Elucidating the molecular and immune interplay between head and neck squamous cell carcinoma and diffuse large B-cell lymphoma through bioinformatics and machine learning.

Background: Head and neck squamous cell carcinoma (HNSCC) contributes significantly to global health challenges, presenting primarily in the oral cavity, pharynx, nasopharynx, and larynx. HNSCC has a high propensity for lymphatic metastasis. Diffuse large B-cell lymphoma (DLBCL), the most common subtype of non-Hodgkin lymphoma, exhibits significant heterogeneity and aggressive behavior, leading to high mortality rates. Epstein-Barr virus (EBV) is notably associated with DLBCL and certain types of HNSCC. The purpose of this study is to elucidate the molecular and immune interplay between HNSCC and DLBCL using bioinformatics and machine learning (ML) to identify shared biomarkers and potential therapeutic targets. Methods: Differentially expressed genes (DEGs) were identified using the "limma" package in R from the HNSCC dataset in The Cancer Genome Atlas (TCGA) database, and relevant modules were selected through weighted gene co-expression network analysis (WGCNA) from a DLBCL dataset in the Gene Expression Omnibus (GEO) database. Based on their intersection genes, functional enrichment analyses were conducted using Gene Ontology (GO), Disease Ontology, and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases. Protein-protein interaction (PPI) networks and ML algorithms were employed to screen for biomarkers. The prognostic value of these biomarkers was evaluated using Kaplan-Meier (K-M) survival analysis and receiver operating characteristic (ROC) curve analyses. The Human Protein Atlas (HPA) database facilitated the examination of messenger RNA (mRNA) and protein expressions. Further analyses of mutations, immune infiltration, drug predictions, and pan-cancer impacts were performed. Additionally, single-cell RNA sequencing (scRNA-seq) data analysis at the cell type level was conducted to provide deeper insights into the tumor microenvironment. Results: From 2,040 DEGs and 1,983 module-related genes, 85 shared genes were identified. PPI analysis with six algorithms proposed 21 prospective genes, followed ML examination yielded 16 candidates. Survival and ROC analyses pinpointed four hub genes- Conclusions: Leveraging bioinformatics and ML, this study identified four pivotal genes with significant diagnostic capabilities for DLBCL and HNSCC. The survival analysis corroborates their diagnostic accuracy, supporting the development of a diagnostic nomogram to assist in clinical decision-making.