Screening immune-related blood biomarkers for DKD-related HCC using machine learning.

Background: Diabetes mellitus is a significant health problem worldwide, often leading to diabetic kidney disease (DKD), which may also influence the occurrence of hepatocellular carcinoma (HCC). However, the relationship and diagnostic biomarkers between DKD and HCC are unclear. Methods: Using public database data, we screened DKD secretory RNAs and HCC essential genes by limma and WGCNA. Potential mechanisms, drugs, and biomarkers for DKD-associated HCC were identified using PPI, functional enrichment, cMAP, and machine learning algorithms, and a diagnostic nomogram was constructed. Then, ROC, calibration, and decision curves were used to evaluate the diagnostic performance of the nomograms. In addition, immune cell infiltration in HCC was explored using CIBERSORT. Finally, the detectability of critical genes in blood was verified by qPCR. Results: 104 DEGs associated with HCC using WGCNA were identified. 101 DEGs from DKD were predicated on secreting into the bloodstream with Exorbase datasets. PPI analysis identified three critical modules considered causative genes for DKD-associated HCC, primarily involved in inflammation and immune regulation. Using lasso and RM, four hub genes associated with DKD-associated HCC were identified, and a diagnostic nomogram confirmed by DCA curves was established. The results of immune cell infiltration showed immune dysregulation in HCC, which was associated with the expression of four essential genes. PLVAP was validated by qPCR as a possible blood-based diagnostic marker for DKD-related HCC. Conclusion: We revealed the inflammatory immune pathways of DKD-related HCC and developed a diagnostic nomogram for HCC based on PLVAP, C7, COL15A1, and MS4A6A. We confirmed with qPCR that PLVAP can be used as a blood marker to assess the risk of HCC in DKD patients.