Abstract: Objective To conduct a pan-cancer analysis of the expression of long non-coding RNA (lncRNA) MIR22HG and explore its association with clinical characteristics. Methods We analyzed the expression of MIR22HG in different tumors and its association with clinical staging, lymph node metastasis, tumor mutation burden (TMB) and microsatellite instability (MSI) using R package based on the Cancer Genome Atlas (TCGA) datasets. The relationship between MIR22HG expression and infiltrating immune cells was analyzed using TIMER algorithm. The association of MIR22HG gene alteration frequency with the clinical outcomes was examined using cBioPortal online software. Data form Genomics of Drug Sensitivity in Cancer (GDSC) were used to analyze the relationship between MIR22HG and the sensitivity of chemotherapy drugs. We specifically analyzed MIR22HG expression in hepatocellular carcinoma (HCC) and its correlation with sorafenib treatment using GEO database and verified the results in 12 pairs of HCC specimens. Kaplan-Meier analysis was performed to analyze the correlation of MIR22HG with the outcomes of sorafenib treatment. We also tested the effects of MIR22HG overexpression and knockdown on IC50 of sorafenib in HCC cells. Results MIR22HG was downregulated in most tumors (P<0.05), where its deletion mutations were frequent, and associated with a poor prognosis (P<0.05). In many tumors, MIR22HG expression level was correlated with clinical stage, lymph node metastasis, TMB, MSI, immune cell infiltration, immune checkpoint-related genes, and sensitivity to common chemotherapeutic drugs (P<0.05). Among the 6 common infiltrating immune cells in cancers, neutrophil infiltration had the strongest correlation with MIR22HG expression level, especially in breast cancer, rectal cancer and kidney renal papillary cell carcinoma (P<0.05). MIR22HG was downregulated in HCC in association with HCC progression (P<0.05). In HCC patients, a low MIR22HG expression was associated with a favorable outcome after sorafenib treatment (HR=2.94, P=0.075) and was capable of predicting the response to sorafenib treatment (AUC=0.8095). Compared with the negative control, MIR22HG overexpression obviously reduced sorafenib sensitivity (with IC50 of 7.731 vs 15.61) while MIR22HG knockdown increased sorafenib sensitivity of HCC cells (with IC50 of 7.986 vs 5.085). Conclusion MIR22HG expression level is correlated with clinical stage, lymph node metastasis, TMB, MSI, immune cell infiltration, and chemosensitivity in most cancer, suggesting its potential as an immunotherapeutic target and also a prognostic biomarker for tumors.

Key words: long non-coding RNA; MIR22HG; pan-cancer analysis; hepatocellular carcinoma; sorafenib