Abstract
The RING finger (RNF) protein family, the largest family of E3 ubiquitin ligases, plays a critical role in tumorigenesis and progression; however, a comprehensive macroscopic analysis of this field is still lacking. Colorectal cancer (CRC) is a highly prevalent malignancy worldwide, and its substantial heterogeneity urgently calls for more precise prognostic biomarkers. This study integrated bibliometric and bioinformatics approaches to systematically delineate the RNF research landscape, construct a CRC prognostic risk model based on the RNF43 interaction network, and explore its association with the tumor microenvironment (TME). Bibliometric analysis revealed that publications in the RNF field have increased rapidly (compound annual growth rate of 28.5%), with research hotspots shifting toward CRC and other diseases, and RNF43 emerging as a core frontier. Analysis of TCGA and GTEx data showed that RNF43 was significantly overexpressed in CRC. Through the STRING database, RNF43-interacting proteins were screened and intersected with CRC-related targets, identifying RNF43-mediated core CRC targets (RNF43, FZD1, FZD5, FZD8, LGR4, LGR5, RSPO1, RSPO2, RSPO3, RSPO4, and AKAP8L). Based on the TCGA-COAD+READ cohort (n = 541), these core targets were incorporated into LASSO regression analysis, resulting in a risk score model comprising LGR4, RSPO4, AKAP8L, and RNF43. In the training set (TCGA-COAD+READ), the high-risk group exhibited significantly shorter overall survival (OS), with AUCs of 0.61, 0.60, and 0.61 for 1-, 3-, and 5-year OS, respectively. The model was externally validated in the GSE39582 cohort (n = 579), although the AUC values remained modest. Multivariate Cox regression analysis demonstrated that the risk score, age, and tumor stage were independent prognostic factors for OS in CRC patients; a nomogram integrating these three factors showed acceptable calibration. TME analysis revealed that the high-risk group displayed a "pro-inflammatory phenotype under immunosuppression": increased infiltration of stromal components, increased CD8⁺ T cells but decreased CD4⁺ memory T cells, unexpectedly elevated M1 macrophages, and no significant difference in overall immune score. This complex imbalance pattern-characterized by increased effector cells, loss of memory cells, and pro-inflammatory polarization accompanied by potential functional suppression-suggests a possible mechanism of immune evasion. In conclusion, this study is the first to adopt a strategy combining "macro-level bibliometric insights" with "micro-level bioinformatics mining," systematically delineating the research trajectory of the RNF family and successfully constructing and validating a CRC prognostic risk model based on the RNF43 interaction network. This model may serve as an auxiliary stratification tool with potential prognostic value and may also reveal associations between the risk score and TME stromal activation and immune dysfunction, providing new insights for precise prognostic assessment and individualized treatment of CRC.