Prognostic significance and pathogenesis of RFC3 gene expression in diffuse large B-cell lymphoma

Data collection

Gene expression profiles and clinical characteristics of DLBCL patients were extracted from the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) database. The RFC3 expression and survival were extracted from the GSE181063 (Painter et al., 2019), GSE10846 (Cardesa-Salzmann et al., 2011), GSE32918 (Care et al., 2014), and GSE31312 (Xu-Monette et al., 2020) datasets, while the GSE32018 (Gomez-Abad et al., 2011) and GSE12453 (Weniger et al., 2018) were utilized for the differential expression analysis of RFC3 in DLBCL. All data were preprocessed and analyzed using the R software (version 4.2) as shown in Table 1. Only the data of patients treated with the R-CHOP chemotherapy regimen who had all relevant clinical characteristics reported in the database were included in this study. RFC3 expression Microarray data extracted from the GEO database (GSE32018, GSE12453) and the Gene Expression Profile Interaction Analysis database (GEPIA, http://gepia.cancer-pku.cn) were used to analyze the differential expression of RFC3 between DLBCL and control samples. The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) projects were used to facilitate the analysis of RFC3 expression in various tumors.

Immunohistochemistry

Fifteen tissue samples of DLBCL lymph nodes and eight samples of normal lymph node tissue were obtained. Immunohistochemistry (IHC) was used to examine the RFC3 expression levels within the tissues as follows. All tissue samples were sectioned into two µm slices and deparaffinized using xylene and ethanol. Antigen retrieval was performed at high temperature in an ethylenediaminetetra-acetic acid (EDTA) solution (pH = 9), followed by cooling to room temperature (25 °C). Endogenous peroxidase activity was blocked using a 3% hydrogen peroxide solution. Sections were then incubated overnight at 4 °C with the primary antibody against RFC3 (1:300; Biodragon). After primary incubation, a secondary antibody was applied for 8 min. The colorimetric signal was developed using diaminobenzidine (DAB), and nuclei were counterstained with hematoxylin for 1 min. The slides were subsequently dehydrated, cleared, and mounted. ImageJ software was used to analyze integrated optical density (IOD) values. Statistical analysis and data visualization were performed using GraphPad Prism 9.0 to compare RFC3 expression between DLBCL and normal lymph node tissues.

Ethical considerations

This study was approved by Ethics Committee of Affiliated Hospital of Zunyi Medical University (Ethics No:KLL-2024-020) (Zunyi, China) (Supplementary Material 1). The research utilized anonymized samples collected from patients previously diagnosed and treated, with no direct contact or intervention involved. Given the retrospective nature and the inability to obtain informed consent from all individuals, the committee granted a waiver of consent. All procedures complied with ethical standards to ensure patient confidentiality and data protection.

Prognostic significance of RFC3 expression in DLBCL

The pROC R package (Robin et al., 2011) was used to calculate the optimal cut-off value for RFC3 expression to stratify DLBCL patients into high and low RFC3 expression groups. Kaplan–Meier survival analysis was then conducted to assess differences in overall survival (OS) between the two groups. To ensure the robustness of the findings, validation analyses were performed using multiple independent datasets, including GSE10846, GSE32918 and GSE31312. Survival analysis and visualization were carried out using the R packages survival (Therneau, 2023) and survminer (Kassambara, Kosinski & Biecek, 2021).

Analysis of the correlation between RFC3 expression and clinical characteristics

The Chi-square test was used to analyze the correlation between RFC3 expression and clinical characteristics of DLBCL patients including age, gender, cell of origin (COO), lactate dehydrogen-ase levels (LDH), Ann Arbor stage, Eastern Cooperative Oncology Group (ECOG) Performance Status, B symptoms (Bs), extronodal sites and IPI. The Crasstable function of the gmodels R package (Warnes et al., 2022) and the ggstatsplot (Patil, 2021) R package were used for statistical analysis and graphical presentation.

GSEA enrichment analysis

GSEA analysis was performed using R packages clusterProfiler (Yu et al., 2012) and msigdbr (Liberzon et al., 2015) to identify signaling pathways associated with high RFC3 expression.

Analysis of immune infiltration and response to immunotherapy

The relationship between RFC3 expression and the tumor immune microenvironment (TIME) in DLBCL was investigated using immune infiltration analyses based on the GSE181063 dataset. The deconvo_tme function from the Immuno-Oncology Biological Research (IOBR) package (version 2.0.0) (Zeng et al., 2024) was used to compare the ImmuneScore, StromalScore, ESTIMATEScore, and Tumorpurity between the high and low RFC3 expression groups. These scores, derived from the ESTIMATE algorithm, quantify key components of the tumor microenvironment: ImmuneScore reflects the level of immune cell infiltration, StromalScore estimates the presence of stromal cells, and ESTIMATEScore combines both to represent the overall non-tumor content in the tumor tissue. Tumorpurity refers to the proportion of cancer cells in a tumor sample relative to non-cancerous cells, such as immune cells, stromal cells, and normal tissue. Additionally, immune cell infiltration differences were further assessed using single-sample gene set enrichment analysis (ssGSEA) via the Gene Set Variation Analysis (GSVA) package (Hanzelmann, Castelo & Guinney, 2013) available on R software, using gene sets representing 28 tumor-infiltrating lymphocytes (TILs) obtained from the Tumor-Immune System Interactions Database (TISIDB) (Barbie et al., 2009). In addition, the correlation between RFC3 expression and immune-related genes was analyzed in DLBCL. Gene sets included immune inhibitor genes, immune stimulation genes, chemokine genes, Chemokine receptor genes, and immune checkpoint genes. These gene sets were selected as they are involved in a range of immunosuppressive, immune-activating, chemotactic, and immune checkpoint pathways. Spearman’s correlation analysis was conducted for each gene in these categories. The immunotherapy response was predicted using the IMvigor210 cohort through the IMvigor210CoreBiologies R package (Nickles & Bourgon, 2019). This dataset includes transcriptomic and clinical data from patients with urothelial carcinoma treated with anti–PD-L1 therapy, allowing us to assess potential associations between RFC3 expression and immunotherapy outcomes.

Drug-susceptibility analysis

The R package oncoPredict (Maeser, Gruener & Huang, 2021) was used to analyze the relationship between RFC3 expression and drug sensitivity in DLBCL. For this analysis, we obtained drug response data from the Genomics of Drug Sensitivity in Cancer (GDSC) data version 1 and 2 and the Cancer Therapeutics Response Portal, version 2 (CTRP2). These datasets include pharmacogenomic profiles of cancer cell lines exposed to a range of therapeutic agents for different RFC3 expression levels. The oncoPredict package was used to estimate the drug concentration required to inhibit 50% of the cell viability (IC50).

Pan-cancer analysis of RFC3

The pan-cancer expression and prognostic value of RFC3 across different tumor types were investigated using the easyTCGA package (Li, 2024). The Cox proportional hazards model was used to evaluate the impact of RFC3 on OS, disease-specific survival (DSS), progression-free interval (PFI) and disease-free interval (DFI). Additionally, the R package TCGAplot (Liao & Wang, 2023) was employed to explore correlations between RFC3 expression and tumor mutational burden (TMB), as well as microsatellite instability (MSI), across multiple cancer types.

Statistical analysis

All statistical analyses and graphical presentations of public data were performed using the R software version 4.2 (R Core Team, 2022). GraphPad Prism 9.0 (GraphPad Software, La Jolla, CA, USA) was used for statistical analysis and visualization of the IHC results. For all statistical tests, a p-value below 0.05 was deemed statistically significant. Significance levels were denoted as follows: *, P < 0.05. **, P < 0.01.***, P < 0.001. ****, P < 0.0001.

Up-regulation of RFC3 expression in DLBCL

The expression profile of RFC3 in DLBCL was systematically analyzed using multiple publicly available datasets and validated through IHC. Analysis of the GSE32018 dataset revealed that RFC3 was significantly overexpressed in DLBCL compared to normal lymphoid tissues (P = 0.0029, Fig. 1A). Similarly, analysis of the GSE112453 dataset showed higher RFC3 expression in DLBCL relative to B-cells (P =1.1e−05, Fig. 1B). Consistent with these findings, data from the GEPIA database demonstrated significantly elevated RFC3 expression in TCGA-DLBC samples compared to normal tissue (P < 0.05, Fig. 1C).

To validate these results at the protein level, IHC was performed on clinical samples, including DLBCL tissues and normal lymph nodes. The analysis confirmed significantly elevated RFC3 expression in DLBCL samples compared to normal lymph node tissues (P < 0.01, Fig. 1D). Notably, overexpression was observed in both germinal center B-cell-like (GCB) (P < 0.05, Fig. 1E) and non-GCB subtypes of DLBCL (P < 0.05, Fig. 1F), but there was no difference in RFC3 expression between these two subtypes. Representative immunohistochemical images are shown in Figs. 1G to 1L, indicating that RFC3 dysregulation is a common feature across molecular subtypes of DLBCL. Collectively, these results establish RFC3 as a consistently overexpressed gene in DLBCL.

High RFC3 expression indicates poor prognosis in DLBCL

Survival analysis revealed that high RFC3 expression was significantly associated with worse overall survival (OS) in the GSE181063 (P = 0.0022, Fig. 2A), GSE10846 (P = 0.0053, Fig. 2B), and GSE32918 (P = 0.0071, Fig. 2C) cohorts. Additionally, in the GSE31312 cohort, both OS (P = 0.014, Fig. 2D) and progression free survival (PFS) (P = 0.029, Fig. 2E) were significantly shorter in DLBCL patients with high RFC3 expression compared to those with low expression. These findings indicate that elevated RFC3 expression could potentially be used as a negative prognostic marker in DLBCL.

Elevated RFC3 expression correlates with aggressive clinical features in DLBCL

The results of the association between RFC3 expression levels and clinical characteristics of DLBCL are shown in Table 2. High RFC3 expression was significantly correlated with several known unfavorable prognostic factors, including ECOG > 1 (P = 0.03, Fig. 3A), raised LDH levels (P = 3.32e−03, Fig. 3B), Ann Arbor stages III–IV (P = 0.02, Fig. 3C), high IPI risk (>2) (P = 4.97e−07, Fig. 3D), extranodal involvement in more than two sites (P = 0.04, Fig. 3E), and MHG subtype (P = 1.12e−13, Fig. 3F). However, no significant association was observed between RFC3 expression and demographic variables such as age and gender.

RFC3 plays a role in various tumor-related signaling pathways in DLBCL

Enrichment analysis revealed that RFC3 expression was significantly correlated with multiple signaling pathways in DLBCL (Table 3). Higher RFC3 levels were notably associated with key tumor-related pathways (Figs. 4A & 4B), including DNA repair, E2F transcription factor targets (E2F_TARGETS), G2/M cell cycle checkpoint (G2M_CHECKPOINT), mechanistic target of rapamycin complex 1 (mTORC1) signaling, MYC proto-oncogene transcription factor targets V1 & V2 (MYC Targets V1 and MYC Targets V2), oxidative phosphorylatio (OXPHOS), unfolded protein response (UPR).

The expression of RFC3 is significantly associated with immune infiltration and various immune factors, indicating a differential response to immunotherapy

Elevated RFC3 expression was correlated with a significant decrease in the ESTIMATEScores (P < 2e−16), ImmuneScore (P < 2e−16), and StromalScore (P = 9.1e−16), along with a marked increase in TumorPurity (P < 2e−16) (Fig. 5A). These results suggest the presence of a “cold” immune microenvironment characterized by a reduction in immune cell populations and an abundance of tumor cells. Immune cell profiling utilizing ssGSEA analysis demonstrated a positive correlation between RFC3 expression and pro-tumorigenic immune subsets, such as activated CD4 T cells, memory B cells, activated B cells, immature B cells, central memory CD8 T cells, gamma delta T cells, type 2 T helper cells, CD56 bright natural killer cells, natural killer cells, eosinophils, type 1 T helper cells, monocytes, effector memory CD4 T cells, and plasmacytoid dendritic cells. Conversely, RFC3 expression was inversely associated with activated dendritic cells, T follicular helper cells, mast cells, natural killer T cells, and neutrophils (Fig. 5B).

Analyses using the TISIDB revealed strong positive correlations between RFC3 expression and immune stimulator genes such as MICB, CD70, CD48, etc., but negatively correlated with TNFSF14, RAET1E, TNFRSF25, TNFRSF4 and multiple co-stimulatory molecules (Fig. 6A). Among the immune inhibitor genes RFC3 positively correlated with PDCD1LG2, but negative associations with KIR2DL1, TIGIT, and others (Fig. 6B). RFC3 shows a positive correlation with the chemokine CX3CL1, while displaying negative correlations with multiple T-cell trafficking-related chemokines (e.g., CCL26, CXCL14, CCL16, CCL24) (Fig. 6C). Among chemokine receptors, RFC3 is positively linked to CXCR5, CCR10, CXCR4 but negatively associated with CXCR1, CCR6, CCR8, and others (Fig. 6D).

RFC3 was positively associated with the immune checkpoint gene PDCD1LG2 but inversely correlated with CTLA4, PDCD1, and TIGIT (Fig. 6E), highlighting its dual role in modulating immune activation and exhaustion. This suggests RFC3 may influence immune cell migration by modulating chemokine signaling pathways. Collectively, these findings position RFC3 as a key orchestrator of immune evasion in DLBCL, attenuating cytotoxic responses while promoting myeloid-driven immunosuppression. Furthermore, analysis using the IMvigor210CoreBiologies R package revealed that patients with high RFC3 expression exhibited significantly higher rates of complete or partial response to chemotherapy (CR/PR: 30% vs. 12%, p < 0.01) (Figs. 7A & 7B).

The expression of RFC3 demonstrated a significant correlation with drug sensitivity

Figures 8A to 8C present a graphical representation of drugs showing a significant association with RFC3 expression (correlation coefficient >0.5 and p < 0.01). The analysis revealed a positive correlation between RFC3 expression and the concentrations of cyclophosphamide and bleomycin, both commonly used in lymphoma treatment (Figs. 9A & 9B), suggesting that elevated RFC3 levels may contribute to resistance against these agents. In contrast, concentrations of newer lymphoma therapies—such as venotoclax, lenalidomide, decitabine, and ibrutinib were negatively correlated with RFC3 expression (Figs. 9C–9F). This inverse relationship indicates that these drugs may be more effective in patients with high RFC3 expression, highlighting their potential benefit in lymphoma treatment.

RFC3 exhibits abnormal expression patterns across a range of tumors and is linked to prognostic outcomes

A comprehensive pan-cancer analysis of RFC3 expression, using matched normal tissues as controls, revealed significant dysregulation across various cancer types. Notable increases in RFC3 levels were observed in cancers originating from epithelial tissues, such as bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), and cervical squamous cell carcinoma (CESC). Additionally, elevated RFC3 expression was found in gastrointestinal cancers such as cholangiocarcinoma (CHOL), colon Adenocarcinoma (COAD), esophageal carcinoma (ESCA), and Stomach adenocarcinoma (STAD). Increased RFC3 levels were also noted in neurological cancers (glioblastoma multiforme (GBM), lower grade glioma (LGG)), thoracic cancers (lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC)), and reproductive system cancers (ovarian serous cystadenocarcinoma (OV), uterine corpus endometrial carcinoma (UCEC), Uterine carcinosarcoma (UCS)). Other cancers exhibiting heightened RFC3 levels included head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), sarcoma (SARC), skin cutaneous melanoma (SKCM), and testicular germ cell tumors (TGCT). Conversely, RFC3 expression was reduced in kidney chromophobe (KICH), acute myeloid leukemia (LAML), thyroid carcinoma (THCA), and thymoma (THYM), indicating that its regulatory functions may differ by tissue type (Fig. 10A).

Survival analysis indicates that the aberrant expression of RFC3 is significantly associated with various survival outcomes across a range of tumor types (Fig. 10B). Specifically, elevated RFC3 expression correlates with diminished OS in malignancies such as ACC and CESC. Furthermore, in tumors for instance ACC, CESC, KICH and KIRC, increased RFC3 expression is significantly linked to a reduction in DSS. Additionally, in malignancies e.g., ACC, BLCA, BRCA, higher levels of RFC3 are significantly correlated with PFI. Moreover, in ACC, BRCA, CESC, COAD, and other cancers, elevated RFC3 expression is associated with an unfavorable DFI.

Additionally, the expression of RFC3 exhibits a positive correlation with TMB across various cancer types, including BLCA, KICH, LGG, LUAD, PRAD, and UCEC (Fig. 10C). Furthermore, RFC3 expression positively correlated with MSI in KIRC, READ, SARC, STAD, and UCEC, while exhibiting a negative correlation with MSI in DLBCL (Fig. 10D). These findings suggest that RFC3 may play distinct roles in genomic instability and the tumor immune microenvironment across several cancer types.