Construction and validation of prognostic prediction established on N6-methyladenosine related genes in cervical squamous cell carcinoma

Danxia Chen; Wenhao Guo; Hailan Yu; Jianhua Yang

doi:10.21037/tcr-22-881

Construction and validation of prognostic prediction established on N6-methyladenosine related genes in cervical squamous cell carcinoma

Transl Cancer Res. 2022 Sep;11(9):3064-3079. doi: 10.21037/tcr-22-881.

Authors

Danxia Chen^#^{1

2}, Wenhao Guo^#^{3

4}, Hailan Yu^{1

2}, Jianhua Yang^{1

2}

Affiliations

¹ Department of Obstetrics and Gynecology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China.
² Key Laboratory of Reproductive Dysfunction Management of Zhejiang Province, Hangzhou, China.
³ Department of Urology, Sir Run‑Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China.
⁴ Department of Urology, Shaoxing Branch of Sir Run‑Run Shaw Hospital, Zhejiang University School of Medicine, Shaoxing, China.

^# Contributed equally.

Abstract

Background: Cervical cancer (CESC) is the second most common cancer death in middle-aged women. The N6-methyladenosine (m6A) plays an essential role in the epitranscriptomics of cancer and affects immune cell infiltration. Our study used The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) data to construct and validate prognostic prediction established on m6A-related genes in CESC.

Methods: We gained gene expression and clinical characteristics from TCGA and GEO. After differentially expression analysis of the m6A-related genes, we identified eight genes of CESC development. Next, we executed consensus clustering to analyze CESC types established on the differential expression of the m6A-related genes and found different subtypes significantly correlate with survival prognosis, immune microenvironment, and PD-L1 expression. Then, based on Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression analysis, a five-gene (IGF2BP1, IGF2BP2, HNRNPA2B1, YTHDF1, RBM15) predictive model was built in the TCGA training cohort. Finally, we checked the predictive model with survival analysis and receiver operating characteristic (ROC) curve both in the training cohort (TCGA) and in the validation cohort (GSE44001). We found the expression and variation of the five genes significantly correlate with immune cell infiltration.

Results: The CESC could be divided into subtypes according to eight expression m6A-related genes. Different subtypes are related to various immune cells, immune scores, and the expression of the PD-L1. We develop a risk prediction model: risk score = (0.023558929) * Exp IGF2BP1 + (0.021148829) * Exp IGF2BP2 + (0.045035491) * Exp HNRNPA2B1 + (-0.106566550) * Exp YTHDF1 + (-0.001037932) * Exp RBM15. Moreover, different m6A-related genes significantly correlated with immune cells.

Conclusions: The m6A-related genes risk prediction model plays an essential role in predicting CESC patients. The m6A-related genes affected the immune cell infiltration in CESC. These results suggest that the expression of m6A-related genes may influence the immune therapy of CESC and be the potential therapeutic target.

Keywords: Cervical squamous cell carcinoma; immune cell infiltration; m6A RNA methylation; prognostic prediction.