Expression characteristics of FAM20A in stomach adenocarcinoma and its potential as a prognostic marker

Introduction

Stomach adenocarcinoma (STAD) is a common malignant tumor of the digestive system, characterized by high incidence and mortality rates worldwide. This disease ranks as the fifth leading cause of cancer-related mortality globally (1). Gastric cancer accounts for nearly one million new cases annually worldwide, resulting in >650,000 mortalities (2). China accounts for 70 and 50% of all newly diagnosed cases of cardia and non-cardia gastric cancer, respectively. The predominant histological subtype is poorly differentiated (including poorly differentiated and moderately-poorly differentiated), accounting for 73.35% of cases. In total, ~50% patients present with already advanced-stage disease (stage III or IV) at initial diagnosis, which not only poses substantial challenges to the patient but also leads to a diminished quality of life and imposes a considerable economic burden on healthcare systems (3). Current treatment modalities for STAD primarily encompass surgical intervention, chemotherapy, radiotherapy and Traditional Chinese Medicine therapy. However, these strategies are frequently constrained by difficulties in early diagnosis and the necessity for personalized treatment approaches (4–8). Consequently, there exists an urgent need to identify and develop suitable novel biomarkers that can enhance prognostic accuracy and inform therapeutic decision-making in STAD.

Family with sequence similarity 20 member A (FAM20A) was originally identified in mouse hematopoietic cells (9). Due to the substitution of the conserved glutamic acid at position 258 of human FAM20A to glutamine, which lacks a negatively charged side chain, human FAM20A does not have all the necessary catalytically active amino acids that human FAM20C possesses. Consequently, FAM20A is considered to be a pseudo-kinase (10,11). Previous investigations on FAM20A have predominantly concentrated on the positive regulation of osteogenesis and osteogenic differentiation (12–14). In addition, it has been demonstrated that FAM20A is involved in the development and progression of lung squamous cell carcinoma (negative correlation with DNA repair proteins may drive radiotherapy resistance) and papillary thyroid carcinoma (highly expressed in tumor cells) (15,16). However, to the best of our knowledge, the specific role of FAM20A in STAD remains to be fully elucidated.

The specific objectives of the present investigation were to examine the expression profiles of FAM20A across various types of cancer, evaluate the association between FAM20A expression and clinical outcomes and explore its functional implications in the context of tumor behavior and the immune microenvironment. By focusing on these objectives, the study sought to provide insights into the intricate relationship between FAM20A and cancer biology, whilst exploring its potential application in clinical settings. The present experimental validation utilized cell culture models to confirm the bioinformatics findings and identify the underlying mechanisms.

Materials and methods

Gene expression and dataset acquisition

Initially, the distribution of FAM20A across various human tissues and organs was determined by accessing the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/). RNA sequencing data were downloaded for bladder urothelial carcinoma (BLCA), breast cancer (BRCA), cervical cancer (17), cholangiocarcinoma (CHOL), colorectal adenocarcinoma (COAD), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney clear cell carcinoma (KIRC), kidney papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PAAD), pheochromocytoma and paraganglioma (PCPG), prostate cancer (PRAD), rectal adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid cancer (THCA) and uterine corpus endometrioid carcinoma (UCEC) from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov) in TPM format, along with corresponding clinical information. The following TCGA data screening protocol was used: Search terms used in the TCGA/GDC portal were the individual project codes ‘TCGA-BLCA’, ‘TCGA-BRCA’, ‘TCGA-CESC’, ‘TCGA-CHOL’, ‘TCGA-COAD’, ‘TCGA-ESCA’, ‘TCGA-GBM’, ‘TCGA-HNSC’, ‘TCGA-KICH’, ‘TCGA-KIRC’, ‘TCGA-KIRP’, ‘TCGA-LIHC’, ‘TCGA-LUAD’, ‘TCGA-LUSC’, ‘TCGA-PAAD’, ‘TCGA-PCPG’, ‘TCGA-PRAD’, ‘TCGA-READ’, ‘TCGA-STAD’, ‘TCGA-THCA’, and ‘TCGA-UCEC’, together with the filters ‘RNA-Seq’ and ‘Gene Expression Quantification’. Criteria for screening datasets for download were as follows: i) Inclusion of the aforementioned cancer types; ii) availability of RNA-seq gene expression data in TPM format; iii) availability of matched clinical information; iv) retention of samples with both expression and clinical data; and v) exclusion of samples with missing expression values or missing clinical annotation. Only datasets meeting these criteria were used for downstream analysis.

In addition, expression data were acquired from the Genotype-Tissue Expression project (GTEx; http://www.gtexportal.org/) for normal tissues to facilitate comparisons with tumor samples. The following GTEx data screening protocol was used: Search terms used in the GTEx portal were ‘GTEx’, ‘RNA-seq’, ‘gene expression’, ‘gene-level’, ‘TPM’, and ‘V8’. Criteria for screening datasets for download were as follows: i) Normal tissue samples with available RNA-seq data; ii) gene-level expression data available in TPM format; iii) availability of sample metadata, including tissue type and donor information; iv) retention of samples with both expression data and tissue annotation; and v) exclusion of samples with missing expression values or missing tissue annotation. Only datasets meeting these criteria were used for downstream analysis. Prior to conducting further analyses, log2(TPM +1) transformation was performed on all transcriptomic data to reduce the skewness of the data and enhance their robustness. This transformation is commonly employed to normalize gene expression data, reduce data skewness, stabilize variance and improve suitability for subsequent parametric statistical tests (such as the Wilcoxon rank-sum test).

For statistical analysis, R (version 4.2.2; http://www.r-project.org/) and the ‘ggplot2’ package (v3.5.2, http://ggplot2.tidyverse.org) were applied to evaluate the expression levels of FAM20A across various tumor types, using median expression levels as thresholds for differential expression analysis. The Wilcoxon rank-sum test was utilized to evaluate differences in expression between tumor and normal tissues.

Survival analysis

In the present study, survival analyses were also conducted using TCGA data only to investigate the impact of FAM20A expression on various tumor outcomes. Utilizing the ‘survival’ package (v3.5–7, http://cran.r-project.org/web/packages/survival/index.html)in R, Kaplan-Meier analysis was performed for adrenocortical cancer (10), BLCA, BRCA, CESC, CHOL, COAD, large B-cell lymphoma (DLBC), ESCA, GBM, HNSC, KICH, KIRC, KIRP, acute myeloid leukemia (LAML), lower grade glioma (LGG), LIHC, LUAD, LUSC, mesothelioma (MESO), ovarian cancer (11), PAAD, PCPG, PRAD, READ, sarcoma (SARC), skin cutaneous melanoma (SKCM), STAD, testicular germ cell tumors (TGCT), THCA, thymoma (THYM), UCEC, uterine carcinosarcoma (UCS) and uveal melanoma (UVM). Cox regression was employed to compare overall survival (OS, time from diagnosis to mortality from any cause or last follow-up), disease-specific survival (DSS, time from diagnosis to mortality attributable to the specific cancer or last follow-up) and the progression-free interval (PFI, time from diagnosis to disease progression or mortality from any cause, whichever occurred first) between high-expression and low-expression groups of FAM20A. In addition, clinical information was integrated from patients with STAD in TCGA database to analyze differences in OS, DSS and PFI among all patients with STAD, in addition to patients with different clinical stages of STAD. Factors with samples meeting the preset P-value threshold (P<0.05) in the univariate analysis were included in the construction of the multivariate Cox regression model. The ‘survival’ package was utilized for proportional hazards assumption tests and fitting survival regression models, whilst the ‘survminer’ (v0.4.9, http://cran.r-project.org/package=survminer) and ‘ggplot2’ packages were employed for visualization.

Functional enrichment analysis and gene set enrichment analysis (GSEA)

To identify genes associated with the expression levels of FAM20A, FAM20A expression data from patients with STAD in TCGA database were extracted. Pearson pairwise correlation analyses were performed between FAM20A and all other genes profiled in TCGA dataset. The top 30 correlated genes were then visualized using the ‘ggplot2’ package. In addition, the top 150 positively correlated genes and the top 150 negatively correlated genes identified by screening were subjected to Gene Ontology (11) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses utilizing the ‘ClusterProfiler’ package (v4.14.4; http://www.bioconductor.org/packages/release/bioc/html/clusterProfiler.html) in R, with ID conversion facilitated by the ‘R-org.Hs.eg.db’ package (v3.20.0; http://www.bioconductor.org/packages/release/data/annotation/html/org.Hs.eg.db.html). Following this, the ‘clusterProfiler’ package was used to perform GSEA on all molecules selected by correlation analysis after their IDs had been converted. Data arising from the GO, KEGG and GSEA analyses were visualized by the ‘ggplot2’ package.

Immune cell infiltration analysis

To comprehensively evaluate the tumor immune microenvironment and its association with FAM20A expression, the ‘estimate’ package (version 1.0.13) (18) in R was first used to calculate stromal cell scores (quantifies the extent of stromal cell infiltration in a tumor sample based on the ssGSEA enrichment score of a predefined stromal gene signature), immune cell scores (quantifies the extent of immune cell infiltration in a tumor sample based on the ssGSEA enrichment score of a predefined immune gene signature) and tumor purity (calculated as the sum of the StromalScore and ImmuneScore and represents the overall level of stromal and immune cell infiltration in the tumor microenvironment) across 33 cancer types (BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LAML, LGG, LIHC, LUAD, LUSC, MESO, OV, PAAD, PCPG, PRAD, READ, SARC, SKCM, STAD, TGCT, THCA, THYM, UCEC, UCS and UVM), where the correlations among these three immunerelated metrics and FAM20A expression were then examined in each cancer type. Subsequently. focusing on STAD, two complementary algorithms were applied to quantify the infiltration levels of a range of immune cell subtypes and assess their correlation with FAM20A expression. The ssGSEAalgorithm [implemented in the R package ‘GSVA’ (version 1.46.0) (19)/] was used to evaluate the enrichment scores of 24 immune cell types [activated dendritic cells (aDC), B cells, CD8+ T cells, cytotoxic cells, DCs, eosinophils, immature DCs (iDC), macrophages, mast cells, neutrophils, natural killer (NK) CD56bright cells, NK CD56dim cells, NK cells, plasmacytoid DCs (pDC), T cells, T helper (Th) cells, T central memory (Tcm), T effector memory (Tem), T follicular helper (TFH) and γδT cells (Tgd), Th1 cells, Th17 cells, Th2 cells and T-regulatory cells (Tregs)].

The CIBERSORT algorithm (based on immune cell markers from the CIBERSORTx website; http://cibersortx.stanford.edu/) was used to estimate the relative proportions of 22 immune cell subtypes (naïve B cells, memory B cells, plasma cells, CD8+ T cells, naïve CD4+ T cells, resting memory CD4+ T cells, activated memory CD4+ T cells, TFH, Tregs, Tgd, resting NK cells, activated NK cells, monocytes, macrophages M0, macrophages M1, macrophages M2, resting DCs, aDC, resting mast cells, activated mast cells, eosinophils and neutrophils) in tumor tissue. Subsequently, the ‘survival’ module in the TIMER database (https://cistrome.shinyapps.io/timer) was utilized to evaluate the specific relationship between OS and the abundance of six immune cell types (B cells, CD4+ T cells, CD8+ T cells, neutrophils, macrophages and DCs) in patients with STAD, where additionally generated Kaplan-Meier survival curves were used to visualize the differences in OS according to immune infiltration levels and FAM20A expression.

Cell culture and overexpression of FAM20A

The STAD cell line HGC-27 (National Collection of Authenticated Cell Cultures) was grown in DMEM (Wisent Biotechnology) containing 10% FBS (Cegrogen Biotech GmbH). HGC-27 cells were cultured in a humidified incubator at 37°C with an atmosphere containing 5% CO2. FAM20A overexpression was achieved by the transfection of a FAM20A overexpression lentivirus (cat. no. D02001, Genepharma Co., Ltd). The lentiviral vector contained the full-length FAM20A cDNA sequence (GenBank accession no. NM_017565.4; nucleotides 702–2,327; corresponding to transcript variant 1). An empty lentiviral vector [LV5(EF-1a/GFP&Puro)] was used as a negative control. To induce FAM20A overexpression, HGC-27 cells at 30–50% confluence were transfected with lentivirus (MOI=20) containing a control vector or lentivirus overexpressing FAM20A following the manufacturer's protocol. The time period between lentiviral transduction and subsequent assays was 48 h.

Total RNA was extracted using a Super FastPure Cell RNA Isolation Kit (Vazyme Biotech Co., Ltd.), where 1 µg RNA was reverse transcribed into cDNA by using HiScript IV All-in-One Ultra RT SuperMix for quantitative PCR (qPCR; Vazyme Biotech Co., Ltd.). The reverse transcription (5) reaction system was as follows: 4X All-in-One Ultra qRT SuperMix (5 µl), template RNA (1 µg of total RNA) and RNase-free ddH2O added to a final volume of 20 µl. The RT conditions were as follows: 50°C for 5 min, followed by 85°C for 5 sec. qPCR was performed using Taq Pro Universal SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd.) with specific primers for FAM20A with GAPDH as the internal control. Primers for FAM20A (forward, 5′-CAACTTTTCCCGAACCGAGC-3′ and reverse, 5′-TGTGAAGCACCTCCCAGAAG-3′) and GAPDH (forward, 5′-AATGGGCAGCCGTTAGGAAA-3′ and reverse, 5′-GCGCCCAATACGACCAAATC-3′) were used for qPCR. The qPCR reaction system was as follows: 2X Taq Pro Universal SYBR qPCR Master Mix (10.0 µl), primer 1 (10 µM, 0.4 µl), primer 2 (10 µM, 0.4 µl), template DNA/cDNA (1 µl), and ddH2O was added to a final volume of 20 µl. The qPCR cycling conditions were as follows: Stage 1 (pre-denaturation) at 95°C for 30 sec, 1 cycle; stage 2 (cycling reaction) at 95°C for 3–10 sec and 60°C for 10–30 sec, 40 cycles; stage 3 (melt curve) using the default melt curve acquisition program of the instrument. qPCR was performed on an ABI VIIA 7 real-time PCR system (Applied Biosystems; Thermo Fisher Scientific, Inc.). Relative expression was calculated using the 2−ΔΔCq method (20).

Cell proliferation assay

The number of viable cells was determined by the Cell Counting Kit-8 assay (Vazyme Biotech Co., Ltd). For this, 2,000 cells per well were seeded into 96-well plates for 24 h, followed by the addition of 10 µl CCK-8 solution to each well. The plates were then incubated at 37°C for 2 h, followed by measurement of the optical density at 450 nm using a Synergy NEO microplate reader (BioTek; Agilent Technologies, Inc.). For each condition, three technical replicate wells were used, and the experiment was repeated three times independently.

Wound-healing assay

HGC-27 cells were seeded into 6-well plates and cultured until they reached confluence, forming a monolayer. A straight line was then drawn at the bottom of each well. After discarding the medium in each well, a 200-µl pipette tip was used with a ruler to smoothly scrape the cells along the diameter of the well. PBS was used to thoroughly wash away cells that detached. Serum-free DMEM medium was added to the wells and the plate was placed in the incubator at 37°C for further culture. The cells were observed and photographed under a microscope (4× magnification; Olympus Corporation) at predetermined time-points (0, 12 and 24 h) before the distance of cell migration was calculated and statistically analyzed. The wound width was measured at the same fixed positions in each well at each time point, and the mean wound width of three technical replicate wells was used for each condition and each time point. The relative migration ability of each group at each time point was calculated as1+ (W0-Wt)/W0, where W0 is the mean wound width at 0 h and Wt is the mean wound width at the indicated time point (12 or 24 h). For each condition, three technical replicate wells were used, and the experiment was repeated three times independently.

Transwell assay

For Transwell assays, 600 µl complete medium (DMEM containing 10% FBS) was added to the wells of a 24-well plate and 200 µl cell suspension (containing 2×104 HGC-27 cells resuspended in serum-free culture medium) was added to the insert. The Transwell insert (MilliporeSigma) used with a pore size of 8 µm and without Matrigel coating. The Transwell system was then placed in the incubator at 37°C for 24 h. The culture medium was discarded from the insert, which was then thoroughly washed with PBS. Fixation was performed with 4% formaldehyde solution at room temperature for 30 min. The insert was washed with PBS to remove the fixative and staining was conducted with 0.5% crystal violet solution for 30 min at room temperature. After staining, the residual staining solution was washed off with PBS and the cells that had not transgressed through the membrane were scraped off with a scraper. Images of five random fields of view were then acquired (magnification, 40×) and statistical analysis was performed. For each condition, three technical replicate wells were used, and the experiment was repeated three times independently.

Statistical analysis

Statistical analyses and the generation of graphs were conducted using R (version 4.0.2) and GraphPad Prism version 9.0 (Dotmatics). The Wilcoxon rank-sum test was used to assess differences in gene expression between different types of tumor tissues and normal tissues. Survival analysis was conducted using the Kaplan-Meier method, with the log-rank test used to compare survival curves between groups. Multivariate Cox proportional hazards regression analysis was performed to evaluate the impact of variables on survival outcomes. The hazard ratio (HR) and its 95% confidence interval (CI) were calculated using the Cox proportional hazards regression model. Specifically, the HR was derived from the exponentiated coefficient [exp(β)] of the Cox model, where β is the regression coefficient for the grouping variable (high vs. low FAM20A expression). The 95% CI for the HR was calculated as exp[β ± 1.96 × SE(β)], where SE(β) is the standard error of the coefficient. The log-rank test was used to compare the survival distributions between the high-expression and low-expression groups. Spearman's or Pearson's algorithm was utilized to investigate the correlation between two variables. Prior to analyzing the outcomes of proliferation, Transwell and wound-healing experiments, the Shapiro-Wilk normality test and Levene's test were respectively employed to examine whether the data were normally distributed, and the variance was homogeneous. When the experimental data were normally distributed and the variance was homogeneous, unpaired Student's t-test was used for statistical analysis. If the data were normally distributed but the variance was not homogeneous, Welch's t-test was used. If the data did not satisfy the normal distribution and a non-parametric test was required, the Wilcoxon rank-sum test was adopted. ImageJ software (version 1.52a) (21) was used to analyze results arising from the wound-healing and Transwell assays. P<0.05 was considered to indicate a statistically significant difference.

Results

Variability in the pan-cancer expression of FAM20A

Initially, the tissue distribution of FAM20A expression was investigated using the HPA database. As shown in Fig. 1A, FAM20A exhibited relatively high transcriptional levels in various organs, such as the salivary gland, liver, stomach and lung. To elucidate the role of FAM20A in tumors, its expression levels were analyzed across multiple distinct tumor types utilizing data from TCGA and GTEx databases. As shown in Fig. 1B and C, group comparisons revealed that when compared with those in normal tissues, the expression levels of FAM20A were significantly reduced in BRCA (P<0.001), CESC (P<0.05), CHOL (P<0.01), KICH (P<0.001), KIRC (P<0.001), KIRP (P<0.001), LUSC (P<0.001), PAAD (P<0.05) and UCEC (P<0.01). By contrast, the expression levels of FAM20A were significantly increased in GBM (P<0.01), LIHC (P<0.001) and THCA (P<0.001). In paired analyses, in comparison with the respective matched non-cancerous tissues, a reduction in the expression levels of FAM20A was observed for BRCA (P<0.001), CHOL (P<0.05), KICH (P<0.001), KIRC (P<0.001), LUSC (P<0.001) and UCEC (P<0.05). However, an increase in the expression levels of FAM20A was identified for COAD (P<0.05), LICH (P<0.05) and THCA (P<0.001). These findings suggest that FAM20A may have diverse functions across different tumor types.

Figure 1. FAM20A expression across pan-cancer and its prognostic value. (A) mRNA expression levels of FAM20A in normal human tissues. (B) Expression levels of FAM20A in non-matched samples from cancerous and non-cancerous tissues across different types of cancer. (C) Expression of FAM20A in paired tissue samples from cancerous and non-cancerous tissues in different types of cancer. Univariate Cox regression analysis for (D) disease-specific survival, (E) overall survival and (F) progression-free interval in pan-cancers. *P<0.05, **P<0.01 and ***P<0.001. ns, no significance; TPM, transcripts per million; HR, hazard ratio; FAM20A, family with sequence similarity 20, member A; BLCA, bladder urothelial carcinoma; BRCA, breast cancer; CESC, cervical carcinoma; CHOL, cholangiocarcinoma; COAD, colorectal adenocarcinoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, chromophobe renal cell carcinoma; KIRC, clear cell renal cell carcinoma; KIRP, kidney papillary cell carcinoma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; PAAD, pancreatic ductal adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, colorectal adenocarcinoma; STAD, stomach adenocarcinoma; THCA, thyroid papillary carcinoma; UCEC, uterine corpus endometrioid carcinoma; ACC, adrenocortical carcinoma; DLBC, large B-cell lymphoma; LAML, acute myeloid leukemia; LGG, lower grade glioma; MESO, mesothelioma; OV, ovarian serous adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; TGCT, testicular germ cell tumors; THYM, thymoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.

Prognostic pan-cancer significance of FAM20A

After confirming the expression of FAM20A in different tumors, the impact of high and low expression levels of FAM20A on OS, DSS and PFI in different tumor types was analyzed (Fig. 1D-F). In terms of OS, FAM20A acted as a risk factor in BLCA, KIRC, LGG and STAD, with HR values of all >1 (all P<0.05) FAM20A also served as a protective factor in LIHC and SKCM, with HR values <1 (all P<0.01). In terms of DSS, FAM20A acted as a risk factor in KIRC, LGG, STAD and UCEC, with HR values all >1 (all P<0.05). In addition, FAM20A was a protective factor for LIHC and SKCM, with HR values <1 (all P<0.05). In terms of PFI, FAM20A was identified as a risk factor in KIRC, LGG and STAD, with HR values all >1 (all P<0.05) FAM20A was a protective factor only for SKCM, with an HR <1 (all P<0.05). Therefore, high expression levels of FAM20A may be associated with the progression and poor prognosis of various tumor types (KIRC, LGG and STAD).

FAM20A expression is associated with the staging of STAD

The initial pan-cancer analysis confirmed that FAM20A can serve as a risk factor for OS, DSS and PFI of STAD. Therefore, the differences in FAM20A expression between STAD and normal tissues were analyzed. As illustrated in Fig. 2A and B, the expression level of FAM20A in STAD tumor samples was found to be significantly higher compared with that in normal tissues, both in unmatched (P<0.001) and matched (P<0.001) sample cohorts. FAM20A expression exhibited an increasing tendency in the higher T stage (T2-T4; P<0.01) and higher N stage (N3; P <0.05), where higher expression levels were detected in deceased patients (P<0.05) and those with recurrence (P<0.05; Fig. 2C-F), suggesting that FAM20A may be associated with a higher clinical stage and worse prognosis. Therefore, the relationship between the expression of FAM20A and patient prognosis was next analyzed. As shown in Fig. 2G-I, high expression of FAM20A was significantly associated with worse prognosis in patients with STAD, as indicated by a shorter OS, DSS and PFI, with HR values all >1 (all P<0.01).

Figure 2. FAM20A expression in STAD and its association with clinical outcomes. Expression levels of FAM20A in (A) non-matched samples and (B) paired tissue samples from cancerous and non-cancerous tissues from patients with STAD. Expression levels of FAM20A in STAD tumor samples at different (C) T and (D) N stages. Expression levels of FAM20A in patients with STAD classified by (E) deceased or alive or (F) death from STAD or patient was censored for DSS. Univariate Cox regression analysis and survival analyses for (G) OS, (H) DSS and (I) PFI in STAD. Univariate Cox regression analysis for (J) OS, (K) DSS and (L) PFI in STAD at T1 and T2 stages. Univariate Cox regression analysis for (M) OS, (N) DSS and (O) PFI in STAD at T3 and T4 stages. Univariate Cox regression analysis for (P) OS, (Q) DSS and (R) PFI in STAD at stages II, III and IV. *P<0.05; **P<0.01; ***P<0.001. FAM20A, family with sequence similarity 20, member A; OS, overall survival; DSS, disease-specific survival; PFI, progression-free interval; STAD, stomach adenocarcinoma; HR, hazard ratio.

When considering the T stage and the expression of FAM20A in patients, no significant differences were found in OS, DSS and PFI of patients in the T1-T2 stage (Fig. 2J-L), whilst significant differences were identified in OS, DSS and PFI in patients in the T3-T4 stage (Fig. 2M-O), with HR values all >1 (all P<0.05). When conducting OS, DSS and PFI analyses for patients with STAD at different clinical stages, it was discovered that higher expression levels of FAM20A significantly reduced the OS, DSS and PFI of patients with STAD with more advanced clinical stages, with HR values all >1 (all P<0.05) (Fig. 2P-R). After demonstrating the association between FAM20A expression levels and clinicopathological features including T stage and N stage in patients with gastric cancer, multivariate Cox regression analysis was conducted among patients with STAD incorporating FAM20A expression levels, T stage, N stage, clinical stage and age. As presented in Table I, FAM20A retained independent significance following adjustment for the effects of other covariates. Furthermore, multivariate Cox regression analysis revealed that, in addition to FAM20A expression, age >65 years also retained independent significance as a prognostic factor for OS (P<0.05), DSS (P<0.01) and PFI (P<0.05) in patients with STAD. These analyses suggest that high expression levels of FAM20A are associated with more severe local tumor invasion and a higher clinical stage in patients with STAD.

Table I. Multivariate Cox regression analysis.

Table I.

Multivariate Cox regression analysis.

Characteristic	Total (n)	Univariate analysis [HR (95% CI)]	Univariate analysis P-value	Multivariate analysis [HR (95% CI)]	Multivariate analysis (P-value)
FAM20A expression	370	1.234 (1.043–1.459)	0.014	1.244 (1.028–1.504)	0.025
Pathological T stage	362
T1	18	Reference		Reference
T2	78	6.725 (0.913–49.524)	0.061	4.998 (0.642–38.937)	0.124
T3	167	9.548 (1.326–68.748)	0.025	6.416 (0.739–55.702)	0.092
T4	99	9.634 (1.323–70.151)	0.025	5.719 (0.639–51.201)	0.119
Pathological N stage	352
N0	107	Reference		Reference
N1	97	1.629 (1.001–2.649)	0.049	1.343 (0.674–2.679)	0.402
N2	74	1.655 (0.979–2.797)	0.060	1.430 (0.608–3.362)	0.412
N3	74	2.709 (1.669–4.396)	<0.001	2.053 (0.874–4.824)	0.099
Pathological stage	347
I	50	Reference		Reference
II	110	1.551 (0.782–3.078)	0.209	0.924 (0.345–2.471)	0.874
III	149	2.381 (1.256–4.515)	0.008	0.870 (0.231–3.275)	0.837
IV	38	3.991 (1.944–8.192)	<0.001	2.118 (0.566–7.928)	0.265
Age, years	367
≤65	163	Reference		Reference
>65	204	1.620 (1.154–2.276)	0.005	1.862 (1.292–2.684)	<0.001

[i] HR, hazard ratio; FAM20A, family with sequence similarity 20, member A.

Functional enrichment analysis of differentially expressed genes associated with FAM20A in STAD

After finding that FAM20A was associated with the prognosis of patients with STAD, the 300 genes most associated with FAM20A were next screened before the top 30 positively and negatively correlated genes were selected to create a heatmap of correlation (Fig. 3A and B). The genes positively correlated with FAM20A were co-upregulated in the same subset of STAD samples in which FAM20A was highly expressed and were expressed at low levels in samples with low FAM20A expression; representative positively correlated genes included HRH2, PI3KR5 and THEMIS2. By contrast, the genes negatively correlated with FAM20A showed the opposite pattern, being highly expressed in samples with low FAM20A expression and expressed at low levels in samples with high FAM20A expression; representative negatively correlated genes included CDK1, DKC1 and RAB25. To investigate how the expression of FAM20A affected changes in the signaling pathways associated with STAD, GO and KEGG analyses were next conducted using the top 300 genes (Fig. 3C-F). The analysis identified enriched pathways in cellular component (such as ‘external side of plasma membrane’, ‘secretory granule membrane’, ‘cell leading edge’ and ‘membrane microdomain’), biological process (such as ‘immune response-regulating signaling pathway’, ‘positive regulation of cytokine production’, ‘leukocyte mediated immunity’, ‘regulation of immune effector process’), molecular function (such as ‘nucleoside-triphosphatase regulator activity’, ‘GTPase regulator activity’, ‘immune receptor activity’ and ‘cytokine receptor activity’) and KEGG pathway categories (such as ‘tuberculosis’, ‘pathogenic Escherichia coli infection’, ‘chemokine signaling pathway’ and ‘neutrophil extracellular trap formation’). Several GO functional terms and KEGG pathways related to immune-related functions were enriched by the FAM20A-correlated genes. (such as ‘immune response-regulating signaling pathway’ and ‘regulation of immune effector process’). After conducting GSEA analysis using the genes selected by correlation analysis with FAM20A, similar conclusions (such as Thelper pathway, primary immunodeficiency, Fcgr activation) were reached (Fig. 3G-J). These data suggest that FAM20A may be associated with the immune response in STAD.

Figure 3. Functional enrichment analysis of FAM20A-associated genes in STAD. Top 30 genes that were most (A) positively and (B) negatively correlated with FAM20A. P-values were adjusted using the Spearman's algorithm. GO/KEGG analysis for the top 300 genes most closely correlated with FAM20A, with the classification of enrichment analysis results according to (C) CC, (D) BP, (E) MF and (F) KEGG pathway categories. Gene set enrichment analysis of genes with absolute values >0.1 with FAM20A, with the classification of enrichment analysis according to (G) BIOCARTA, (H) KEGG, (I) PID and (J) REACTOME categories. FAM20A, family with sequence similarity 20, member A; KEGG, Kyoto Encyclopedia of Genes and Genomes; CC, cellular component; BP, biological process; MF, molecular function; P adj, adjusted P-value.

High infiltration of macrophages and DCs correlated with FAM20A expression and poor prognosis in STAD

Based on the present hypothesis that FAM20A may be associated with the immune response in STAD the estimate, ssGSEA and CIBERSORT algorithms were then used to analyze the correlations among FAM20A expression and infiltration by various tumor microenvironmental stromal cells and immune cells.

FAM20A expression showed a strong association with both stromal cell (P<0.001) and immune cell infiltration (P<0.001) in STAD, where FAM20A expression was also significantly association with tumor purity (P<0.001) (Fig. 4A). These data suggest a role of FAM20A in the tumor microenvironment, where high levels of FAM20A expression may be associated with poor prognosis in STAD due to the ability of FAM20A to regulate the tumor microenvironment. When analyzed using the ssGSEA algorithm, FAM20A expression was positively correlated with all cell types (including macrophage, DC, iDC, pDC and aDC), except for Th2 and Th17 cells (Fig. 4B). When analyzed using the CIBERSORT algorithm, FAM20A was positively correlated with M2 macrophages (P<0.001), monocytes (P<0.01), resting mast cells (P<0.01), CD8+ T cells (P<0.01), activated CD4+ memory T cells (P<0.05) and eosinophils (P<0.01), whilst being negatively correlated with resting CD4+ memory T cells (P<0.05), activated mast cells (P<0.05), M0 macrophages (P<0.05), activated NK cells (P<0.01) and Tregs (P<0.001; Fig. 4C). As shown in Fig. 4D and E, patients were further divided into FAM20A high and low expression groups for analysis with the ssGSEA algorithm, which revealed increased infiltration by the majority of immune cell types in the FAM20A-high group. The CIBERSORT algorithm revealed an increase of CD8+ T cells (P<0.001), activated CD4+ memory T cells (P<0.01), M2 macrophages (P<0.01) and resting mast cells (P<0.05) in cells with high expression levels of FAM20A, while Tregs (P<0.01) and activated mast cells (P<0.05) were decreased. Collectively, these results suggest that macrophages and DCs with high FAM20A expression levels may be associated with the tumor microenvironment in patients with STAD. Therefore, the relationship between high and low expression levels of FAM20A and B cell, T cell, macrophage and DC infiltration, along with survival of patients with STAD, was analyzed using the TIMER database. As shown in Fig. 4F, the infiltration of levels (according to the median infiltration level of each immune cell type) of macrophages (P<0.01) and DC cells (P<0.05) into gastric cancer, were significantly associated with a shorter survival time in patients with STAD. As depicted in Fig. 5A and B, when the range of the analysis was extended to multiple malignant tumors, the estimate algorithm demonstrated that the expression of FAM20A exhibited a pronounced correlation with the stromal cell score, immune cell score and tumor purity in various types of malignant tumor (such as ACC, BLCA and BRCA). When analyzed using the ssGSEA algorithm, FAM20A demonstrated a significant correlation with immune cells in the majority of malignant tumors, except for a few malignant tumor types, including CHOL, LIHC and MESO. This result suggests that FAM20A may be associated with the tumor immune microenvironment beyond STAD, providing a direction for future research.

Figure 4. Correlation between FAM20A expression and immune cell infiltration in STAD. (A) Correlation between FAM20A expression and stromal cells, immune cell infiltration and tumor purity in STAD. The correlations between FAM20A expression and various immune cells were analyzed with the (B) ssGSEA and (C) CIBERSORT algorithms in STAD. The enrichment score for immune cells with high and low expression levels of FAM20A in tumors were analyzed with the (D) ssGSEA and (E) CIBERSORT algorithms. (F) Impact of enrichment of immune cells with high and low expression levels of FAM20A in tumors on the overall survival of patients with STAD. P-values were adjusted using the Pearson algorithm.*P<0.05, **P<0.01 and ***P<0.001. FAM20A, family with sequence similarity 20, member A; ssGSEA, single sample gene set enrichment analysis; STAD, stomach adenocarcinoma; Cor, correlation coefficient; aDC, activated dendritic cells; DCs, dendritic cells; iDC, immature DCs; NK, natural killer; pDC, plasmacytoid DCs; Th, T helper; Tcm, T central memory; Tem, T effector memory; TFH, T follicular helper; Tgd, gamma delta T cells; Th1, type 1 T-helper; Th17, Th17 cells; Th2, Th2 cells; Tregs, T-regulatory cells.

Figure 5. Correlation between FAM20A expression and tumor microenvironment components across multiple cancer types. (A) Correlation between FAM20A expression and stromal cells, immune cell infiltration and tumor purity in different types of cancer. (B) The correlation between FAM20A expression and various immune cells in different types of cancer were analyzed by the ssGSEA algorithm. P-values were adjusted using the Spearman's algorithm.*P<0.05, **P<0.01 and ***P<0.001. FAM20A, family with sequence similarity 20, member A; Cor, correlation coefficient; aDC, activated dendritic cells; DCs, dendritic cells; iDC, immature DCs; NK, natural killer; pDC, plasmacytoid DCs; Th, T helper; Tcm, T central memory; Tem, T effector memory; TFH, T follicular helper; Tgd, gamma delta T cells; Th1, type 1 T-helper; Th17, Th17 cells; Th2, Th2 cells; Tregs, T-regulatory cells BLCA, bladder urothelial carcinoma; BRCA, breast cancer; CESC, cervical carcinoma; CHOL, cholangiocarcinoma; COAD, colorectal adenocarcinoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, chromophobe renal cell carcinoma; KIRC, clear cell renal cell carcinoma; KIRP, kidney papillary cell carcinoma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; PAAD, pancreatic ductal adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, colorectal adenocarcinoma; STAD, stomach adenocarcinoma; THCA, thyroid papillary carcinoma; UCEC, uterine corpus endometrioid carcinoma; ACC, adrenocortical carcinoma; DLBC, large B-cell lymphoma; LAML, acute myeloid leukemia; LGG, lower grade glioma; MESO, mesothelioma; OV, ovarian serous adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; TGCT, testicular germ cell tumors; THYM, thymoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.

Impact of FAM20A on STAD cells

To investigate the influence of FAM20A expression on gastric cancer cells, a lentivirus overexpressing FAM20A was transfected into HGC-27 cells (Fig. 6A). Compared with that in the control group, overexpression of FAM20A significantly promoted the proliferation of HGC-27 cells at 48 h (relative proliferation, 10.826±0.439 vs. 8.199±0.460, P<0.001; Fig. 6B). After confirming that the overexpression of FAM20A could enhance the proliferation of gastric cancer cells, the impact of FAM20A on the migration and invasion of gastric cancer cells was next assessed by wound healing and Transwell assays, respectively. As depicted in Fig. 6C, compared with that in the control group, overexpression of FAM20A significantly strengthened the invasive ability of gastric cancer cells (2.235 vs. 1.020; P<0.001). In the wound healing assay, overexpression of FAM20A resulted in the wound closing faster over 24 h (1.926±0.286 vs. 1.370±0.133, P<0.01 at 12 h; 3.566±0.336 vs. 2.103±0.400, P<0.001 at 24 h; Fig. 6D). These results indicate that overexpression of FAM20A possibly increased the metastatic capabilities of tumors by promoting the proliferation, migration and invasion of gastric cancer cells.

Figure 6. FAM20A overexpression promotes the proliferation and migration of STAD cells. (A) Reverse transcription-quantitative PCR was used to confirm the successful transfection of FAM20A-OE lentivirus. (B) Cell Counting Kit-8 assays were used to investigate the viability of control and FAM20A-overexpressing HGC-27 cells at 12 and 24 h after transfection. (C) The influence of FAM20A on the migratory ability of HGC-27 cells was assessed by Transwell assays (scale bars, 50 µm). (D) The influence of FAM20A on the migratory capacity of HGC-27 cells was assessed by wound-healing assays (scale bars, 200 µm). **P<0.01; ***P<0.001. FAM20A, family with sequence similarity 20, member A; OE, overexpression.

Discussion

STAD represents a significant global health challenge due to its high incidence and mortality rates. This disease is the fifth most common cancer type worldwide and ranks fifth for all cancer-related mortalities, with >700,000 mortalities each year (1). Current therapeutic strategies, including surgery, chemotherapy and radiation, typically yield limited success (22–24), highlighting the urgent need to identify novel biomarkers for early detection and prognostic improvement.

In the present study, survival analysis demonstrated an association between high FAM20A expression levels and unfavorable outcomes across multiple types of cancer, including gastric cancer, bladder cancer and kidney cancers. This relationship is particularly concerning, as it suggests that elevated levels of FAM20A may not only serve as a prognostic indicator but could also be implicated in the aggressiveness of tumors. Nonetheless, FAM20A has been shown to be associated with improved prognosis in patients with SKCM and LIHC. This suggests that FAM20A can modulate the development of various tumor types through multiple signaling pathways or varying activation levels within the same pathway.

The association between FAM20A expression and immune cell infiltration patterns adds another layer of complexity to its role in cancer biology. Accumulating evidence has demonstrated that the infiltration of tumor-associated macrophages (TAMs) and DCs in gastric cancer exhibits a well-defined association with the prognosis of patients (25–27). The findings demonstrate that high FAM20A expression correlates with increased infiltration of macrophages and DCs. Notably, the infiltration of these two cell types in gastric cancer is also associated with poor prognosis, as they are commonly implicated in the tumor microenvironment (28–32). This suggests that FAM20A may contribute to immune evasion mechanisms utilized by tumors, thereby negatively impacting patient outcomes. The present results showed that high expression levels of FAM20A in tumors correlated with the increased infiltration of DCs, which is associated with a significantly unfavorable prognosis for patients with STAD. Considering the subtypes of DCs, FAM20A expression exhibited a positive correlation with the infiltration of iDCs and pDCs. iDCs are known to differentiate into tolerogenic DCs under the influence of IL-10, thereby inhibiting T-cell responses (33). Furthermore, pDCs are frequently associated with unfavorable prognoses of patients with cutaneous malignancies (34), thus aligning with the current findings. In STAD tissues characterized by high levels of FAM20A expression, there was an increase in both total macrophage and M2 macrophage infiltration. This is potentially key because previous studies found that TAMs and M2 macrophages are implicated in tumor progression (such as HNSC) (35,36). These data suggest that elevated levels of FAM20A within tumors may facilitate the recruitment of additional macrophages and DCs, thereby promoting tumor growth within the tumor microenvironment. Understanding the interplay between FAM20A and the immune landscape of tumors may provide crucial insights into the development of novel immunotherapeutic strategies. This, however, remains contingent upon further investigations at the single-cell level in future studies.

The present in vitro experiments demonstrated that overexpression of FAM20A significantly enhanced the proliferation and migration of gastric cancer cells. The HGC-27 cell line was selected because the majority of newly diagnosed gastric cancer cases in China are of low or moderately low differentiation (3). HGC-27 originates from a poorly differentiated tumor (37), optimally reflecting the characteristics of the Chinese patient population. These findings provide experimental support to the notion that FAM20A may function as an oncogene in gastric cancer that promotes tumor aggressiveness and metastatic potential. The mechanisms underlying these effects warrant further investigation, particularly with regard to the signaling pathways activated by FAM20A that contribute to these phenotypic changes. Understanding these pathways could ultimately lead to the identification of novel therapeutic targets for the treatment of gastric cancer, particularly for patients with high levels of FAM20A expression. Furthermore, these results emphasize the importance of integrating both in vitro and clinical data to develop comprehensive models for the role of FAM20A in cancer biology, thus facilitating the translation of these findings into clinical practice.

According to previous reports, FAM20A is a pseudokinase that forms a functional complex with Fam20C, which then potentiates the phosphorylation of exogenous proteins (ameloblastin and amelotin) within the secretory pathway (10). It appears that the function of FAM20A is completely dependent on FAM20C. However, based on data acquired from the BioGRID database (v4.4.230; http://thebiogrid.org), FAM20A can interact with parkin RBR E3 ubiquitin protein ligase (PRKN), although FAM20C is a protein known to interact with PRKN (38). The gene products of PRKN were initially identified as being mutated in human Parkinson's disease, which were subsequently demonstrated to function in concert to promote mitophagy, thereby implicating dysfunctional mitochondria in the etiology of Parkinson's disease (39). PRKN-mediated mitophagy has also been reported to participate in the occurrence and development of multiple types of tumor (LIHC, BRCA, COAD and PRAD) (40–43). However, in the present study, an analysis of TCGA database suggested that the expression of PRKN is not associated with the prognosis of patients with STAD (data not shown).

There are certain limitations to the present study that need to be considered. The study primarily relied on publicly available databases (such as TCGA and GTEx), which may have introduced inherent biases in sample selection, data processing and clinical annotation. To address database bias, future studies should validate FAM20A expression and its correlation with immune infiltration in independent, multi-center clinical cohorts using real-world patient samples with matched RNA-sequencing or reverse transcription-quantitative PCR data. According to the HPA database, FAM20A expression is undetectable in HGC-27 cells, making small interfering RNA/short hairpin RNA knockdown experiments unfeasible in this cell line. As a result, the present study could only rely on exogenous overexpression to assess gain-of-function phenotypes. Furthermore, when reviewing the HPA database, it was found that only a limited number of gastric cancer cell lines (NCC-StC-K140, TGBC11TKB, GSU) express FAM20A, which are not routinely available from standard commercial cell repositories and can only be obtained through direct requests to the originating laboratories or specialized cell banks. This is a limitation of the present study. To overcome the lack of experimental validation, future research should employ in vitro gain/loss-of-function assays (such as CRISPR-Cas9 knockout or overexpression in STAD cell lines) combined with functional tests (proliferation, apoptosis, migration and invasion) and in vivo xenograft models to directly confirm the oncogenic role of FAM20A. In addition, the specific molecular mechanisms by which tumors with high expression levels of FAM20A recruit immune cells or how immune cells with high expression levels of FAM20A influence tumor progression remain undefined and further research is still required. Futures studies also needs to determine whether there are more proteins other than FAM20C that can interact with FAM20A and thereby affect multiple signaling pathways. Specifically, understanding the molecular interactions and regulatory networks involving FAM20A could elucidate its function in distinct tumor microenvironments, thereby facilitating the development of targeted therapies or nanomedicines aimed at modulating its expression or activity in various cancers (44). Furthermore, the identification of FAM20A as a potential prognostic marker may enhance clinical decision-making processes. Therefore, prospective studies with larger, well-annotated patient cohorts (including treatment history and survival follow-up) are needed to evaluate its independent prognostic value in multivariate models and its potential for guiding personalized therapy.

In conclusion, the present comprehensive investigation of FAM20A revealed its significant association with tumor biology and patient prognosis across various types of cancer. The expression patterns identified herein emphasize the potential of FAM20A as a prognostic biomarker and therapeutic target, particularly for STAD. By elucidating the interplay between FAM20A expression and the tumor immune microenvironment, the present study provides a foundation for future investigations aimed at developing novel therapeutic strategies.

Acknowledgements

Not applicable.

Funding

The present study was financially supported by the Shengli Oilfield Central Hospital (grant no. zxyy-ky-202510).

Availability of data and materials

The data generated in the present study may be requested from the corresponding author.

Authors' contributions

TN contributed to writing the original draft, bioinformatics analyses and formal analysis. ZL was involved in data curation, including data extraction, cleaning and dataset verification, formal analysis, including statistical analysis using R and PRISM, writing the original draft and investigation. ZF was responsible for writing the original draft, in vitro experiments and visualization. QL obtained resources and was involved in the investigation. LL was responsible for data curation (organization and verification of experimental data) and formal analysis (statistical analysis of the experimental data). JW was involved in the acquisition and interpretation of data for bioinformatics analysis, provided resources, and contributed to visualization and supervision. YL performed conceptualization, supervision, project administration, reviewing and editing, and funding acquisition. JL made substantial contributions to the conception and design of the study and to the acquisition, analysis and interpretation of data, including the bioinformatics results and experimental data, and was responsible for data curation, including the verification of bioinformatics and experimental data; JL also contributed to funding acquisition. All authors read and approved the final version of the manuscript. YL and JL confirm the authenticity of all the raw data.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Glossary

Abbreviations

Abbreviations:

References