Mitochondrial gene expression-based model for the prediction of thyroid cancer prognosis.

Abstract

The roles of mitochondrial genes in the development of thyroid cancer (TC) and the associated tumor microenvironment remain to be elucidated. Based on 64 dysregulated mitochondrial genes, unsupervised consensus clustering analysis was performed using TC datasets from The Cancer Genome Atlas and integrated gene expression databases. A dysregulated mitochondrial-based prognostic model was constructed using machine learning. Fourteen prognostic genes were identified, and the correlation between semaphorin 7A (SEMA7A) and immune cell infiltration was validated. The functions of SEMA7A were investigated through plate cloning, scratch wound healing, and Transwell invasion assays. This study constructed a prognostic model for predicting overall survival by identifying 14 mitochondrial dysregulated-based prognostic genes (APOD, ATP2C2, FAM111B, GJB4, GZMA, HHIPL2, IFIT3, IL18, MCEMP1, PRRT4, RGS16, SEMA7A, SLC2A3, and TDRD9). Its effectiveness was validated in an independent cohort. The model accurately predicted OS across cancer stages. SEMA7A expression was significantly upregulated in papillary thyroid carcinoma (PTC) tissues and PTC cell lines. Functional experiments showed that SEMA7A promoted TC progression, including cell proliferation, migration, and invasion. Compared with the SEMA7A low-expression group, the high-expression group showed a larger FOXP3+ regulatory T cell (Treg) population. Although a higher ratio of SEMA7A + CD8-positive cells/SEMA7A + FOXP3-positive cells was observed, increased Treg abundance is known to shift the immune context toward a less cytotoxic and more exhausted state. Thus, oncogenic SEMA7A in TC promotes malignant functions in TC cells and may play an important role in shaping the immunosuppressive microenvironment. Consequently, our mitochondrial dysregulation-based prognostic risk model for TC may offer good clinical application value.