD1, and TIGIT in the AAMP high-expression group was higher than AAMP low expression group, which showed that AAMP was the vital immunotherapeutic target in LIHC. TIDE is considered an indicator of cancer's immune intelligence response rate. The patients with higher TIDE scores have lower immune responses. In our study, the patients with increased expression of AAMP have low immune response rates and cannot benefit from immune checkpoint inhibitors. This may be because high AAMP expression induces a decrease in B and T cells, while low lymphocyte counts indicate a poor host anti-tumor immune response [
36].
To further investigate the biological process of AAMP in LIHC, we analyzed GSEA in TCGA and ICGC databases. Interestingly, the highly expressed AAMP in the two databases is mainly concentrated in cell cycle, DNA replication, and mismatch repair, indicating that AAMP causes liver cancer through the above pathways.
We discussed the expression pattern and prognostic significance of AAMP in pan-cancer and LIHC from a bioinformatics perspective, providing a basis for further research on the mechanism of AAMP for LIHC. However, our study had some limitations. We studied AAMP expression in pan-cancer and LIHC only by bioinformatics. Furthermore, many experiments to explore the mechanism of AAMP for LIHC will help developing a more accurate prognosis model for patients and providing a basis for more personalized treatment.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.