Datasets processing

The GSE62254 datasets was downloaded from the gene expression omnibus (GEO) database of the National Center for Biotechnology Research (http://www.ncbi.nlm.nih.gov/gds). In Table 1, we present 298 gastric cancer patients and their clinical and pathological characteristics, as well as survival data materials (Table 1). Clinical materials including the age, gender, TNM stage, tumor grade, site of tumor metastasis and survival time. Patients with incomplete information such as survival or pathological stage were excluded. For module detection, the RNA-seq data was included for subsequent analysis and we eliminate two outlier samples (GSM1523817, GSM1523984) according to sample network from our datasets (Additional Figs. 1B and 1C). In Additional Fig. 1A, we demonstrate the research process of this study to make our research ideas easier to understand. RNA information is sourced from the STAD-PRJEB25780 dataset and can be downloaded from the STAD website. The TCGA-STAD dataset was downloaded from TCGA database (https://cancergenome.nih.gov).

Table 1 Characteristics of GSE62254 cohortFig. 1

Go ontology and KEGG pathway enrichment analyses of the genes in the yellow module. A GO functional (Biological Process, BP) enrichment analyses of genes in the yellow module. B GO functional (Cellular Component, CC) pathway enrichment analyses of genes in the yellow module. C GO functional (Molecular Function, MF) pathway enrichment analyses of genes in the yellow module. D KEGG pathway enrichment analyses of genes in the yellow module. (The –log10 (P-value) of each term is colored according to the legend). E PPI network of genes in yellow module. F Hub network of top 50 hub genes. G The significant sub-module was identified by MCODE

Network construction

In order to screen modules with highly correlated genes and construct a weighted gene co- expression network, we applied WGCNA for analysis [16]. We first used WGCNA to identify important modules from thousands of genes. Further determine the key basis for GC peritoneal metastasis based on the correlation between gene expression and sample traits. In this study, the weighted adjacency matrix was created with the formula amn =|cmn|β (where amn: adjacency between gene m and gene n, cmn: Pearson’s correlation, and β: soft-power threshold). The modules were clustered by calculated the TOM matrix. We defined the minimal gene module size as 30 to obtain appropriate modules, and the threshold to merge similar modules was set to 0.25.

Identifying modules and functional enrichment analysis

Correlation between modules and different clinical features was assessed by calculating the module eigengene (ME). Pearson’s correlation test was used to analyse the relationship between gene significance (GS) and clinical features by linear regression, P < 0.05 indicated significant correlation.

The construction of protein–protein interaction network

In this work, we constructed a PPI network by using the online database STRING (https://cn.string-db.org/). The Cytoscape software was then applied to analyze the interaction of the candidate proteins. A plug-in named The Molecular Complex Detection (MCODE) was used to score and explore parameters that have been optimized to produce the best results for the network. For this, we set the following parameters in MCODE: Degree Cut-off = 2, Node Score Cut-off = 0.2, K-Core = 2 and Max Depth = 100 [17, 18].

Establishment of the nomogram

The nomogram was established with T, N, M, gender, grade, age and HECTD2 expression by R “rms” and “survival” packages. Calibration curves were further used to assess the accuracy of nomograms in differentiating patient groups.

The intratumor immune landscape and cancer antigenome

The Cancer ImmunoTome Atlas (https://tcia.at/) describes the intratumor immune landscape and cancer antigenome of gastric cancer [19]. The difference of immune score in different groups according to the difference of HECTD2 expression level were detected.

Tumor Mutational Burden (TMB)

The RNA sequencing data, gene mutation data, and clinical data of STAD were downloaded from the TCGA database (https://gdc-portal.nci.nih.gov/). The correlation analysis was conducted between the expression level of HECTD2 and TMB in each sample.

GSEA enrichment analysis

The gene set enrichment analysis (GSEA) of the high and low expression subgroups of HECTD2 in the GSE62254 dataset was performed by R “clusterprofiler” package. We further used R "enrichplot" package to draw gene set enrichment maps. P < 0.05 was considered statistically significant.

Immunoinfiltration analysis

SsGSEA method was used to investigate the immune cells, immune-related functions, immune-related pathways and different infiltration of immune celltypes in GSE62254 database by Rpacket "GSVA" [20]. Based on the expression of RNA-seq, the Stromal Score, Immune Score, ESTIMATE Score and Tumor Purity of every sample in the GSE62254 database were evaluated by R package "estimate" [21]. The R package "ggpubr" was used to show violplots of Stromal Score, Immune Score and ESTIMATE Score. In order to investigate the differences in immune cell subtypes, we further applied R package "CIBERSORT" to calculate the proportion of 22 immune cells in all GC samples [22].

Enrichment analysis

In order to explore the role of HECTD2 in gastric cancer, the cBioPortal for cancer genomics (http://www.cbioportal.org) was used to get genes that are positively correlated with HECTD2 in gastric cancer, then screen for gene sets with a correlation coefficient greater than or equal to 0.3 [23, 24]. Then, we employed Metascape (http://metascape.org) [25] to perform enrichment analysis on HECTD2 related genes obtained from cBioPortal.

Survival analysis and HECTD2 expressionin different subtypes and pathways of gastric cancer

GSCA Lite online tool (http://bioinfo.life.hust.edu.cn/web/GSCALite/) [26] was used to analyze the HECTD2 expression in different subtypes and pathways of gastric cancer. The difference was evaluated by Spearman’s test.

Database used to explore HECTD2 coexpression networks

The LinkedOmics database (http://www.linkedomics.org/login.php) [27] was used to determine the HECTD2 coexpression genes by using Pearson’s correlation coefficient and showed the results via heat maps. Then, we explored the Gene Ontology biological process (GO_BP), and KEGG pathways of HECTD2 and its coexpression genes by using gene set enrichment analysis (GSEA).

Patients

From January 2018 to December 2022, 26 patients undergoing gastrectomy for gastric cancer and 24 advanced GC patients receiving ICIs and chemotherapy in Changzhou No.2 People's hospital. 26 patients diagnosed with gastric cancer were enrolled as the following criteria: GC with stage I–III; no neoadjuvant therapy before surgery; Received standard adjuvant therapy after operation; Provide standard treatment after disease progression. For 24 inoperable patients with advanced gastric cancer, the biopsy specimens have been confirmed to have a definite diagnosis of gastric cancer. All the above patients were followed up in Changzhou No.2 People’s Hospital. The total follow-up period was 70 months. The clinicopathological characteristics of these patients were summarized in Tables 2 and 3. Cancerous and adjacent normal tissues were collected during surgery or puncturation, and histopathologically confirmed and staged according to the Union for International Cancer Control. Patients’ written informed consents and approval from the Ethics Committees of Changzhou No.2 People’s Hospital (No.2017-C-015-01) were obtained for the use of these clinical materials.

Table 2 Clinicopathological characteristics of patients with gastric cancerTable 3 Clinicopathological characteristics of patients treated with ICIsMultiplex immunofluorescence

Multiplex immunofluorescence was performed with the TSA kit (H-D110061,yuanxibio) according to the protocol of the manufacturer. And perform microwave treatment to remove primary and secondary antibodies while maintaining a complete fluorescence signal. Repeat the process until all antigens are stained with their respective fluorophores.

The antibodies were diluted as follows: Anti-CD68 (#BX50031-C3, Biolynx), anti-HLA-DR (#ab92511, Abcam), anti-PanCK (#GM351507, Gene Tech), DAPI (FP1490A, PerkinElmer), working fluid.

The stained slides were scanned to obtain multispectral images using the Pannoramic MIDI imaging system (3D HISTECH). Images were analyzed by using Indica Halo software.

Immunohistochemistry (IHC)

Tissue samples from human gastric cancer and adjacent tissues were originally fixed with 10% formalin for 48 h at room temperature followed with ethanol tissue dehydration and replacement with Xylene before embedding tissues into paraffin blocks. The paraffin tissue sections were cut at 5-μm and deparaffinized with xylene and rehydrated with a graded series of ethanol. Set the oven to a temperature of 65 °C and bake at a constant temperature for 120 min. Performing according to immunohistochemistry kit. The staining was evaluated by scanning the entire tissue specimen under low magnification (× 10) and confirmed under high magnification (× 20 and × 40). The protein expression was visualized and classified based on the percentage of positive cells and the intensity of staining. After 30 min blocking with the universal blocking serum (Dako Diagnostics, Carpinteria, CA), the sections were incubated with anti-HECTD2 antibody at 4 °C overnight and washed 3 times with PBS at room temperature. Then a secondary antibody was added for 30 min incubation (Dako Diagnostics). The samples were washed 3 times with PBS and developed using DAB followed by counterstaining with hematoxylin. Dehydration was performed following a standard procedure and the slides were sealed with cover slips. Images were scanned with a digital pathology slide scanner (KF-BIO, CHINA).

HECTD2 immunostaining signals were evaluated by two researchers, with the clinical information blinded to them, and scored. Brown cytoplasmic staining for HECTD2 was considered positive. The percentage of HECTD2-positive cells was scored with the following four categories: 1 (< 25%), 2 (25–50%), 3 (50–75%), and 4 (> 75%). The staining intensity of positive cells was scored as 0 (absent), 1 (weak infiltration), 2 (moderate infiltration), and 3 (strong infiltration). The final score was the product of the intensity and the percentage.

Survival analyses

Statistical analyses was performed by SPSS software 21.0. It was determined that P < 0.05 was statistical significant.